repo_name
stringlengths
6
97
path
stringlengths
3
341
text
stringlengths
8
1.02M
gaybro8777/klio
devtools/src/klio_devtools/commands/develop.py
<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import os import docker import dockerpty from klio_cli.commands import base class DevelopKlioContainer(base.BaseDockerizedPipeline): DOCKER_LOGGER_NAME = "klio.develop" PIP_CMD = "pip install --use-feature=2020-resolver -e /usr/src/{pkg}" # order is important - klio-core doesn't depend on any other # internal libraries; klio depends on klio-core; klio-exec depends # on klio-core & klio PKGS = ("klio-core", "klio", "klio-exec", "klio-audio") def __init__( self, job_dir, klio_config, docker_runtime_config, klio_path, exclude_pkgs, ): super().__init__(job_dir, klio_config, docker_runtime_config) self.klio_path = klio_path self.install_pkgs = [ pkg for pkg in self.PKGS if pkg not in exclude_pkgs ] self.requires_config_file = False def _run_docker_container(self, runflags): """ Create & run job container, install klio packages as editable, and attach to the container with an interactive terminal. """ container = self._docker_client.containers.create(**runflags) container.start() def run_and_log(cmd): _, output = container.exec_run(cmd, tty=True, stream=True) for line in output: try: self._docker_logger.info(line.decode("utf-8").strip("\n")) except Exception: # sometimes there's a decode error for a log line, but it # shouldn't stop the setup pass run_and_log("pip install --upgrade pip setuptools") for pkg in self.install_pkgs: run_and_log(self.PIP_CMD.format(pkg=pkg)) # need to use lower-level Docker API client in order to start # an interactive terminal inside the running container self._docker_logger.info( "\nConnecting to job's container. Use CTRL+C to stop." ) pty_client = docker.APIClient(base_url="unix://var/run/docker.sock") dockerpty.start(pty_client, container.attrs) def _get_volumes(self): volumes = super(DevelopKlioContainer, self)._get_volumes() local_exec_path = os.path.join(self.klio_path, "exec") volumes[local_exec_path] = {"bind": "/usr/src/klio-exec", "mode": "rw"} local_core_path = os.path.join(self.klio_path, "core") volumes[local_core_path] = {"bind": "/usr/src/klio-core", "mode": "rw"} local_lib_path = os.path.join(self.klio_path, "lib") volumes[local_lib_path] = {"bind": "/usr/src/klio", "mode": "rw"} local_audio_path = os.path.join(self.klio_path, "audio") volumes[local_audio_path] = { "bind": "/usr/src/klio-audio", "mode": "rw", } return volumes def _get_docker_runflags(self, *args, **kwargs): return { "image": self._full_image_name, "entrypoint": "/bin/bash", "volumes": self._get_volumes(), "environment": self._get_environment(), "stdin_open": True, "tty": True, }
gaybro8777/klio
exec/tests/unit/commands/utils/test_profile_utils.py
<filename>exec/tests/unit/commands/utils/test_profile_utils.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import numpy as np import pytest from klio_exec.commands.utils import profile_utils def test_get_profiling_data(mocker): profiling_data = "MEM 100 100001\n" "MEM 200 100002\n" "MEM 300 100003\n" m_open = mocker.mock_open(read_data=profiling_data) mock_open = mocker.patch( "klio_exec.commands.utils.profile_utils.open", m_open ) act_ret = profile_utils._get_profiling_data("foo.txt") mock_open.assert_called_once_with("foo.txt", "r") exp_ret = { "data": [100.0, 200.0, 300.0], "timestamp": [100001.0, 100002.0, 100003.0], } assert exp_ret == act_ret def test_get_profiling_data_raises_no_file(mocker, caplog): with pytest.raises(SystemExit): ret = profile_utils._get_profiling_data("i_should_not_exist.txt") assert ret is None assert 1 == len(caplog.records) assert "ERROR" == caplog.records[0].levelname assert "Could not read profiling data" in caplog.records[0].message def test_get_profiling_data_malformed_data(mocker): profiling_data = ( "MEM 100 100001\n" "foobar baz\n" "\n" "MEM 1 2 3\n" "MEM 300 100003\n" ) m_open = mocker.mock_open(read_data=profiling_data) mock_open = mocker.patch( "klio_exec.commands.utils.profile_utils.open", m_open ) act_ret = profile_utils._get_profiling_data("foo.txt") mock_open.assert_called_once_with("foo.txt", "r") exp_ret = {"data": [100.0, 300.0], "timestamp": [100001.0, 100003.0]} assert exp_ret == act_ret def test_get_profiling_data_raises_no_data(mocker, caplog): m_open = mocker.mock_open(read_data="\n") mock_open = mocker.patch( "klio_exec.commands.utils.profile_utils.open", m_open ) with pytest.raises(SystemExit): ret = profile_utils._get_profiling_data("foo.txt") assert ret is None mock_open.assert_called_once_with("foo.txt", "r") assert 1 == len(caplog.records) assert "ERROR" == caplog.records[0].levelname assert "No samples to parse in" in caplog.records[0].message def test_plot(mocker, monkeypatch): parsed_data = { "data": [100.0, 200.0, 300.0], "timestamp": [100001.0, 100002.0, 100003.0], } mock_get_profiling_data = mocker.Mock() mock_get_profiling_data.return_value = parsed_data monkeypatch.setattr( profile_utils, "_get_profiling_data", mock_get_profiling_data ) mock_plt = mocker.Mock() monkeypatch.setattr(profile_utils, "plt", mock_plt) profile_utils.plot( "input.txt", "output.png", "x-label", "y-label", "title test" ) mock_get_profiling_data.assert_called_once_with("input.txt") mock_plt.figure.assert_called_once_with(figsize=(14, 6), dpi=90) # can't directly compare (or assert called with) np arrays; # grabbing the args/kwargs then using np.allclose with a 0 tolerance exp_data = np.asarray(parsed_data["data"]) ts = np.asarray(parsed_data["timestamp"]) exp_t = ts - float(ts[0]) act_call_args, act_call_kwargs = mock_plt.plot.call_args assert 3 == len(act_call_args) assert not act_call_kwargs assert np.allclose(exp_t, act_call_args[0], rtol=0) assert np.allclose(exp_data, act_call_args[1], rtol=0) assert "+-c" == act_call_args[2] mock_plt.xlabel.assert_called_once_with("x-label") mock_plt.ylabel.assert_called_once_with("y-label") mock_plt.title.assert_called_once_with("title test") mock_plt.grid.assert_called_once_with() mock_plt.savefig.assert_called_once_with("output.png")
gaybro8777/klio
lib/tests/unit/metrics/test_dispatcher.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import pytest from klio.metrics import dispatcher class GenericDispatcher(dispatcher.BaseMetricDispatcher): METRIC_TYPE = "generic" def _setup_metric_relay(self, relay_clients): pass @pytest.fixture def metric_params(): return { "name": "my-metric", "value": 0, "transform": "my-transform", "kwargs": {"tags": {"key-tag": "value-tag"}}, } @pytest.fixture def metric(mocker): mock = mocker.Mock() mock.update = mocker.Mock() return mock @pytest.fixture def relay_client(mocker, metric): client = mocker.Mock() client.emit = mocker.Mock() client.counter.return_value = metric client.gauge.return_value = metric client.timer.return_value = metric return client @pytest.fixture def counter_dispatcher( relay_client, metric_params, metric, mocker, monkeypatch ): counter = dispatcher.CounterDispatcher( relay_clients=[relay_client], **metric_params ) mock_submit_to_thread = mocker.Mock() monkeypatch.setattr(counter, "submit", mock_submit_to_thread) return counter @pytest.fixture def gauge_dispatcher(relay_client, metric_params, metric, mocker, monkeypatch): gauge = dispatcher.GaugeDispatcher( relay_clients=[relay_client], **metric_params ) mock_submit_to_thread = mocker.Mock() monkeypatch.setattr(gauge, "submit", mock_submit_to_thread) return gauge @pytest.fixture def default_timer(mocker, monkeypatch): mock = mocker.Mock() mock.default_timer.side_effect = [0.0, 1.0] monkeypatch.setattr(dispatcher, "timeit", mock) return mock @pytest.fixture def timer_dispatcher( relay_client, metric_params, metric, default_timer, mocker, monkeypatch ): timer = dispatcher.TimerDispatcher( relay_clients=[relay_client], **metric_params ) mock_submit_to_thread = mocker.Mock() monkeypatch.setattr(timer, "submit", mock_submit_to_thread) return timer @pytest.fixture def generic_dispatcher(relay_client, metric_params): return GenericDispatcher(relay_clients=[relay_client], **metric_params) def test_base_metric_dispatcher_raises(relay_client, metric_params): class FakeMetricDispatcher(dispatcher.BaseMetricDispatcher): pass with pytest.raises(NotImplementedError): FakeMetricDispatcher(relay_clients=[relay_client], **metric_params) @pytest.mark.parametrize( "has_transform,exp_key", ((True, "generic_my-metric_my-transform"), (False, "generic_my-metric")), ) def test_base_metric_setup_metric_key( has_transform, exp_key, relay_client, metric_params ): if not has_transform: metric_params.pop("transform") gd = GenericDispatcher(relay_clients=[relay_client], **metric_params) assert exp_key == gd.metric_key def test_base_metric_logger(generic_dispatcher): assert not getattr( generic_dispatcher._thread_local, "klio_metrics_dispatcher_logger", None, ) # sanity check logger = generic_dispatcher.logger assert logging.getLogger("klio.metrics.dispatcher") == logger assert ( logger == generic_dispatcher._thread_local.klio_metrics_dispatcher_logger ) def test_base_metric_submit_callback(generic_dispatcher, mocker, caplog): fut = mocker.Mock() fut.result.return_value = None generic_dispatcher._submit_callback(fut) fut.result.assert_called_once_with() assert not len(caplog.records) def test_base_metric_submit_callback_exception( generic_dispatcher, mocker, caplog ): fut = mocker.Mock() fut.result.side_effect = Exception("fuu") generic_dispatcher._submit_callback(fut) fut.result.assert_called_once_with() assert 1 == len(caplog.records) assert logging.WARNING == caplog.records[0].levelno def test_base_metric_submit( generic_dispatcher, relay_client, mocker, monkeypatch ): mock_thread_pool, mock_future = mocker.Mock(), mocker.Mock() mock_thread_pool.submit.return_value = mock_future monkeypatch.setattr(generic_dispatcher, "_thread_pool", mock_thread_pool) metric = mocker.Mock() metric.name, metric.transform = "my-metric", "my-transform" expected_key = "generic_my-metric_my-transform" generic_dispatcher.submit(relay_client.emit, metric) mock_thread_pool.submit.assert_called_once_with(relay_client.emit, metric) assert expected_key == mock_future.metric_key mock_future.add_done_callback.assert_called_once_with( generic_dispatcher._submit_callback ) def test_counter_setup_metric_relay(relay_client, metric, metric_params): counter = dispatcher.CounterDispatcher( relay_clients=[relay_client], **metric_params ) expected_relay_to_metric = [(relay_client, metric)] assert expected_relay_to_metric == counter.relay_to_metric relay_client.counter.assert_called_once_with(**metric_params) def test_counter_inc(counter_dispatcher, relay_client, metric): assert 0 == counter_dispatcher.value # sanity check counter_dispatcher.inc() assert 1 == counter_dispatcher.value metric.update.assert_called_once_with(1) counter_dispatcher.submit.assert_called_once_with( relay_client.emit, metric ) def test_gauge_setup_metric_relay(relay_client, metric, metric_params): gauge = dispatcher.GaugeDispatcher( relay_clients=[relay_client], **metric_params ) expected_relay_to_metric = [(relay_client, metric)] assert expected_relay_to_metric == gauge.relay_to_metric relay_client.gauge.assert_called_once_with(**metric_params) def test_gauge_set(gauge_dispatcher, relay_client, metric): assert 0 == gauge_dispatcher.value # sanity check gauge_dispatcher.set(5) assert 5 == gauge_dispatcher.value metric.update.assert_called_once_with(5) gauge_dispatcher.submit.assert_called_once_with(relay_client.emit, metric) def test_timer_setup_metric_relay(relay_client, metric, metric_params): metric_params["timer_unit"] = "ns" timer = dispatcher.TimerDispatcher( relay_clients=[relay_client], **metric_params ) expected_relay_to_metric = [(relay_client, metric)] assert expected_relay_to_metric == timer.relay_to_metric relay_client.timer.assert_called_once_with(**metric_params) def test_timer_start(timer_dispatcher, metric): assert 0 == timer_dispatcher.value # sanity check assert not timer_dispatcher._start_time # sanity check timer_dispatcher.start() assert 0.0 == timer_dispatcher._start_time metric.update.assert_not_called() timer_dispatcher.submit.assert_not_called() def test_timer_stop(timer_dispatcher, relay_client, metric): assert 0 == timer_dispatcher.value # sanity check assert not timer_dispatcher._start_time # sanity check timer_dispatcher.start() assert 0.0 == timer_dispatcher._start_time metric.update.assert_not_called() timer_dispatcher.submit.assert_not_called() timer_dispatcher.stop() # seconds -> nanoseconds assert 1.0 * 1e9 == timer_dispatcher.value metric.update.assert_called_once_with(timer_dispatcher.value) timer_dispatcher.submit.assert_called_once_with(relay_client.emit, metric) def test_timer_stop_no_start(timer_dispatcher, metric, caplog): assert not timer_dispatcher._start_time # sanity check timer_dispatcher.stop() assert 1 == len(caplog.records) assert logging.WARNING == caplog.records[0].levelno metric.update.assert_not_called() timer_dispatcher.submit.assert_not_called() def test_timer_context_manager( relay_client, metric, metric_params, default_timer, mocker, monkeypatch ): timer = dispatcher.TimerDispatcher( relay_clients=[relay_client], **metric_params ) mock_submit_to_thread = mocker.Mock() monkeypatch.setattr(timer, "submit", mock_submit_to_thread) with timer: assert 1 == 1 assert 0 == timer._start_time assert 1 * 1e9 == timer.value metric.update.assert_called_once_with(timer.value) timer.submit.assert_called_once_with(relay_client.emit, metric)
gaybro8777/klio
core/tests/test_dataflow.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import os import pytest from klio_core import dataflow from klio_core import utils HERE = os.path.abspath(os.path.join(os.path.abspath(__file__), os.path.pardir)) FIXTURE_PATH = os.path.join(HERE, "fixtures") @pytest.fixture def mock_discovery_client(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(dataflow.discovery, "build", lambda x, y: mock) return mock def list_jobs(): list_job_file = os.path.join(FIXTURE_PATH, "list_all_active_jobs.json") with open(list_job_file, "r") as f: return json.load(f) def list_jobs_with_duplicates(): list_job_file = os.path.join( FIXTURE_PATH, "list_all_active_jobs_duplicates.json" ) with open(list_job_file, "r") as f: return json.load(f) def get_job_detail(): get_job_detail_file = os.path.join(FIXTURE_PATH, "get_job_detail.json") with open(get_job_detail_file, "r") as f: return json.load(f) def get_job_detail_no_kind(): return {"steps": [{"kind": "SomethingElse"}]} def get_job_detail_no_username_value(): return { "steps": [ { "kind": "ParallelRead", "properties": {"username": "SomethingElse"}, } ] } def get_job_detail_no_topic(): return { "steps": [ { "kind": "ParallelRead", "properties": {"username": "ReadFromPubSub/Read"}, } ] } @pytest.mark.parametrize( "region,side_effect,exp_call_count,exp_log_count", ( (None, [list_jobs_with_duplicates(), {}, {}], 13, 11), ("foo-region", [list_jobs(), {}, {}], 1, 0), (None, Exception("nojob4u"), 13, 13), ), ) def test_find_job_by_name( region, side_effect, exp_call_count, exp_log_count, mock_discovery_client, caplog, ): client = dataflow.DataflowClient() locs = mock_discovery_client.projects.return_value.locations.return_value request = locs.jobs.return_value.list.return_value request.execute.side_effect = side_effect client.find_job_by_name("benchmark-beats", "gcp-project", region) assert exp_call_count == request.execute.call_count assert exp_log_count == len(caplog.records) @pytest.mark.parametrize( "found_job,response,exp_log_count", ( (True, Exception("foo"), 1), (True, get_job_detail(), 0), (False, None, 0), ), ) def test_get_job_detail( found_job, response, exp_log_count, mock_discovery_client, caplog, mocker, monkeypatch, ): client = dataflow.DataflowClient() mock_find_job_by_name = mocker.Mock() monkeypatch.setattr(client, "find_job_by_name", mock_find_job_by_name) if found_job: mock_find_job_by_name.return_value = list_jobs().get("jobs")[0] else: mock_find_job_by_name.return_value = None locs = mock_discovery_client.projects.return_value.locations.return_value request = locs.jobs.return_value.get.return_value if isinstance(response, Exception): request.execute.side_effect = response exp_response = None else: request.execute.return_value = response exp_response = response ret_response = client.get_job_detail("job-name", "gcp-project") mock_find_job_by_name.assert_called_once_with( "job-name", "gcp-project", None ) if found_job: assert 1 == request.execute.call_count assert exp_response == ret_response assert exp_log_count == len(caplog.records) @pytest.mark.parametrize( "ret_job_info,exp_topic", ( (lambda: None, None), ( get_job_detail, "projects/sigint/topics/benchmark-audio-download-output", ), (get_job_detail_no_kind, None), (get_job_detail_no_username_value, None), (get_job_detail_no_topic, None), ), ) def test_get_job_input_topic( ret_job_info, exp_topic, mock_discovery_client, mocker, monkeypatch ): client = dataflow.DataflowClient() mock_get_job_detail = mocker.Mock() monkeypatch.setattr(client, "get_job_detail", mock_get_job_detail) mock_get_job_detail.return_value = ret_job_info() ret_topic = client.get_job_input_topic("job-name", "gcp-project") assert exp_topic == ret_topic @pytest.mark.parametrize( "api_version,exp_version", ((None, "v1b3"), ("v2", "v2")) ) def test_get_dataflow_client(api_version, exp_version, mock_discovery_client): client = dataflow.get_dataflow_client(api_version) assert isinstance(client, dataflow.DataflowClient) global_client = utils.get_global("dataflow_client_%s" % exp_version) assert global_client == client delattr(utils, "klio_global_state_dataflow_client_%s" % exp_version)
gaybro8777/klio
integration/audio-spectrograms/run.py
<reponame>gaybro8777/klio<gh_stars>0 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import warnings import apache_beam as beam from klio_audio.transforms import io as aio from klio_audio.transforms import audio import transforms warnings.simplefilter("ignore") loggers_to_mute = ( "apache_beam.io.filebasedsink", "apache_beam.runners.worker.statecache", "apache_beam.runners.portability.fn_api_runner", "apache_beam.runners.portability.fn_api_runner_transforms", "apache_beam.internal.gcp.auth", "oauth2client.transport", "oauth2client.client", # The concurrency logs may be different for every machine, so let's # just turn them off "klio.concurrency", ) for logger in loggers_to_mute: logging.getLogger(logger).setLevel(logging.ERROR) logging.getLogger("klio").setLevel(logging.DEBUG) def run(in_pcol, job_config): # load 5 seconds of audio and get STFT stft = ( in_pcol | aio.GcsLoadBinary() | audio.LoadAudio(offset=10, duration=5) | audio.GetSTFT() ) # get magnitude of audio magnitude = ( stft | "Get magnitude" >> beam.ParDo(transforms.GetMagnitude()).with_outputs() ) # map the result to a key (the KlioMessage element) # so we can group all results by key magnitude_key = ( magnitude.spectrogram | "element to spec" >> beam.Map(transforms.create_key_from_element) ) # get nearest neighbors and map the result to a key (the KlioMessage element) nn_filter = ( magnitude.spectrogram | "Get nn filter" >> beam.ParDo(transforms.FilterNearestNeighbors()) | "element to filter" >> beam.Map(transforms.create_key_from_element) ) # map together the full magnitude with its filter by key (the KlioMessage element) merge = ( {"full": magnitude_key, "nnfilter": nn_filter} | "merge" >> beam.CoGroupByKey() ) # calc the difference between full magnitude and the filter net = merge | beam.Map(transforms.subtract_filter_from_full) # create a mask from the filter minus the difference of full & filter first_mask = ( {"first": nn_filter, "second": net, "full": magnitude_key} | "first mask group" >> beam.CoGroupByKey() | "first mask" >> beam.ParDo(transforms.GetSoftMask(margin=2)) ) # create another mask from the difference of full & filter minus the filter second_mask = ( {"first": net, "second": nn_filter, "full": magnitude_key} | "second mask group" >> beam.CoGroupByKey() | "second mask" >> beam.ParDo(transforms.GetSoftMask(margin=10)) ) # plot the full magnitude spectrogram magnitude_out = ( magnitude.spectrogram | "full spec" >> audio.GetSpec() | "plot full spec" >> audio.SpecToPlot(title="Full Spectrogam for {element}", y_axis="log") | "save full" >> aio.GcsUploadPlot(suffix="-full") ) # plot the first mask (background) spectrogram background_out = ( first_mask | "background spec" >> audio.GetSpec() | "plot background spec" >> audio.SpecToPlot(title="Background Spectrogam for {element}", y_axis="log") | "save background" >> aio.GcsUploadPlot(suffix="-background") ) # plot the second mask (foreground) spectrogram foreground_out = ( second_mask | "foreground spec" >> audio.GetSpec() | "plot forground spec" >> audio.SpecToPlot(title="Foreground Spectrogam for {element}", y_axis="log") | "save foreground" >> aio.GcsUploadPlot(suffix="-foreground") ) return ( (magnitude_out, background_out, foreground_out) | "flatten output paths" >> beam.Flatten() | "remove dups" >> beam.Distinct() )
gaybro8777/klio
core/src/klio_core/config/_converters.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging logger = logging.getLogger("klio") class _UnsetRequiredValue(object): """ Sentinel class used on as the default for attributes that require a value. ``_UnsetRequiredValue`` is a singleton. There is only ever one of it. """ _singleton = None def __new__(cls): if _UnsetRequiredValue._singleton is None: _UnsetRequiredValue._singleton = super( _UnsetRequiredValue, cls ).__new__(cls) return _UnsetRequiredValue._singleton def __repr__(self): return "UNSET_REQUIRED_VALUE" UNSET_REQUIRED_VALUE = _UnsetRequiredValue() """ This is set as the default for an attribute when we want no default. This way attrs itself doesn't raise an exception and instead the value is passed into our converter which will raise a properly formatted error """ class ConfigValueConverter(object): """Base class for converters automatically added to any `attrs` `attrib` created in a class decorated with `config_object`. These converters properly handle raising Exceptions when a required value is not provided, as well as more strict type checking and type coercion. """ def __init__(self, key): self.key = key def validate(self, value): if value == UNSET_REQUIRED_VALUE: raise Exception("missing required key: {}".format(self.key)) elif value is None: return value return self._validate_value(value) def _validate_value(self, value): return value class TypeCheckingConverter(ConfigValueConverter): def __init__(self, key, type): super().__init__(key) self.type = type def _wrong_type_msg(self, wrong_value): return ( "{}: expected value of type '{}'" ", got a '{}' value instead" ).format(self.key, self.type.__name__, type(wrong_value).__name__) class SingleValueConverter(TypeCheckingConverter): def validate(self, value): if isinstance(value, dict) or isinstance(value, list): self._wrong_type_message(value) return super().validate(value) class IntConverter(SingleValueConverter): def __init__(self, key_prefix): super().__init__(key_prefix, int) def _validate_value(self, value): try: return int(value) except Exception: raise Exception( "{}: expected numeric value, got '{}'".format(self.key, value) ) class BoolConverter(SingleValueConverter): def __init__(self, key_prefix): super().__init__(key_prefix, bool) def _validate_value(self, value): if value is None: return value try: new_value = bool(value) if not isinstance(value, bool): logger.warning( ( "{}: found non-bool value '{}'" ", converting to bool '{}'" ).format(self.key, value, new_value) ) return new_value except Exception: raise Exception( "{}: expected boolean value, got '{}'".format(self.key, value) ) class StringConverter(SingleValueConverter): def __init__(self, key_prefix): super().__init__(key_prefix, str) def _validate_value(self, value): if value is None: return value if not isinstance(value, str): logger.warning( "{}: found non-string value '{}', converting to str".format( self.key, value ) ) return str(value) class Converters(object): TYPES = { int: IntConverter, str: StringConverter, bool: BoolConverter, } @classmethod def for_type(cls, type, key): if type in cls.TYPES: return cls.TYPES[type](key).validate else: return ConfigValueConverter(key).validate
gaybro8777/klio
lib/src/klio/metrics/client.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Metrics registry to manage metrics for all configured relay clients. The :class:`MetricsRegistry` class is the main client for which to create and emit metrics. For example: .. code-block:: python class MyTransform(KlioBaseTransform): def __init__(self): self.my_counter = self._klio.metrics.counter( name="foo", value=2, transform="mytransform", tags={"tag1": "value1"} ) def process(self, element): # user code # immediately emits a metric self.my_counter.inc() """ import logging from klio.metrics import dispatcher class MetricsRegistry(object): """Main client to create and emit metrics. Args: relay_clients (list(klio.metrics.base.AbstractRelayClient)): configured relay clients. """ def __init__(self, relay_clients, transform_name): self._relays = relay_clients self._transform_name = transform_name self._registry = {} def counter(self, name, value=0, **kwargs): """Get or create a counter. Creates a new counter if one is not found with a key of ``"counter_{name}_{transform}"`` of the given ``transform``. New counters will be stored in memory for simple caching. Args: name (str): name of counter value (int): starting value of counter; defaults to 0 kwargs (dict): keyword arguments passed to each configured relay clients' counter object. Returns: dispatcher.CounterDispatcher: instance of a counter dispatcher """ transform_name = kwargs.pop("transform", self._transform_name) key = "counter_{}_{}".format(name, transform_name) if key in self._registry: return self._registry[key] counter = dispatcher.CounterDispatcher( relay_clients=self._relays, name=name, value=value, transform=transform_name, **kwargs ) self._registry[key] = counter return counter def gauge(self, name, value=0, **kwargs): """Get or create a gauge. Creates a new gauge if one is not found with a key of "gauge_{name}_{transform}". New gauges will be stored in memory for simple caching. Args: name (str): name of gauge value (int): starting value of gauge; defaults to 0 kwargs (dict): keyword arguments passed to each configured relay clients' gauge object. Returns: dispatcher.GaugeDispatcher: instance of a gauge dispatcher """ transform_name = kwargs.pop("transform", self._transform_name) key = "gauge_{}_{}".format(name, transform_name) if key in self._registry: return self._registry[key] gauge = dispatcher.GaugeDispatcher( relay_clients=self._relays, name=name, value=value, transform=transform_name, **kwargs ) self._registry[key] = gauge return gauge def timer(self, name, value=0, timer_unit="ns", **kwargs): """Get or create a timer. Creates a new timer if one is not found with a key of "timer_{name}_{transform}". New timers will be stored in memory for simple caching. Args: name (str): name of timer value (int): starting value of timer; defaults to 0 timer_unit (str): desired unit of time; defaults to ns kwargs (dict): keyword arguments passed to each configured relay clients' timer object. Returns: dispatcher.TimerDispatcher: instance of a timer dispatcher """ transform_name = kwargs.pop("transform", self._transform_name) key = "timer_{}_{}".format(name, transform_name) if key in self._registry: return self._registry[key] timer = dispatcher.TimerDispatcher( relay_clients=self._relays, name=name, value=value, transform=transform_name, timer_unit=timer_unit, **kwargs ) self._registry[key] = timer return timer def marshal(self, metric): """Create a dictionary-representation of a given metric. Used when metric objects need to be pickled. Args: metric (dispatcher.BaseMetricDispatcher): metric instance to marshal. Returns: dict: the metric's data in dictionary form """ data = { "type": metric.METRIC_TYPE, "name": metric.name, "value": metric.value, } data.update(metric.kwargs) return data def unmarshal(self, metric_data): """Create a metric instance based off of a dictionary. If "type" is not specified or is not one of "counter", "gauge", or "timer", it defaults to a gauge-type metric. Used when metrics objects need to be unpickled. Args: metric_data (dict): dictionary-representation of a given metric. Returns: dispatcher.BaseMetricDispatcher: a dispatcher relevant to metric type. """ metric_type = metric_data.pop("type", None) name = metric_data.pop("name", None) value = metric_data.pop("value", None) kwargs = metric_data metric_method = { "counter": self.counter, "gauge": self.gauge, "timer": self.timer, }.get(metric_type) if not metric_method: msg = "Metric type '{}' not supported. Defaulting to gauge.".format( metric_type ) logging.getLogger("klio.metrics").warning(msg) metric_method = self.gauge return metric_method(name=name, value=value, **kwargs)
gaybro8777/klio
lib/tests/unit/utils/test_thread_limiter.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import multiprocessing import pytest from klio.utils import _thread_limiter N_CPU = multiprocessing.cpu_count() BASE_LOG = "Initial semaphore value:" @pytest.mark.parametrize( "max_thread_count, exp_count_log", ( (1, f"{BASE_LOG} 1"), (_thread_limiter.ThreadLimit.NONE, "Using unlimited semaphore"), (_thread_limiter.ThreadLimit.DEFAULT, f"{BASE_LOG} {N_CPU}"), (lambda: 2 * 3, f"{BASE_LOG} 6"), ), ) def test_thread_limiter(max_thread_count, exp_count_log, caplog): _thread_limiter.ThreadLimiter(max_thread_count=max_thread_count) assert 1 == len(caplog.messages) assert exp_count_log in caplog.messages[0] @pytest.mark.parametrize( "max_thread_count", (2, _thread_limiter.ThreadLimit.NONE) ) def test_thread_limiter_ctx_mgr(max_thread_count, mocker, monkeypatch, caplog): mock_semaphore = mocker.Mock() limiter = _thread_limiter.ThreadLimiter(max_thread_count=max_thread_count) monkeypatch.setattr(limiter, "_semaphore", mock_semaphore) with limiter: 3 * 3 mock_semaphore.acquire.assert_called_once_with() mock_semaphore.release.assert_called_once_with() assert 3 == len(caplog.messages)
gaybro8777/klio
exec/tests/unit/commands/utils/test_plugin_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.utils import plugin_utils @pytest.fixture def mock_inspect(mocker, monkeypatch): mock = mocker.Mock() mock.getfile.return_value = "/path/to/package" monkeypatch.setattr(plugin_utils, "inspect", mock) return mock @pytest.fixture def mock_plugins(mocker): mock_dist = mocker.Mock() mock_dist.project_name = "package-name" mock_dist.parsed_version = "1.2.3" mock_plugins = [] for i in range(2): mock_loaded = mocker.Mock() mock_loaded.AUDIT_STEP_NAME = "step-{}".format(i) mock_plugin = mocker.Mock(name=i) mock_plugin.load.return_value = mock_loaded mock_plugin.dist = mock_dist mock_plugins.append(mock_plugin) return mock_plugins @pytest.fixture def mock_iter_entry_points(mock_plugins, mocker, monkeypatch): mock = mocker.Mock() mock.return_value = mock_plugins monkeypatch.setattr(plugin_utils.pkg_resources, "iter_entry_points", mock) return mock def test_load_plugins_by_namespace(mock_plugins, mock_iter_entry_points): ret = plugin_utils.load_plugins_by_namespace("a.namespace") mock_iter_entry_points.assert_called_once_with("a.namespace") for plugin in mock_plugins: plugin.load.assert_called_once_with() assert ret == [p.load.return_value for p in mock_plugins] @pytest.mark.parametrize( "get_desc,docstr,exp_desc", ( ("a description", None, "a description"), (None, "a docstring desc", "a docstring desc"), (None, None, "No description."), ("a description", "a docstring desc", "a description"), ), ) def test_get_plugins_by_namespace( get_desc, docstr, exp_desc, mock_inspect, mock_plugins, mock_iter_entry_points, mocker, ): for mp in mock_plugins: mp.load.return_value.get_description.return_value = get_desc mp.load.return_value.__doc__ = docstr exp_plugins = [] for i in range(2): exp_p = plugin_utils.KlioPlugin( plugin_name="step-{}".format(i), description=exp_desc, package_name="package-name", package_version="1.2.3", module_path="/path/to/package", ) exp_plugins.append(exp_p) actual_plugins = list( plugin_utils._get_plugins_by_namespace("a.namespace") ) mock_iter_entry_points.assert_called_once_with("a.namespace") for plugin in mock_plugins: plugin.load.assert_called_once_with() exp_calls = [mocker.call(plugin.load.return_value)] mock_inspect.getfile.assert_has_calls(exp_calls) assert len(mock_plugins) == mock_inspect.getfile.call_count assert exp_plugins == actual_plugins @pytest.mark.parametrize("tw", (True, False)) def test_print_plugins(tw, mock_terminal_writer, mocker, monkeypatch): loaded_plugins = [ plugin_utils.KlioPlugin( plugin_name="plugin_name", description="a desc", package_name="package-name", package_version="1.2.3", module_path="/path/to/package", ) ] mock_get_plugins_by_ns = mocker.Mock() mock_get_plugins_by_ns.return_value = loaded_plugins monkeypatch.setattr( plugin_utils, "_get_plugins_by_namespace", mock_get_plugins_by_ns ) exp_plugin_meta_str = " -- via package-name (v1.2.3) -- /path/to/package\n" exp_plugin_desc = "\ta desc\n\n" exp_calls = [ mocker.call("plugin_name", blue=True, bold=True), mocker.call(exp_plugin_meta_str, green=True), mocker.call(exp_plugin_desc), ] kwargs = {"tw": None} if tw: kwargs["tw"] = mock_terminal_writer plugin_utils.print_plugins("a.namespace", **kwargs) mock_get_plugins_by_ns.assert_called_once_with("a.namespace") mock_terminal_writer.write.assert_has_calls(exp_calls) assert 3 == mock_terminal_writer.write.call_count
gaybro8777/klio
exec/tests/unit/commands/audit_steps/test_numpy_broken_blas.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.audit_steps import numpy_broken_blas @pytest.mark.parametrize( "worker_threads,platform,has_numpy,version,exp_emit_calls", ( (1, None, False, None, 0), (2, "Darwin", False, None, 0), (0, "Linux", False, None, 0), (2, "Linux", False, None, 0), (2, "Linux", True, "1.17.0", 0), (2, "Linux", True, "1.16.3", 0), (2, "Linux", True, "1.16.2", 1), ), ) def test_numpy_broken_blas_usage( worker_threads, platform, has_numpy, version, exp_emit_calls, klio_config, mock_emit_error, mocker, monkeypatch, ): if worker_threads: klio_config.pipeline_options.experiments = [ "worker_threads={}".format(worker_threads) ] monkeypatch.setattr(numpy_broken_blas.platform, "system", lambda: platform) if has_numpy is False: mocker.patch.dict("sys.modules", {"numpy": None}) else: monkeypatch.setattr("numpy.version.short_version", version) numpy_broken_blas_usage = numpy_broken_blas.NumPyBrokenBLASUsage( "job/dir", klio_config, "term_writer" ) numpy_broken_blas_usage.after_tests() # don't care about the actual message assert exp_emit_calls == mock_emit_error.call_count
gaybro8777/klio
exec/tests/unit/commands/utils/test_cpu_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.utils import cpu_utils @pytest.mark.parametrize("interval,exp_sleep", ((None, 0.1), (0.5, 0.5))) def test_get_cpu_usage(interval, exp_sleep, mocker, monkeypatch, capsys): mock_proc = mocker.Mock() # wanting at least 50 lines to hit the flush proc_side_effect = [None] * 51 proc_side_effect.append(True) mock_proc.poll.side_effect = proc_side_effect mock_time = mocker.Mock() timestamps = range(1, 53) mock_time.time.side_effect = timestamps monkeypatch.setattr(cpu_utils, "time", mock_time) cpu_measurements = (29.0, 51.4, 78.1, 61.2) * 13 mock_cpu_percent = mocker.Mock(side_effect=cpu_measurements) monkeypatch.setattr(cpu_utils.psutil, "cpu_percent", mock_cpu_percent) cpu_utils.get_cpu_usage(mock_proc, interval=interval) assert 52 == mock_time.sleep.call_count assert mocker.call(exp_sleep) == mock_time.sleep.call_args captured = capsys.readouterr() data = zip(cpu_measurements, timestamps) exp_out = "\n".join("CPU {} {}.0000".format(d[0], d[1]) for d in data) exp_out = exp_out + "\n" assert exp_out == captured.out
gaybro8777/klio
cli/src/klio_cli/options.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import click from klio_core import options as core_options def _verify_gcs_uri(ctx, param, value): if value and not value.startswith("gs://"): raise click.BadParameter( "Unsupported location type. Please provide a GCS location with " "the `gs://` prefix." ) return value ##### # common options ##### def force_build(func): return click.option( "--force-build", default=False, is_flag=True, help="Build Docker image even if you already have it.", )(func) def runtime(func): func = force_build(func) func = core_options.direct_runner(func) func = core_options.update(func) return func def job_name(*args, **kwargs): def wrapper(func): return click.option("--job-name", **kwargs)(func) # allows @options.foo to be used without parens (i.e. no need to do # `@options.foo()`) when there are no args/kwargs provided if args: return wrapper(args[0]) return wrapper def region(func): return click.option( "--region", help=( "Region of job, if neither ``--job-dir`` nor ``--config-file`` is " "not provided." ), )(func) def gcp_project(*args, **kwargs): def wrapper(func): return click.option("--gcp-project", **kwargs)(func) # allows @options.foo to be used without parens (i.e. no need to do # `@options.foo()`) when there are no args/kwargs provided if args: return wrapper(args[0]) return wrapper ##### # options for `klio job verify ##### def create_resources(func): return click.option( "--create-resources", default=False, is_flag=True, help=( "Create missing GCP resources based on ``klio-info.yaml``." " Default: ``False``" ), )(func) ##### # options for `klio job audit` ##### def list_steps(func): return click.option( "--list", "list_steps", is_flag=True, default=False, help="List available audit steps (does not run any audits).", )(func) ##### # options for `klio job create` ##### def output(func): return click.option( "--output", help="Output directory. Defaults to current working directory.", )(func) def use_defaults(func): return click.option( "--use-defaults", default=False, is_flag=True, help="Accept default values.", )(func) ##### # common options for `klio job profile <subcommand>` ##### def input_file(*args, **kwargs): mutually_exclusive = kwargs.pop("mutex", []) def wrapper(func): return click.option( "-i", "--input-file", type=click.Path(exists=False, dir_okay=False, readable=True), cls=core_options.MutuallyExclusiveOption, mutually_exclusive=mutually_exclusive, **kwargs, )(func) # allows @options.foo to be used without parens (i.e. no need to do # `@options.foo()`) when there are no args/kwargs provided if args: return wrapper(args[0]) return wrapper def output_file(*args, **kwargs): mutually_exclusive = kwargs.pop("mutex", []) def wrapper(func): return click.option( "-o", "--output-file", type=click.Path(exists=False, dir_okay=False, writable=True), cls=core_options.MutuallyExclusiveOption, mutually_exclusive=mutually_exclusive, **kwargs, )(func) # allows @options.foo to be used without parens (i.e. no need to do # `@options.foo()`) when there are no args/kwargs provided if args: return wrapper(args[0]) return wrapper ##### # options for `klio job profile collect-profiling-data` ##### def gcs_location(func): return click.option( "--gcs-location", default=None, show_default="`pipeline_options.profile_location` in `klio-job.yaml`", help="GCS location of cProfile data.", cls=core_options.MutuallyExclusiveOption, mutually_exclusive=["job_dir", "input_file", "config_file"], callback=_verify_gcs_uri, )(func) def since(func): return click.option( "--since", default=None, show_default="1 hour ago", help=( "Start time, relative or absolute (interpreted by " "``dateparser.parse``). " ), )(func) def until(func): return click.option( "--until", default=None, show_default="now", help=( "End time, relative or absolute (interpreted by " "``dateparser.parse``). " ), )(func) # https://docs.python.org/3.6/library/profile.html#pstats.Stats.sort_stats # unfortunately, can't get this dynamically <3.7 (@lynn) SORT_STATS_KEY = [ "calls", "cumulative", "cumtime", "file", "filename", "module", "ncalls", "pcalls", "line", "name", "nfl", "stdname", "time", "tottime", ] def sort_stats(func): return click.option( "--sort-stats", multiple=True, default=["tottime"], show_default=True, type=click.Choice(SORT_STATS_KEY, case_sensitive=False), help=( "Sort output of profiling statistics as supported by " "`sort_stats <https://docs.python.org/3/library/profile.html#" "pstats.Stats.sort_stats>`_. Multiple ``--sort-stats`` invocations " "are supported." ), )(func) ##### # options for `klio message publish` ##### def force(func): return click.option( "-f", "--force", default=False, is_flag=True, help="Force processing even if output data exists.", )(func) def ping(func): return click.option( "-p", "--ping", default=False, is_flag=True, help="Skip the processing of an ID to trace the path the ID takes.", )(func) def top_down(func): return click.option( "-t", "--top-down", default=False, is_flag=True, help=( "Trigger an apex node and all child nodes below. Default: " "``False``" ), )(func) def bottom_up(func): return click.option( "-b", "--bottom-up", # default is actually true, but initially conflicts w top-down logic default=False, is_flag=True, help=( "Trigger work for only this job and any required parent jobs, " "but no sibling or child jobs. Default: ``True``" ), )(func) def non_klio(func): return click.option( "-n", "--non-klio", default=False, is_flag=True, help=( "Publish a free-form, non-Klio message to the targeted job. The " "targeted job must also support non-Klio messages. Mutually " "exclusive with ``--force``, ``--ping``, and ``--bottom-up``. " "Default: ``False``" ), )(func)
gaybro8777/klio
exec/tests/unit/commands/audit_steps/conftest.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.audit_steps import base @pytest.fixture def klio_config(mocker): conf = mocker.Mock() conf.pipeline_options = mocker.Mock() conf.pipeline_options.experiments = [] return conf @pytest.fixture def mock_emit_warning(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(base.BaseKlioAuditStep, "emit_warning", mock) return mock @pytest.fixture def mock_emit_error(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(base.BaseKlioAuditStep, "emit_error", mock) return mock
gaybro8777/klio
cli/src/klio_cli/commands/job/run.py
<reponame>gaybro8777/klio # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import atexit import logging import docker from klio_cli import __version__ as klio_cli_version from klio_cli.commands import base from klio_cli.utils import docker_utils from klio_cli.utils import stackdriver_utils as sd_utils GCP_CRED_FILE = "gcloud/application_default_credentials.json" class RunPipeline(base.BaseDockerizedPipeline): DOCKER_LOGGER_NAME = "klio.job.run" def __init__( self, job_dir, klio_config, docker_runtime_config, run_job_config ): super().__init__(job_dir, klio_config, docker_runtime_config) self.run_job_config = run_job_config self.requires_config_file = True @staticmethod def _try_container_kill(container): try: container.kill() except docker.errors.APIError as e: logging.debug("Exception while killing container", exc_info=e) def _run_docker_container(self, runflags): container = self._docker_client.containers.run(**runflags) if self.run_job_config.direct_runner: atexit.register(self._try_container_kill, container) # TODO: container.logs(stream=True) redirects stderr to stdout. # We should use appropriate streams so it's obvious to the use. # (@jpvelez) for line in container.logs(stream=True): self._docker_logger.info(line.decode("utf-8").strip("\n")) exit_status = container.wait()["StatusCode"] if exit_status == 0 and not self.run_job_config.direct_runner: dashboard_name = sd_utils.DASHBOARD_NAME_TPL.format( job_name=self.klio_config.job_name, region=self.klio_config.pipeline_options.region, ) base_err_msg = ( "Could not find a Stackdriver dashboard for job '%s' that " "matched the name %s" % (self.klio_config.job_name, dashboard_name) ) try: dashboard_url = sd_utils.get_stackdriver_group_url( self.klio_config.pipeline_options.project, self.klio_config.job_name, self.klio_config.pipeline_options.region, ) except Exception as e: logging.warning("%s: %s" % (base_err_msg, e)) else: if dashboard_url: logging.info( "View the job's dashboard on Stackdriver: %s" % dashboard_url ) else: logging.warning(base_err_msg) return exit_status def _get_environment(self): envs = super()._get_environment() envs[ "GOOGLE_CLOUD_PROJECT" ] = self.klio_config.pipeline_options.project envs["COMMIT_SHA"] = self.run_job_config.git_sha envs["KLIO_CLI_VERSION"] = klio_cli_version return envs def _get_command(self): command = ["run"] if self.docker_runtime_config.image_tag: command.extend( ["--image-tag", self.docker_runtime_config.image_tag] ) if self.run_job_config.direct_runner: command.append("--direct-runner") if self.run_job_config.update is True: command.append("--update") elif ( self.run_job_config.update is False ): # don't do anything if `None` command.append("--no-update") return command def _setup_docker_image(self): super()._setup_docker_image() if not self.run_job_config.direct_runner: logging.info("Pushing worker image to GCR") docker_utils.push_image_to_gcr( self._full_image_name, self.docker_runtime_config.image_tag, self._docker_client, )
gaybro8777/klio
core/tests/config/test_preprocessing.py
<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import pytest import yaml from klio_core import exceptions from klio_core.config import _preprocessing @pytest.fixture def kcp(): return _preprocessing.KlioConfigPreprocessor @pytest.fixture def job_raw_config(): return { "allow_non_klio_messages": False, "events": { "inputs": { "file0": { "type": "file", "location": "gs://test/yesterday.txt", }, "file1": {"type": "file", "location": "gs://test/today.txt"}, } }, } @pytest.fixture def job_raw_config_templated(): return { "events": { "inputs": { "bigquery0": { "type": "bigquery", "table": "testing.top_tracks_global", "partition": "$YESTERDAY", }, "bigquery1": { "type": "bigquery", "table": "testing.top_artists_global", "partition": "$TODAY", }, "bigquery2": { "type": "bigquery", "table": "testing.${GENRE}_join_$COUNTRY", "partition": "$TODAY", }, } } } @pytest.mark.parametrize( "overrides_dict,expected", ( # No overrides given - no changes in returned dict ( {}, { "allow_non_klio_messages": False, "events": { "inputs": { "file0": { "type": "file", "location": "gs://test/yesterday.txt", }, "file1": { "type": "file", "location": "gs://test/today.txt", }, } }, }, ), # Override dict given but no changes - no changes in returned dict ( {"allow_non_klio_messages": True}, { "allow_non_klio_messages": True, "events": { "inputs": { "file0": { "type": "file", "location": "gs://test/yesterday.txt", }, "file1": { "type": "file", "location": "gs://test/today.txt", }, } }, }, ), # Override current values ( { "events.inputs.file0.location": "gs://test/12-31-2019.txt", "events.inputs.file1.location": "gs://test/01-01-2020.txt", }, { "allow_non_klio_messages": False, "events": { "inputs": { "file0": { "type": "file", "location": "gs://test/12-31-2019.txt", }, "file1": { "type": "file", "location": "gs://test/01-01-2020.txt", }, } }, }, ), # Add new values ( { "events.inputs.file2.location": "gs://test/01-01-2020.txt", "events.inputs.file2.type": "file", }, { "allow_non_klio_messages": False, "events": { "inputs": { "file0": { "type": "file", "location": "gs://test/yesterday.txt", }, "file1": { "type": "file", "location": "gs://test/today.txt", }, "file2": { "type": "file", "location": "gs://test/01-01-2020.txt", }, } }, }, ), ), ) def test_apply_overrides(kcp, job_raw_config, overrides_dict, expected): # One key override new_dict = kcp._apply_overrides(job_raw_config, overrides_dict) assert expected == new_dict @pytest.mark.parametrize( "template_dict,expected", ( # One or more missing templates should raise a KeyError ( { "YESTERDAY": "12-31-2019", "TODAY": "01-01-2020", "COUNTRY": "sto", }, None, ), ( { "YESTERDAY": "12-31-2019", "TODAY": "01-01-2020", "GENRE": "rock", "COUNTRY": "sto", }, { "events": { "inputs": { "bigquery0": { "type": "bigquery", "table": "testing.top_tracks_global", "partition": "12-31-2019", }, "bigquery1": { "type": "bigquery", "table": "testing.top_artists_global", "partition": "01-01-2020", }, "bigquery2": { "type": "bigquery", "table": "testing.rock_join_sto", "partition": "01-01-2020", }, } } }, ), ), ) def test_apply_templates( kcp, job_raw_config_templated, template_dict, expected ): raw_config_str = json.dumps(job_raw_config_templated) if not expected: with pytest.raises( exceptions.KlioConfigTemplatingException, match=r"'GENRE' .*" ): kcp._apply_templates(raw_config_str, template_dict) else: new_dict = kcp._apply_templates(raw_config_str, template_dict) assert isinstance(new_dict, str) assert expected == json.loads(new_dict) def test_transform_io(kcp): config = { "job_config": { "events": { "inputs": [ {"type": "bq", "key": "value"}, {"type": "bq", "key": "value2"}, {"type": "gcs", "gcskey": "gcsvalue"}, ], "outputs": [ {"type": "bq", "key": "value", "name": "mybq"}, {"type": "bq", "key": "value2"}, ], }, "data": { "inputs": [], "outputs": [ {"type": "bq", "key": "value", "name": "mybq"}, {"type": "bq", "key": "value2"}, ], }, } } expected = { "job_config": { "events": { "inputs": { "bq0": {"type": "bq", "key": "value"}, "bq1": {"type": "bq", "key": "value2"}, "gcs0": {"type": "gcs", "gcskey": "gcsvalue"}, }, "outputs": { "mybq": {"type": "bq", "key": "value"}, "bq0": {"type": "bq", "key": "value2"}, }, }, "data": { "inputs": {}, "outputs": { "mybq": {"type": "bq", "key": "value"}, "bq0": {"type": "bq", "key": "value2"}, }, }, } } actual = kcp._transform_io_sections(config) assert actual == expected def test_apply_plugin_preprocessing(mocker, kcp): config = { "version": 1, "job_name": "test_job", "job_config": {}, } raw_config = yaml.dump(config) def add_field(config_dict): config_dict["version"] = 2 return config_dict expected_config = config.copy() expected_config["version"] = 2 mocker.patch.object(kcp, "PLUGIN_PREPROCESSORS", [add_field]) processed_config = kcp.process(raw_config, [], []) assert expected_config == processed_config
gaybro8777/klio
cli/tests/commands/job/test_verify.py
<reponame>gaybro8777/klio<filename>cli/tests/commands/job/test_verify.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import pytest from google.api_core import exceptions as api_ex from google.api_core import page_iterator from google.cloud import exceptions from googleapiclient import errors as google_errors from klio_core import config as kconfig from klio_cli.commands.job import verify from klio_cli.utils import stackdriver_utils as sd_utils @pytest.fixture def mock_publisher(mocker): return mocker.patch.object(verify.pubsub_v1, "PublisherClient") @pytest.fixture def mock_sub(mocker): return mocker.patch.object(verify.pubsub_v1, "SubscriberClient") @pytest.fixture def mock_storage(mocker): return mocker.patch.object(verify.storage, "Client") @pytest.fixture def mock_bucket(mocker): return mocker.patch.object(verify.storage.Client, "bucket") @pytest.fixture def mock_blob(mocker): return mocker.patch.object(verify.storage, "blob") @pytest.fixture def mock_discovery_client(mocker, monkeypatch): mock = mocker.MagicMock() monkeypatch.setattr(verify.discovery, "build", lambda x, y: mock) return mock @pytest.fixture def mock_iterator(mocker): return mocker.patch.object(page_iterator, "HTTPIterator") @pytest.fixture def config(): return { "job_name": "klio-job-name", "job_config": { "inputs": [ { "topic": "test-in-topic", "subscription": "foo", "data_location": "gs://test-in-data", } ], "outputs": [ { "topic": "test-out-topic", "data_location": "gs://test-out-data", } ], }, "pipeline_options": { "streaming": True, "worker_harness_container_image": "a-worker-image", "experiments": ["beam_fn_api"], "project": "test-gcp-project", "zone": "europe-west1-c", "region": "europe-west1", "staging_location": "gs://test-gcp-project-dataflow-tmp/staging", "temp_location": "gs://test-gcp-project-dataflow-tmp/temp", "max_num_workers": 2, "autoscaling_algorithm": "NONE", "disk_size_gb": 32, "worker_machine_type": "n1-standard-2", "runner": "DataflowRunner", }, } @pytest.fixture def klio_config(config): return kconfig.KlioConfig(config) @pytest.fixture def mock_get_sd_group_url(mocker): return mocker.patch.object(sd_utils, "get_stackdriver_group_url") @pytest.fixture def mock_create_sd_group(mocker): return mocker.patch.object(sd_utils, "create_stackdriver_group") @pytest.mark.parametrize( "expected,bindings,create_resources", [ ( True, [ { "role": "roles/monitoring.metricWriter", "members": ["serviceAccount:the-default-svc-account"], } ], False, ), ( False, [ { "role": "roles/monitoring.metricWriter", "members": ["serviceAccount:some-other-account"], } ], False, ), ( False, [ { "role": "roles/monitoring.metricWriter", "members": ["serviceAccount:some-other-account"], } ], True, ), ( False, [{"role": "roles/monitoring.metricWriter", "members": []}], False, ), (False, [], False), ], ) def test_verify_iam_roles( caplog, expected, bindings, create_resources, mock_discovery_client, klio_config, ): compute_client = mock_discovery_client.build("compute") compute_client.projects().get().execute.return_value = { "defaultServiceAccount": "the-default-svc-account" } iam_client = mock_discovery_client.build("cloudresourcemanager") job = verify.VerifyJob(klio_config, create_resources) job._compute_client = compute_client job._iam_client = iam_client gcp_project = job.klio_config.pipeline_options.project iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.return_value = {"bindings": bindings} result = job._verify_iam_roles() if create_resources: assert ( "--create-resources is not able to add these roles" in caplog.records[-1].msg ) compute_client.projects().get( project=gcp_project ).execute.assert_called_once_with() iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.assert_called_once_with() assert result is expected def test_verify_iam_roles_editor(caplog, klio_config, mock_discovery_client): bindings = [ { "role": "roles/monitoring.metricWriter", "members": ["serviceAccount:the-default-svc-account"], }, { "role": "roles/editor", "members": ["serviceAccount:the-default-svc-account"], }, ] gcp_project = klio_config.pipeline_options.project compute_client = mock_discovery_client.build("compute") compute_client.projects().get().execute.return_value = { "defaultServiceAccount": "the-default-svc-account" } iam_client = mock_discovery_client.build("cloudresourcemanager") iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.return_value = {"bindings": bindings} job = verify.VerifyJob(klio_config, False) job._compute_client = compute_client job._iam_client = iam_client result = job._verify_iam_roles() compute_client.projects().get( project=gcp_project ).execute.assert_called_once_with() iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.assert_called_once_with() assert result is True with caplog.at_level(logging.WARNING): assert len(caplog.records) == 3 msg = caplog.records[1].msg assert "unsafe project editor or owner permissions" in msg def test_verify_iam_roles_with_svc_account(klio_config, mock_discovery_client): "If the user configures a SA, verify it instead of the default compute SA" job = verify.VerifyJob(klio_config, False) job.klio_config.pipeline_options.service_account_email = ( "<EMAIL>" ) bindings = [ { "role": "roles/monitoring.metricWriter", "members": ["serviceAccount:<EMAIL>"], }, { "role": "roles/editor", "members": ["serviceAccount:the-default-svc-account"], }, ] gcp_project = job.klio_config.pipeline_options.project compute_client = mock_discovery_client.build("compute") compute_client.projects().get().execute.return_value = { "defaultServiceAccount": "the-default-svc-account" } iam_client = mock_discovery_client.build("cloudresourcemanager") iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.return_value = {"bindings": bindings} job._compute_client = compute_client job._iam_client = iam_client result = job._verify_iam_roles() # Assert that we don't fetch the default SA since we don't need it compute_client.projects().get( project=gcp_project ).execute.assert_not_called() iam_client.projects().getIamPolicy( resource=gcp_project, body={} ).execute.assert_called_once_with() assert result is True def test_verify_iam_roles_http_error(klio_config, mock_discovery_client): compute_client = mock_discovery_client.build("compute") err = google_errors.HttpError compute_client.projects().get().execute.side_effect = err( "some resp", "some content".encode() ) iam_client = mock_discovery_client.build("cloudresourcemanager") job = verify.VerifyJob(klio_config, False) job._compute_client = compute_client job._iam_client = iam_client result = job._verify_iam_roles() assert result is False compute_client = mock_discovery_client.build("compute") compute_client.projects().get().execute.side_effect = None compute_client.projects().get().execute.return_value = { "defaultServiceAccount": "the-default-svc-account" } iam_client = mock_discovery_client.build("cloudresourcemanager") iam_client.projects().getIamPolicy( resource=job.klio_config.pipeline_options.project, body={} ).execute.side_effect = google_errors.HttpError( "some resp", "some content".encode() ) job._compute_client = compute_client job._iam_client = iam_client result = job._verify_iam_roles() assert result is False @pytest.mark.parametrize( "create_resources,dashboard_url", ((True, None), (False, None), (False, "dashboard/url")), ) def test_verify_stackdriver_dashboard( klio_config, mock_get_sd_group_url, mock_create_sd_group, create_resources, dashboard_url, ): mock_get_sd_group_url.return_value = dashboard_url job = verify.VerifyJob(klio_config, create_resources) actual = job._verify_stackdriver_dashboard() mock_get_sd_group_url.assert_called_once_with( "test-gcp-project", "klio-job-name", "europe-west1" ) if create_resources: mock_create_sd_group.assert_called_once_with( "test-gcp-project", "klio-job-name", "europe-west1" ) assert actual is True elif dashboard_url: assert actual is True else: assert actual is False def test_verify_stackdriver_dashboard_raises( klio_config, mock_get_sd_group_url, caplog ): mock_get_sd_group_url.side_effect = Exception("error") with pytest.raises(Exception, match="error"): job = verify.VerifyJob(klio_config, False) job._verify_stackdriver_dashboard() assert 2 == len(caplog.records) def test_verify_stackdriver_dashboard_errors( klio_config, mock_get_sd_group_url, mock_create_sd_group, caplog ): mock_get_sd_group_url.return_value = None mock_create_sd_group.return_value = None job = verify.VerifyJob(klio_config, True) actual = job._verify_stackdriver_dashboard() assert actual is False assert 3 == len(caplog.records) @pytest.mark.parametrize("create_resources", (True, False)) def test_verify_gcs_bucket(klio_config, mock_storage, create_resources): test_path = "gs://bucket/blob" job = verify.VerifyJob(klio_config, create_resources) job._storage_client = mock_storage actual = job._verify_gcs_bucket(test_path) if create_resources: mock_storage.create_bucket.assert_called_once_with("bucket") else: mock_storage.get_bucket.assert_called_once_with("bucket") assert actual is True def test_verify_gcs_bucket_invalid_name(klio_config, mock_storage, caplog): job = verify.VerifyJob(klio_config, True) job._storage_client = mock_storage assert not job._verify_gcs_bucket("a/b/c") assert 2 == len(caplog.records) def test_verify_gcs_bucket_exists(klio_config, mock_storage): test_path = "gs://bucket/blob" mock_storage.create_bucket.side_effect = api_ex.Conflict("test") job = verify.VerifyJob(klio_config, True) job._storage_client = mock_storage actual = job._verify_gcs_bucket(test_path) assert actual is True @pytest.mark.parametrize("not_found", (True, False)) def test_verify_gcs_bucket_exceptions(klio_config, mock_storage, not_found): test_path = "gs://bucket/blob" if not_found: mock_storage.get_bucket.side_effect = exceptions.NotFound("test") else: mock_storage.get_bucket.side_effect = Exception job = verify.VerifyJob(klio_config, False) job._storage_client = mock_storage actual = job._verify_gcs_bucket(test_path) assert actual is False @pytest.mark.parametrize("create_resources", (True, False)) def test_verify_pub_topic(klio_config, mock_publisher, create_resources): test_topic = "test" job = verify.VerifyJob(klio_config, create_resources) job._publisher_client = mock_publisher actual = job._verify_pub_topic(test_topic, input) if create_resources: mock_publisher.create_topic.assert_called_once_with( request={"name": test_topic} ) else: mock_publisher.get_topic.assert_called_once_with( request={"topic": test_topic} ) assert actual is True def test_verify_pub_topic_exists( klio_config, mock_publisher, ): test_topic = "test" mock_publisher.create_topic.side_effect = api_ex.AlreadyExists("test") job = verify.VerifyJob(klio_config, True) job._publisher_client = mock_publisher actual = job._verify_pub_topic(test_topic, input) assert actual is True @pytest.mark.parametrize("not_found", (True, False)) def test_verify_pub_topic_exceptions(klio_config, mock_publisher, not_found): test_topic = "test" if not_found: mock_publisher.get_topic.side_effect = exceptions.NotFound("test") else: mock_publisher.get_topic.side_effect = Exception job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher actual = job._verify_pub_topic(test_topic, input) assert actual is False @pytest.mark.parametrize("create_resources", (True, False)) def test_verify_subscription_and_topic( klio_config, mock_publisher, mock_sub, create_resources ): test_sub = "test" upstream_topic = "Some" job = verify.VerifyJob(klio_config, create_resources) job._publisher_client = mock_publisher job._subscriber_client = mock_sub if create_resources: actual = job._verify_subscription_and_topic(test_sub, upstream_topic,) mock_sub.create_subscription.assert_called_once_with( request={"name": test_sub, "topic": upstream_topic} ) else: actual = job._verify_subscription_and_topic(test_sub, upstream_topic,) mock_sub.get_subscription.assert_called_once_with( request={"subscription": test_sub} ) expected = True, True assert expected == actual def test_verify_subscription_and_topic_exists( klio_config, mock_publisher, mock_sub ): test_sub = "test" upstream_topic = "Some" job = verify.VerifyJob(klio_config, True) job._publisher_client = mock_publisher job._subscriber_client = mock_sub mock_sub.create_subscription.side_effect = api_ex.AlreadyExists("test") actual = job._verify_subscription_and_topic(test_sub, upstream_topic) expected = True, True assert expected == actual @pytest.mark.parametrize( "not_found, no_topic", ((True, False), (False, False), (False, True)) ) def test_verify_subscription_and_topic_exceptions( klio_config, mock_publisher, mock_sub, not_found, no_topic ): test_sub = "test" if no_topic: upstream_topic = None expected = False, True else: upstream_topic = "Some" expected = True, False if not_found: mock_sub.get_subscription.side_effect = exceptions.NotFound("test") else: mock_sub.get_subscription.side_effect = Exception job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._subscriber_client = mock_sub actual = job._verify_subscription_and_topic(test_sub, upstream_topic) assert expected == actual mock_gcs_event_config = { "type": "gcs", "location": "gs://some/events.txt", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.INPUT, } mock_gcs_event = kconfig._io.KlioReadFileConfig.from_dict( mock_gcs_event_config ) mock_bq_event_config = { "type": "bq", "project": "sigint", "dataset": "test-data", "table": "test-table", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.INPUT, } mock_bq_event = kconfig._io.KlioBigQueryEventInput.from_dict( mock_bq_event_config ) mock_avro_event_config = { "type": "avro", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.INPUT, } mock_avro_event = kconfig._io.KlioReadAvroEventConfig.from_dict( mock_avro_event_config ) @pytest.mark.parametrize( "mock_event_input", ((mock_gcs_event), (mock_bq_event), (mock_avro_event),) ) def test_unverified_event_inputs( mocker, caplog, mock_storage, mock_publisher, mock_sub, klio_config, mock_event_input, ): mock_verify_gcs_bucket = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) mock_verify_sub = mocker.patch.object( verify.VerifyJob, "_verify_subscription_and_topic" ) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job._subscriber_client = mock_sub job.klio_config.pipeline_options.project = "sigint" job.klio_config.job_config.events.inputs = [mock_event_input] data_config = { "type": "gcs", "location": "test", "io_type": kconfig._io.KlioIOType.DATA, "io_direction": kconfig._io.KlioIODirection.INPUT, } job.klio_config.job_config.data.inputs = [ kconfig._io.KlioGCSInputDataConfig.from_dict(data_config) ] mock_verify_gcs_bucket.return_value = True actual = job._verify_inputs() mock_verify_sub.assert_not_called() mock_verify_gcs_bucket.assert_called_with("test") assert actual assert 3 == len(caplog.records) @pytest.mark.parametrize( "unverified_bucket, unverified_topic, unverified_sub", ( (False, False, False), (True, False, False), (True, True, True), (False, True, True), (False, False, True), (False, True, False), (False, False, False), ), ) def test_verify_inputs( mocker, unverified_bucket, unverified_topic, unverified_sub, klio_config, mock_storage, mock_publisher, mock_sub, ): mock_verify_gcs_bucket = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) mock_verify_sub = mocker.patch.object( verify.VerifyJob, "_verify_subscription_and_topic" ) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job._subscriber_client = mock_sub job.klio_config.pipeline_options.project = "sigint" event_config = { "type": "pubsub", "topic": "test", "subscription": "test", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.INPUT, } job.klio_config.job_config.events.inputs = [ kconfig._io.KlioPubSubEventInput.from_dict(event_config) ] data_config = { "type": "gcs", "location": "test", "io_type": kconfig._io.KlioIOType.DATA, "io_direction": kconfig._io.KlioIODirection.INPUT, } job.klio_config.job_config.data.inputs = [ kconfig._io.KlioGCSInputDataConfig.from_dict(data_config) ] if unverified_topic and unverified_sub and unverified_bucket: mock_verify_gcs_bucket.return_value = False mock_verify_sub.return_value = False, False actual = job._verify_inputs() expected = False assert expected == actual elif unverified_topic and unverified_sub: mock_verify_sub.return_value = False, False actual = job._verify_inputs() expected = False assert expected == actual elif unverified_topic: mock_verify_sub.return_value = False, True actual = job._verify_inputs() expected = False assert expected == actual elif unverified_sub: mock_verify_sub.return_value = True, False actual = job._verify_inputs() expected = False assert expected == actual elif unverified_bucket: mock_verify_gcs_bucket.return_value = False mock_verify_sub.return_value = True, True actual = job._verify_inputs() expected = False assert expected == actual else: mock_verify_gcs_bucket.return_value = True mock_verify_sub.return_value = True, True actual = job._verify_inputs() expected = True assert expected == actual mock_verify_gcs_bucket.assert_called_with("test") mock_verify_sub.assert_called_with("test", "test") @pytest.mark.parametrize( "data_dict, event_dict, expected_log_count", ( ({"location": "test"}, {"topic": "test", "subscription": "test"}, 3), ({"location": None}, {"topic": "test", "subscription": "test"}, 4), ({"location": "test"}, {"topic": "test", "subscription": None}, 4), ({"location": None}, {"topic": "test", "subscription": None}, 5), (None, None, 5), ), ) def test_verify_inputs_logs( data_dict, event_dict, expected_log_count, klio_config, mock_storage, mock_publisher, mock_sub, mocker, caplog, ): mocker.patch.object(verify.VerifyJob, "_verify_gcs_bucket") mock_verify_sub = mocker.patch.object( verify.VerifyJob, "_verify_subscription_and_topic" ) mock_verify_sub.return_value = (False, False) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job._subscriber_client = mock_sub job.klio_config.pipeline_options.project = "sigint" if event_dict: event_dict["type"] = "pubsub" event_dict["io_type"] = kconfig._io.KlioIOType.EVENT event_dict["io_direction"] = kconfig._io.KlioIODirection.OUTPUT job.klio_config.job_config.events.inputs = [ kconfig._io.KlioPubSubEventInput.from_dict(event_dict) ] else: job.klio_config.job_config.events.inputs = [] if data_dict: data_dict["type"] = "gcs" data_dict["io_type"] = kconfig._io.KlioIOType.DATA data_dict["io_direction"] = kconfig._io.KlioIODirection.OUTPUT job.klio_config.job_config.data.inputs = [ kconfig._io.KlioGCSOutputDataConfig.from_dict(data_dict) ] else: job.klio_config.job_config.data.inputs = [] job._verify_inputs() assert expected_log_count == len(caplog.records) mock_gcs_output_event_config = { "type": "gcs", "location": "gs://some/events.txt", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } mock_gcs_output_event = kconfig._io.KlioWriteFileConfig.from_dict( mock_gcs_event_config ) mock_bq_output_event_config = { "type": "bq", "project": "sigint", "dataset": "test-data", "table": "test-table", "schema": {"fields": [{"name": "n", "type": "t", "mode": "m"}]}, "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } mock_bq_output_event = kconfig._io.KlioBigQueryEventOutput.from_dict( mock_bq_output_event_config ) @pytest.mark.parametrize( "mock_event_output", ((mock_gcs_output_event), (mock_bq_output_event),) ) def test_unverified_event_outputs( mocker, caplog, klio_config, mock_event_output, mock_storage, mock_publisher, ): mock_verify_gcs_bucket = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) mock_verify_pub_topic = mocker.patch.object( verify.VerifyJob, "_verify_pub_topic" ) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job.klio_config.job_config.events.outputs = [mock_event_output] data_config = { "type": "gcs", "location": "test", "io_type": kconfig._io.KlioIOType.DATA, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } job.klio_config.job_config.data.outputs = [ kconfig._io.KlioGCSOutputDataConfig.from_dict(data_config) ] mock_verify_gcs_bucket.return_value = True actual = job._verify_outputs() assert actual mock_verify_pub_topic.assert_not_called() mock_verify_gcs_bucket.assert_called_with("test") assert 3 == len(caplog.records) @pytest.mark.parametrize( "unverified_gcs, unverified_topic", ((False, False), (True, False), (True, True), (False, True)), ) def test_verify_outputs( mocker, klio_config, unverified_gcs, unverified_topic, mock_storage, mock_publisher, ): mock_verify_gcs_bucket = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) mock_verify_pub_topic = mocker.patch.object( verify.VerifyJob, "_verify_pub_topic" ) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage event_config = { "type": "pubsub", "topic": "test", "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } job.klio_config.job_config.events.outputs = [ kconfig._io.KlioPubSubEventOutput.from_dict(event_config) ] data_config = { "type": "gcs", "location": "test", "io_type": kconfig._io.KlioIOType.DATA, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } job.klio_config.job_config.data.outputs = [ kconfig._io.KlioGCSOutputDataConfig.from_dict(data_config) ] if unverified_gcs and unverified_topic: mock_verify_gcs_bucket.return_value = False mock_verify_pub_topic.return_value = False actual = job._verify_outputs() expected = False assert expected == actual elif unverified_topic: mock_verify_gcs_bucket.return_value = True mock_verify_pub_topic.return_value = False actual = job._verify_outputs() expected = False assert expected == actual elif unverified_gcs: mock_verify_gcs_bucket.return_value = False mock_verify_pub_topic.return_value = True actual = job._verify_outputs() expected = False assert expected == actual else: mock_verify_gcs_bucket.return_value = True mock_verify_pub_topic.return_value = True actual = job._verify_outputs() expected = True assert expected == actual mock_verify_gcs_bucket.assert_called_with("test") mock_verify_pub_topic.assert_called_with("test", "output") @pytest.mark.parametrize( "event_topic, data_location, expected_log_count", ( ("test", "test", 3), (None, "test", 4), ("test", None, 4), (None, None, 5), ), ) def test_verify_outputs_logs( event_topic, data_location, expected_log_count, mocker, klio_config, mock_storage, mock_publisher, caplog, ): mock_verify_gcs = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) mock_verify_pub = mocker.patch.object( verify.VerifyJob, "_verify_pub_topic" ) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage event_config = { "type": "pubsub", "topic": event_topic, "io_type": kconfig._io.KlioIOType.EVENT, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } job.klio_config.job_config.events.outputs = [ kconfig._io.KlioPubSubEventOutput.from_dict(event_config) ] data_config = { "type": "gcs", "location": data_location, "io_type": kconfig._io.KlioIOType.DATA, "io_direction": kconfig._io.KlioIODirection.OUTPUT, } job.klio_config.job_config.data.outputs = [ kconfig._io.KlioGCSOutputDataConfig.from_dict(data_config) ] job._verify_outputs() assert expected_log_count == len(caplog.records) if data_location is not None: mock_verify_gcs.assert_called_with(data_location) if event_topic is not None: mock_verify_pub.assert_called_with(event_topic, "output") @pytest.mark.parametrize( "unverified_staging, unverified_temp", ((False, False), (True, True), (True, False), (False, True)), ) def test_verify_tmp_files( mocker, mock_storage, klio_config, unverified_staging, unverified_temp ): job = verify.VerifyJob(klio_config, True) job._storage_client = mock_storage job.klio_config.pipeline_options.staging_location = "test" job.klio_config.pipeline_options.temp_location = "test2" mock_verify_gcs = mocker.patch.object( verify.VerifyJob, "_verify_gcs_bucket" ) if unverified_staging and unverified_temp: mock_verify_gcs.side_effect = [False, False] actual = job._verify_tmp_files() assert actual is False elif unverified_staging: mock_verify_gcs.side_effect = [False, True] actual = job._verify_tmp_files() assert actual is False elif unverified_temp: mock_verify_gcs.side_effect = [True, False] actual = job._verify_tmp_files() assert actual is False else: mock_verify_gcs.side_effect = [True, True] actual = job._verify_tmp_files() assert actual is True mock_verify_gcs.assert_any_call("test") mock_verify_gcs.assert_any_call("test2") assert 2 == mock_verify_gcs.call_count def test_verify( mocker, klio_config, mock_storage, mock_publisher, mock_sub, caplog ): caplog.set_level(logging.INFO) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job._subscriber_client = mock_sub mock_verify_inputs = mocker.patch.object(job, "_verify_inputs") mock_verify_outputs = mocker.patch.object(job, "_verify_outputs") mock_verify_tmp = mocker.patch.object(job, "_verify_tmp_files") mock_verify_iam_roles = mocker.patch.object(job, "_verify_iam_roles") mock_verify_dashboard = mocker.patch.object( job, "_verify_stackdriver_dashboard" ) mocker.patch.object(verify, "discovery") mock_verify_config = mocker.patch.object(kconfig, "KlioConfig") mock_verify_config.return_value = klio_config mock_storage.return_value = mock_storage mock_publisher.return_value = mock_publisher mock_sub.return_value = mock_sub mock_verify_inputs.return_value = True mock_verify_outputs.return_value = True mock_verify_tmp.return_value = True mock_verify_iam_roles.return_value = True job.verify_job() mock_verify_inputs.assert_called_once_with() mock_verify_outputs.assert_called_once_with() mock_verify_tmp.assert_called_once_with() assert mock_verify_iam_roles.called mock_verify_dashboard.assert_called_once_with() assert 1 == len(caplog.records) def test_verify_raises_exception_raises_system_exit(mocker, klio_config): job = verify.VerifyJob(klio_config, False) mock_verify_all = mocker.patch.object(job, "_verify_all") mock_verify_all.return_value = False mock_verify_all.side_effect = Exception with pytest.raises(SystemExit): job.verify_job() def test_verify_raises_system_exit( mocker, klio_config, mock_storage, mock_publisher, mock_sub, caplog ): caplog.set_level(logging.INFO) job = verify.VerifyJob(klio_config, False) job._publisher_client = mock_publisher job._storage_client = mock_storage job._subscriber_client = mock_sub mock_verify_inputs = mocker.patch.object(job, "_verify_inputs") mock_verify_outputs = mocker.patch.object(job, "_verify_outputs") mock_verify_tmp = mocker.patch.object(job, "_verify_tmp_files") mock_verify_iam_roles = mocker.patch.object(job, "_verify_iam_roles") mock_verify_dashboard = mocker.patch.object( job, "_verify_stackdriver_dashboard" ) mocker.patch.object(verify, "discovery") mock_verify_config = mocker.patch.object(kconfig, "KlioConfig") mock_verify_config.return_value = klio_config mock_storage.return_value = mock_storage mock_publisher.return_value = mock_publisher mock_sub.return_value = mock_sub mock_verify_inputs.return_value = True mock_verify_outputs.return_value = False mock_verify_iam_roles.return_value = True mock_verify_tmp.return_value = True with pytest.raises(SystemExit): job.verify_job() mock_verify_inputs.assert_called_once_with() mock_verify_outputs.assert_called_once_with() mock_verify_tmp.assert_called_once_with() mock_verify_dashboard.assert_called_once_with()
gaybro8777/klio
cli/src/klio_cli/utils/docker_utils.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import logging import os import docker import emoji import requests from klio_cli.utils import multi_line_terminal_writer def check_docker_connection(docker_client): try: docker_client.ping() except (docker.errors.APIError, requests.exceptions.ConnectionError): logging.error(emoji.emojize("Could not reach Docker! :whale:")) logging.error("Is it installed and running?") raise SystemExit(1) def check_dockerfile_present(job_dir): dockerfile_path = job_dir + "/Dockerfile" if not os.path.exists(dockerfile_path): logging.error("Klio can't run job without a Dockerfile.") logging.error("Please supply \033[4m{}\033[4m".format(dockerfile_path)) raise SystemExit(1) def docker_image_exists(name, client): try: client.images.get(name) exists = True except docker.errors.ImageNotFound: exists = False except docker.errors.APIError as e: msg = ( "Docker ran into the error checking if image {}" "has already been built:\n{}".format(name, e) ) logging.error(msg) raise SystemExit(1) return exists def build_docker_image(job_dir, image_name, image_tag, config_file=None): """Build given Docker image. Note: This uses the python Docker SDK's low-level API in order to capture and emit build logs as they are generated by Docker. Using the high-level API, you only get access to logs at the end of the build, which creates a bad user experience. Args: job_dir (str): Relative path to directory containing Dockerfile. image_name (str): Name to build the image with (forms a ‘name:tag’ pair) image_tag (str): Tag to build the image with (forms a ‘name:tag’ pair) Raises: SystemExit(1) If Docker build errors out, process terminates. """ def clean_logs(log_generator): # Loop through lines containing log JSON objects. # Example line: {"stream":"Starting build..."}\r\n{"stream":"\\n"}\n for line in log_generator: if isinstance(line, bytes): line = line.decode("utf-8") # Some lines contain multiple whitespace-separated objects. # Split them so json.loads doesn't choke. for log_obj in line.split("\r\n"): # Some log objects only wrap newlines. # Split sometimes produces '' char. # Remove these artifacts. if log_obj != '{"stream":"\\n"}' and log_obj != "": yield log_obj def print_log(log): if "stream" in log: logging.info(log["stream"].strip("\n")) if "error" in log: fail_color = "\033[91m" end_color = "\033[0m" logging.info( "{}{}{}".format( fail_color, log["errorDetail"]["message"], end_color ) ) logging.error("\nDocker hit an error while building job image.") logging.error( "Please fix your Dockerfile: {}/Dockerfile".format(job_dir) ) raise SystemExit(1) build_flag = { "path": job_dir, "tag": "{}:{}".format(image_name, image_tag), "rm": True, "buildargs": { "tag": image_tag, "KLIO_CONFIG": config_file or "klio-job.yaml", }, } # Remove intermediate build containers. logs = docker.APIClient(base_url="unix://var/run/docker.sock").build( **build_flag ) for log_obj in clean_logs(logs): log = json.loads(log_obj) print_log(log) def _get_layer_id_and_message(clean_line): line_json = json.loads(clean_line) layer_id = line_json.get("id") # very first log message doesn't have an id msg_pfx = "" if layer_id: msg_pfx = "{}: ".format(layer_id) msg = "{prefix}{status}{progress}".format( prefix=msg_pfx, status=line_json.get("status", ""), progress=line_json.get("progress", ""), ) return layer_id, msg def push_image_to_gcr(image, tag, client): kwargs = {"repository": image, "tag": tag, "stream": True} writer = multi_line_terminal_writer.MultiLineTerminalWriter() for raw_line in client.images.push(**kwargs): clean_line = raw_line.decode("utf-8").strip("\r\n") clean_lines = clean_line.split("\r\n") for line in clean_lines: layer_id, msg = _get_layer_id_and_message(line) writer.emit_line(layer_id, msg.strip()) def get_docker_image_client(job_dir, image_tag, image_name, force_build): """Returns the docker image and client for running klio commands. Args: job_dir (str): Relative path to directory containing Dockerfile. image_tag (str): Tag to build the image with (forms a ‘name:tag’ pair) image_name (str): Name to build the image with (forms a ‘name:tag’ pair) force_build(bool): Flag to force a new docker image build. Raises: Valid docker image and client. """ image = "{}:{}".format(image_name, image_tag) client = docker.from_env() check_docker_connection(client) check_dockerfile_present(job_dir) if not docker_image_exists(image, client) or force_build: logging.info("Building worker image: {}".format(image)) build_docker_image(job_dir, image_name, image_tag) else: logging.info("Found worker image: {}".format(image)) return image, client
gaybro8777/klio
docs/src/_ext/collapsible_admon.py
<reponame>gaybro8777/klio # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ Custom Sphinx extension that defines a ``collapsible`` admonition. This directive will look like a normal admonition with the ability to toggle its content to show/hide. Example: .. collapsible:: A title for my admonition Some body text. It can include other directives within and will be parsed like other admonitions, e.g.: .. code-block:: python print("I should be properly syntax highlighted") """ from docutils.parsers.rst.directives import admonitions from sphinx.util import docutils class CollapsibleAdmonition(admonitions.Admonition, docutils.SphinxDirective): """Add CSS classes to `.. collapsible::` admonition.""" def run(self): if not self.options.get("class"): self.options["class"] = ["admonition-collapsible"] else: self.options["class"].append("admonition-collapsible") return super().run() def setup(app): app.add_directive("collapsible", CollapsibleAdmonition) return { "version": "1.0", "parallel_read_safe": True, "parallel_write_safe": True, }
gaybro8777/klio
examples/snippets/custom_ext_checks.py
<reponame>gaybro8777/klio # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Example of custom data existence checks implementation. Please see the full walk-through here: https://docs.klio.io/en/latest/userguide/ examples/custom_data_ext_checks.html """ import apache_beam as beam from apache_beam import pvalue from apache_beam.io.gcp import gcsio from klio.transforms import decorators class BonjourInputCheck(beam.DoFn): @decorators.set_klio_context def setup(self): self.client = gcsio.Client() @decorators.handle_klio def process(self, data): element = data.element.decode("utf-8") ic = self._klio.config.job_config.data.inputs[0] subdirs = ("subdir1", "subdir2") inputs_exists = [] for subdir in subdirs: path = f"{ic.location}/{subdir}/{element}.{ic.file_suffix}" exists = self.client.exists(path) inputs_exists.append(exists) if all(inputs_exists): yield data else: self._klio.logger.info(f"Skipping {element}: input data not found") class BonjourOutputCheck(beam.DoFn): @decorators.set_klio_context def setup(self): self.client = gcsio.Client() @decorators.handle_klio def process(self, data): element = data.element.decode("utf-8") oc = self._klio.config.job_config.data.outputs[0] subdirs = ("subdir1", "subdir2") outputs_exist = [] for subdir in subdirs: path = f"{oc.location}/{subdir}/{element}.{oc.file_suffix}" exists = self.client.exists(path) outputs_exist.append(exists) if all(outputs_exist): yield pvalue.TaggedOutput("not_found", data) else: yield pvalue.TaggedOutput("found", data) ##### # example of run.py ##### from klio.transforms import helpers def run(input_pcol, config): output_data = input_pcol | beam.ParDo(BonjourOutputCheck()).with_outputs() output_force = output_data.found | helpers.KlioFilterForce() to_input_check = ( (output_data.not_found, output_force.process) | beam.Flatten() ) to_process = to_input_check | beam.ParDo(BonjourInputCheck()) # continue on with the job-related logic output_pcol = to_process | ... return output_pcol
gaybro8777/klio
cli/src/klio_cli/commands/job/audit.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from klio_cli.commands import base class AuditPipeline(base.BaseDockerizedPipeline): DOCKER_LOGGER_NAME = "klio.job.audit" def _get_environment(self): envs = super()._get_environment() envs["KLIO_TEST_MODE"] = "true" return envs def _get_command(self, list_steps): command = ["audit"] if list_steps: command.append("--list") return command
gaybro8777/klio
examples/audio_spectrograms/transforms.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ Implement custom transforms using utilities for Klio-fication. """ import io import os import pickle import apache_beam as beam from apache_beam import pvalue import librosa import numpy as np from klio_audio import decorators as audio_decorators from klio_core.proto import klio_pb2 from klio.transforms import decorators as tfm_decorators #### # Helper funcs for handling klio & numpy de/serialization when working # with pcolls that are grouped by key #### def _unpickle_from_klio_message(item): kmsg = klio_pb2.KlioMessage() kmsg.ParseFromString(item) return pickle.loads(kmsg.data.payload) def _dump_to_klio_message(key, payload): kmsg = klio_pb2.KlioMessage() kmsg.data.element = key out = io.BytesIO() np.save(out, payload) kmsg.data.payload = out.getvalue() return kmsg.SerializeToString() ##### # Transforms ##### class GetMagnitude(beam.DoFn): """Get the magnitude of a song given its STFT.""" @tfm_decorators.handle_klio @audio_decorators.handle_binary(load_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Computing the magnitude spectrogram for {}".format(element) ) stft = item.payload spectrogram, phase = librosa.magphase(stft) yield pvalue.TaggedOutput("phase", phase) yield pvalue.TaggedOutput("spectrogram", spectrogram) class FilterNearestNeighbors(beam.DoFn): @tfm_decorators.handle_klio @audio_decorators.handle_binary def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Filtering nearest neighbors for {}".format(element) ) spectrogram = item.payload nn_filter = librosa.decompose.nn_filter( spectrogram, aggregate=np.median, metric="cosine", width=int(librosa.time_to_frames(2)), ) # The output of the filter shouldn't be greater than the input # if we assume signals are additive. Taking the pointwise minimium # with the input spectrum forces this. nn_filter = np.minimum(spectrogram, nn_filter) yield nn_filter class GetSoftMask(beam.DoFn): def __init__(self, margin=1, power=2): self.margin = margin self.power = power @tfm_decorators.set_klio_context def process(self, item): key, data = item first_data = data["first"][0] second_data = data["second"][0] full_data = data["full"][0] first = _unpickle_from_klio_message(first_data) second = _unpickle_from_klio_message(second_data) full = _unpickle_from_klio_message(full_data) self._klio.logger.debug("Getting softmask for {}".format(key)) mask = librosa.util.softmask( first, self.margin * second, power=self.power ) ret = mask * full yield _dump_to_klio_message(key, ret) def create_key_from_element(item): kmsg = klio_pb2.KlioMessage() kmsg.ParseFromString(item) return (kmsg.data.element, item) # key_pair looks like # (element, {"full": [<serialized numpy array>], # "nnfilter": [<serialized numpy array>]}) def subtract_filter_from_full(key_pair): key, pair_data = key_pair full = _unpickle_from_klio_message(pair_data["full"][0]) nn_filter = _unpickle_from_klio_message(pair_data["nnfilter"][0]) net = full - nn_filter payload = pickle.dumps(net) kmsg = klio_pb2.KlioMessage() kmsg.data.element = key kmsg.data.payload = payload return (key, kmsg.SerializeToString())
gaybro8777/klio
cli/tests/commands/job/test_delete.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_core import config as kconfig from klio_cli.commands.job import delete as delete_job @pytest.fixture def config_data(): return { "job_name": "test-job", "version": 1, "pipeline_options": {"project": "foo"}, "job_config": { "events": { "inputs": [ { "type": "pubsub", "topic": "foo-topic", "subscription": "foo-sub", } ], "outputs": [{"type": "pubsub", "topic": "foo-topic-output"}], }, "data": { "inputs": [{"type": "gcs", "location": "foo-input-location"}], "outputs": [ {"type": "gcs", "location": "foo-output-location"} ], }, }, } @pytest.fixture def klio_config(config_data): return kconfig.KlioConfig(config_data) @pytest.fixture def delete_job_inst(klio_config): return delete_job.DeleteJob(klio_config) @pytest.fixture def mock_confirm(mocker): return mocker.patch.object(delete_job.click, "confirm") @pytest.fixture def mock_prompt(mocker): return mocker.patch.object(delete_job.click, "prompt") @pytest.fixture def mock_publisher(mocker): return mocker.patch.object( delete_job.pubsub_v1, "PublisherClient", autospec=True ) @pytest.fixture def mock_subscriber(mocker): return mocker.patch.object( delete_job.pubsub_v1, "SubscriberClient", autospec=True ) @pytest.fixture def mock_storage(mocker): return mocker.patch.object(delete_job.storage, "Client", autospec=True) @pytest.mark.parametrize( "confirmation,prompt,name,expected,effect", [ ([False], [None], {}, False, None), ([False], ["foo"], "foo", False, None), ([True], ["foo"], "foo", True, None), ([True], ["pancake"], "milkshake", None, ValueError), ], ) def test_confirmation_dialog( confirmation, prompt, name, expected, effect, delete_job_inst, mock_confirm, mock_prompt, ): mock_confirm.side_effect = confirmation mock_prompt.side_effect = prompt if effect: with pytest.raises(ValueError): delete_job_inst._confirmation_dialog("meh", name) else: assert expected == delete_job_inst._confirmation_dialog("meh", name) @pytest.mark.parametrize( "confirm,expected", [ ( # do not confirm to delete an resource # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [False, False, False, False, False, False], { "topic": [], "subscription": [], "location": [], "stackdriver_group": False, }, ), ( # only confirm stackdriver # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [False, False, False, False, False, True], { "topic": [], "subscription": [], "location": [], "stackdriver_group": True, }, ), ( # gcs buckets & stackdriver # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [False, False, False, True, True, True], { "topic": [], "subscription": [], "location": ["foo-input-location", "foo-output-location"], "stackdriver_group": True, }, ), ( # output topic, gcs buckets & stackdriver # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [False, False, True, True, True, True], { "topic": ["foo-topic-output"], "subscription": [], "location": ["foo-input-location", "foo-output-location"], "stackdriver_group": True, }, ), ( # subscription, output topic, gcs buckets & stackdriver # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [False, True, True, True, True, True], { "topic": ["foo-topic-output"], "subscription": ["foo-sub"], "location": ["foo-input-location", "foo-output-location"], "stackdriver_group": True, }, ), ( # everything! # (2 topics, 1 sub, 2 gcs locations, 1 stackdriver) [True, True, True, True, True, True], { "topic": ["foo-topic", "foo-topic-output"], "subscription": ["foo-sub"], "location": ["foo-input-location", "foo-output-location"], "stackdriver_group": True, "stackdriver_group": True, }, ), ], ) def test_get_resources(confirm, expected, delete_job_inst, mocker): mock_sd = mocker.patch.object(delete_job.sd_utils, "generate_group_meta") mock_sd.return_value = (None, "fake-stackdriver-group") mock_confirmation = mocker.patch.object( delete_job_inst, "_confirmation_dialog" ) mock_confirmation.side_effect = confirm resources = delete_job_inst._get_resources() assert resources == expected @pytest.mark.parametrize("effect,record_count", [(Exception, 6), (None, 3)]) def test_delete_subscription( effect, record_count, mocker, mock_subscriber, caplog, delete_job_inst ): mock_subscriber.return_value.delete_subscription.side_effect = effect delete_job_inst._delete_subscriptions(["a", "b", "c"]) mock_subscriber.return_value.delete_subscription.assert_has_calls( [ mocker.call(request={"subscription": "a"}), mocker.call(request={"subscription": "b"}), mocker.call(request={"subscription": "c"}), ] ) assert 3 == mock_subscriber.return_value.delete_subscription.call_count assert record_count == len(caplog.records) @pytest.mark.parametrize("effect,record_count", [(Exception, 6), (None, 3)]) def test_delete_topic( effect, record_count, mocker, mock_publisher, caplog, delete_job_inst ): mock_publisher.return_value.delete_topic.side_effect = effect delete_job_inst._delete_topics(["a", "b", "c"]) mock_publisher.return_value.delete_topic.assert_has_calls( [ mocker.call(request={"topic": "a"}), mocker.call(request={"topic": "b"}), mocker.call(request={"topic": "c"}), ] ) assert 3 == mock_publisher.return_value.delete_topic.call_count assert record_count == len(caplog.records) @pytest.mark.parametrize("effect,record_count", [(Exception, 6), (None, 4)]) def test_delete_buckets( effect, record_count, mocker, mock_storage, caplog, delete_job_inst ): mock_storage.return_value.get_bucket.return_value.delete.side_effect = ( effect ) delete_job_inst._delete_buckets( "test-project", ["gs://a", "gs://b/d", "c"] ) mock_storage.assert_called_once_with("test-project") mock_storage.return_value.get_bucket.assert_has_calls( [mocker.call("a"), mocker.call("b")], any_order=True ) assert 2 == mock_storage.return_value.get_bucket.call_count mock_storage.return_value.get_bucket.return_value.delete.assert_has_calls( [mocker.call(force=True), mocker.call(force=True)] ) assert ( 2 == mock_storage.return_value.get_bucket.return_value.delete.call_count ) assert record_count == len(caplog.records) @pytest.mark.parametrize("sd_group", [True, False]) def test_delete(sd_group, mocker, mock_confirm, mock_prompt, klio_config): mock_get_resources = mocker.patch.object( delete_job.DeleteJob, "_get_resources" ) mock_get_resources.return_value = { "topic": ["a", "d"], "subscription": ["b", "e"], "location": ["c", "f"], "stackdriver_group": sd_group, } mock_delete_subscriptions = mocker.patch.object( delete_job.DeleteJob, "_delete_subscriptions" ) mock_delete_topics = mocker.patch.object( delete_job.DeleteJob, "_delete_topics" ) mock_delete_buckets = mocker.patch.object( delete_job.DeleteJob, "_delete_buckets" ) mock_delete_stackdriver_group = mocker.patch.object( delete_job.sd_utils, "delete_stackdriver_group" ) delete_job.DeleteJob(klio_config).delete() mock_delete_subscriptions.assert_called_once_with(["b", "e"]) mock_delete_topics.assert_called_once_with(["a", "d"]) mock_delete_buckets.assert_called_once_with("foo", ["c", "f"]) if sd_group: mock_delete_stackdriver_group.assert_called_once_with( "foo", "test-job", "europe-west1" ) else: mock_delete_stackdriver_group.assert_not_called()
gaybro8777/klio
integration/read-bq-write-bq/it/common.py
<filename>integration/read-bq-write-bq/it/common.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Variables and functions shared between the stages of integration testing.""" import os import yaml from klio_core import config def get_config(): """Load KlioConfig object.""" config_path = os.path.join(os.path.dirname(__file__), "..", "klio-job.yaml") try: with open(config_path) as f: cfg_dict = yaml.safe_load(f) return config.KlioConfig(cfg_dict) except IOError as e: logging.error(e) raise SystemExit(1) entity_ids = ['1', '2', '3', '4', '5']
gaybro8777/klio
exec/tests/unit/commands/utils/test_memory_utils.py
<reponame>gaybro8777/klio<filename>exec/tests/unit/commands/utils/test_memory_utils.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.utils import memory_utils @pytest.fixture def k_profiler(): return memory_utils.KMemoryLineProfiler() @pytest.fixture def mock_show_results(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(memory_utils.mp, "show_results", mock) return mock def dummy_func(x): return x + 10 def dummy_gen(x): yield x + 10 yield x + 20 def test_wrap_per_element(k_profiler, mock_show_results, mocker, monkeypatch): # don't want to mock the whole class since we need to test a class method, # just mocking out the creation of a new instance to return what we can # control mock_new = mocker.Mock(return_value=k_profiler) monkeypatch.setattr(memory_utils.KMemoryLineProfiler, "__new__", mock_new) wrapper = memory_utils.KMemoryLineProfiler.wrap_per_element(dummy_func) mock_new.assert_not_called() mock_show_results.assert_not_called() assert "dummy_func" == wrapper.__name__ result = wrapper(10) assert 20 == result mock_new.assert_called_once_with( memory_utils.KMemoryLineProfiler, backend="psutil" ) mock_show_results.assert_called_once_with(k_profiler, stream=None) mock_new.reset_mock() mock_show_results.reset_mock() wrapper = memory_utils.KMemoryLineProfiler.wrap_per_element(dummy_gen) mock_new.assert_not_called() mock_show_results.assert_not_called() assert "dummy_gen" == wrapper.__name__ result = wrapper(10) assert 20 == next(result) assert 30 == next(result) with pytest.raises(StopIteration): next(result) mock_new.assert_called_once_with( memory_utils.KMemoryLineProfiler, backend="psutil" ) mock_show_results.assert_called_once_with(k_profiler, stream=None) def test_wrap_maximum(k_profiler, mocker, monkeypatch): wrapper = memory_utils.KMemoryLineProfiler.wrap_maximum( k_profiler, dummy_func ) assert "dummy_func" == wrapper.__name__ result = wrapper(10) assert 20 == result wrapper = memory_utils.KMemoryLineProfiler.wrap_maximum( k_profiler, dummy_gen ) assert "dummy_gen" == wrapper.__name__ result = wrapper(10) assert 20 == next(result) assert 30 == next(result) with pytest.raises(StopIteration): next(result)
gaybro8777/klio
exec/tests/unit/commands/utils/test_wrappers.py
<filename>exec/tests/unit/commands/utils/test_wrappers.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands.utils import wrappers class DummyTransformGenerator(object): def process(self, x): yield x + 10 yield x + 20 class DummyTransformGeneratorRaises(object): def process(self, x): yield x + 10 yield x + 20 raise Exception("catch me") class DummyTransformFunc(object): def process(self, x): return x + 10 class DummyTransformFuncRaises(object): def process(self, *args): raise Exception("catch me") class TestProfiler(wrappers.KLineProfilerMixin): def add_function(self, *args, **kwargs): pass def enable_by_count(self, *args, **kwargs): pass def disable_by_count(self, *args, **kwargs): pass @pytest.fixture def k_profiler(): return TestProfiler() @pytest.fixture def mock_add_func(k_profiler, mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(k_profiler, "add_function", mock) return mock @pytest.fixture def mock_enable_by_count(k_profiler, mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(k_profiler, "enable_by_count", mock) return mock @pytest.fixture def mock_disable_by_count(k_profiler, mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(k_profiler, "disable_by_count", mock) return mock def dummy_func(x): return x + 10 def dummy_gen(x): yield x + 10 yield x + 20 @pytest.mark.parametrize( "transform,print_msg", ( (DummyTransformGenerator, ""), (DummyTransformGeneratorRaises, "WARN: Error caught while profiling"), ), ) def test_print_user_exceptions_generators(transform, print_msg, capsys): transform, *_ = list(wrappers.print_user_exceptions([transform])) # fails without functools.wraps(func) assert "process" == transform.process.__name__ # first argument is "self" of the process method result = transform.process(None, 10) assert 20 == next(result) assert 30 == next(result) with pytest.raises(StopIteration): next(result) captured = capsys.readouterr() assert print_msg in captured.out @pytest.mark.parametrize( "transform,print_msg,exp_ret", ( (DummyTransformFunc, "", 20), (DummyTransformFuncRaises, "WARN: Error caught while profiling", None), ), ) def test_print_user_exceptions_funcs(transform, print_msg, exp_ret, capsys): transform, *_ = list(wrappers.print_user_exceptions([transform])) # fails without functools.wraps(func) assert "process" == transform.process.__name__ # first argument is "self" of the process method result = transform.process(None, 10) assert exp_ret == result captured = capsys.readouterr() assert print_msg in captured.out def test_call( k_profiler, mock_add_func, mock_enable_by_count, mock_disable_by_count ): wrapper = k_profiler.__call__(dummy_func) mock_add_func.assert_called_once_with(dummy_func) mock_enable_by_count.assert_not_called() mock_disable_by_count.assert_not_called() assert "dummy_func" == wrapper.__name__ result = wrapper(10) assert 20 == result mock_enable_by_count.assert_called_once_with() mock_disable_by_count.assert_called_once_with() mock_add_func.reset_mock() mock_enable_by_count.reset_mock() mock_disable_by_count.reset_mock() wrapper = k_profiler.__call__(dummy_gen) mock_add_func.assert_called_once_with(dummy_gen) mock_enable_by_count.assert_not_called() mock_disable_by_count.assert_not_called() assert "dummy_gen" == wrapper.__name__ result = wrapper(10) assert 20 == next(result) assert 30 == next(result) with pytest.raises(StopIteration): next(result) mock_enable_by_count.assert_called_once_with() mock_disable_by_count.assert_called_once_with()
gaybro8777/klio
cli/tests/commands/job/utils/fixtures/expected/__init__.py
<filename>cli/tests/commands/job/utils/fixtures/expected/__init__.py """ Notice: Do not delete this file! Klio will not be able to run your job! """
gaybro8777/klio
integration/audio-spectrograms/transforms.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import io import os import pickle import apache_beam as beam from apache_beam import pvalue import librosa import numpy as np from klio_audio import decorators from klio_core.proto import klio_pb2 from klio.transforms import decorators as tfm_decorators #### # Helper funcs for handling klio & numpy de/serialization when working # with pcolls that are grouped by key # TODO: it'd prob be helpful to provide utilities for making use of # beam.(Co)GroupByKey #### def _load_from_msg(item): kmsg = klio_pb2.KlioMessage() kmsg.ParseFromString(item) return pickle.loads(kmsg.data.payload) def _dump_to_klio_message(key, payload): kmsg = klio_pb2.KlioMessage() kmsg.data.element = key out = io.BytesIO() np.save(out, payload) kmsg.data.payload = out.getvalue() return kmsg.SerializeToString() # Transforms class GetMagnitude(beam.DoFn): @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Computing the magnitude spectrogram for {}".format(element) ) stft = item.payload spectrogram, phase = librosa.magphase(stft) # yield "phase" to show multi-yields w/ tagged outputswork, but # we're only concerned about the spectrogram in our integration # test pipeline yield pvalue.TaggedOutput("phase", spectrogram) yield pvalue.TaggedOutput("spectrogram", spectrogram) class FilterNearestNeighbors(beam.DoFn): @tfm_decorators._handle_klio @decorators.handle_binary def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Filtering nearest neighbors for {}".format(element) ) spectrogram = item.payload nn_filter = librosa.decompose.nn_filter( spectrogram, aggregate=np.median, metric="cosine", width=int(librosa.time_to_frames(2)), ) # The output of the filter shouldn't be greater than the input # if we assume signals are additive. Taking the pointwise minimium # with the input spectrum forces this. nn_filter = np.minimum(spectrogram, nn_filter) yield nn_filter # TODO: this could be useful enough to make a generic "group by klio element" def create_key_from_element(item): kmsg = klio_pb2.KlioMessage() kmsg.ParseFromString(item) return (kmsg.data.element, item) def subtract_filter_from_full(key_pair): # key_pair looks like # (element, {"full": [<serialized numpy array>], "nnfilter": [<serialized numpy array>]}) key, pair_data = key_pair full = _load_from_msg(pair_data["full"][0]) nn_filter = _load_from_msg(pair_data["nnfilter"][0]) net = full - nn_filter payload = pickle.dumps(net) kmsg = klio_pb2.KlioMessage() kmsg.data.element = key kmsg.data.payload = payload return (key, kmsg.SerializeToString()) class GetSoftMask(beam.DoFn): def __init__(self, margin=1, power=2): self.margin = margin self.power = power @tfm_decorators._set_klio_context def process(self, item): key, data = item first_data = data["first"][0] second_data = data["second"][0] full_data = data["full"][0] first = _load_from_msg(first_data) second = _load_from_msg(second_data) full = _load_from_msg(full_data) self._klio.logger.debug("Getting softmask for {}".format(key)) mask = librosa.util.softmask( first, self.margin * second, power=self.power ) ret = mask * full yield _dump_to_klio_message(key, ret)
gaybro8777/klio
cli/tests/commands/job/utils/fixtures/expected/test_transforms.py
<reponame>gaybro8777/klio<filename>cli/tests/commands/job/utils/fixtures/expected/test_transforms.py """ Notice: the code below is just a simple example of how to write unit tests for a transform. Feel free to import what's needed, rewrite tests, etc. """ import pytest from klio_core.proto import klio_pb2 import transforms @pytest.fixture def klio_msg(): msg = klio_pb2.KlioMessage() msg.data.element = b"hello" msg.version = klio_pb2.Version.V2 return msg.SerializeToString() def test_process(klio_msg): """Assert process method yields expected data.""" helloklio_dofn = transforms.HelloKlio() output = helloklio_fn.process(klio_msg) assert klio_msg == list(output)[0]
gaybro8777/klio
docs/src/conf.py
<filename>docs/src/conf.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # Configuration file for the Sphinx documentation builder. # # This file only contains a selection of the most common options. For a full # list see the documentation: # https://www.sphinx-doc.org/en/master/usage/configuration.html # -- Path setup -------------------------------------------------------------- # If extensions (or modules to document with autodoc) are in another directory, # add these directories to sys.path here. If the directory is relative to the # documentation root, use os.path.abspath to make it absolute, like shown here. # import codecs import os import re import sys sys.path.append(os.path.abspath("./_ext")) # -- Helper funcs def read(*parts): """ Build an absolute path from *parts* and and return the contents of the resulting file. Assume UTF-8 encoding. """ here = os.path.abspath(os.path.dirname(__file__)) with codecs.open(os.path.join(here, *parts), "rb", "utf-8") as f: return f.read() def find_version(*file_paths): """ Build a path from *file_paths* and search for a ``__version__`` string inside. """ version_file = read(*file_paths) version_match = re.search( r"^__version__ = ['\"]([^'\"]*)['\"]", version_file, re.M ) if version_match: return version_match.group(1) raise RuntimeError("Unable to find version string.") # -- Project information ----------------------------------------------------- project = "klio" copyright = "2020, Spotify AB" author = "The klio developers" versions = { "klio_cli_release": find_version("../../cli/src/klio_cli/__init__.py"), "klio_core_release": find_version("../../core/src/klio_core/__init__.py"), "klio_devtools_release": find_version("../../devtools/src/klio_devtools/__init__.py"), "klio_exec_release": find_version("../../exec/src/klio_exec/__init__.py"), "klio_release": find_version("../../lib/src/klio/__init__.py"), "klio_audio_release": find_version("../../audio/src/klio_audio/__init__.py"), } # Define ``rst_prolog`` to make custom roles available at the beginning of each # rst file. The `/` is 'relative' to the main source directory / where conf.py # is, otherwhise sphinx will try to do relative includes rst_prolog = """ .. include:: /.custom_roles.rst """ # Define ``rst_epilog`` to make variables globally-available to compiled .rst files rst_epilog = """ .. |klio-cli-version| replace:: {klio_cli_release} .. |klio-version| replace:: {klio_release} .. |klio-audio-version| replace:: {klio_audio_release} .. |klio-exec-version| replace:: {klio_exec_release} .. |klio-core-version| replace:: {klio_core_release} .. |klio-devtools-version| replace:: {klio_devtools_release} """.format(**versions) # -- General configuration --------------------------------------------------- # Add any Sphinx extension module names here, as strings. They can be # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom # ones. extensions = [ "sphinx.ext.todo", # highlight TODO items "sphinx.ext.intersphinx", # interlink between other projects w/ sphinx docs "sphinx.ext.autodoc", # auto-generate docs from docstrings "sphinx.ext.napoleon", # handle Google-style docstrings "sphinx.ext.autosummary", # auto-gen summaries "collapsible_admon", # custom extension from _ext dir "sphinxcontrib.images", # thumbnail images "sphinxcontrib.spelling", # spell check "sphinx_click", # auto-docs for Click commands ] # Add any paths that contain templates here, relative to this directory. templates_path = ["_templates"] # List of patterns, relative to source directory, that match files and # directories to ignore when looking for source files. # This pattern also affects html_static_path and html_extra_path. exclude_patterns = [] # sphinx-build has a "nitpick" mode (used during CI docs workflow and # `make stricthtml`). We inherit some docs from Apache Beam, and some of # their docstrings fail to compile in nitpick mode. # https://www.sphinx-doc.org/en/master/usage/configuration.html#confval-nitpick_ignore nitpick_ignore = [ # <-- nitpick docstrings from beam.io.gcp.WriteToBigQuery ("py:class", "callable"), ("py:class", "ValueProvider"), ("py:class", "apache_beam.io.gcp.internal.clients.bigquery.bigquery_v2_messages.TableSchema"), ("py:class", "BigQueryDisposition"), ("py:attr", "BigQueryDisposition.CREATE_IF_NEEDED"), ("py:attr", "BigQueryDisposition.CREATE_NEVER"), ("py:attr", "BigQueryDisposition.WRITE_TRUNCATE"), ("py:attr", "BigQueryDisposition.WRITE_APPEND"), ("py:attr", "BigQueryDisposition.WRITE_EMPTY"), # --> # <-- nitpick docstrings from beam.io.textio.WriteToText ("py:class", "WriteToText"), ("py:class", "apache_beam.io.filesystem.CompressionTypes.AUTO"), # --> # <-- nitpick drom beam.io.ReadFromText ("py:class", "ReadFromText"), # --> # <-- nitpick docstrings that reference other Klio objects that are # not yet documented ("py:class", "klio_core.config.KlioConfig"), ("py:class", "klio_core.proto.klio_pb2.KlioMessage"), ("py:exc", "klio_core.proto.klio_pb2._message.DecodeError"), # --> ] # sphinx-build -b linkcheck will error out if links in docstrings are broken, # including inherited docstrings (i.e. Beam) # https://www.sphinx-doc.org/en/master/usage/configuration.html#confval-linkcheck_ignore linkcheck_ignore = [ r"https://cloud.google.com/bigquery/docs/reference/rest/v2/jobs#configuration.load", # ignore local links r"\./.+\.html", # temp ignore reference to keps r"https://github.com/spotify/klio/tree/master/docs/src/keps", r"https://docs.klio.io/en/latest/keps/", ] # -- Options for HTML output ------------------------------------------------- # The theme to use for HTML and HTML Help pages. See the documentation for # a list of builtin themes. # html_theme = "pydata_sphinx_theme" html_logo = "_static/images/logo.png" html_theme_options = { "github_url": "https://github.com/spotify/klio", "twitter_url": "https://twitter.com/SpotifyEng", "show_prev_next": False, # when there's group of thumbnails (like User Guide > Pipelines > Transforms) # don't navigate to the next page automatically; allow to key left/right # to view thumbnail carousel "navigation_with_keys": False, } # Add any paths that contain custom static files (such as style sheets) here, # relative to this directory. They are copied after the builtin static files, # so a file named "default.css" will overwrite the builtin "default.css". html_static_path = ["_static"] html_css_files = ["css/fonts.css", "css/custom.css", "css/colors.css", "css/roles.css"] # relative to _static html_js_files = ["js/custom.js"] # relative to _static html_favicon = "_static/images/favicon.png" pygments_style = "vs" # -- Extention configuration # TODO ext: https://www.sphinx-doc.org/en/master/usage/extensions/todo.html todo_include_todos = True todo_emit_warnings = False todo_link_only = False # don't render todos in production docs if os.environ.get("READTHEDOCS"): todo_include_todos = False # -- Autodoc config autodoc_default_options = { "inherited-members": False, } autodoc_mock_imports = ["librosa", "numpy", "matplotlib"] autodoc_member_order = "bysource" # -- intersphinx mapping # This will auto-generate links to Python's docs when referenced (e.g. # :func:`pdb.set_trace` will link to the set_trace docs) intersphinx_mapping = { "https://docs.python.org/3": None, "https://beam.apache.org/releases/pydoc/current": None, "https://librosa.org/doc/latest": None, "https://numpy.org/doc/stable/": None, "https://matplotlib.org/": None, "https://googleapis.dev/python/pubsub/latest/": None, } # -- sphinxcontrib.spelling config spelling_word_list_filename="spelling_wordlist.txt"
gaybro8777/klio
cli/src/klio_cli/commands/image/build.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import os import docker from klio_cli.utils import docker_utils def build(job_dir, conf, config_file, image_tag): image_name = conf.pipeline_options.worker_harness_container_image client = docker.from_env() if config_file: basename = os.path.basename(config_file) image_tag = "{}-{}".format(image_tag, basename) docker_utils.check_docker_connection(client) docker_utils.check_dockerfile_present(job_dir) docker_utils.build_docker_image( job_dir, image_name, image_tag, config_file )
gaybro8777/klio
lib/src/klio/metrics/base.py
<reponame>gaybro8777/klio # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ Base classes from which a metrics consumer (i.e. Stackdriver, ffwd, etc.) will need to implement. New consumers are required to implement the :class:`AbstractRelayClient`, and three metrics objects based off of :class:`BaseMetric`: a counter, a gauge, and a timer. """ import abc import six class _DummyAttribute(object): # for the ability to do `FOO_ATTR = abstract_attr()` as well as # decorate a property method pass def abstract_attr(obj=None): """Set an attribute or a property as abstract. Supports class-level attributes as well as methods defined as a ``@property``. Usage: .. code-block:: python class Foo(object): my_foo_attribute = abstract_attr() @property @abstract_attr def my_foo_property(self): pass Args: obj (callable): Python object to "decorate", i.e. a class method. If none is provided, a dummy object is created in order to attach the ``__isabstractattr__`` attribute (similar to ``__isabstractmethod__`` from ``abc.abstractmethod``). Returns object with ``__isabstractattr__`` attribute set to ``True``. """ if not obj: obj = _DummyAttribute() obj.__isabstractattr__ = True return obj def _has_abstract_attributes_implemented(cls, name, bases): """Verify a given class has its abstract attributes implemented.""" for base in bases: abstract_attrs = getattr(base, "_klio_metrics_abstract_attributes", []) class_attrs = getattr(cls, "_klio_metrics_all_attributes", []) for attr in abstract_attrs: if attr not in class_attrs: err_str = ( "Error instantiating class '{0}'. Implementation of " "abstract attribute '{1}' from base class '{2}' is " "required.".format(name, attr, base.__name__) ) raise NotImplementedError(err_str) def _get_all_attributes(clsdict): return [name for name, val in six.iteritems(clsdict) if not callable(val)] def _get_abstract_attributes(clsdict): return [ name for name, val in six.iteritems(clsdict) if not callable(val) and getattr(val, "__isabstractattr__", False) ] class _ABCBaseMeta(abc.ABCMeta): """Enforce behavior upon implementations of ABC classes.""" def __init__(cls, name, bases, clsdict): _has_abstract_attributes_implemented(cls, name, bases) def __new__(metaclass, name, bases, clsdict): clsdict[ "_klio_metrics_abstract_attributes" ] = _get_abstract_attributes(clsdict) clsdict["_klio_metrics_all_attributes"] = _get_all_attributes(clsdict) cls = super(_ABCBaseMeta, metaclass).__new__( metaclass, name, bases, clsdict ) return cls class AbstractRelayClient(six.with_metaclass(_ABCBaseMeta)): """Abstract base class for all metric consumer relay clients. Each new consumer (i.e. Stackdriver, ffwd, logging-based metrics) will need to implement this relay class. Attributes: RELAY_CLIENT_NAME (str): must match the key in ``klio-job.yaml`` under ``job_config.metrics``. """ RELAY_CLIENT_NAME = abstract_attr() def __init__(self, klio_config): self.klio_config = klio_config @abc.abstractmethod def unmarshal(self, metric): """Returns a dictionary-representation of the ``metric`` object""" pass @abc.abstractmethod def emit(self, metric): """Emit the given metric object to the particular consumer. ``emit`` will be run in a threadpool separate from the transform, and any errors raised from the method will be logged then ignored. """ pass @abc.abstractmethod def counter(self, name, value=0, transform=None, **kwargs): """Return a newly instantiated counter-type metric specific for the particular consumer. Callers to the ``counter`` method will store new counter objects returned in memory for simple caching. """ pass @abc.abstractmethod def gauge(self, name, value=0, transform=None, **kwargs): """Return a newly instantiated gauge-type metric specific for the particular consumer. Callers to the ``gauge`` method will store new gauge objects returned in memory for simple caching. """ pass @abc.abstractmethod def timer(self, name, transform=None, **kwargs): """Return a newly instantiated timer-type metric specific for the particular consumer. Callers to the ``timer`` method will store new timer objects returned in memory for simple caching. """ pass class BaseMetric(object): """Base class for all metric types. A consumer must implement a counter metric, a gauge metric, and a timer metric. """ def __init__(self, name, value=0, transform=None, **kwargs): self.name = name self.value = value self.transform = transform def update(self, value): self.value = value
gaybro8777/klio
integration/read-bq-write-bq/it/after.py
<reponame>gaybro8777/klio<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Cleanup after each integration test. Deletes both input and output tables. """ import os from apache_beam.io.gcp import bigquery_tools as beam_bq_tools import common def delete_tables(): klio_cfg = common.get_config() input_table_cfg = klio_cfg.job_config.events.inputs[0] output_table_cfg = klio_cfg.job_config.events.outputs[0] bq_client = beam_bq_tools.BigQueryWrapper() bq_client._delete_table(input_table_cfg.project, input_table_cfg.dataset, input_table_cfg.table) bq_client._delete_table(output_table_cfg.project, output_table_cfg.dataset, output_table_cfg.table) def restore_original_config(): klio_cfg_file_path = os.path.join(os.path.dirname(__file__), "..", "klio-job.yaml") klio_save_file_path = os.path.join(os.path.dirname(__file__), "..", "klio-job.yaml.save") os.rename(klio_save_file_path, klio_cfg_file_path) if __name__ == '__main__': delete_tables() restore_original_config()
gaybro8777/klio
cli/src/klio_cli/utils/cli_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import subprocess def get_git_sha(cwd=None, image_tag=None): cmd = "git describe --match=NeVeRmAtCh --always --abbrev=8 --dirty" try: return ( subprocess.check_output( # pipe to devnull to suppress the error msgs from git itself cmd.split(), cwd=cwd, stderr=subprocess.DEVNULL, ) .decode() .strip() ) except subprocess.CalledProcessError: if not image_tag: logging.error( "The directory from which you are running this is not a git " "directory, or has no commits yet. The latest commit is used " "to tag the Docker image that is built by this command. " "Consider overriding this value using the --image-tag flag " "until such a time as commits are available." ) raise SystemExit(1) # TODO: Move this to KlioConfig validation # once overriding & templates are done def validate_dataflow_runner_config(klio_config): pipeline_opts = klio_config.pipeline_options.as_dict() mandatory_gcp_keys = [ "project", "staging_location", "temp_location", "region", ] is_gcp = all( pipeline_opts.get(key) is not None for key in mandatory_gcp_keys ) if not is_gcp: logging.error( "Unable to verify the mandatory configuration fields for" " DataflowRunner. Please fix job configuration or run via direct" "runner." ) raise SystemExit(1) def is_direct_runner(klio_config, direct_runner): if not direct_runner: validate_dataflow_runner_config(klio_config) return direct_runner
gaybro8777/klio
cli/tests/utils/test_multi_line_terminal_writer.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from klio_cli.utils import multi_line_terminal_writer def test_simple_output(capsys): writer = multi_line_terminal_writer.MultiLineTerminalWriter() writer.emit_line("line one", "initial contents of line one") captured = capsys.readouterr() assert "initial contents of line one\n" == captured.out def test_multiline_output(capsys): writer = multi_line_terminal_writer.MultiLineTerminalWriter() writer.emit_line("line one", "initial contents of line one") writer.emit_line("line two", "initial contents of line two") writer.emit_line("line one", "updated line one") captured = capsys.readouterr() expected_output = ( "initial contents of line one\n" # Write the first line "initial contents of line two\n" # Write the second line "\x1b[2F" # Move up two lines "updated line one\u001b[0K\n" # Overwrite the first line "\x1b[1E" # Move back down one line (as we already wrote a newline) ) assert expected_output == captured.out
gaybro8777/klio
cli/tests/utils/test_cli_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import subprocess import pytest from klio_cli.utils import cli_utils @pytest.mark.parametrize("job_dir", (None, "/foo/bar/klio-job.yaml")) def test_get_git_sha(mocker, job_dir): check_output_mock = mocker.patch.object(subprocess, "check_output") check_output_mock.return_value = b"" result = cli_utils.get_git_sha(job_dir) cmd = "git describe --match=NeVeRmAtCh --always --abbrev=8 --dirty" check_output_mock.assert_called_once_with( cmd.split(), cwd=job_dir, stderr=subprocess.DEVNULL ) assert "" == result def test_get_git_sha_called_process_error(mocker): check_output_mock = mocker.patch.object(subprocess, "check_output") check_output_mock.side_effect = subprocess.CalledProcessError(1, "") with pytest.raises(SystemExit): cli_utils.get_git_sha() @pytest.mark.parametrize( "pipeline_options,will_raise", [ ({}, True), ( { # missing one key "project": "p", "staging_location": "s", "region": "r", }, True, ), ( { # has all keys "project": "p", "staging_location": "s", "temp_location": "t", "region": "r", }, False, ), ], ) def test_validate_dataflow_runner_config( mocker, caplog, pipeline_options, will_raise ): mock_klio_cfg = mocker.Mock() mock_klio_cfg.pipeline_options.as_dict.return_value = pipeline_options if will_raise: with pytest.raises(SystemExit): cli_utils.validate_dataflow_runner_config(mock_klio_cfg) assert 1 == len(caplog.records) assert "ERROR" == caplog.records[0].levelname else: cli_utils.validate_dataflow_runner_config(mock_klio_cfg) @pytest.mark.parametrize("direct_runner", [False, True]) def test_is_direct_runner(mocker, monkeypatch, direct_runner): mock_klio_cfg = mocker.Mock() mock_validate_df_config = mocker.Mock() monkeypatch.setattr( cli_utils, "validate_dataflow_runner_config", mock_validate_df_config ) if not direct_runner: assert mock_validate_df_config.called_once_with(mock_klio_cfg) assert ( cli_utils.is_direct_runner(mock_klio_cfg, direct_runner) == direct_runner )
gaybro8777/klio
exec/src/klio_exec/commands/audit_steps/base.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import traceback def _get_relevant_frames(tb): """Remove up to and including pytest-related frames from tb. Args: list(traceback.FrameSummary): complete traceback of code executed by pytest Returns: list(traceback.FrameSummary) of traceback frames not including the invocation of pytest """ last_index = 0 for i, frame in enumerate(reversed(tb)): if "pytest" in str(frame): last_index = len(tb) - i break return tb[last_index:] class BaseKlioAuditStep(object): """ A base class to be subclassed when creating new audit steps for auditing Klio jobs. Subclasses have access to the following properties: - self.job_dir - self.klio_config All other inspection or modification of the Python environment should happen using global state, as pytest itself is run in between before_tests and after_tests. """ AUDIT_STEP_NAME = None def __init__(self, job_dir, klio_config, tw): self.klio_config = klio_config self.job_dir = job_dir self.tw = tw self.errored = False self.warned = False @staticmethod def get_description(): """ User-facing description of audit step shown in `klio job audit --list`. If not implemented, the audit step class's docstring is used. If there is no class-level docstring, description defaults to "No description". """ pass def _emit(self, message, tb=None, **kw): msg = "[{}]: {}\n".format(self.AUDIT_STEP_NAME, message) if tb: tb = _get_relevant_frames(tb) tb_fmtd = "\n".join(traceback.format_list(tb)) msg = "{}{}\n".format(msg, tb_fmtd) self.tw.write(msg, **kw) def emit_warning(self, warning, tb=None, **kw): """Emit audit warning and set `self.warned` to True. Excludes any pytest related stacktraces in tb. Args: warning(str): Warning to emit. Prefixed to tb. tb(traceback object): Traceback to emit. kw(dict): Any attributes to pass onto `TerminalWriter.write`. """ self.warned = True kwargs = {"yellow": True} kwargs.update(kw) self._emit(warning, tb=tb, **kwargs) def emit_error(self, error, tb=None, **kw): """Emit audit error and set `self.errored` to True. Excludes any pytest related stacktraces in tb. Args: warning(str): Warning to emit. Prefixed to tb. tb(traceback object): Traceback to emit. kw(dict): Any attributes to pass onto `TerminalWriter.write`. """ self.errored = True kwargs = {"red": True} kwargs.update(kw) self._emit(error, tb=tb, **kwargs) def before_tests(self): """Any setup needed for test.""" pass def after_tests(self): """Evaluate the results of auditing and emit results. Users can use the `emit_warning` and `emit_error` methods here to simplify reporting results. Implementation required. """ raise NotImplementedError("Subclasses must implement after_tests!")
gaybro8777/klio
audio/src/klio_audio/transforms/audio.py
<reponame>gaybro8777/klio # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import librosa import librosa.display import numpy as np from klio.transforms import decorators as tfm_decorators from klio_audio import decorators from klio_audio.transforms import _base class LoadAudio(_base.KlioAudioBaseDoFn): """Load audio into memory as a :class:`numpy.ndarray`. This transform wraps :func:`librosa.load` takes in a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` with the payload of the ``KlioMessage`` a file-like object or a path to a file, and returns a ``PCollection`` of ``KlioMessages`` where the payload is a :class:`numpy.ndarray`. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() # other transforms ) Args: librosa_kwargs (dict): Instantiate the transform with keyword arguments to pass into :func:`librosa.load`. """ def __init__(self, *_, **librosa_kwargs): self.librosa_kwargs = librosa_kwargs @tfm_decorators._handle_klio @decorators.handle_binary(save_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Loading {} into memory as a numpy array.".format(element) ) audio, _ = librosa.load(item.payload, **self.librosa_kwargs) yield audio class GetSTFT(_base.KlioAudioBaseDoFn): """Calculate Short-time Fourier transform from a :class:`numpy.ndarray`. This transform wraps :func:`librosa.stft` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` and the output is the same with the ``stft`` calculation applied. The Short-time Fourier transform (STFT) is a Fourier-related transform used to determine the sinusoidal frequency and phase content of local sections of a signal as it changes over time. STFT provides the time-localized frequency information for situations in which frequency components of a signal vary over time, whereas the standard Fourier transform provides the frequency information averaged over the entire signal time interval. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetSTFT # other transforms ) Args: librosa_kwargs (dict): Instantiate the transform with keyword arguments to pass into :func:`librosa.stft`. """ def __init__(self, *_, **librosa_kwargs): self.librosa_kwargs = librosa_kwargs @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True, save_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Calculating the short-time Fourier transform for {}".format( element ) ) yield librosa.stft(y=item.payload, **self.librosa_kwargs) class GetSpec(_base.KlioAudioBaseDoFn): """Generate a dB-scaled spectrogram from a :class:`numpy.ndarray`. This transform wraps :func:`librosa.amplitude_to_db` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` and the output is the same with the ``amplitude_to_ db`` function applied. A spectrogram shows the the intensity of frequencies over time. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetSpec() # other transforms ) Args: librosa_kwargs (dict): Instantiate the transform with keyword arguments to pass into :func:`librosa.amplitude_to_db`. """ def __init__(self, *_, **librosa_kwargs): self.librosa_kwargs = librosa_kwargs @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True, save_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Generating a spectrogram for {}".format(element) ) stft = item.payload yield librosa.amplitude_to_db( np.abs(stft), ref=np.max(np.abs(stft)), **self.librosa_kwargs ) class GetMelSpec(_base.KlioAudioBaseDoFn): """Generate a spectrogram from a :class:`numpy.ndarray` using the mel scale. This transform wraps :func:`librosa.feature.melspectrogram` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` and the output is the same with the ``melspectrogram`` function applied. The mel scale is a non-linear transformation of frequency scale based on the perception of pitches. The mel scale is calculated so that two pairs of frequencies separated by a delta in the mel scale are perceived by humans as being equidistant. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetMelSpec() # other transforms ) Args: librosa_kwargs (dict): Instantiate the transform with keyword arguments to pass into :func:`librosa.feature.melspectrogram`. """ def __init__(self, *_, **librosa_kwargs): self.librosa_kwargs = librosa_kwargs @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True, save_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Generating a Mel spectrogram for {}".format(element) ) yield librosa.feature.melspectrogram( y=item.payload, **self.librosa_kwargs ) class GetMFCC(_base.KlioAudioBaseDoFn): """Calculate MFCCs from a :class:`numpy.ndarray`. This transform wraps :func:`librosa.power_to_db` followed by :func:`librosa.feature.mfcc` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` and the output is the same with the ``mfcc`` function applied. The Mel frequency cepstral coefficients (MFCCs) of a signal are a small set of features (usually about 10–20) which describe the overall shape of a spectral envelope. It's is often used to describe timbre or model characteristics of human voice. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetMFCC() # other transforms ) Args: librosa_kwargs (dict): Instantiate the transform with keyword arguments to pass into :func:`librosa.feature.mfcc`. """ def __init__(self, *_, **librosa_kwargs): self.librosa_kwargs = librosa_kwargs @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True, save_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") self._klio.logger.debug( "Generating Mel frequency cepstral coefficients for {}".format( element ) ) # melspectrogram by default returns a power'ed (**2) spectrogram # so we need to convert to decibel units (if it wasn't a power'ed # spec, then we'd use amplitude_to_db) Sdb = librosa.power_to_db(item.payload, ref=np.max) yield librosa.feature.mfcc(S=Sdb, **self.librosa_kwargs) class SpecToPlot(_base.KlioPlotBaseDoFn): """Generate a matplotlib figure of the spectrogram of a :class:`numpy.ndarray`. This transform wraps :func:`librosa.display.specshow` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` of a spectrogram and the output is a :class:`matplotlib.figure.Figure` instance. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetSpec() | audio.SpecToPlot() # other transforms ) Args: title (str): Title of spectrogram plot. Default: ``Spectrogram of {KlioMessage.data.element}``. plot_args (dict): keyword arguments to pass to :func:`librosa.display.specshow`. """ DEFAULT_TITLE = "Spectrogram of {element}" def __init__(self, *_, title=None, **plot_args): super(SpecToPlot, self).__init__(self, title=title, **plot_args) self.plot_args["x_axis"] = self.plot_args.get("x_axis", "time") self.plot_args["y_axis"] = self.plot_args.get("y_axis", "linear") def _plot(self, item, fig): librosa.display.specshow(item.payload, ax=fig.gca(), **self.plot_args) class MelSpecToPlot(_base.KlioPlotBaseDoFn): """Generate a matplotlib figure of the mel spectrogram of a a :class:`numpy.ndarray`. This transform wraps :func:`librosa.power_to_db` followed by :func:`librosa.display.specshow` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` of a melspectrogram and the output is a :class:`matplotlib.figure.Figure` instance. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetMelSpec() | audio.SpecToPlot() # other transforms ) Args: title (str): Title of spectrogram plot. Default: ``Mel-freqency Spectrogram of {KlioMessage.data.element}``. plot_args (dict): keyword arguments to pass to :func:`librosa.display.specshow`. """ DEFAULT_TITLE = "Mel-frequency Spectrogram of {element}" def __init__(self, *_, title=None, **plot_args): super(MelSpecToPlot, self).__init__(self, title=title, **plot_args) self.plot_args["y_axis"] = "mel" self.plot_args["x_axis"] = self.plot_args.get("x_axis", "time") self.plot_args["fmax"] = self.plot_args.get("fmax", 8000) def _plot(self, item, fig): Sdb = librosa.power_to_db(item.payload, ref=np.max) librosa.display.specshow(Sdb, ax=fig.gca(), **self.plot_args) class MFCCToPlot(_base.KlioPlotBaseDoFn): """Generate a matplotlib figure of the MFCCs as a :class:`numpy.ndarray`. This transform wraps :func:`librosa.display.specshow` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` of the MFCCs of an audio and the output is a :class:`matplotlib.figure.Figure` instance. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.GetMFCC() | audio.MFCCToPlot() # other transforms ) Args: title (str): Title of spectrogram plot. Default: ``MFCCs of {KlioMessage.data.element}``. plot_args (dict): keyword arguments to pass to :func:`librosa.display.specshow`. """ DEFAULT_TITLE = "MFCCs of {element}" def __init__(self, *_, title=None, **plot_args): super(MFCCToPlot, self).__init__(self, title=title, **plot_args) self.plot_args["x_axis"] = self.plot_args.get("x_axis", "time") def _plot(self, item, fig): librosa.display.specshow(item.payload, ax=fig.gca(), **self.plot_args) class WaveformToPlot(_base.KlioAudioBaseDoFn): """Generate a matplotlib figure of the wave form of a :class:`numpy.ndarray`. This transform wraps :func:`librosa.display.waveplot` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`numpy.ndarray` of a loaded audio file the output is a :class:`matplotlib.figure.Figure` instance. Example: .. code-block:: python # run.py import apache_beam as beam from klio.transforms import decorators from klio_audio.transforms import audio @decorators.handle_klio def element_to_filename(ctx, data): filename = data.element.decode("utf-8") return f"file:///path/to/audio/{filename}.wav" def run(in_pcol, job_config): return ( in_pcol | beam.Map(element_to_filename) | audio.LoadAudio() | audio.WaveformToPlot() # other transforms ) Args: num_samples (int): Number of samples to plot. Default: ``5000``. title (str): Title of spectrogram plot. Default: ``Waveplot of {KlioMessage.data.element}``. plot_args (dict): keyword arguments to pass to :func:`librosa.display.waveplot`. """ DEFAULT_TITLE = "Waveplot of {element}" def __init__(self, *_, num_samples=5000, title=None, **plot_args): super(WaveformToPlot, self).__init__(self, title=title, **plot_args) self.num_samples = num_samples def _plot(self, item, fig): librosa.display.waveplot( item.payload[: self.num_samples], ax=fig.gca(), **self.plot_args )
gaybro8777/klio
exec/src/klio_exec/commands/audit.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import time import pytest from py import io from klio_exec.commands.utils import plugin_utils PLUGIN_NAMESPACE = "klio.plugins.audit" def _run_pytest(tw): tw.write("Running tests for audit validation...\n") # Test-style verification exit_code = pytest.main(["-qq"]) pytest_failed = exit_code != 0 if pytest_failed: tw.write("PyTest failed!\n", yellow=True) else: tw.write("PyTest passed!\n") return pytest_failed def _get_audit_steps(job_dir, config_obj, tw): loaded_steps = plugin_utils.load_plugins_by_namespace(PLUGIN_NAMESPACE) return [step(job_dir, config_obj, tw) for step in loaded_steps] def list_audit_steps(ctx, param, value): # this is a click.option callback where it automatically passes in # context, param name, and param value if not value or ctx.resilient_parsing: return tw = io.TerminalWriter() tw.hasmarkup = True tw.sep("=", "Installed audit steps") plugin_utils.print_plugins(PLUGIN_NAMESPACE, tw) ctx.exit() def audit(job_dir, config_obj): tw = io.TerminalWriter() tw.hasmarkup = True tw.write("Auditing your Klio job...\n") try: audit_steps = _get_audit_steps(job_dir, config_obj, tw) for step in audit_steps: step.before_tests() pytest_failed = _run_pytest(tw) tw.sep("=", "audit session starts", bold=True) start = time.time() # TODO: Currently, we're running all of the audit steps in # the same Python process and there's no isolation. Do we want # to run the tests once per step instead? for step in audit_steps: step.after_tests() if not any([step.errored, step.warned]): msg = "[{}]: audit step passed!\n".format(step.AUDIT_STEP_NAME) tw.write(msg, green=True) end = time.time() failed_steps = [step for step in audit_steps if step.errored] warned_steps = [step for step in audit_steps if step.warned] error_count = len(failed_steps) warning_count = len(warned_steps) finished_msg = "{0} errors, {1} warnings in {2:.3f} seconds".format( error_count, warning_count, end - start ) color = "green" if error_count: color = "red" elif warning_count: color = "yellow" kwargs = { color: True, "bold": True, } tw.sep("=", finished_msg, **kwargs) except Exception: # use logging instead of tw to easily get traceback info logging.error( "Unable to run the audit command due to:\n", exc_info=True ) raise SystemExit(1) if not pytest_failed and not error_count: if not warning_count: tw.write("Good job! Your job does not exhibit any known issues.\n") else: tw.write( "Cool! Your job has warnings, but no errors. " "Please check the warnings.\n", yellow=True, ) else: tw.sep( "You have errors in your job. Please fix and try again.\n", red=True, ) raise SystemExit(1)
gaybro8777/klio
exec/tests/unit/commands/test_stop.py
<reponame>gaybro8777/klio # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from googleapiclient import errors as gerrors from klio_exec.commands import stop @pytest.fixture def mock_discovery_client(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(stop.discovery, "build", lambda x, y: mock) return mock @pytest.fixture def config(mocker): pipeline_options = mocker.Mock( project="test-project", region="europe-west1" ) return mocker.Mock(job_name="test-job", pipeline_options=pipeline_options) @pytest.fixture def jobs_response(): return { "jobs": [ {"name": "not-the-test-job"}, { "id": "1234", "name": "test-job", "projectId": "test-project", "location": "europe-west1", }, ] } @pytest.fixture def job(jobs_response): return jobs_response["jobs"][1] @pytest.fixture def mock_sleep(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(stop.time, "sleep", mock) return mock @pytest.mark.parametrize("api_version", (None, "v1b3", "v2")) def test_set_dataflow_client(mock_discovery_client, api_version): assert stop._client is None stop._set_dataflow_client(api_version) assert stop._client is not None assert mock_discovery_client == stop._client # cleanup setattr(stop, "_client", None) # return the desired job, no jobs at all, or no jobs matching job name @pytest.mark.parametrize("returns_jobs", (True, False, None)) def test_check_job_running( mock_discovery_client, returns_jobs, jobs_response, config, monkeypatch ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.list req.return_value.execute.return_value = {} if returns_jobs: req.return_value.execute.return_value = jobs_response elif returns_jobs is False: req.return_value.execute.return_value = { "jobs": [{"name": "not-the-test-job"}] } ret = stop._check_job_running(config) if returns_jobs: assert jobs_response["jobs"][1] == ret else: assert ret is None req.assert_called_once_with( projectId="test-project", location="europe-west1", filter="ACTIVE" ) req.return_value.execute.assert_called_once_with() def test_check_job_running_errors( mock_discovery_client, config, monkeypatch, caplog ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.list req.return_value.execute.side_effect = Exception("foo") stop._check_job_running(config) req.assert_called_once_with( projectId="test-project", location="europe-west1", filter="ACTIVE" ) req.return_value.execute.assert_called_once_with() assert 2 == len(caplog.records) @pytest.mark.parametrize( "state,pyver", (("drain", None), ("cancel", None), (None, 2), (None, 3)) ) def test_update_job_state( state, pyver, mock_discovery_client, job, monkeypatch ): monkeypatch.setattr(stop, "_client", mock_discovery_client) exp_state = state if not state: if pyver == 2: exp_state = "JOB_STATE_DRAINED" else: exp_state = "JOB_STATE_CANCELLED" monkeypatch.setitem(stop.JOB_STATE_MAP, "default", exp_state) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update req.return_value.execute.return_value = None stop._update_job_state(job, state) job["requestedState"] = exp_state req.assert_called_once_with( jobId="1234", projectId="test-project", location="europe-west1", body=job, ) req.return_value.execute.assert_called_once_with() def test_update_job_state_400_error( mock_discovery_client, job, mock_sleep, mocker, monkeypatch, caplog ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update mock_resp = mocker.Mock(status=400) req.return_value.execute.side_effect = gerrors.HttpError(mock_resp, b"foo") with pytest.raises(SystemExit): stop._update_job_state(job, "drain") assert 1 == req.return_value.execute.call_count assert 1 == len(caplog.records) assert not mock_sleep.call_count def test_update_job_state_500_error( mock_discovery_client, job, mock_sleep, mocker, monkeypatch, caplog ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update mock_resp = mocker.Mock(status=500) req.return_value.execute.side_effect = gerrors.HttpError(mock_resp, b"foo") with pytest.raises(SystemExit): stop._update_job_state(job, "drain") assert 4 == req.return_value.execute.call_count assert 4 == len(caplog.records) assert 3 == mock_sleep.call_count def test_update_job_state_error( mock_discovery_client, job, mock_sleep, monkeypatch, caplog ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update req.return_value.execute.side_effect = Exception("foo") with pytest.raises(SystemExit): stop._update_job_state(job, "cancel") assert 4 == req.return_value.execute.call_count assert 4 == len(caplog.records) assert 3 == mock_sleep.call_count @pytest.mark.parametrize( "exec_side_effect", ( ( {"currentState": "JOB_STATE_CANCELLING"}, {"currentState": "JOB_STATE_CANCELLED"}, ), (Exception("foo"), {"currentState": "JOB_STATE_CANCELLED"}), ), ) def test_watch_job_state( mock_discovery_client, mock_sleep, monkeypatch, caplog, job, exec_side_effect, ): monkeypatch.setattr(stop, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.get req.return_value.execute.side_effect = exec_side_effect stop._watch_job_state(job) assert 2 == req.return_value.execute.call_count mock_sleep.assert_called_once_with(5) assert 1 == len(caplog.records) def test_watch_job_state_raises( mock_discovery_client, monkeypatch, caplog, job ): monkeypatch.setattr(stop, "_client", mock_discovery_client) with pytest.raises(SystemExit): stop._watch_job_state(job, timeout=0) assert 1 == len(caplog.records) @pytest.mark.parametrize("has_running_job", (True, False)) def test_stop(has_running_job, config, mocker, monkeypatch, job): mock_set_dataflow_client = mocker.Mock() monkeypatch.setattr(stop, "_set_dataflow_client", mock_set_dataflow_client) ret_val = None if has_running_job: ret_val = job mock_check_job_running = mocker.Mock(return_value=ret_val) monkeypatch.setattr(stop, "_check_job_running", mock_check_job_running) mock_update_job_state = mocker.Mock() monkeypatch.setattr(stop, "_update_job_state", mock_update_job_state) mock_watch_job_state = mocker.Mock() monkeypatch.setattr(stop, "_watch_job_state", mock_watch_job_state) stop.stop(config, "cancel") mock_set_dataflow_client.assert_called_once_with() mock_check_job_running.assert_called_once_with(config) if has_running_job: mock_update_job_state.assert_called_once_with(job, req_state="cancel") mock_watch_job_state.assert_called_once_with(job) else: mock_update_job_state.assert_not_called() mock_watch_job_state.assert_not_called()
gaybro8777/klio
cli/tests/commands/message/test_publish.py
<gh_stars>100-1000 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import functools import pytest from google.api_core import exceptions as gapi_exceptions from klio_core import config from klio_core.proto.v1beta1 import klio_pb2 from klio_cli.commands.message import publish @pytest.fixture def mock_publisher(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(publish.pubsub, "PublisherClient", mock) return mock @pytest.fixture def klio_job_config(): conf = { "job_name": "test-job", "version": 1, "pipeline_options": {"project": "test-gcp-project"}, "job_config": { "events": { "inputs": [ { "type": "pubsub", "topic": "an-input-topic", "subscription": "a-subscription", } ], "outputs": [{"type": "pubsub", "topic": "foo-topic-output"}], }, "data": { "inputs": [ {"type": "gcs", "location": "gs://a-test-input/location"} ], "outputs": [ {"type": "gcs", "location": "foo-output-location"} ], }, }, } return config.KlioConfig(conf) @pytest.fixture def expected_klio_job(klio_job_config): klio_job = klio_pb2.KlioJob() klio_job.job_name = "test-job" klio_job.gcp_project = "test-gcp-project" return klio_job @pytest.fixture def expected_klio_message(expected_klio_job): message = klio_pb2.KlioMessage() message.metadata.downstream.extend([expected_klio_job]) return message def test_create_publisher(mock_publisher): client = mock_publisher.return_value ret_publisher = publish._create_publisher("a-topic") mock_publisher.assert_called_once_with() client.get_topic.assert_called_once_with(request={"topic": "a-topic"}) expected = functools.partial(client.publish, topic="a-topic") assert expected.func == ret_publisher.func def test_create_publisher_topic_not_found(mock_publisher): client = mock_publisher.return_value client.get_topic.side_effect = gapi_exceptions.NotFound("foo") with pytest.raises(SystemExit): publish._create_publisher("a-topic") mock_publisher.assert_called_once_with() client.get_topic.assert_called_once_with(request={"topic": "a-topic"}) def test_create_publisher_raises(mock_publisher): client = mock_publisher.return_value client.get_topic.side_effect = Exception("foo") with pytest.raises(Exception, match="foo"): publish._create_publisher("a-topic") mock_publisher.assert_called_once_with() client.get_topic.assert_called_once_with(request={"topic": "a-topic"}) def test_get_current_klio_job(klio_job_config, expected_klio_job): ret_job = publish._get_current_klio_job(klio_job_config) assert expected_klio_job == ret_job @pytest.mark.parametrize( "force,ping,top_down,version", ( (True, True, True, 1), (True, False, False, 1), (False, True, False, 1), (False, False, False, 1), (False, False, False, 2), ), ) def test_create_pubsub_message( force, ping, top_down, version, expected_klio_job ): entity_id = "s0m3-ent1ty-1D" expected_klio_message = klio_pb2.KlioMessage() expected_klio_message.metadata.force = force expected_klio_message.metadata.ping = ping expected_klio_message.version = version if version == 1: expected_klio_message.data.entity_id = entity_id if not top_down: expected_klio_message.metadata.downstream.extend( [expected_klio_job] ) else: expected_klio_message.data.element = bytes(entity_id, "utf-8") if not top_down: rec = expected_klio_message.metadata.intended_recipients rec.limited.recipients.extend([expected_klio_job]) ret_msg = publish._create_pubsub_message( entity_id, expected_klio_job, force, ping, top_down, version ) assert expected_klio_message.SerializeToString() == ret_msg @pytest.mark.parametrize( "force,ping,top_down,non_klio", ( (True, True, True, False), (True, False, False, False), (False, True, False, False), (False, False, False, False), (False, False, True, True), ), ) def test_private_publish_messages( klio_job_config, mock_publisher, expected_klio_job, force, ping, top_down, non_klio, ): entity_id = "s0m3-ent1ty-1D" msg_version = 1 if non_klio: exp_data = bytes(entity_id.encode("utf-8")) else: expected_klio_message = klio_pb2.KlioMessage() expected_klio_message.metadata.force = force expected_klio_message.metadata.ping = ping expected_klio_message.data.entity_id = entity_id expected_klio_message.version = msg_version if not top_down: expected_klio_message.metadata.downstream.extend( [expected_klio_job] ) exp_data = expected_klio_message.SerializeToString() ret_success, ret_fail = publish._publish_messages( klio_job_config, [entity_id], force, ping, top_down, non_klio, msg_version, ) mock_publisher.return_value.publish.assert_called_once_with( topic="an-input-topic", data=exp_data ) assert 1 == len(ret_success) assert not len(ret_fail) def test_private_publish_messages_raises( mock_publisher, klio_job_config, caplog ): client = mock_publisher.return_value client.publish.side_effect = Exception("foo") ret_success, ret_fail = publish._publish_messages( klio_job_config, ["s0m3-ent1ty-1D"], True, False, False, False, 1 ) assert not len(ret_success) assert 1 == len(ret_fail) assert 1 == len(caplog.records) def test_publish_messages( klio_job_config, expected_klio_message, mock_publisher, caplog ): entity_id = "s0m3-ent1ty-1D" msg_version = 1 expected_klio_message.metadata.force = False expected_klio_message.metadata.ping = False expected_klio_message.data.entity_id = entity_id expected_klio_message.version = msg_version publish.publish_messages(klio_job_config, [entity_id]) mock_publisher.return_value.publish.assert_called_once_with( data=expected_klio_message.SerializeToString(), topic="an-input-topic" ) assert 2 == len(caplog.records) assert "INFO" == caplog.records[0].levelname assert "INFO" == caplog.records[1].levelname def test_publish_messages_fails( klio_job_config, expected_klio_message, mock_publisher, caplog ): client = mock_publisher.return_value client.publish.side_effect = Exception("foo") entity_id = "s0m3-ent1ty-1D" msg_version = 1 expected_klio_message.metadata.force = False expected_klio_message.metadata.ping = False expected_klio_message.data.entity_id = entity_id expected_klio_message.version = msg_version publish.publish_messages(klio_job_config, [entity_id]) mock_publisher.return_value.publish.assert_called_once_with( data=expected_klio_message.SerializeToString(), topic="an-input-topic" ) assert 3 == len(caplog.records) assert "INFO" == caplog.records[0].levelname assert "WARNING" == caplog.records[1].levelname assert "WARNING" == caplog.records[2].levelname def test_publish_messages_raises(klio_job_config, monkeypatch, caplog): monkeypatch.setattr(klio_job_config.job_config.events, "inputs", None) with pytest.raises(SystemExit): publish.publish_messages(klio_job_config, ["s0m3-ent1ty-1D"]) assert 1 == len(caplog.records)
gaybro8777/klio
lib/src/klio/transforms/io.py
<reponame>gaybro8777/klio<filename>lib/src/klio/transforms/io.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import os import apache_beam as beam from apache_beam.io import avroio as beam_avroio from apache_beam.io.gcp import bigquery as beam_bq from fastavro import parse_schema from klio_core.proto import klio_pb2 class BaseKlioIOException(Exception): """Base IO exception.""" class KlioMissingConfiguration(BaseKlioIOException): """Required configuration is missing.""" class _KlioReadWrapper(beam.io.Read): """Klio-ified `beam.io.Read` class. Handle the case if a user invokes an already-wrapped Klio IO transform (via `_KlioTransformMixin`) with `beam.io.Read`, as this is not possible to double-wrap. """ def __init__(self, *args, **kwargs): if not len(args): # I would think we'd never get here, but this is just in case # something real funky happens, like this semi-private class # is not instantiated with an IO transform super(_KlioReadWrapper, self).__init__(*args, **kwargs) self.__wrapped_transform = args[0].__class__.__name__ super(_KlioReadWrapper, self).__init__(*args, **kwargs) def is_bounded(self, *args, **kwargs): # beam.io.Read invokes beam.io.Read.source.is_bounded # (beam.io.Read.source == self/this instance) and throw an # AttributeError if is_bounded not defined. Here we define # the method but raise our own expection telling the user not to # invoke an alread-wrapped Klio IO Transform with `beam.io.Read()` raise TypeError( "Error reading from `{}`. If the transform was wrapped with " "`apache_beam.io.Read`, remove that wrapper and try again.".format( self.__wrapped_transform ) ) class _KlioWrapIOMetaclass(type): def __call__(self, *args, **kwargs): # Some IO transforms require invocation to be wrapped with # `beam.io.Read()`, and others do not. This allows for the API # to be the same no matter what by wrapping `beam.io.Read` if # the transform requires it. If a user wraps a Klio IO transform # again (i.e. `beam.io.Read(_AlreadyWrappedKlioTransform))`), a # human-friendly `TypeError` will raise (see # `_KlioReadWrapper.is_bounded`) if self._REQUIRES_IO_READ_WRAP: return _KlioReadWrapper( super(_KlioWrapIOMetaclass, self).__call__(*args, **kwargs) ) return super(_KlioWrapIOMetaclass, self).__call__(*args, **kwargs) class _KlioTransformMixin(metaclass=_KlioWrapIOMetaclass): """Common properties for klio v2 IO transforms.""" # whether or not the transform needs to be invoked by beam.io.Read() _REQUIRES_IO_READ_WRAP = False class _KlioReadFromTextSource(beam.io.textio._TextSource): """Parses a text file as newline-delimited elements. Supports newline delimiters '\n' and '\r\n Returns: (str) KlioMessage serialized as a string """ def read_records(self, file_name, range_tracker): records = super(_KlioReadFromTextSource, self).read_records( file_name, range_tracker ) for record in records: record_as_bytes = record.encode("utf-8") message = klio_pb2.KlioMessage() message.version = klio_pb2.Version.V2 message.metadata.intended_recipients.anyone.SetInParent() message.data.element = record_as_bytes yield message.SerializeToString() class KlioReadFromText(beam.io.ReadFromText, _KlioTransformMixin): """Read from a local or GCS file with each new line as a ``KlioMessage.data.element``. """ _source_class = _KlioReadFromTextSource class _KlioReadFromBigQueryMapper(object): """Wrapper class to provide a ``beam.Map`` object that converts a row of a ``ReadFromBigQuery`` to a properly formatted ``KlioMessage``. """ def __init__(self, klio_message_columns=None): self.__klio_message_columns = klio_message_columns def _generate_klio_message(self): message = klio_pb2.KlioMessage() message.version = klio_pb2.Version.V2 message.metadata.intended_recipients.anyone.SetInParent() # TODO: this is where we should add (relevant) KlioMessage.metadata; # (1) One thing to figure out is the klio_pb2.KlioJob definition, # particularly the JobInput definition, in light of KlioConfig v2. # Once that's figured out, we should at least populate the # job audit log. # (2) Another thing to figure out is force/ping. In streaming, messages # are individually marked as force or ping when needed. However, # users aren't able to tag individual messages generated from a row # of BQ data as force/ping, and it's probably very difficult for us # to provide a way to do that. So, should we allow users to at least # globally set force/ping on their event input config in klio-job.yaml? # Potentially. return message def _map_row_element(self, row): # NOTE: this assumes that the coder being used (default is # beam.io.gcp.bigquery_tools.RowAsDictJsonCoder, otherwise set in # klio-job.yaml) is JSON serializable (since the default is just # a plain dictionary). This assumption might break if someone # provides a different coder. # NOTE: We need to have the row elements be bytes, so if it is # a dictionary, we json.dumps into a str to convert to bytes, # but that may need to change if we want to support other coders data = {} if self.__klio_message_columns: if len(self.__klio_message_columns) == 1: data = row[self.__klio_message_columns[0]] else: for key, value in row.items(): if key in self.__klio_message_columns: data[key] = value data = json.dumps(data) else: data = json.dumps(row) return data def _map_row(self, row): message = self._generate_klio_message() message.data.element = bytes(self._map_row_element(row), "utf-8") return message.SerializeToString() def as_beam_map(self): return "Convert to KlioMessage" >> beam.Map(self._map_row) # Note: copy-pasting the docstrings of `ReadFromBigQuery` so that we can # include our added parameter (`klio_message_columns`) in the API # documentation (via autodoc). If we don't do this, then just the parent # documentation will be shown, excluding our new parameter. class KlioReadFromBigQuery(beam.PTransform, _KlioTransformMixin): """Read data from BigQuery. This PTransform uses a BigQuery export job to take a snapshot of the table on GCS, and then reads from each produced file. File format is Avro by default. Args: table (str, callable, ValueProvider): The ID of the table, or a callable that returns it. The ID must contain only letters ``a-z``, ``A-Z``, numbers ``0-9``, or underscores ``_``. If dataset argument is :data:`None` then the table argument must contain the entire table reference specified as: ``'DATASET.TABLE'`` or ``'PROJECT:DATASET.TABLE'``. If it's a callable, it must receive one argument representing an element to be written to BigQuery, and return a TableReference, or a string table name as specified above. dataset (str): The ID of the dataset containing this table or :data:`None` if the table reference is specified entirely by the table argument. project (str): The ID of the project containing this table. klio_message_columns (list): A list of fields (``str``) that should be assigned to ``KlioMessage.data.element``. .. note:: If more than one field is provided, the results including the column names will be serialized to JSON before assigning to ``KlioMessage.data.element``. (e.g. ``'{"field1": "foo", "field2": bar"}'``). If only one field is provided, just the value will be assigned to ``KlioMessage.data.element``. query (str, ValueProvider): A query to be used instead of arguments table, dataset, and project. validate (bool): If :data:`True`, various checks will be done when source gets initialized (e.g., is table present?). This should be :data:`True` for most scenarios in order to catch errors as early as possible (pipeline construction instead of pipeline execution). It should be :data:`False` if the table is created during pipeline execution by a previous step. coder (~apache_beam.coders.coders.Coder): The coder for the table rows. If :data:`None`, then the default coder is _JsonToDictCoder, which will interpret every row as a JSON serialized dictionary. use_standard_sql (bool): Specifies whether to use BigQuery's standard SQL dialect for this query. The default value is :data:`False`. If set to :data:`True`, the query will use BigQuery's updated SQL dialect with improved standards compliance. This parameter is ignored for table inputs. flatten_results (bool): Flattens all nested and repeated fields in the query results. The default value is :data:`True`. kms_key (str): Optional Cloud KMS key name for use when creating new temporary tables. gcs_location (str, ValueProvider): The name of the Google Cloud Storage bucket where the extracted table should be written as a string or a :class:`~apache_beam.options.value_provider.ValueProvider`. If :data:`None`, then the temp_location parameter is used. bigquery_job_labels (dict): A dictionary with string labels to be passed to BigQuery export and query jobs created by this transform. See: https://cloud.google.com/bigquery/docs/reference/rest/v2/\ Job#JobConfiguration use_json_exports (bool): By default, this transform works by exporting BigQuery data into Avro files, and reading those files. With this parameter, the transform will instead export to JSON files. JSON files are slower to read due to their larger size. When using JSON exports, the BigQuery types for DATE, DATETIME, TIME, and TIMESTAMP will be exported as strings. This behavior is consistent with BigQuerySource. When using Avro exports, these fields will be exported as native Python types (datetime.date, datetime.datetime, datetime.datetime, and datetime.datetime respectively). Avro exports are recommended. To learn more about BigQuery types, and Time-related type representations, see: https://cloud.google.com/bigquery/docs/reference/standard-sql/\ data-types To learn more about type conversions between BigQuery and Avro, see: https://cloud.google.com/bigquery/docs/loading-data-cloud-\ storage-avro#avro_conversions """ def __init__(self, *args, klio_message_columns=None, **kwargs): self.__reader = beam_bq.ReadFromBigQuery(*args, **kwargs) self.__mapper = _KlioReadFromBigQueryMapper(klio_message_columns) def expand(self, pcoll): return pcoll | self.__reader | self.__mapper.as_beam_map() class KlioWriteToBigQuery(beam.io.WriteToBigQuery, _KlioTransformMixin): """Writes to BigQuery table with each row as ``KlioMessage.data.element``. """ # Note: Not using BigQuerySink due to it only being available for # batch. See https://beam.apache.org/releases/pydoc/2.22.0/ # apache_beam.io.gcp.bigquery.html?highlight=bigquerysink # #apache_beam.io.gcp.bigquery.BigQuerySink _REQUIRES_IO_READ_WRAP = False def __unwrap(self, encoded_element): message = klio_pb2.KlioMessage() message.ParseFromString(encoded_element) data = json.loads(message.data.payload) return data def expand(self, pcoll): return super().expand(pcoll | beam.Map(self.__unwrap)) class _KlioTextSink(beam.io.textio._TextSink): """A :class:`~apache_beam.transforms.ptransform.PTransform` for writing to text files. Takes a PCollection of KlioMessages and writes the elements to a textfile """ def write_record(self, file_handle, encoded_element): """Writes a single encoded record. Args: file_handle (str): a referential identifier that points to an audio file found in the configured output data location. encoded_element (KlioMessage): KlioMessage """ message = klio_pb2.KlioMessage() message.ParseFromString(encoded_element) record = message.data.element super(_KlioTextSink, self).write_encoded_record(file_handle, record) class KlioWriteToText(beam.io.textio.WriteToText): """Write to a local or GCS file with each new line as ``KlioMessage.data.element``. """ def __init__(self, *args, **kwargs): self._sink = _KlioTextSink(*args, **kwargs) # note: fast avro is default for py3 on beam class _KlioFastAvroSource(beam_avroio._FastAvroSource): def read_records(self, file_name, range_tracker): records = super(_KlioFastAvroSource, self).read_records( file_name=file_name, range_tracker=range_tracker ) for record in records: message = klio_pb2.KlioMessage() message.version = klio_pb2.Version.V2 message.metadata.intended_recipients.anyone.SetInParent() # If an element is sent then we set the element # to handle event reading # If "element" is not present then we stuff the record # into the message element message.data.element = ( record["element"] if "element" in record else bytes(json.dumps(record).encode("utf-8")) ) yield message.SerializeToString() # define an I/O transform using the klio-specific avro source # note: fast avro is default for py3 on beam # Note: copy-pasting the docstrings of `ReadFromAvro` so that we can # include our added parameter (`location`) in the API # documentation (via autodoc) and drop `use_fastavro` since we default to # True. If we don't do this, then just the parent documentation will be shown, # excluding our new parameter and including an unavailable parameter # (`location` and `use_fastavro` respectively) class KlioReadFromAvro(beam.io.ReadFromAvro): """Read avro from a local directory or GCS bucket. Data from avro is dumped into JSON and assigned to ``KlioMessage.data. element``. Args: file_pattern (str): the file glob to read. location (str): local or GCS path of file(s) to read. min_bundle_size (int): the minimum size in bytes, to be considered when splitting the input into bundles. validate (bool): flag to verify that the files exist during the pipeline creation time. """ _REQUIRES_IO_READ_WRAP = True def __init__( self, file_pattern=None, location=None, min_bundle_size=0, validate=True, ): file_pattern = self._get_file_pattern(file_pattern, location) super(KlioReadFromAvro, self).__init__( file_pattern=file_pattern, min_bundle_size=min_bundle_size, validate=validate, use_fastavro=True, ) self._source = _KlioFastAvroSource( file_pattern, min_bundle_size, validate=validate ) def _get_file_pattern(self, file_pattern, location): # TODO: this should be a validator in klio_core.config if not any([file_pattern, location]): raise KlioMissingConfiguration( "Must configure at least one of the following keys when " "reading from avro: `file_pattern`, `location`." ) if all([file_pattern, location]): file_pattern = os.path.join(location, file_pattern) elif file_pattern is None: file_pattern = location return file_pattern # note: fast avro is default for py3 on beam class _KlioFastAvroSink(beam_avroio._FastAvroSink): def write_record(self, writer, encoded_element): message = klio_pb2.KlioMessage() message.ParseFromString(encoded_element) record = {"element": message.data.element} super(_KlioFastAvroSink, self).write_record( writer=writer, value=record ) # Note of caution: In the past problems have arisen due to # changes to internal beam classes. # If this occurs, consider writing this as a custom PTransform # that bundles a ``beam.Map`` with the standard ``WriteToAvro`` # Refer to ``KlioReadFromBigQuery`` as an example. class KlioWriteToAvro(beam.io.WriteToAvro): """Write avro to a local directory or GCS bucket. ``KlioMessage.data.element`` data is parsed out and dumped into arvo format. Args: file_path_prefix (str): The file path to write to location (str): local or GCS path to write to schema (str): The schema to use, as returned by avro.schema.parse codec (str): The codec to use for block-level compression. defaults to 'deflate' file_name_suffix (str): Suffix for the files written. num_shards (int): The number of files (shards) used for output. shard_name_template (str): template string for shard number and count mime_type (str): The MIME type to use for the produced files. Defaults to "application/x-avro" """ KLIO_SCHEMA_OBJ = { "namespace": "klio.avro", "type": "record", "name": "KlioMessage", "fields": [{"name": "element", "type": "bytes"}], } def __init__( self, file_path_prefix=None, location=None, schema=parse_schema(KLIO_SCHEMA_OBJ), codec="deflate", file_name_suffix="", num_shards=0, shard_name_template=None, mime_type="application/x-avro", ): file_path = self._get_file_path(file_path_prefix, location) super(KlioWriteToAvro, self).__init__( file_path_prefix=file_path, schema=schema, codec=codec, file_name_suffix=file_name_suffix, num_shards=num_shards, shard_name_template=shard_name_template, mime_type=mime_type, use_fastavro=True, ) self._sink = _KlioFastAvroSink( file_path, schema, codec, file_name_suffix, num_shards, shard_name_template, mime_type, ) def _get_file_path(self, file_path_prefix, location): # TODO: this should be a validator in klio_core.config if not any([file_path_prefix, location]): raise KlioMissingConfiguration( "Must configure at least one of the following keys when " "writing to avro: `file_path_prefix`, `location`." ) if all([file_path_prefix, location]): file_path_prefix = os.path.join(location, file_path_prefix) elif file_path_prefix is None: file_path_prefix = location return file_path_prefix
gaybro8777/klio
cli/src/klio_cli/utils/multi_line_terminal_writer.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # from contextlib import contextmanager CLEAR_TO_END_OF_LINE = "\u001b[0K" def _move_n_lines(to_move: int) -> None: if to_move == 0: return # F to move the cursor up w/i the console; E to move # the cursor down ansi_code = "E" if to_move > 0 else "F" print("\033[%d%c" % (abs(to_move), ansi_code), end="", flush=True) class MultiLineTerminalWriter(object): """ A class to manage writing multi-line output to the terminal, with each line tagged with a line_id. Useful for displaying progress of multiple conncurrent items, like a multi-layer Docker push. """ def __init__(self): self._line_id_to_index = {} self._lines_printed = 0 def emit_line(self, line_id, contents): """ Emit a string at the given line_id. If this line_id has not yet been printed, it will be placed at the bottom of the current terminal output. """ writing_new_line_at_bottom = line_id not in self._line_id_to_index if writing_new_line_at_bottom: self._line_id_to_index[line_id] = len(self._line_id_to_index) with self._at_line(self._line_id_to_index[line_id]): if not writing_new_line_at_bottom: contents += CLEAR_TO_END_OF_LINE print(contents, flush=True) @contextmanager def _at_line(self, index): """ A context manager that moves the cursor to a given line, assuming that the cursor is already placed at the bottom of the list of lines, and returns the cursor to the bottom after writing one line. """ distance_from_bottom = self._lines_printed - index if distance_from_bottom: _move_n_lines(-distance_from_bottom) # Print one line. yield # Assume that a newline was printed when we yielded. # If we had to move up to get to this line, we have to move down # by one fewer line. If we didn't have to move up to get to this # line, we don't need to move down at all as we're already at # the new bottom. if distance_from_bottom != 0: distance_from_bottom -= 1 _move_n_lines(distance_from_bottom) self._lines_printed = max(self._lines_printed, index + 1)
gaybro8777/klio
exec/tests/unit/commands/test_profile.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import sys import pytest from klio_exec.commands import profile as profile_cmd @pytest.fixture def config(mocker): return mocker.Mock() @pytest.fixture def klio_pipeline(config): return profile_cmd.KlioPipeline(config) class DummyTransformGenerator(object): def process(self, x): yield x + 10 yield x + 20 class DummyTransformFunc(object): def process(self, x): return x + 10 class DummyTransformFuncRaises(object): def process(self, *args): raise Exception("catch me") @pytest.fixture def transforms(): return [ DummyTransformGenerator, DummyTransformFunc, DummyTransformFuncRaises, ] @pytest.mark.parametrize("filename", (None, "-", "input.txt")) def test_smart_open(filename, mocker): m_open = mocker.mock_open() mock_open = mocker.patch("klio_exec.commands.profile.open", m_open) with profile_cmd.smart_open(filename, fmode="r") as act_ret: pass if filename and filename != "-": mock_open.assert_called_once_with(filename, "r") assert act_ret.closed else: mock_open.assert_not_called() assert act_ret == sys.stdout @pytest.mark.parametrize( "what,output_file,temp_out,exp_ret", ( ("mem", "output.txt", False, "output.png"), ("mem", "output.txt", True, "klio_profile_mem_1234.png"), ("mem", "output", False, "output.png"), ), ) def test_get_output_png_file( what, output_file, temp_out, exp_ret, klio_pipeline, monkeypatch ): monkeypatch.setattr(klio_pipeline, "_now_str", "1234") monkeypatch.setattr(klio_pipeline, "output_file", output_file) act_ret = klio_pipeline._get_output_png_file(what, temp_out) assert exp_ret == act_ret @pytest.mark.parametrize( "output_file,plot_graph,exp_temp_output", ((None, True, True), (None, False, False), ("output.txt", True, False)), ) def test_smart_temp_create( output_file, plot_graph, exp_temp_output, klio_pipeline, mocker, monkeypatch, ): monkeypatch.setattr(klio_pipeline, "output_file", output_file) mock_named_temp_file = mocker.Mock() mock_file = mocker.Mock() mock_file.name = "temp-1234" mock_named_temp_file.return_value = mock_file monkeypatch.setattr( profile_cmd.tempfile, "NamedTemporaryFile", mock_named_temp_file ) with klio_pipeline._smart_temp_create( "mem", plot_graph ) as act_temp_output: pass assert exp_temp_output is act_temp_output if exp_temp_output: assert klio_pipeline.output_file == "temp-1234" @pytest.mark.parametrize( "show_logs,input_file", ((False, None), (True, "input.txt")) ) def test_get_subproc( show_logs, input_file, klio_pipeline, mocker, monkeypatch ): mock_subproc_open = mocker.Mock() monkeypatch.setattr(profile_cmd.subprocess, "Popen", mock_subproc_open) monkeypatch.setattr(klio_pipeline, "input_file", input_file) monkeypatch.setattr(klio_pipeline, "entity_ids", ("foo", "bar")) kwargs = {"show_logs": show_logs} klio_pipeline._get_subproc(**kwargs) exp_call_args = ["klioexec", "profile", "run-pipeline"] if show_logs: exp_call_args.append("--show-logs") if input_file: exp_call_args.extend(["--input-file", input_file]) else: exp_call_args.extend(["foo", "bar"]) mock_subproc_open.assert_called_once_with(exp_call_args) @pytest.mark.parametrize("output_file", (None, "output.txt")) def test_profile_wall_time_per_line( output_file, klio_pipeline, mocker, monkeypatch ): monkeypatch.setattr(klio_pipeline, "output_file", output_file) mock_line_prof = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_cpu_line_profiler", mock_line_prof ) mock_run_pipeline = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_run_pipeline", mock_run_pipeline) klio_pipeline._profile_wall_time_per_line(iterations=1) if output_file: mock_line_prof.return_value.print_stats.assert_called_once_with( klio_pipeline.output_file, output_unit=1 ) else: mock_line_prof.return_value.print_stats.assert_called_once_with( output_unit=1 ) @pytest.mark.parametrize("get_maximum,exp_fmode", ((True, "w"), (False, "a"))) def test_profile_memory_per_line( get_maximum, exp_fmode, klio_pipeline, mocker, monkeypatch ): mock_memory_profiler = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_memory_line_profiler", mock_memory_profiler ) mock_memory_wrapper = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_memory_line_wrapper", mock_memory_wrapper ) mock_smart_open = mocker.patch.object(profile_cmd, "smart_open") mock_smart_open.return_value.__enter__.return_value = "opened_file" mock_run_pipeline = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_run_pipeline", mock_run_pipeline) mock_show_results = mocker.Mock() monkeypatch.setattr( profile_cmd.memory_profiler, "show_results", mock_show_results ) klio_pipeline._profile_memory_per_line(get_maximum=get_maximum) mock_memory_profiler.assert_called_once_with() mock_memory_wrapper.assert_called_once_with( mock_memory_profiler.return_value, get_maximum ) mock_smart_open.assert_called_once_with( klio_pipeline.output_file, fmode=exp_fmode ) assert "opened_file" == klio_pipeline._stream mock_run_pipeline.assert_called_once_with() if get_maximum: mock_show_results.assert_called_once_with( mock_memory_profiler.return_value, stream="opened_file" ) else: mock_show_results.assert_not_called() @pytest.mark.parametrize("plot_graph", (True, False)) def test_profile_memory(plot_graph, klio_pipeline, mocker, monkeypatch): mock_get_subproc = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_get_subproc", mock_get_subproc) mock_smart_temp_create = mocker.patch.object( klio_pipeline, "_smart_temp_create" ) mock_smart_temp_create.return_value.__enter__.return_value = True mock_smart_open = mocker.patch.object(profile_cmd, "smart_open") mock_smart_open.return_value.__enter__.return_value.name = "bar" mock_memory_usage = mocker.Mock() monkeypatch.setattr( profile_cmd.memory_profiler, "memory_usage", mock_memory_usage ) mock_get_output_png = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_output_png_file", mock_get_output_png ) mock_plot = mocker.Mock() monkeypatch.setattr(profile_cmd.profile_utils, "plot", mock_plot) kwargs = {"plot_graph": plot_graph} act_ret = klio_pipeline._profile_memory(**kwargs) if plot_graph: mock_get_output_png.assert_called_once_with("memory", True) assert mock_get_output_png.return_value == act_ret else: assert act_ret is None @pytest.mark.parametrize("plot_graph", (True, False)) def test_profile_cpu(plot_graph, klio_pipeline, mocker, monkeypatch): mock_get_subproc = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_get_subproc", mock_get_subproc) mock_smart_temp_create = mocker.patch.object( klio_pipeline, "_smart_temp_create" ) mock_smart_temp_create.return_value.__enter__.return_value = True mock_smart_open = mocker.patch.object(profile_cmd, "smart_open") mock_smart_open.return_value.__enter__.return_value.name = "bar" mock_get_cpu_usage = mocker.Mock() monkeypatch.setattr( profile_cmd.cpu_utils, "get_cpu_usage", mock_get_cpu_usage ) mock_get_output_png = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_output_png_file", mock_get_output_png ) mock_plot = mocker.Mock() monkeypatch.setattr(profile_cmd.profile_utils, "plot", mock_plot) kwargs = {"plot_graph": plot_graph} act_ret = klio_pipeline._profile_cpu(**kwargs) if plot_graph: mock_get_output_png.assert_called_once_with("cpu", True) assert mock_get_output_png.return_value == act_ret else: assert act_ret is None @pytest.mark.parametrize("input_file", (None, "input.txt")) def test_get_io_mapper( input_file, klio_pipeline, transforms, mocker, monkeypatch ): entity_ids = ["id1", "id2", "id3"] serialized_messages = [ klio_pipeline._entity_id_to_message(i).SerializeToString() for i in entity_ids ] if input_file: monkeypatch.setattr(klio_pipeline, "input_file", input_file) else: monkeypatch.setattr(klio_pipeline, "entity_ids", entity_ids) mock_read_from_text = mocker.Mock() monkeypatch.setattr( profile_cmd.beam.io, "ReadFromText", mock_read_from_text ) mock_create = mocker.Mock() monkeypatch.setattr(profile_cmd.beam, "Create", mock_create) mock_flatmap = mocker.Mock() mock_transform = mocker.MagicMock() mock_flatmap.return_value = mock_transform monkeypatch.setattr(profile_cmd.beam, "FlatMap", mock_flatmap) mapper = klio_pipeline._get_io_mapper(1) assert isinstance(mapper, profile_cmd.StubIOMapper) if input_file: mock_read_from_text.assert_called_once_with(klio_pipeline.input_file) else: mock_create.assert_called_once_with(serialized_messages) def test_run_pipeline(klio_pipeline, mocker, monkeypatch): mock_get_io_mapper = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_get_io_mapper", mock_get_io_mapper) mock_get_user_pipeline = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_user_pipeline", mock_get_user_pipeline ) mock_get_user_config = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_get_user_config", mock_get_user_config ) klio_pipeline._run_pipeline() mock_get_io_mapper.assert_called_once_with(1) mock_get_user_config.assert_called_once_with() mock_get_user_pipeline.assert_called_once_with( mock_get_user_config.return_value, mock_get_io_mapper.return_value ) mock_get_user_pipeline.return_value.run.assert_called_once_with() @pytest.mark.parametrize( "what", ("run", "cpu", "timeit", "memory", "memory_per_line", "foo") ) def test_profile(what, klio_pipeline, mocker, monkeypatch): mock_run_pipeline = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_run_pipeline", mock_run_pipeline) mock_profile_cpu = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_profile_cpu", mock_profile_cpu) mock_profile_memory = mocker.Mock() monkeypatch.setattr(klio_pipeline, "_profile_memory", mock_profile_memory) mock_memory_per_line = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_profile_memory_per_line", mock_memory_per_line ) mock_wall_time_per_line = mocker.Mock() monkeypatch.setattr( klio_pipeline, "_profile_wall_time_per_line", mock_wall_time_per_line ) klio_pipeline.profile(what) if what == "run": mock_run_pipeline.assert_called_once_with() mock_profile_cpu.assert_not_called() mock_profile_memory.assert_not_called() mock_memory_per_line.assert_not_called() mock_wall_time_per_line.assert_not_called() elif what == "cpu": mock_run_pipeline.assert_not_called() mock_profile_cpu.assert_called_once_with() mock_profile_memory.assert_not_called() mock_memory_per_line.assert_not_called() mock_wall_time_per_line.assert_not_called() elif what == "memory": mock_run_pipeline.assert_not_called() mock_profile_cpu.assert_not_called() mock_profile_memory.assert_called_once_with() mock_memory_per_line.assert_not_called() mock_wall_time_per_line.assert_not_called() elif what == "memory_per_line": mock_run_pipeline.assert_not_called() mock_profile_cpu.assert_not_called() mock_profile_memory.assert_not_called() mock_memory_per_line.assert_called_once_with() mock_wall_time_per_line.assert_not_called() elif what == "timeit": mock_run_pipeline.assert_not_called() mock_profile_cpu.assert_not_called() mock_profile_memory.assert_not_called() mock_memory_per_line.assert_not_called() mock_wall_time_per_line.assert_called_once_with() else: mock_run_pipeline.assert_not_called() mock_profile_cpu.assert_not_called() mock_profile_memory.assert_not_called() mock_memory_per_line.assert_not_called() mock_wall_time_per_line.assert_not_called()
gaybro8777/klio
lib/tests/unit/metrics/test_stackdriver.py
<filename>lib/tests/unit/metrics/test_stackdriver.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from googleapiclient import errors as gapi_errors from klio.metrics import stackdriver as sd_metrics @pytest.fixture def mock_stackdriver_client(mocker, monkeypatch): mock_client = mocker.Mock() monkeypatch.setattr( sd_metrics.discovery, "build", lambda x, y: mock_client ) return mock_client @pytest.fixture def client(klio_config, monkeypatch, mocker, mock_stackdriver_client): return sd_metrics.StackdriverLogMetricsClient(klio_config) @pytest.fixture def mock_discovery_build(mocker, monkeypatch): mock_discovery_build = mocker.Mock() monkeypatch.setattr(sd_metrics.discovery, "build", mock_discovery_build) return mock_discovery_build @pytest.fixture def counter(): kwargs = { "name": "my-counter", "job_name": "test-job", "project": "test-project", } return sd_metrics.StackdriverLogMetricsCounter(**kwargs) def test_stackdriver_client(client, mock_discovery_build): # sanity check assert not getattr(client._thread_local, "stackdriver_client", None) client._stackdriver_client assert getattr(client._thread_local, "stackdriver_client", None) mock_discovery_build.assert_called_once_with("logging", "v2") @pytest.mark.parametrize("value", (None, 5)) def test_client_counter(client, value, mocker, monkeypatch, caplog): mock_init_metric = mocker.Mock() monkeypatch.setattr( sd_metrics.StackdriverLogMetricsCounter, "_init_metric", mock_init_metric, ) kwargs = {} if value: kwargs["value"] = value counter = client.counter(name="my-counter", **kwargs) mock_init_metric.assert_called_once_with(client._stackdriver_client) assert isinstance(counter, sd_metrics.StackdriverLogMetricsCounter) if value: assert 1 == len(caplog.records) assert "WARNING" == caplog.records[0].levelname def test_client_gauge(client, caplog): gauge = client.gauge(name="my-gauge") assert isinstance(gauge, sd_metrics.StackdriverLogMetricsGauge) assert 1 == len(caplog.records) assert "WARNING" == caplog.records[0].levelname def test_client_timer(client, caplog): gauge = client.timer(name="my-timer") assert isinstance(gauge, sd_metrics.StackdriverLogMetricsTimer) assert 1 == len(caplog.records) assert "WARNING" == caplog.records[0].levelname def test_counter_get_filter(counter): exp_filter = ( 'resource.type="dataflow_step" ' 'logName="projects/test-project/logs/' 'dataflow.googleapis.com%2Fworker" ' 'jsonPayload.message:"[my-counter]"' ) actual_filter = counter._get_filter() assert exp_filter == actual_filter @pytest.mark.parametrize("raises", (False, True)) def test_counter_init_metric(raises, counter, mock_stackdriver_client, caplog): _mock_projects = mock_stackdriver_client.projects.return_value mock_req = _mock_projects.metrics.return_value.create if raises: mock_req.return_value.execute.side_effect = Exception("foo") counter._init_metric(mock_stackdriver_client) mock_req.assert_called_once_with(parent=counter.parent, body=counter.body) mock_req.return_value.execute.assert_called_once_with() if raises: assert 1 == len(caplog.records) assert "ERROR" == caplog.records[0].levelname @pytest.mark.parametrize( "status,exp_log_level", ((409, "DEBUG"), (403, "ERROR")) ) def test_counter_init_metric_api_error( status, exp_log_level, counter, mock_stackdriver_client, mocker, caplog ): mock_resp = mocker.Mock() mock_resp.status = status error = gapi_errors.HttpError(resp=mock_resp, content=b"foo") _mock_projects = mock_stackdriver_client.projects.return_value mock_req = _mock_projects.metrics.return_value.create mock_req.return_value.execute.side_effect = error counter._init_metric(mock_stackdriver_client) mock_req.assert_called_once_with(parent=counter.parent, body=counter.body) mock_req.return_value.execute.assert_called_once_with() assert 1 == len(caplog.records) assert exp_log_level == caplog.records[0].levelname
gaybro8777/klio
audio/src/klio_audio/transforms/io.py
<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import io import os from apache_beam.io.gcp import gcsio from klio.transforms import decorators as tfm_decorators from klio_audio import decorators from klio_audio.transforms import _base # TODO: handle multiple data inputs class GcsLoadBinary(_base.KlioAudioBaseDoFn): """Download binary file from GCS into memory. This transform uses the ``job_config.data.inputs`` configuration in a job's ``klio-job.yaml`` file. This transform uses Apache's native :class:`GCS client <apache_beam.io.gcp.gcsio.GcsIO>` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>`, and returns with a payload of the downloaded binary as a file-like bytes object. Example: .. code-block:: python # run.py from klio_audio.transforms import audio from klio_audio.transforms import io def run(in_pcol, job_config): return ( in_pcol | io.GcsLoadBinary() | audio.LoadAudio() # other transforms ) .. code-block:: yaml # klio-job.yaml # <-- snip --> job_config: events: # <-- snip --> data: inputs: - type: file location: gs://my-bucket/input file_suffix: .ogg # <-- snip --> """ def setup(self): self.client = gcsio.GcsIO() @tfm_decorators._handle_klio @decorators.handle_binary(skip_load=True) def process(self, item): element = item.element.decode("utf-8") input_data_config = self._klio.config.job_config.data.inputs # raise a runtime error so it actually crashes klio/beam rather than # just continue processing elements if len(input_data_config) == 0: raise RuntimeError( "The `klio_audio.transforms.io.GcsLoadBinary` transform " "requires a data input to be configured in " "`klio-job.yaml::job_config.data.inputs`." ) # raise a runtime error so it actually crashes klio/beam rather than # just continue processing elements if len(input_data_config) > 1: raise RuntimeError( "The `klio_audio.transforms.io.GcsLoadBinary` transform " "does not support multiple configured inputs in " "`klio-job.yaml::job_config.data.inputs`." ) input_data = input_data_config[0] file_suffix = input_data.file_suffix if not file_suffix.startswith("."): file_suffix = "." + file_suffix filename = element + file_suffix input_path = os.path.join(input_data.location, filename) self._klio.logger.debug( "Downloading {} from {}".format(filename, input_data.location) ) with self.client.open(input_path, "rb") as source: out = io.BytesIO(source.read()) self._klio.logger.debug("Downloaded {}".format(filename)) yield out # TODO: handle multiple data outputs class GcsUploadPlot(_base.KlioAudioBaseDoFn): """Upload a matplotlib :class:`figure <matplotlib.figure.Figure>` to GCS. This transform uses the ``job_config.data.outputs`` configuration in a job's ``klio-job.yaml`` file. This transform wraps :class:`savefig <matplotlib.figure.Figure>` and expects a :class:`PCollection <apache_beam.pvalue.PCollection>` of :ref:`KlioMessages <klio-message>` where the payload is a :class:`matplotlib.figure.Figure` and returns with a payload of the uploaded file location as ``bytes``. Example: .. code-block:: python # run.py from klio_audio.transforms import audio from klio_audio.transforms import io def run(in_pcol, job_config): return ( in_pcol | io.GcsLoadBinary() | audio.LoadAudio() | audio.GetSpec() | audio.SpecToPlot() | io.GcsUploadPlot() ) .. code-block:: yaml # klio-job.yaml # <-- snip --> job_config: # <-- snip --> data: inputs: # <-- snip --> outputs: - type: file location: gs://my-bucket/output file_suffix: .png Args: prefix (str): filename prefix. Default: ``""`` suffix (str): filename suffix. Default: ``""`` file_format (str): plot format (e.g. png). Defaults to the file suffix as configured in ``klio-job.yaml::job_config.data.outputs[].file_suffix``. plt_kwargs (dict): keyword arguments to pass to :class:`savefig <matplotlib.figure.Figure>`. """ def __init__(self, prefix="", suffix="", file_format=None, **plt_kwargs): self.prefix = prefix self.suffix = suffix self.file_format = file_format self.plt_kwargs = plt_kwargs def setup(self): self.client = gcsio.GcsIO() @tfm_decorators._handle_klio @decorators.handle_binary(skip_dump=True) def process(self, item): element = item.element.decode("utf-8") output_data_config = self._klio.config.job_config.data.outputs # raise a runtime error so it actually crashes klio/beam rather than # just continue processing elements if len(output_data_config) == 0: raise RuntimeError( "The `klio_audio.transforms.io.GcsUploadPlot` transform " "requires a data output to be configured in " "`klio-job.yaml::job_config.data.outputs`." ) # raise a runtime error so it actually crashes klio/beam rather than # just continue processing elements if len(output_data_config) > 1: raise RuntimeError( "The `klio_audio.transforms.io.GcsUploadPlot` transform " "does not support multiple configured outputs in " "`klio-job.yaml::job_config.data.outputs`." ) output_data = output_data_config[0] file_suffix = output_data.file_suffix if not file_suffix.startswith("."): file_suffix = "." + file_suffix filename = self.prefix + element + self.suffix + file_suffix output_path = os.path.join(output_data.location, filename) source = io.BytesIO() fig = item.payload fig_format = self.file_format or file_suffix.lstrip(".") self._klio.logger.debug( "Saving plot as {} for {}".format(fig_format, element) ) fig.savefig(source, format=fig_format, **self.plt_kwargs) with self.client.open(output_path, "wb") as out: out.write(source.getvalue()) self._klio.logger.debug("Saved plot to {}".format(output_path)) yield output_path
gaybro8777/klio
exec/src/klio_exec/commands/audit_steps/numpy_broken_blas.py
<reponame>gaybro8777/klio # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import platform from packaging import version from klio_exec.commands.audit_steps import base class NumPyBrokenBLASUsage(base.BaseKlioAuditStep): """Detect numpy version for potential threading issues.""" AUDIT_STEP_NAME = "numpy_broken_blas" MINIMUM_NUMPY_VERSION = version.parse("1.16.3") @staticmethod def get_description(): # tmp turn off black formatting to skip long URL # fmt: off return ( "The 1.16.3 version of numpy links against a newer version of " "OpenBLAS that fixes some important threading issues - notably, " "the `dot` function that calls into OpenBLAS' _dgemv function, " "which on older versions, is non-reentrant and can cause both " "incorrect results and deadlocks.\n\n" "See:\n" "\t- https://github.com/numpy/numpy/blob/2f70544179e24b0ebc0263111f36e6450bbccf94/doc/source/release/1.16.3-notes.rst#numpy-1163-release-notes\n" # noqa: E501 "\t- https://github.com/xianyi/OpenBLAS/issues/1844\n" "\t- https://github.com/numpy/numpy/issues/12394\n" "\t- https://github.com/xianyi/OpenBLAS/pull/1865\n" ) # fmt: on @property def _is_job_single_threaded_per_container(self): exps = self.klio_config.pipeline_options.experiments return "worker_threads=1" in exps @staticmethod def _get_current_numpy_version(): try: import numpy return version.parse(numpy.version.short_version) except ImportError: return None def after_tests(self): if self._is_job_single_threaded_per_container: return if platform.system().lower() != "linux": return numpy_version = self._get_current_numpy_version() if numpy_version is None: return if numpy_version < NumPyBrokenBLASUsage.MINIMUM_NUMPY_VERSION: msg = ( "Multiple threads are used, but a NumPy version older than %s " "was detected. Older versions of NumPy are known to be " "thread-unsafe due to a broken OpenBLAS dependency on Linux." ) % NumPyBrokenBLASUsage.MINIMUM_NUMPY_VERSION self.emit_error(msg) _init = NumPyBrokenBLASUsage
gaybro8777/klio
cli/tests/commands/job/utils/fixtures/expected/transforms.py
<filename>cli/tests/commands/job/utils/fixtures/expected/transforms.py """ Notice: the code below is just an example of what can be done. Feel free to import what's needed, including third-party libraries or other self-written modules. """ import apache_beam as beam from klio.transforms import decorators class HelloKlio(beam.DoFn): """A simple DoFn.""" @decorators.handle_klio def process(self, data): """Main entrypoint to a transform. Any errors raised here (explicitly or not), Klio will catch and log + drop the Klio message. For information on the Klio message, see https://docs.klio.io/en/latest/userguide/pipeline/message.html For information on yielding other information other than ``data``, see https://docs.klio.io/en/latest/userguide/pipeline/state.html Args: data (KlioMessage.data): The data of the Klio message, which contains two attributes: ``data.element`` and ``data.payload``. Yields: KlioMessage.data: the same Klio message data object received. """ element = data.element.decode("utf-8") self._klio.logger.info( "Received '%s' from Pub/Sub topic '%s'" % (element, self._klio.config.job_config.events.inputs[0].topic) ) yield data
gaybro8777/klio
core/src/klio_core/config/_io.py
<reponame>gaybro8777/klio<gh_stars>1-10 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import enum import importlib import logging import attr logger = logging.getLogger("klio") class KlioIODirection(enum.Enum): INPUT = 1 OUTPUT = 2 class KlioIOType(enum.Enum): EVENT = 1 DATA = 2 class KlioJobMode(enum.Enum): STREAMING = 1 BATCH = 2 def supports(*type_directions): """Decorator for subclasses of KlioIOConfig to indicate what IO types and directions the Config supports. In many cases the same config object can be used for configuring either input or output, and likewise event I/O and data I/O. Adding respective `KlioIOType` and `KlioIODirection` values via this decorator will inform Klio about which class to use when parsing config. """ def wrapper(cls): io_types = [t for t in type_directions if isinstance(t, KlioIOType)] io_directions = [ t for t in type_directions if isinstance(t, KlioIODirection) ] # IMPORTANT - new lists must be created here, otherwise all subclasses # end up sharing the same lists cls.SUPPORTED_TYPES = [t for t in cls.SUPPORTED_TYPES] cls.SUPPORTED_DIRECTIONS = [d for d in cls.SUPPORTED_DIRECTIONS] for t in io_types: if t not in cls.SUPPORTED_TYPES: cls.SUPPORTED_TYPES.append(t) for d in io_directions: if d not in cls.SUPPORTED_DIRECTIONS: cls.SUPPORTED_DIRECTIONS.append(d) return cls return wrapper @attr.attrs class KlioIOConfig(object): io_type = attr.attrib(type=KlioIOType) io_direction = attr.attrib(type=KlioIODirection) # these must be filled in by subclasses so Klio knows what supports what SUPPORTED_TYPES = [] SUPPORTED_DIRECTIONS = [] # NOTICE! any new `attr.attrib`s added to this class should be added in # `ATTRIBS_TO_SKIP` below in order to be filtered out when calling # `as_dict`, leaving only `attr.attrib`s related to the specific # config instance ATTRIBS_TO_SKIP = ["io_type", "io_direction"] @classmethod def supports_type(cls, io_type): return io_type in cls.SUPPORTED_TYPES @classmethod def supports_direction(cls, io_direction): return io_direction in cls.SUPPORTED_DIRECTIONS @classmethod def from_dict(cls, config_dict, *args, **kwargs): copy = config_dict.copy() if "type" in copy: del copy["type"] return cls(*args, **copy, **kwargs) def _as_dict(self): """Return a dictionary-representation of the config object. Useful for instantiating objects where the config keys map 1:1 to the object init arguments/keyword arguments, i.e. KlioReadFromBigQuery(**config.job_config.events.inputs[0].as_dict()) """ # filter parameter is used as "to include"/"filter-in", # not "to exclude"/"filter-out" return attr.asdict( self, filter=lambda x, _: x.name not in self.ATTRIBS_TO_SKIP ) def as_dict(self): config_dict = self._as_dict() # since dicts preserve order by default in py3, let's force # type to be first - particularly helpful/useful for dumping # config via `klio job config show` copy = {"type": self.name} copy.update(config_dict) return copy def to_io_kwargs(self): return self._as_dict() @attr.attrs(frozen=True) class IOFlags(object): io_direction = attr.attrib(type=KlioIODirection) io_type = attr.attrib(type=KlioIOType) job_mode = attr.attrib(type=KlioJobMode) @attr.attrs(frozen=True) class KlioEventInput(KlioIOConfig): skip_klio_read = attr.attrib(type=bool) @classmethod def from_dict(cls, config_dict, *args, **kwargs): # work-around with attrs - since attrs will not allow attributes # to be defined without a default after those that have defaults, # we're inserting the default value here if the user doesn't have # it already in their config if "skip_klio_read" not in config_dict: copy = config_dict.copy() copy["skip_klio_read"] = False return super().from_dict(copy, *args, **kwargs) return super().from_dict(config_dict, *args, **kwargs) def to_io_kwargs(self): kwargs = super().to_io_kwargs() kwargs.pop("skip_klio_read", None) return kwargs @attr.attrs(frozen=True) class KlioEventOutput(KlioIOConfig): skip_klio_write = attr.attrib(type=bool) @classmethod def from_dict(cls, config_dict, *args, **kwargs): # work-around with attrs - since attrs will not allow attributes # to be defined without a default after those that have defaults, # we're inserting the default value here if the user doesn't have # it already in their config if "skip_klio_write" not in config_dict: copy = config_dict.copy() copy["skip_klio_write"] = False return super().from_dict(copy, *args, **kwargs) return super().from_dict(config_dict, *args, **kwargs) def to_io_kwargs(self): kwargs = super().to_io_kwargs() kwargs.pop("skip_klio_write", None) return kwargs @attr.attrs(frozen=True) class KlioDataIOConfig(KlioIOConfig): skip_klio_existence_check = attr.attrib(type=bool) @classmethod def from_dict(cls, config_dict, *args, **kwargs): # work-around with attrs - since attrs will not allow attributes # to be defined without a default after those that have defaults, # we're inserting the default value here if the user doesn't have # it already in their config if "skip_klio_existence_check" not in config_dict: copy = config_dict.copy() copy["skip_klio_existence_check"] = False return super().from_dict(copy, *args, **kwargs) return super().from_dict(config_dict, *args, **kwargs) def to_io_kwargs(self): kwargs = super().to_io_kwargs() kwargs.pop("skip_klio_existence_check", None) return kwargs @attr.attrs(frozen=True) class KlioPubSubConfig(object): name = "pubsub" topic = attr.attrib(type=str) @staticmethod def _from_dict(config_dict): copy = config_dict.copy() copy.pop("data_location", None) return copy @supports(KlioIODirection.INPUT, KlioIOType.EVENT) @attr.attrs(frozen=True) class KlioPubSubEventInput(KlioEventInput, KlioPubSubConfig): subscription = attr.attrib(type=str) @classmethod def from_dict(cls, config_dict, *args, **kwargs): config_dict = super()._from_dict(config_dict) # work-around with attrs - since attrs will not allow attributes # to be defined without a default after those that have defaults, # we're inserting the default value here if the user doesn't have # it already in their config if "topic" not in config_dict: config_dict["topic"] = None if "subscription" not in config_dict: config_dict["subscription"] = None return super().from_dict(config_dict, *args, **kwargs) @subscription.validator def __assert_topic_subscription(self, attribute, value): # either topic or subscription is required if not self.topic and not self.subscription: raise ValueError("One of 'topic', 'subscription' required") def to_io_kwargs(self): kwargs = super().to_io_kwargs() # pubsub only allows either topic or subscription, not both. We'll # prioritize subscription if kwargs["topic"] is not None and kwargs["subscription"] is None: kwargs.pop("subscription", None) else: kwargs.pop("topic", None) return kwargs @supports(KlioIODirection.OUTPUT, KlioIOType.EVENT) @attr.attrs(frozen=True) class KlioPubSubEventOutput(KlioEventOutput, KlioPubSubConfig): @classmethod def from_dict(cls, config_dict, *args, **kwargs): config_dict = super()._from_dict(config_dict) return super().from_dict(config_dict, *args, **kwargs) class KlioFileConfig(object): name = "file" @attr.attrs(frozen=True) class KlioReadFileConfig(KlioEventInput, KlioFileConfig): file_pattern = attr.attrib(type=str) @classmethod def from_dict(cls, config_dict, *args, **kwargs): if "location" in config_dict: copy = config_dict.copy() copy["file_pattern"] = copy.pop("location", None) return super().from_dict(copy, *args, **kwargs) try: return super().from_dict(config_dict, *args, **kwargs) except TypeError as e: if "file_pattern" in str(e): raise KeyError( "I/O configuration for `type: file` is missing 'location' " "key." ) def as_dict(self): config_dict = super().as_dict() copy = config_dict.copy() copy["location"] = copy.pop("file_pattern", None) return copy @attr.attrs(frozen=True) class KlioWriteFileConfig(KlioEventOutput, KlioFileConfig): file_path_prefix = attr.attrib(type=str) @classmethod def from_dict(cls, config_dict, *args, **kwargs): if "location" in config_dict: copy = config_dict.copy() copy["file_path_prefix"] = copy.pop("location", None) return super().from_dict(copy, *args, **kwargs) return super().from_dict(config_dict, *args, **kwargs) def as_dict(self): config_dict = super().as_dict() copy = config_dict.copy() copy["location"] = copy.pop("file_path_prefix", None) return copy @supports(KlioIODirection.INPUT, KlioIOType.EVENT) @attr.attrs(frozen=True) class KlioReadFileEventConfig(KlioReadFileConfig): pass @supports(KlioIODirection.OUTPUT, KlioIOType.EVENT) @attr.attrs(frozen=True) class KlioWriteFileEventConfig(KlioWriteFileConfig): pass @attr.attrs(frozen=True) @supports(KlioIODirection.INPUT, KlioIOType.DATA) class KlioFileInputDataConfig(KlioDataIOConfig, KlioFileConfig): location = attr.attrib(type=str) ping = attr.attrib(type=bool, default=False) file_suffix = attr.attrib(type=str, default="") @attr.attrs(frozen=True) @supports(KlioIODirection.OUTPUT, KlioIOType.DATA) class KlioFileOutputDataConfig(KlioDataIOConfig, KlioFileConfig): location = attr.attrib(type=str) file_suffix = attr.attrib(type=str, default="") force = attr.attrib(type=bool, default=False) class KlioAvroConfig(object): name = "avro" @attr.attrs(frozen=True) class KlioReadAvroConfig(KlioAvroConfig): # TODO: Add validation # Potentially could set a general file pattern default, i.e. *.avro, too # but either file_pattern or location must be set file_pattern = attr.attrib(type=str, default=None) location = attr.attrib(type=str, default=None) min_bundle_size = attr.attrib(type=int, default=0) validate = attr.attrib(type=bool, default=True) @supports(KlioIODirection.INPUT, KlioIOType.EVENT) @attr.attrs(frozen=True) class KlioReadAvroEventConfig(KlioEventInput, KlioReadAvroConfig): pass # TODO: integrate into @dsimon's converter logic once his PR#154 is merged def _convert_bigquery_input_coder(coder_str): # direct runner seems to call this multiple times, prob with pickling; # subsequent times seem to call with a None value (not sure why...) if coder_str is None: return coder_path_stems = coder_str.split(".") coder_kls_str = coder_path_stems.pop() module_str = ".".join(coder_path_stems) module = importlib.import_module(module_str) # should raise if not avail return getattr(module, coder_kls_str) # should raise if not avail @attr.attrs(frozen=True) class KlioBigQueryConfig(object): name = "bq" project = attr.attrib(type=str, default=None) dataset = attr.attrib(type=str, default=None) table = attr.attrib(type=str, default=None) @attr.attrs(frozen=True) @supports(KlioIODirection.INPUT, KlioIOType.EVENT) class KlioBigQueryEventInput(KlioEventInput, KlioBigQueryConfig): # Set to true by default to fail early (check table existence before # starting pipeline) (Beam sets it to false by default but suggests it # to be set to true in most scenarios) validate = attr.attrib(type=bool, default=True) # TODO: use @dsimon's converter logic once his PR#154 is merged in coder = attr.attrib( type=str, default=None, converter=_convert_bigquery_input_coder ) kms_key = attr.attrib(type=str, default=None) # Optional, but only applies to project+dataset+table; # Klio handles filtering data after a `SELECT *` query klio_message_columns = attr.attrib(type=list, default=None) # Mutually exclusive with project+dataset+table & klio_message_columns query = attr.attrib(type=str, default=None) use_standard_sql = attr.attrib(type=bool, default=False) flatten_results = attr.attrib(type=bool, default=True) @klio_message_columns.validator def __assert_project_dataset_table(self, attribute, value): has_project_dataset_table = all( [self.project, self.dataset, self.table] ) if value is not None: if not has_project_dataset_table: raise ValueError( "Must include `project`, `dataset` and `table` if " "selecting `columns` for BigQuery." ) @query.validator def __assert_query_or_project_dataset_table(self, attribute, value): has_project_dataset_table = any( [self.project, self.dataset, self.table] ) if value is not None and has_project_dataset_table: raise ValueError( "`query` is mutually exclusive with `project`, `dataset` " "and `table`." ) @classmethod def from_dict(cls, config_dict, *args, **kwargs): if "columns" in config_dict: copy = config_dict.copy() copy["klio_message_columns"] = copy.pop("columns", None) return super().from_dict(copy, *args, **kwargs) return super().from_dict(config_dict, *args, **kwargs) def as_dict(self): config_dict = super().as_dict() copy = config_dict.copy() copy["columns"] = copy.pop("klio_message_columns", None) return copy @attr.attrs(frozen=True) @supports(KlioIODirection.OUTPUT, KlioIOType.EVENT) class KlioBigQueryEventOutput(KlioEventOutput, KlioBigQueryConfig): # schema field is optional; assumes form of # {"fields": [{"name": ...,"type": ..., "mode": ...}, ... ] schema = attr.attrib(type=dict, default=None) create_disposition = attr.attrib(type=str, default="CREATE_IF_NEEDED") write_disposition = attr.attrib(type=str, default="WRITE_EMPTY") @schema.validator def check(self, attribute, value): def valid_field(field_dict): is_dict = isinstance(field_dict, dict) required_keys = [ "name", "type", "mode", ] # are all of these required? return is_dict and all(k in field_dict for k in required_keys) def contains_invalid_fields(field_list): return any(valid_field(f) is False for f in field_list) has_fields = value.get("fields") if not has_fields or contains_invalid_fields(value.get("fields", [])): raise ValueError( "Must be a dict with the key `fields` set to a list of" "dict, each of which have the keys " "`name`, `type`, and `mode`" ) @attr.attrs(frozen=True) class KlioGCSConfig(KlioIOConfig): name = "gcs" location = attr.attrib(type=str) @staticmethod def _from_dict(config_dict): copy = config_dict.copy() copy.pop("topic", None) copy.pop("subscription", None) data_location = copy.pop("data_location", None) copy["location"] = copy.get("location", data_location) return copy @attr.attrs(frozen=True) @supports(KlioIODirection.INPUT, KlioIOType.DATA) class KlioGCSInputDataConfig(KlioDataIOConfig, KlioGCSConfig): file_suffix = attr.attrib(type=str, default="") ping = attr.attrib(type=bool, default=False) @classmethod def from_dict(cls, config_dict, *args, **kwargs): config_dict = super()._from_dict(config_dict) return super().from_dict(config_dict, *args, **kwargs) @attr.attrs(frozen=True) @supports(KlioIODirection.OUTPUT, KlioIOType.DATA) class KlioGCSOutputDataConfig(KlioDataIOConfig, KlioGCSConfig): file_suffix = attr.attrib(type=str, default="") force = attr.attrib(type=bool, default=False) @classmethod def from_dict(cls, config_dict, *args, **kwargs): config_dict = super()._from_dict(config_dict) return super().from_dict(config_dict, *args, **kwargs)
gaybro8777/klio
lib/tests/unit/transforms/test_decorators.py
<filename>lib/tests/unit/transforms/test_decorators.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_core.proto import klio_pb2 from klio.transforms import decorators from klio.utils import _thread_limiter @pytest.fixture def kmsg(): msg = klio_pb2.KlioMessage() msg.data.element = b"3l3m3nt" return msg def test_retry(kmsg, mocker, mock_config): mock_function = mocker.Mock() @decorators._handle_klio @decorators._retry(tries=2) def func(*args, **kwargs): mock_function(*args, **kwargs) raise Exception("fuu") func(kmsg.SerializeToString()) assert 2 == mock_function.call_count def test_retry_custom_catch(kmsg, mocker, mock_config): # Assert retry on custom exception class CustomCatchException(Exception): pass mock_function = mocker.Mock() raise_custom = True @decorators._handle_klio @decorators._retry(tries=3, exception=CustomCatchException) def func(*args, **kwargs): mock_function(*args, **kwargs) if raise_custom: raise CustomCatchException("custom fuu") raise Exception("fuu") func(kmsg.SerializeToString()) assert 3 == mock_function.call_count # Assert retry on multiple custom exceptions class AnotherCustomCatchException(Exception): pass exc_tuple = (CustomCatchException, AnotherCustomCatchException) @decorators._handle_klio @decorators._retry(tries=3, exception=exc_tuple) def func(*args, **kwargs): mock_function(*args, **kwargs) if raise_custom: raise CustomCatchException("custom fuu") raise Exception("fuu") mock_function.reset_mock() func(kmsg.SerializeToString()) assert 3 == mock_function.call_count # Assert retries do not happen for an exception that isn't provided mock_function.reset_mock() raise_custom = False func(kmsg.SerializeToString()) assert 1 == mock_function.call_count def test_retry_raises_runtime_parents(kmsg, mocker, mock_config): # Need to call @retry with parens with pytest.raises(RuntimeError): @decorators._handle_klio @decorators.retry def func(*args, **kwargs): pass func(kmsg.SerializeToString()) @pytest.mark.parametrize( "invalid_tries", (-2, 0.5, "1", {"a": "dict"}, ["a", "list"], lambda x: x) ) def test_retry_raises_runtime_invalid_tries( invalid_tries, kmsg, mocker, mock_config ): # Assert `tries` as a valid integer with pytest.raises(RuntimeError): @decorators._handle_klio @decorators._retry(tries=invalid_tries) def func(*args, **kwargs): pass func(kmsg.SerializeToString()) @pytest.mark.parametrize( "invalid_delay", (-2, {"a": "dict"}, ["a", "list"], lambda x: x) ) def test_retry_raises_runtime_invalid_delay( invalid_delay, kmsg, mocker, mock_config ): # Assert `delay` as a valid int/float with pytest.raises(RuntimeError): @decorators._handle_klio @decorators._retry(tries=1, delay=invalid_delay) def func(*args, **kwargs): pass func(kmsg.SerializeToString()) @pytest.mark.parametrize( "max_thread_count,patch_str", ( (None, "threading.BoundedSemaphore"), (_thread_limiter.ThreadLimit.DEFAULT, "threading.BoundedSemaphore"), (_thread_limiter.ThreadLimit.NONE, "_DummySemaphore"), ), ) def test_thread_limiting( max_thread_count, patch_str, kmsg, mock_config, mocker, monkeypatch ): mock_function = mocker.Mock() mock_semaphore = mocker.Mock() monkeypatch.setattr( f"klio.utils._thread_limiter.{patch_str}", mock_semaphore ) kwargs = {} if max_thread_count is not None: kwargs["max_thread_count"] = max_thread_count @decorators._handle_klio(**kwargs) def func(*args, **kwargs): mock_function(*args, **kwargs) return func(kmsg.SerializeToString()) assert 1 == mock_function.call_count mock_semaphore.return_value.acquire.assert_called_once_with() mock_semaphore.return_value.release.assert_called_once_with() def test_thread_limiting_custom_limiter( kmsg, mock_config, mocker, monkeypatch ): mock_function = mocker.Mock() mock_semaphore = mocker.Mock() limiter = _thread_limiter.ThreadLimiter(max_thread_count=1) monkeypatch.setattr(limiter, "_semaphore", mock_semaphore) @decorators._handle_klio(thread_limiter=limiter) def func(*args, **kwargs): mock_function(*args, **kwargs) return func(kmsg.SerializeToString()) assert 1 == mock_function.call_count mock_semaphore.acquire.assert_called_once_with() mock_semaphore.release.assert_called_once_with() def test_thread_limiting_raises_mutex_args(kmsg, mocker, mock_config): limiter = _thread_limiter.ThreadLimiter(max_thread_count=1) with pytest.raises(RuntimeError): @decorators._handle_klio(max_thread_count=1, thread_limiter=limiter) def func(*args, **kwargs): pass func(kmsg.SerializeToString()) def test_thread_limiting_raises_invalid_limiter(kmsg, mocker, mock_config): limiter = "not an instance of ThreadLimiter" with pytest.raises(RuntimeError): @decorators._handle_klio(thread_limiter=limiter) def func(*args, **kwargs): pass func(kmsg.SerializeToString()) @pytest.mark.parametrize( "invalid_max_thread_count", (-2, {"a": "dict"}, ["a", "list"]) ) def test_thread_limiting_raises_invalid_max( invalid_max_thread_count, kmsg, mocker, mock_config ): with pytest.raises(RuntimeError): @decorators._handle_klio(max_thread_count=invalid_max_thread_count) def func(*args, **kwargs): pass func(kmsg.SerializeToString())
gaybro8777/klio
lib/src/klio/transforms/core.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import __main__ import glob import logging import os import threading import yaml from klio_core import config from klio_core.proto import klio_pb2 from klio.metrics import client as metrics_client from klio.metrics import logger as metrics_logger from klio.metrics import stackdriver class RunConfig(object): _thread_local = threading.local() @classmethod def _load_config_from_file(cls): # [Klio v2] this may get expensive, to always be reading config # from a file. Can this be replaced by something in memory # that's also globally accessible? klio_job_file = "/usr/src/config/.effective-klio-job.yaml" # for backwards compatibility, and user is using setup.py and we # have to find it somewhere... if not os.path.exists(klio_job_file): # use iterator so we don't waste time searching everywhere upfront files = glob.iglob("/usr/**/klio-job.yaml", recursive=True) for f in files: klio_job_file = f # only grab the first one break with open(klio_job_file, "r") as f: all_config_data = yaml.safe_load(f) return config.KlioConfig(all_config_data) # NOTE: for now this approach is not being used (and may be removed in the # future) @classmethod def _get_via_main_session(cls): if hasattr(__main__, "run_config"): return __main__.run_config else: raise Exception( "Attempt to access RunConfig before it was set. This likely" " means something was imported before RunConfig was set." ) @classmethod def _get_via_thread_local(cls): klio_config = getattr(cls._thread_local, "klio_config", None) if not klio_config: cls._thread_local.klio_config = cls._load_config_from_file() return cls._thread_local.klio_config @classmethod def get(cls): return cls._get_via_thread_local() @classmethod def set(cls, config): __main__.run_config = config class KlioContext(object): """Context related to the currently running job. Available to transforms via one of the :ref:`KlioContext decorators <klio-context-decorators>`. """ _thread_local = threading.local() def __init__(self): self.__transform_name = None def _create_klio_job_obj(self): klio_job = klio_pb2.KlioJob() klio_job.job_name = self.config.job_name klio_job.gcp_project = self.config.pipeline_options.project klio_job_str = klio_job.SerializeToString() return klio_job_str def _get_metrics_registry(self): clients = [] use_logger, use_stackdriver = None, None metrics_config = self.config.job_config.metrics # use_logger and use_stackdriver could be False (turn off), # None (use default config), or a dict of configured values use_logger = metrics_config.get("logger") use_stackdriver = metrics_config.get("stackdriver_logger") # TODO: set runner in OS environment (via klio-exec), since # the runner defined in config could be overwritten via # `--direct-runner`. # i.e.: runner = os.getenv("BEAM_RUNNER", "").lower() runner = self.config.pipeline_options.runner if "dataflow" in runner.lower(): # Must explicitly compare to `False` since `None` could be # the user accepting default config. # If explicitly false, then just disable logger underneath SD if use_stackdriver is not False: sd_client = stackdriver.StackdriverLogMetricsClient( self.config ) clients.append(sd_client) else: # if use_stackdriver is explicitly false, then make sure # logger client is disabled since the stackdriver client # inherits the logger client use_logger = False if not len(clients): # setup default client disabled = False # User might disable the logger, but we still need a relay # client if all other relay clients are disabled. This allows # folks to silence metrics but not need to remove code that # interacts with `_klio.metrics`. # Must explicitly compare to `False` since `None` could be # the user accepting default config if use_logger is False: disabled = True logger_client = metrics_logger.MetricsLoggerClient( self.config, disabled=disabled ) clients.append(logger_client) return metrics_client.MetricsRegistry( clients, transform_name=self._transform_name ) @property def config(self): """A ``KlioConfig`` instance representing the job's configuration.""" return RunConfig.get() @property def job(self): """An instance of :ref:`kliojob` of the current job.""" klio_job = getattr(self._thread_local, "klio_job", None) if not klio_job: self._thread_local.klio_job = self._create_klio_job_obj() return self._thread_local.klio_job @property def logger(self): """A namespaced logger. Equivalent to ``logging.getLogger("klio")``. """ klio_logger = getattr(self._thread_local, "klio_logger", None) if not klio_logger: self._thread_local.klio_logger = logging.getLogger("klio") return self._thread_local.klio_logger @property def metrics(self): """A metrics registry instance. See :ref:`metrics <metrics>` for more information.""" metrics_registry = getattr(self._thread_local, "klio_metrics", None) if not metrics_registry: self._thread_local.klio_metrics = self._get_metrics_registry() return self._thread_local.klio_metrics # <-- private/internal attributes --> @property def _transform_name(self): return self.__transform_name @_transform_name.setter def _transform_name(self, name): self.__transform_name = name
gaybro8777/klio
cli/tests/commands/job/test_test.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_cli.commands.job import test as test_job @pytest.fixture def mock_os_environ(mocker): patch = {"HOME": "/home", "USER": "cookiemonster"} return mocker.patch.dict("os.environ", patch) @pytest.fixture def test_pipeline(mock_os_environ): return test_job.TestPipeline("job/dir", "klio_config", "docker_config") def test_get_environment(test_pipeline): gcreds = "/usr/gcloud/application_default_credentials.json" exp_envs = { "PYTHONPATH": "/usr/src/app", "GOOGLE_APPLICATION_CREDENTIALS": gcreds, "USER": "cookiemonster", "KLIO_TEST_MODE": "true", } assert exp_envs == test_pipeline._get_environment() def test_get_command(test_pipeline): assert ["test", "py", "args"] == test_pipeline._get_command(["py", "args"]) def test_requires_config_setting(test_pipeline): assert not test_pipeline.requires_config_file
gaybro8777/klio
core/src/klio_core/dataflow.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Module for interacting with the `Dataflow REST API <https://cloud.google.com/dataflow/docs/reference/rest>`_.""" import functools import logging import random from googleapiclient import discovery from klio_core import utils from klio_core import variables class DataflowClient(object): """Client to interact with Dataflow REST API. Args: api_version (str): Version of Dataflow REST API. Defaults to ``v1b3``. """ def __init__(self, api_version=None): _api_version = api_version or "v1b3" self.client = discovery.build("dataflow", _api_version) self.logger = logging.getLogger("klio") def find_job_by_name(self, job_name, gcp_project, region=None): """Search Dataflow for a job given its name and GCP project. Args: job_name (str): Name of Dataflow job. gcp_project (str): GCP project in which to search. region (str): Region in which to search. Defaults to searching all regions in :attr:`klio_core.variables.DATAFLOW_REGIONS`. Returns: dict or None: If found, ``dict`` of job summary results. Otherwise, ``None``. """ if not region: regions = variables.DATAFLOW_REGIONS else: regions = (region,) base_request = self.client.projects().locations().jobs() all_matching_jobs = [] # TODO: no batch requesting from Google's side, but should add # threading to send multiple requests concurrently. @lynn for region in regions: # Note: the parameter `view="JOB_VIEW_ALL"` does not return # the same information in this `.list()` call as it # does in the `.get()` call in `get_job_detail` below. request = base_request.list( projectId=gcp_project, location=region, filter="ACTIVE" ) try: response = request.execute() # general catch all since the handling would be the same no matter # of the exception except Exception as e: self.logger.warning( "Error listing active jobs in project '%s' in region '%s':" " %s" % (gcp_project, region, e) ) continue job_results = response.get("jobs", []) if job_results: for result in job_results: if result["name"] == job_name: all_matching_jobs.append(result) # Note: job names are unique within regions, but not across # regions :grimace: if len(all_matching_jobs) > 1: self.logger.info( "More than one parent job found for job name '%s' under " "project '%s'. Selecting one at random." ) return random.choice(all_matching_jobs) if all_matching_jobs: return all_matching_jobs[0] def get_job_detail(self, job_name, gcp_project, region=None): """Get verbose job detail given a job name. Args: job_name (str): Name of Dataflow job. gcp_project (str): GCP project in which to search. region (str): Region in which to search. Defaults to searching all regions in :attr:`klio_core.variables.DATAFLOW_REGIONS`. Returns: dict or None: If found, ``dict`` of detailed job results. Otherwise, ``None``. """ basic_job = self.find_job_by_name(job_name, gcp_project, region) if not basic_job: return None job_id = basic_job["id"] job_location = basic_job["location"] request = ( self.client.projects() .locations() .jobs() .get( projectId=gcp_project, location=job_location, jobId=job_id, view="JOB_VIEW_ALL", ) ) try: response = request.execute() # general catch all since the handling would be the same no matter # of the exception except Exception as e: self.logger.warning( "Error getting job detail for '%s' in project '%s' in " "region '%s': %s" % (job_name, gcp_project, job_location, e) ) return return response def get_job_input_topic(self, job_name, gcp_project, region=None): """Get input topic of a particular job. Args: job_name (str): Name of Dataflow job. gcp_project (str): GCP project in which to search. region (str): Region in which to search. Defaults to searching all regions in :attr:`klio_core.variables.DATAFLOW_REGIONS`. Returns: str or None: If found, input topic of job. Otherwise, ``None``. """ job_info = self.get_job_detail(job_name, gcp_project, region=None) if not job_info: return None read_pubsub_user_name = "ReadFromPubSub/Read" for step in job_info.get("steps", []): if step.get("kind") == "ParallelRead": props = step.get("properties", {}) user_name = props.get("user_name", {}) if user_name.get("value") == read_pubsub_user_name: # TODO: support multiple input topics; will need to # see how Google's response json renders it. @lynn return props.get("pubsub_topic", {}).get("value") def get_dataflow_client(api_version=None): """Get an initialized :class:`DataflowClient`. This function will first check if there is an already initialized client in the global namespace. Otherwise, initialize one then set it in the global namespace to avoid redundant initialization. Args: api_version (str): Version of Dataflow REST API. Defaults to ``v1b3``. Returns: DataflowClient: A client to interact with the Dataflow REST API. """ if not api_version: api_version = "v1b3" key = "dataflow_client_{}".format(api_version) initializer = functools.partial(DataflowClient, api_version) return utils.get_or_initialize_global(key, initializer)
gaybro8777/klio
cli/tests/commands/job/utils/fixtures/expected/run.py
<gh_stars>100-1000 """ Notice: the code within `run` is just an example of what can be done. Feel free to import what's needed, including third-party libraries or other self-written modules. """ import apache_beam as beam import transforms def run(input_pcol, config): """REQUIRED: Main entrypoint in running a job's transform(s). Any Beam transforms that need to happen after a message is consumed from PubSub from an upstream job, and before publishing a message to a downstream job (if needed/configured). Args: input_pcol: A Beam PCollection returned from ``beam.io.ReadFromPubSub``. config (klio.KlioConfig): Configuration as defined in ``klio-job.yaml``. Returns: apache_beam.pvalue.PCollection: PCollection that will be passed to the output transform for the configured event output (if any). """ output_pcol = input_pcol | beam.ParDo(transforms.HelloKlio()) # <-- multiple Klio-based ParDo transforms are supported here --> return output_pcol
gaybro8777/klio
cli/tests/utils/test_docker_utils.py
<reponame>gaybro8777/klio # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import os import docker import pytest from docker import errors as docker_errors from requests import exceptions as requests_exceptions from klio_cli.utils import docker_utils @pytest.fixture def mock_docker_client(mocker): return mocker.Mock() @pytest.fixture def mock_json_loads(mocker): return mocker.patch.object(json, "loads") @pytest.fixture def mock_docker_api_client(mocker): return mocker.patch.object(docker, "APIClient") @pytest.fixture def mock_client(mocker): mock_client = mocker.Mock() return mock_client @pytest.fixture def mock_os_getcwd(monkeypatch): test_dir = "/test/dir" monkeypatch.setattr(os, "getcwd", lambda: test_dir) return "/test/dir" @pytest.fixture def mock_os_environ(mocker): return mocker.patch.dict("os.environ", {"HOME": "/home"}) def test_check_docker_connection(mock_docker_client): docker_utils.check_docker_connection(mock_docker_client) mock_docker_client.ping.assert_called_once_with() @pytest.mark.parametrize( "error", [docker_errors.APIError("msg"), requests_exceptions.ConnectionError()], ) def test_check_docker_connection_with_errors( mock_docker_client, error, caplog ): mock_docker_client.ping.side_effect = error with pytest.raises(SystemExit): docker_utils.check_docker_connection(mock_docker_client) assert 2 == len(caplog.records) @pytest.mark.parametrize("path_exists", [True, False]) def test_check_dockerfile_present(monkeypatch, path_exists, caplog): monkeypatch.setattr(os.path, "exists", lambda x: path_exists) job_dir = "my/job/dir" if not path_exists: with pytest.raises(SystemExit): docker_utils.check_dockerfile_present(job_dir) assert 2 == len(caplog.records) else: docker_utils.check_dockerfile_present(job_dir) def test_docker_image_exists(mocker, mock_client): mock_images = mocker.Mock() mock_images.get = mocker.Mock() images_mock = mocker.PropertyMock(return_value=mock_images) mock_client.images = images_mock image_tag = "gcr.io/sigint/test-image-name" exists = docker_utils.docker_image_exists(image_tag, mock_client) mock_client.images.get.assert_called_once_with(image_tag) assert exists def test_docker_image_exists_with_image_not_found_error(mocker, mock_client): mock_images = mocker.Mock() mock_images.get = mocker.Mock() images_mock = mocker.PropertyMock(return_value=mock_images) mock_client.images = images_mock image_tag = "gcr.io/sigint/test-image-name" mock_client.images.get.side_effect = docker_errors.ImageNotFound( "pew", "pew" ) exists = docker_utils.docker_image_exists(image_tag, mock_client) assert not exists def test_docker_image_exists_with_api_error(mocker, mock_client, caplog): mock_images = mocker.Mock() mock_images.get = mocker.Mock() images_mock = mocker.PropertyMock(return_value=mock_images) mock_client.images = images_mock image_tag = "gcr.io/sigint/test-image-name" mock_client.images.get.side_effect = docker_errors.APIError("pew") with pytest.raises(SystemExit): docker_utils.docker_image_exists(image_tag, mock_client) assert 1 == len(caplog.records) @pytest.mark.parametrize( "config_file,exp_config_file", ((None, "klio-job.yaml"), ("klio-job2.yaml", "klio-job2.yaml"),), ) def test_build_docker_image( config_file, exp_config_file, mocker, mock_json_loads, mock_docker_api_client, caplog, ): mock_api_client = mocker.Mock() mock_docker_api_client.return_value = mock_api_client mock_api_client.build.return_value = ( item for item in (b"BYTELOGS", "LOGS", "", '{"stream":"\\n"}') ) mock_json_loads.return_value = {"stream": "SUCCESS"} job_dir = "/test/dir/jobs/test_run_job" image_name = "gcr.io/sigint/test-image-name" image_tag = "foobar" image_name_and_tag = "{}:{}".format(image_name, image_tag) build_flag = { "path": job_dir, "tag": image_name_and_tag, "rm": True, "buildargs": {"tag": image_tag, "KLIO_CONFIG": exp_config_file}, } docker_utils.build_docker_image( job_dir, image_name, image_tag, config_file ) docker.APIClient.assert_called_once_with( base_url="unix://var/run/docker.sock" ) mock_api_client.build.assert_called_once_with(**build_flag) assert 2 == len(caplog.records) def test_build_docker_image_with_errors( mocker, mock_json_loads, mock_docker_api_client, caplog ): mock_api_client = mocker.Mock() mock_docker_api_client.return_value = mock_api_client mock_api_client.build.return_value = "LOGS" mock_json_loads.return_value = { "error": "FAILURE", "errorDetail": {"message": "FAILURE"}, } job_dir = "/test/dir/jobs/test_run_job" image_name = "gcr.io/sigint/test-image-name" image_tag = "foobar" image_name_and_tag = "{}:{}".format(image_name, image_tag) build_flag = { "path": job_dir, "tag": image_name_and_tag, "rm": True, "buildargs": {"tag": image_tag, "KLIO_CONFIG": "klio-job.yaml"}, } with pytest.raises(SystemExit): docker_utils.build_docker_image(job_dir, image_name, image_tag) docker.APIClient.assert_called_once_with( base_url="unix://var/run/docker.sock" ) mock_api_client.build.assert_called_once_with(**build_flag) assert 3 == len(caplog.records) def test_push_image_to_gcr(mocker, capsys): image_name = "my.img.repo" tag = "my-tag" mock_client = mocker.Mock() mock_push = mocker.Mock() mock_client.images.push = mock_push mock_push.return_value = [ b'{"status": "foo", "progress": {}}', b'{"id": "layerid1", "status": "bar", "progress": "some progress"}', ( b'{"id": "layerid1", "status": "baz", "progress": "some more ' b'progress"}\r\n{"id": "layerid2", "status": "foo", "progress": ' b'"some other progress"}' ), ] exp_stdout = ( "foo{}\nlayerid1: barsome progress\n\x1b[1Flayerid1: " "bazsome more progress\u001b[0K\nlayerid2: foosome other progress\n" ) docker_utils.push_image_to_gcr(image_name, tag, mock_client) mock_push.assert_called_once_with( repository=image_name, tag=tag, stream=True ) captured = capsys.readouterr() assert exp_stdout == captured.out @pytest.mark.parametrize( "docker_image_exists, docker_image_build_raises, check_docker, " "check_dockerfile, force_build", [ (True, False, False, False, True), (True, False, False, False, False), (False, False, False, False, False), (False, True, False, False, True), (False, True, False, False, False), (False, False, True, False, False), (False, False, False, True, False), ], ) def test_get_docker_image_client( mocker, docker_image_exists, docker_image_build_raises, check_docker, check_dockerfile, force_build, caplog, ): mock_docker_from_env = mocker.patch.object(docker, "from_env") mock_check_docker_connection = mocker.patch.object( docker_utils, "check_docker_connection" ) mock_check_dockerfile_present = mocker.patch.object( docker_utils, "check_dockerfile_present" ) mock_docker_image_exists = mocker.patch.object( docker_utils, "docker_image_exists" ) mock_build_docker_image = mocker.patch.object( docker_utils, "build_docker_image" ) job_dir = "/test/dir/jobs/test_run_job" image_name = "gcr.io/sigint/test-image-name" image_tag = "foobar" image_name_and_tag = "{}:{}".format(image_name, image_tag) client = 'A "Docker Client" object.' mock_docker_from_env.return_value = client mock_docker_image_exists.return_value = docker_image_exists if docker_image_build_raises or check_docker or check_dockerfile: mock_check_docker_connection.side_effect = ( SystemExit(1) if check_docker else None ) mock_check_dockerfile_present.side_effect = ( SystemExit(1) if check_dockerfile else None ) mock_build_docker_image.side_effect = ( SystemExit(1) if docker_image_build_raises else None ) with pytest.raises(SystemExit) as _exec: docker_utils.get_docker_image_client( job_dir, image_tag, image_name, force_build ) assert 1 == _exec.value.code else: actual_image, actual_client = docker_utils.get_docker_image_client( job_dir, image_tag, image_name, force_build ) mock_check_docker_connection.assert_called_once_with(client) mock_check_dockerfile_present.assert_called_once_with(job_dir) if not docker_image_exists or force_build: assert mock_build_docker_image.called else: assert 1 == len(caplog.records) assert image_name_and_tag == actual_image assert client == actual_client
gaybro8777/klio
exec/src/klio_exec/commands/utils/cpu_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import sys import time import line_profiler as lp import psutil from klio_exec.commands.utils import wrappers class KLineProfiler(wrappers.KLineProfilerMixin, lp.LineProfiler): pass # adapted from memory_profiler's approach: # github.com/pythonprofilers/memory_profiler/blob/master/memory_profiler.py def get_cpu_usage(proc, interval=None, stream=None): """Measure the cpu % usage at an interval.""" if not interval: interval = 0.1 if not stream: stream = sys.stdout line_count = 0 while True: cpu_usage = psutil.cpu_percent() timestamp = time.time() # Same format as memory_profiler stream.write("CPU {:.1f} {:.4f}\n".format(cpu_usage, timestamp)) time.sleep(interval) line_count += 1 # flush every 50 lines. Make 'tail -f' usable on profile file if line_count > 50: line_count = 0 stream.flush() if proc.poll() is not None: break
gaybro8777/klio
exec/src/klio_exec/options.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import click ##### # options for `klioexec run` ##### def config_file(func): return click.option( "-c", "--config-file", type=click.Path(exists=True, resolve_path=True, dir_okay=False), help=( "Path to config filename. Defaults to `klio-job.yaml` in the " "current working directory. " ), )(func) def blocking(func): return click.option( "--blocking/--no-blocking", default=None, is_flag=True, help="Wait for Dataflow job to finish before returning", )(func) ##### # options for `klioexec profile` ##### def input_file(func): # mutually exclusive with entity ID click args return click.option( "-i", "--input-file", type=click.Path(exists=True, dir_okay=False, readable=True), help=( "File of entity IDs (separated by a new line character) with " "which to profile a Klio job." ), required=False, )(func) def output_file(func): return click.option( "-o", "--output-file", type=click.Path(exists=False, dir_okay=False, writable=True), default=None, show_default="stdout", help="Output file for results.", )(func)
gaybro8777/klio
cli/tests/commands/image/test_build.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_core import config from klio_cli.commands.image import build as build_image @pytest.mark.parametrize( "conf_file,exp_image_tag", ( (None, "v1"), ("klio-job2.yaml", "v1-klio-job2.yaml"), ("bar/klio-job2.yaml", "v1-klio-job2.yaml"), ), ) def test_build(conf_file, exp_image_tag, mocker, monkeypatch): mock_docker = mocker.Mock() mock_client = mocker.Mock() mock_docker.from_env.return_value = mock_client monkeypatch.setattr(build_image, "docker", mock_docker) mock_docker_utils = mocker.Mock() monkeypatch.setattr(build_image, "docker_utils", mock_docker_utils) mock_config = { "job_name": "test-job", "version": 1, "pipeline_options": { "worker_harness_container_image": "gcr.register.io/squad/feature" }, "job_config": { "inputs": [ { "topic": "foo-topic", "subscription": "foo-sub", "data_location": "foo-input-location", } ], "outputs": [ { "topic": "foo-topic-output", "data_location": "foo-output-location", } ], }, } conf_obj = config.KlioConfig(mock_config) job_dir = "jerbs" image_tag = "v1" build_image.build(job_dir, conf_obj, conf_file, image_tag) mock_docker.from_env.assert_called_once_with() mock_docker_utils.check_docker_connection.assert_called_once_with( mock_client ) mock_docker_utils.check_dockerfile_present.assert_called_once_with(job_dir) mock_docker_utils.build_docker_image.assert_called_once_with( job_dir, conf_obj.pipeline_options.worker_harness_container_image, exp_image_tag, conf_file, )
gaybro8777/klio
cli/tests/utils/test_stackdriver_utils.py
<filename>cli/tests/utils/test_stackdriver_utils.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import collections import pytest from klio_cli.utils import stackdriver_utils as sd_utils stackdriver_group = collections.namedtuple( "StackdriverGroup", ["name", "display_name"] ) @pytest.fixture def sd_client(mocker, monkeypatch): client = mocker.Mock() monkeypatch.setattr(sd_utils.monitoring, "GroupServiceClient", client) return client.return_value @pytest.fixture def groups(): base_display_name = "test-job-name-{}-test-region-klio-dashboard" base_name = "projects/test-gcp-project/groups/123456789{}" ret_groups = [] for i in range(3): ret_groups.append( stackdriver_group( name=base_name.format(i), display_name=base_display_name.format(i), ) ) return ret_groups def test_generate_group_meta(): assert sd_utils.generate_group_meta("a", "b", "c") == ( "projects/a", "b-c-klio-dashboard", ) @pytest.mark.parametrize( "job_name,exp_ret", [("test-job-name-1", True), ("test-job-name-4", False)] ) def test_get_stackdriver_group_url(sd_client, groups, job_name, exp_ret): sd_client.list_groups.return_value = groups ret_url = sd_utils.get_stackdriver_group_url( "test-gcp-project", job_name, "test-region" ) exp_url = None if exp_ret: exp_url = ( "https://app.google.stackdriver.com/groups/1234567891/" "test-job-name-1-test-region-klio-dashboard?" "project=test-gcp-project" ) assert exp_url == ret_url sd_client.list_groups.assert_called_once_with( request={"name": "projects/test-gcp-project"} ) def test_get_stackdriver_group_url_raises(sd_client): sd_client.list_groups.side_effect = Exception("fuuuuu") with pytest.raises(Exception): sd_utils.get_stackdriver_group_url( "test-gcp-project", "test-job-name-1", "test-region" ) def test_create_stackdriver_group(sd_client, groups, caplog): sd_client.create_group.return_value = groups[0] ret_url = sd_utils.create_stackdriver_group( "test-gcp-project", "test-job-name-0", "test-region" ) exp_url = ( "https://app.google.stackdriver.com/groups/1234567890/" "test-job-name-0-test-region-klio-dashboard?" "project=test-gcp-project" ) assert exp_url == ret_url group_arg = { "display_name": "test-job-name-0-test-region-klio-dashboard", "filter": "resource.metadata.name=starts_with(test-job-name-0)", } sd_client.create_group.assert_called_once_with( request={"name": "projects/test-gcp-project", "group": group_arg} ) assert 1 == len(caplog.records) def test_create_stackdriver_group_errors(sd_client, caplog): sd_client.create_group.side_effect = Exception("fuuuu") ret_url = sd_utils.create_stackdriver_group( "test-gcp-project", "test-job-name-0", "test-region" ) assert not ret_url assert 1 == len(caplog.records) @pytest.mark.parametrize( "job_name,side_effect,exp_log_level", ( ("test-job-name-1", None, "INFO"), ("test-job-name-1", Exception("meow"), "ERROR"), ("kittehs", None, "WARNING"), ), ) def test_delete_stackdriver_group( job_name, side_effect, exp_log_level, groups, sd_client, caplog ): sd_client.list_groups.return_value = groups sd_client.list_groups.side_effect = side_effect sd_utils.delete_stackdriver_group( "test-gcp-project", job_name, "test-region" ) assert 1 == len(caplog.records) assert exp_log_level == caplog.records[0].levelname
gaybro8777/klio
lib/src/klio/metrics/logger.py
<reponame>gaybro8777/klio<filename>lib/src/klio/metrics/logger.py<gh_stars>100-1000 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ Klio ships with a default :class:`klio.metrics.base.AbstractRelayClient` implementation, which outputs metrics via the standard library ``logging`` module through the :class:`MetricsLoggerClient` below. This implementation is used by default if no other metrics consumers are configured. It must be explicitly turned off. The default configuration in ``klio-info.yaml`` can be overwritten: .. code-block:: yaml job_config: metrics: logger: # Logged metrics are emitted at the `debug` level by default. level: info # Default timer unit is ns/nanoseconds; available # options include `s` or `seconds`, `ms` or `milliseconds`, # `us` or `microseconds`, and `ns` or `nanoseconds`. timer_unit: s To turn off logging-based metrics: .. code-block:: yaml job_config metrics: logger: false """ import logging import threading from klio.metrics import base TIMER_UNIT_MAP = { "nanoseconds": "ns", "microseconds": "us", "milliseconds": "ms", "seconds": "s", "ns": "ns", "us": "us", "ms": "ms", "s": "s", } """Map of supported measurement units to shorthand for :class:`LoggerTimer`. """ class MetricsLoggerClient(base.AbstractRelayClient): """Logging client for transform metrics. Intended to be instantiated by :class:`klio.metrics.client.MetricsRegistry` and not by itself. Args: klio_config (klio_core.config.KlioConfig): the job's configuration. disabled (bool): whether or not to disable the Python ``logger`` Default: ``False``. """ RELAY_CLIENT_NAME = "logger" DEFAULT_LEVEL = logging.DEBUG DEFAULT_TIME_UNIT = "ns" _thread_local = threading.local() def __init__(self, klio_config, disabled=False): super(MetricsLoggerClient, self).__init__(klio_config) self.logger_config = self.klio_config.job_config.metrics.get( "logger", {} ) self.disabled = disabled self.log_level = self._set_log_level() self.timer_unit = self._set_timer_unit() def _set_log_level(self): log_level = MetricsLoggerClient.DEFAULT_LEVEL if isinstance(self.logger_config, dict): log_level_str = self.logger_config.get("level") if log_level_str: log_level = getattr(logging, log_level_str.upper(), log_level) return log_level def _set_timer_unit(self): timer_unit = MetricsLoggerClient.DEFAULT_TIME_UNIT if isinstance(self.logger_config, dict): _timer_unit = self.logger_config.get("timer_unit") if _timer_unit: timer_unit = TIMER_UNIT_MAP.get(_timer_unit, timer_unit) return timer_unit @property def logger(self): """Python logger associated with the job which this client will use to emit metrics. """ klio_metrics_logger = getattr( self._thread_local, "klio_metrics_logger", None ) if not klio_metrics_logger: logger = logging.getLogger("klio.metrics") logger.disabled = self.disabled self._thread_local.klio_metrics_logger = logger return self._thread_local.klio_metrics_logger def unmarshal(self, metric): """Return a dict-representation of a given metric. Args: metric (LoggerMetric): logger-specific metrics object Returns: dict(str, str): metric data """ return { "name": metric.name, "value": metric.value, "transform": metric.transform, "tags": metric.tags, } def emit(self, metric): """Log a given metric. Args: metric (LoggerMetric): logger-specific metrics object """ metric_data = self.unmarshal(metric) self.logger.log( self.log_level, metric.DEFAULT_LOG_FORMAT.format(**metric_data) ) def counter(self, name, value=0, transform=None, tags=None, **kwargs): """Create a :class:`LoggerCounter` object. Args: name (str): name of counter value (int): starting value of counter; defaults to 0 transform (str): transform the counter is associated with tags (dict): any tags of additional contextual information to associate with the counter Returns: LoggerCounter: a log-based counter """ return LoggerCounter( name=name, value=value, transform=transform, tags=tags ) def gauge(self, name, value=0, transform=None, tags=None, **kwargs): """Create a :class:`LoggerGauge` object. Args: name (str): name of gauge value (int): starting value of gauge; defaults to 0 transform (str): transform the gauge is associated with tags (dict): any tags of additional contextual information to associate with the gauge Returns: LoggerGauge: a log-based gauge """ return LoggerGauge( name=name, value=value, transform=transform, tags=tags ) def timer( self, name, value=0, transform=None, tags=None, timer_unit=None, **kwargs ): """Create a :class:`LoggerTimer` object. Args: name (str): name of timer value (int): starting value of timer; defaults to 0 transform (str): transform the timer is associated with tags (dict): any tags of additional contextual information to associate with the timer timer_unit (str): timer unit; defaults to configured value in `klio-job.yaml`, or "ns". See module-level docs of `klio.metrics.logger` for supported values. Returns: LoggerTimer: a log-based timer """ if timer_unit: # Note: this should probably have better validation if it does # not recognize the unit given. Instead of erroring out, we'll # just use the default (@lynn) timer_unit = TIMER_UNIT_MAP.get(timer_unit, self.timer_unit) else: timer_unit = self.timer_unit return LoggerTimer( name=name, value=value, transform=transform, tags=tags, timer_unit=timer_unit, ) class LoggerMetric(base.BaseMetric): """Base metric type for loggers. Args: name (str): name of counter value (int): initial value. Default: ``0``. transform (str): Name of transform associated with metric, if any. tags (dict): Tags to associate with metric. """ LOGGER_METRIC_TAGS = None DEFAULT_LOG_FORMAT = ( "[{name}] value: {value} transform: '{transform}' tags: {tags}" ) def __init__(self, name, value=0, transform=None, tags=None): super(LoggerMetric, self).__init__( name, value=value, transform=transform ) self.tags = tags if tags else {} self.tags.update(self.LOGGER_METRIC_TAGS) class LoggerCounter(LoggerMetric): """Log-based counter metric. Args: name (str): name of counter value (int): initial value. Default: ``0``. transform (str): Name of transform associated with counter, if any. tags (dict): Tags to associate with counter. Note: ``{"metric_type": "counter"}`` will always be an included tag. """ LOGGER_METRIC_TAGS = {"metric_type": "counter"} class LoggerGauge(LoggerMetric): """Log-based gauge metric. Args: name (str): name of gauge value (int): initial value. Default: ``0``. transform (str): Name of transform associated with gauge, if any. tags (dict): Tags to associate with gauge. Note: ``{"metric_type": "gauge"}`` will always be an included tag. """ LOGGER_METRIC_TAGS = {"metric_type": "gauge"} class LoggerTimer(LoggerMetric): """Log-based timer metric. Args: name (str): name of timer value (int): initial value. Default: ``0``. transform (str): Name of transform associated with timer, if any. tags (dict): Tags to associate with timer. Note: ``{"metric_type": "timer"}`` will always be an included tag. timer_unit (str): Unit of measurement. Options: :attr:`TIMER_UNIT_MAP`. Default: ``ns`` (nanoseconds). """ LOGGER_METRIC_TAGS = {"metric_type": "timer"} def __init__( self, name, value=0, transform=None, tags=None, timer_unit="ns" ): self.LOGGER_METRIC_TAGS.update({"unit": timer_unit}) super(LoggerTimer, self).__init__( name, value=value, transform=transform, tags=tags ) self.timer_unit = timer_unit
gaybro8777/klio
integration/audio-spectrograms/integration_test.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # # To be run after `klio job run --direct-runner` (not within job container) import os import unittest import yaml from google.cloud import storage from klio_core import config HERE = os.path.abspath(os.path.join(os.path.abspath(__file__), os.path.pardir)) EXPECTED_LOGS = os.path.join(HERE, "expected_job_output.txt") ACTUAL_LOGS = os.path.join(HERE, "job_output.log") BATCH_IDS = os.path.join(HERE, "batch_track_ids.txt") class TestExpectedOutput(unittest.TestCase): @classmethod def _load_klio_config(cls): config_file_path = os.path.join(os.path.dirname(__file__), "klio-job.yaml") with open(config_file_path) as f: return config.KlioConfig(yaml.safe_load(f)) @classmethod def setUpClass(self): config = self._load_klio_config() self.gcs_bucket, self.gcs_object_path = os.path.split(config.job_config.data.inputs[0].location) self.gcs_bucket = self.gcs_bucket.split("/")[-1] self.project = config.pipeline_options.project self.client = storage.Client(project=self.project) self.output_dir = "/".join(config.job_config.data.outputs[0].location.split("/")[3:]) with open(EXPECTED_LOGS, "r") as f: self.expected_logs = f.readlines() if not os.path.exists(ACTUAL_LOGS): # tox deletes the file after the test is done so that tests # don't pass accidentally from a previously successful run/ # cached results raise Exception( "The job's output does not exist. Rerun the job to produce " "the required output." ) with open(ACTUAL_LOGS, "r") as f: self.actual_logs = f.readlines() with open(BATCH_IDS, "r") as f: self.batch_ids = f.readlines() suffixes = ["full.png", "background.png", "foreground.png"] self.exp_files = ["-".join([i.strip(), s]) for i in self.batch_ids for s in suffixes] def test_expected_output(self): # sort them since the order of some parts of the pipeline are not # deterministic self.assertEqual(sorted(self.expected_logs), sorted(self.actual_logs)) def _exists(self, expected_blob): bucket = self.client.lookup_bucket(self.gcs_bucket) if not bucket: return False object_path = os.path.join(self.output_dir, expected_blob) blob = bucket.get_blob(object_path) return blob is not None def test_expected_gcs_files(self): for exp_file in self.exp_files: exists = self._exists(exp_file) self.assertTrue(exists, msg=f"{exp_file} does not exist") @classmethod def tearDownClass(self): bucket = self.client.get_bucket(self.gcs_bucket) for exp_file in self.exp_files: object_path = os.path.join(self.output_dir, exp_file) bucket.delete_blob(object_path) if __name__ == '__main__': unittest.main()
gaybro8777/klio
lib/tests/unit/transforms/test_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio.transforms import _helpers from klio.transforms import _utils class BaseKlassMeta(object): __metaclass__ = _helpers._KlioBaseDataExistenceCheck class KlassMeta(BaseKlassMeta): def process(self): pass def not_process(self): pass @pytest.mark.parametrize( "clsdict,bases,expected", ( ( {"process": KlassMeta.process}, (_helpers._KlioBaseDataExistenceCheck,), True, ), ( {"process": KlassMeta.not_process}, (_helpers._KlioBaseDataExistenceCheck,), False, ), ( {"process": "not a callable"}, (_helpers._KlioBaseDataExistenceCheck,), False, ), ({"process": KlassMeta.process}, (BaseKlassMeta,), False), ({}, (BaseKlassMeta,), False), ), ) def test_is_original_process_func(clsdict, bases, expected): actual = _utils.is_original_process_func( clsdict, bases, base_class="_KlioBaseDataExistenceCheck" ) assert expected == actual
gaybro8777/klio
examples/catvdog/test_transform.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_core import config from klio_core.proto import klio_pb2 import transforms @pytest.fixture(autouse=True) def mock_config(mocker, monkeypatch): mock = mocker.Mock(autospec=config.KlioConfig) monkeypatch.setattr(transforms.CatVDog, "_klio.config", mock, raising=False) return mock @pytest.fixture(autouse=True) def mock_gcs_client(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(transforms.gcsio, "GcsIO", mock) return mock @pytest.fixture def mock_model(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(transforms.tf.keras.models, "load_model", mock) return mock @pytest.fixture def klio_msg(): msg = klio_pb2.KlioMessage() msg.version = klio_pb2.Version.V2 msg.data.element = b"1234" return msg.SerializeToString() @pytest.mark.parametrize("prediction,exp_folder", ((0, "cat"), (1, "dog"))) def test_process( prediction, exp_folder, klio_msg, mock_model, mocker, monkeypatch ): mock_model.return_value.predict_classes.return_value = [[prediction]] mock_download_image = mocker.Mock() mock_download_image.return_value.name = "/tmp/tmp_abcd.jpg" monkeypatch.setattr( transforms.CatVDog, "download_image", mock_download_image ) mock_load_image = mocker.Mock() monkeypatch.setattr(transforms.CatVDog, "load_image", mock_load_image) mock_upload_image = mocker.Mock() monkeypatch.setattr(transforms.CatVDog, "upload_image", mock_upload_image) dofn_inst = transforms.CatVDog() dofn_inst.setup() ret_data = next(dofn_inst.process(klio_msg)) filename = "1234.jpg" mock_download_image.assert_called_once_with(filename) mock_load_image.assert_called_once_with( mock_download_image.return_value.name ) mock_model.return_value.predict_classes.assert_called_once_with( mock_load_image.return_value ) mock_upload_image.assert_called_once_with( mock_download_image.return_value, exp_folder, filename ) assert klio_msg == ret_data
gaybro8777/klio
examples/catvdog/run.py
<gh_stars>100-1000 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import apache_beam as beam from klio.transforms import helpers import transforms def run(input_pcol, config): """Main entrypoint in running a job's transform(s). Run any Beam transforms that need to happen after a message is consumed from PubSub from an upstream job (if not an apex job), and before publishing a message to any downstream job (if needed/configured). Args: input_pcol: A Beam PCollection returned from ``beam.io.ReadFromPubSub``. config (klio.KlioConfig): Job-related configuration as defined in ``klio-job.yaml``. Returns: A Beam PCollection that will be passed to ``beam.io.WriteToPubSub``. """ output_data = ( input_pcol | beam.ParDo(transforms.CatVDogOutputCheck()).with_outputs() ) output_force = output_data.found | helpers.KlioFilterForce() to_input_check = ( (output_data.not_found, output_force.process) | beam.Flatten() ) to_process = to_input_check | helpers.KlioGcsCheckInputExists() return to_process.found | beam.ParDo(transforms.CatVDog())
gaybro8777/klio
cli/tests/commands/job/test_stop.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from googleapiclient import errors as gerrors from klio_cli.commands.job import stop as stop_job @pytest.fixture def stop_job_inst(): return stop_job.StopJob() @pytest.fixture def mock_discovery_client(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(stop_job.discovery, "build", lambda x, y: mock) return mock @pytest.fixture def job(): return { "id": "1234", "name": "test-job", "projectId": "test-project", "location": "europe-west1", } @pytest.fixture def mock_sleep(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(stop_job.time, "sleep", mock) return mock @pytest.mark.parametrize("api_version", (None, "v1b3", "v2")) def test_set_dataflow_client( mock_discovery_client, api_version, stop_job_inst ): assert stop_job_inst._client is not None stop_job_inst._set_dataflow_client(api_version) assert stop_job_inst._client is not None assert mock_discovery_client == stop_job_inst._client # cleanup setattr(stop_job_inst, "_client", None) @pytest.mark.parametrize( "jobs_response, result_is_none", ( ( { "jobs": [ { "id": "1234", "name": "test-job", "projectId": "test-project", "location": "europe-west1", } ] }, False, ), ( { "jobs": [ { "id": "1234", "name": "not-this-job", "projectId": "test-project", "location": "europe-west1", } ] }, True, ), ({"jobs": []}, True), ), ) def test_check_job_running( mock_discovery_client, jobs_response, result_is_none, monkeypatch, stop_job_inst, job, ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.list req.return_value.execute.return_value = jobs_response ret = stop_job_inst._check_job_running( job["name"], job["projectId"], job["location"] ) exp_ret_val = jobs_response["jobs"][0] if not result_is_none else None assert exp_ret_val == ret req.assert_called_once_with( projectId="test-project", location="europe-west1", filter="ACTIVE" ) req.return_value.execute.assert_called_once_with() def test_check_job_running_errors( mock_discovery_client, monkeypatch, caplog, stop_job_inst, job ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.list req.return_value.execute.side_effect = Exception("foo") stop_job_inst._check_job_running( job["name"], job["projectId"], job["location"] ) req.assert_called_once_with( projectId="test-project", location="europe-west1", filter="ACTIVE" ) req.return_value.execute.assert_called_once_with() assert 2 == len(caplog.records) @pytest.mark.parametrize( "state,pyver", (("drain", None), ("cancel", None), (None, 2), (None, 3)) ) def test_update_job_state( state, pyver, mock_discovery_client, job, monkeypatch, stop_job_inst ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) exp_state = state if not state: if pyver == 2: exp_state = "JOB_STATE_DRAINED" else: exp_state = "JOB_STATE_CANCELLED" monkeypatch.setitem(stop_job.JOB_STATE_MAP, "default", exp_state) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update req.return_value.execute.return_value = None stop_job_inst._update_job_state(job, state) job["requestedState"] = exp_state req.assert_called_once_with( jobId="1234", projectId="test-project", location="europe-west1", body=job, ) req.return_value.execute.assert_called_once_with() def test_update_job_state_400_error( mock_discovery_client, job, mock_sleep, mocker, monkeypatch, caplog, stop_job_inst, ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update mock_resp = mocker.Mock(status=400) req.return_value.execute.side_effect = gerrors.HttpError(mock_resp, b"foo") with pytest.raises(SystemExit): stop_job_inst._update_job_state(job, "drain") assert 1 == req.return_value.execute.call_count assert 1 == len(caplog.records) assert not mock_sleep.call_count def test_update_job_state_500_error( mock_discovery_client, job, mock_sleep, mocker, monkeypatch, caplog, stop_job_inst, ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update mock_resp = mocker.Mock(status=500) req.return_value.execute.side_effect = gerrors.HttpError(mock_resp, b"foo") with pytest.raises(SystemExit): stop_job_inst._update_job_state(job, "drain") assert 4 == req.return_value.execute.call_count assert 4 == len(caplog.records) assert 3 == mock_sleep.call_count def test_update_job_state_error( mock_discovery_client, job, mock_sleep, monkeypatch, caplog, stop_job_inst ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.update req.return_value.execute.side_effect = Exception("foo") with pytest.raises(SystemExit): stop_job_inst._update_job_state(job, "cancel") assert 4 == req.return_value.execute.call_count assert 4 == len(caplog.records) assert 3 == mock_sleep.call_count @pytest.mark.parametrize( "exec_side_effect", ( ( {"currentState": "JOB_STATE_CANCELLING"}, {"currentState": "JOB_STATE_CANCELLED"}, ), (Exception("foo"), {"currentState": "JOB_STATE_CANCELLED"}), ), ) def test_watch_job_state( mock_discovery_client, mock_sleep, monkeypatch, caplog, job, exec_side_effect, stop_job_inst, ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) _projects_req = mock_discovery_client.projects.return_value req = _projects_req.locations.return_value.jobs.return_value.get req.return_value.execute.side_effect = exec_side_effect stop_job_inst._watch_job_state(job) assert 2 == req.return_value.execute.call_count mock_sleep.assert_called_once_with(5) assert 1 == len(caplog.records) def test_watch_job_state_raises( mock_discovery_client, monkeypatch, caplog, job, stop_job_inst ): monkeypatch.setattr(stop_job_inst, "_client", mock_discovery_client) with pytest.raises(SystemExit): stop_job_inst._watch_job_state(job, timeout=0) assert 1 == len(caplog.records) def test_stop(mocker, monkeypatch, mock_discovery_client, job, stop_job_inst): mock_set_dataflow_client = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_set_dataflow_client", mock_set_dataflow_client ) mock_check_job_running = mocker.Mock(return_value=job) monkeypatch.setattr( stop_job_inst, "_check_job_running", mock_check_job_running ) mock_update_job_state = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_update_job_state", mock_update_job_state ) mock_watch_job_state = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_watch_job_state", mock_watch_job_state ) stop_job_inst.stop( job["name"], job["projectId"], job["location"], "cancel" ) mock_set_dataflow_client.assert_called_once_with(None) mock_check_job_running.assert_called_once_with( "test-job", "test-project", "europe-west1" ) mock_update_job_state.assert_called_once_with(job, req_state="cancel") mock_watch_job_state.assert_called_once_with(job) def test_stop_no_running_job( mocker, monkeypatch, mock_discovery_client, stop_job_inst, job ): mock_set_dataflow_client = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_set_dataflow_client", mock_set_dataflow_client ) mock_check_job_running = mocker.Mock(return_value=None) monkeypatch.setattr( stop_job_inst, "_check_job_running", mock_check_job_running ) mock_update_job_state = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_update_job_state", mock_update_job_state ) mock_watch_job_state = mocker.Mock() monkeypatch.setattr( stop_job_inst, "_watch_job_state", mock_watch_job_state ) stop_job_inst.stop( job["name"], job["projectId"], job["location"], "cancel" ) mock_set_dataflow_client.assert_called_once_with(None) mock_check_job_running.assert_called_once_with( "test-job", "test-project", "europe-west1" ) mock_update_job_state.assert_not_called() mock_watch_job_state.assert_not_called()
gaybro8777/klio
cli/src/klio_cli/commands/message/publish.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import functools import logging import emoji from google.api_core import exceptions as gapi_exceptions from google.cloud import pubsub from klio_core.proto.v1beta1 import klio_pb2 def _create_publisher(topic): client = pubsub.PublisherClient() try: client.get_topic(request={"topic": topic}) except gapi_exceptions.NotFound: msg = ( ":persevere: Topic '{}' not found. Is there a running job " "subscribed to this topic? or is there a typo in the configured " "topic?".format(topic) ) logging.error(emoji.emojize(msg, use_aliases=True)) raise SystemExit(1) except Exception: raise return functools.partial(client.publish, topic=topic) def _get_current_klio_job(config): klio_job = klio_pb2.KlioJob() klio_job.job_name = config.job_name klio_job.gcp_project = config.pipeline_options.project return klio_job # [batch dev] TODO: rename entity_id variable in this module def _create_pubsub_message(entity_id, job, force, ping, top_down, msg_version): kmsg = klio_pb2.KlioMessage() kmsg.version = msg_version if msg_version == 1: kmsg.data.entity_id = entity_id if not top_down: kmsg.metadata.downstream.extend([job]) elif msg_version == 2: kmsg.data.element = bytes(entity_id, "utf-8") if not top_down: kmsg.metadata.intended_recipients.limited.recipients.extend([job]) else: kmsg.metadata.intended_recipients.anyone.SetInParent() kmsg.metadata.ping = ping kmsg.metadata.force = force return kmsg.SerializeToString() def _publish_messages( config, entity_ids, ping, force, top_down, allow_non_klio, msg_version ): current_job = _get_current_klio_job(config) publish = _create_publisher(config.job_config.events.inputs[0].topic) success_ids = [] fail_ids = [] for entity_id in entity_ids: if not allow_non_klio: message = _create_pubsub_message( entity_id, current_job, ping, force, top_down, msg_version ) else: # TODO: should rename argument to something more abstract (@lynn) message = bytes(entity_id.encode("utf-8")) try: publish(data=message) success_ids.append(entity_id) except Exception as e: msg = "Failed to publish message for entity '%s': %s" % ( entity_id, e, ) logging.warning(msg) fail_ids.append(entity_id) return success_ids, fail_ids def publish_messages( config, entity_ids, force=False, ping=False, top_down=False, allow_non_klio=False, msg_version=None, ): # [batch dev] TODO: if we use KlioConfig, we don't have to find the # version # [batch dev] maintaining backwards compatibility if this code path # is executed by some other way other than via cli.py::publish. if msg_version is None: msg_version = config.version if not config.job_config.events.inputs: msg = "No input topics configured for {} :-1:".format(config.job_name) logging.error(emoji.emojize(msg, use_aliases=True)) raise SystemExit(1) logging.info( "Publishing {} messages to {}'s input topic {}".format( len(entity_ids), config.job_name, # [batch dev] should we support multiple inputs in the future? config.job_config.events.inputs[0].topic, ) ) success, fail = _publish_messages( config, entity_ids, force, ping, top_down, allow_non_klio, msg_version ) if success: msg = ":boom: Successfully published {} messages.".format(len(success)) logging.info(emoji.emojize(msg, use_aliases=True)) if fail: msg = ( ":persevere: Failed to publish the following {} entity " "IDs: {}".format(len(fail), ", ".join(fail)) ) logging.warning(emoji.emojize(msg, use_aliases=True))
gaybro8777/klio
integration/read-bq-write-bq/it/before.py
<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """Setup for integration test * Rewrites klio-job.yaml so that the input and output BigQuery tables have names unique to the run * Creates BigQuery tables that need to exist before the test run starts. """ import os import apache_beam as beam import yaml import common def _append_build_id(base_name): """Returns base_name with BQ-friendly `GITHUB_SHA` appended""" build_id = os.environ.get("GITHUB_SHA", None) if not build_id: raise Exception("Unable to get build id; env var GITHUB_SHA not set") # valid BQ table names only allow underscores and alphanumeric chars # https://cloud.google.com/bigquery/docs/tables#table_naming table_name = "{}_{}".format(base_name, build_id.replace("-", "_")) return table_name def rewrite_klio_config_yaml(): """Rewrite `klio-job.yaml` with tablenames that have the GITHUB_SHA appended""" klio_cfg_file_path = os.path.join(os.path.dirname(__file__), "..", "klio-job.yaml") klio_save_file_path = os.path.join(os.path.dirname(__file__), "..", "klio-job.yaml.save") with open(klio_cfg_file_path) as f: config_dict = yaml.safe_load(f) # save the original with open(klio_save_file_path, "w") as g: g.write(yaml.safe_dump(config_dict)) new_input_table_name = _append_build_id(config_dict["job_config"]["events"]["inputs"][0]["table"]) config_dict["job_config"]["events"]["inputs"][0]["table"] = new_input_table_name new_output_table_name = _append_build_id(config_dict["job_config"]["events"]["outputs"][0]["table"]) config_dict["job_config"]["events"]["outputs"][0]["table"] = new_output_table_name with open(klio_cfg_file_path, "w") as g: g.write(yaml.safe_dump(config_dict)) def populate_bigquery_table(): """Create & populate input table based on what is configured for event input in klio-job.yaml This needs to run before `klio job run` is called, which is why """ table_schema = {"fields": [{ 'name': 'entity_id', 'type': 'STRING', 'mode': 'NULLABLE' }]} klio_cfg = common.get_config() input_table_cfg = klio_cfg.job_config.events.inputs[0] table_name = "{}:{}.{}".format(input_table_cfg.project, input_table_cfg.dataset, input_table_cfg.table) with beam.Pipeline() as p: def create_record(v): return { 'entity_id': v, } record_ids = p | 'CreateIDs' >> beam.Create(common.entity_ids) records = record_ids | 'CreateRecords' >> beam.Map(lambda x: create_record(x)) records | 'write' >> beam.io.WriteToBigQuery( table_name, schema=table_schema, create_disposition=beam.io.BigQueryDisposition.CREATE_IF_NEEDED, write_disposition=beam.io.BigQueryDisposition.WRITE_TRUNCATE) if __name__ == '__main__': rewrite_klio_config_yaml() populate_bigquery_table()
gaybro8777/klio
audio/src/klio_audio/transforms/_base.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import apache_beam as beam import matplotlib.pyplot as plt from klio.transforms import decorators as tfm_decorators from klio_audio import decorators class KlioAudioDoFnMetaclass(type): """Enforce behavior upon subclasses of `KlioAudioBaseDoFn`.""" def __call__(self, *args, **kwargs): # automatically wrap DoFn in a beam.ParDo so folks can just do # `pcoll | SomeAudioTransform()` rather than # `pcoll | beam.ParDo(SomeAudioTransform())` return beam.ParDo( super(KlioAudioDoFnMetaclass, self).__call__(*args, **kwargs) ) class KlioAudioBaseDoFn(beam.DoFn, metaclass=KlioAudioDoFnMetaclass): pass class KlioPlotBaseDoFn(KlioAudioBaseDoFn): DEFAULT_TITLE = "" def __init__(self, *_, title=None, **plot_args): if "ax" in plot_args: raise RuntimeError( "Invalid keyword `ax`: Specifying the plot's axes is not " "supported." ) self.title = title or self.DEFAULT_TITLE self.plot_args = plot_args def _plot(self, *args, **kwargs): pass @tfm_decorators._handle_klio @decorators.handle_binary(load_with_numpy=True) def process(self, item): element = item.element.decode("utf-8") title = self.title.format(element=element) self._klio.logger.debug("Generating plot '{}'".format(title)) fig = plt.figure() fig.suptitle(title) self._plot(item, fig) yield fig
gaybro8777/klio
cli/tests/commands/job/test_create.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import datetime import os import click import mock import pytest import yaml from klio_cli.commands.job import create HERE = os.path.abspath(os.path.join(os.path.abspath(__file__), os.path.pardir)) FIXTURE_PATH = os.path.join(HERE, "utils", "fixtures") @pytest.fixture def job(): return create.CreateJob() # pytest prevents monkeypatching datetime directly class MockDatetime(datetime.datetime): @classmethod def now(cls): return datetime.datetime(2019, 1, 1) def test_get_environment(job): env = job._get_environment() expected_templates = [ "MANIFEST.in.tpl", "README.md.tpl", "dockerfile.tpl", "init.py.tpl", "job-requirements.txt.tpl", "klio-job-batch.yaml.tpl", "klio-job.yaml.tpl", "run.py.tpl", "setup.py.tpl", "test_transforms.py.tpl", "transforms-batch.py.tpl", "transforms.py.tpl", ] assert expected_templates == sorted(env.list_templates()) def test_create_job_directory(tmpdir, job): output_base_dir = tmpdir.mkdir("testing") output_dir = os.path.join(str(output_base_dir), "test_job") job._create_job_directory(output_dir) assert os.path.exists(output_dir) @pytest.mark.parametrize("error_code", (1, 17)) def test_create_job_directory_raises(monkeypatch, tmpdir, error_code, job): output_base_dir = tmpdir.mkdir("testing") output_dir = os.path.join(str(output_base_dir), "test_job") def mock_mkdir(*args, **kwargs): raise OSError(error_code, "some error message", output_dir) monkeypatch.setattr(create.os, "mkdir", mock_mkdir) if error_code != 17: with pytest.raises(OSError) as e: job._create_job_directory(output_dir) assert e.match("some error message") else: # should not re-raise if file already existsgit job._create_job_directory(output_dir) def test_write_template(tmpdir, job): data = u"this is some test data" output_dir = tmpdir.mkdir("testing") output_file = "foo.txt" job._write_template(output_dir.strpath, output_file, data) ret_file = output_dir.join(output_file) ret_contents = ret_file.read() assert data == ret_contents @pytest.fixture def batch_job_context(): return { "inputs": [ { "event_location": "test-job_input_elements.txt", "data_location": "test-job-input", } ], "outputs": [ { "event_location": "test-job_output_elements", "data_location": "test-job-output", } ], } @pytest.fixture def context(): base_gcs = "gs://test-gcp-project-dataflow-tmp/test-job" gcr_url = "gcr.io/test-gcp-project/test-job-worker" return { "job_name": "test-job", "job_type": "streaming", "python_version": "3.6", "pipeline_options": { "project": "test-gcp-project", "region": "europe-west1", "worker_harness_container_image": gcr_url, "experiments": ["beam_fn_api"], "staging_location": base_gcs + "/staging", "temp_location": base_gcs + "/temp", "num_workers": 2, "max_num_workers": 2, "autoscaling_algorithm": "NONE", "disk_size_gb": 32, "worker_machine_type": "n1-standard-2", }, "job_options": { "inputs": [ { "topic": ( "projects/test-parent-gcp-project/topics/" "test-parent-job-output" ), "subscription": ( "projects/test-gcp-project/subscriptions/" "test-parent-job-output-test-job" ), "data_location": ( "gs://test-parent-gcp-project-output/test-parent-job" ), } ], "outputs": [ { "topic": "projects/test-gcp-project/topics/test-job-output", "data_location": "gs://test-gcp-project-output/test-job", } ], "dependencies": [ { "job_name": "test-parent-job", "gcp_project": "test-parent-gcp-project", "input_topics": [ ( "projects/test-grandparent-gcp-project/topics/" "test-grandparent-job-output" ) ], "region": "us-central1", } ], }, } @pytest.fixture def default_context(): base_gcs = "gs://test-gcp-project-dataflow-tmp/test-job" gcr_url = "gcr.io/test-gcp-project/test-job-worker" return { "job_name": "test-job", "job_type": "streaming", "python_version": "3.6", "use_fnapi": False, "create_resources": False, "pipeline_options": { "project": "test-gcp-project", "region": "europe-west1", "worker_harness_container_image": gcr_url, "experiments": [], "staging_location": base_gcs + "/staging", "temp_location": base_gcs + "/temp", "num_workers": 2, "max_num_workers": 2, "autoscaling_algorithm": "NONE", "disk_size_gb": 32, "worker_machine_type": "n1-standard-2", }, "job_options": { "inputs": [ { "topic": ( "projects/test-gcp-project/topics/test-job-input" ), "subscription": ( "projects/test-gcp-project/subscriptions/" "test-job-input-test-job" ), "data_location": ("gs://test-gcp-project-input/test-job"), } ], "outputs": [ { "topic": "projects/test-gcp-project/topics/test-job-output", "data_location": "gs://test-gcp-project-output/test-job", } ], "dependencies": [], }, } @pytest.mark.parametrize("job_type", ("batch", "streaming")) @pytest.mark.parametrize("use_fnapi", (True, False)) def test_create_job_config( use_fnapi, context, batch_job_context, tmpdir, monkeypatch, job, job_type ): output_dir = ( tmpdir.mkdir("testing") .mkdir("jobs") .mkdir("test-job-{}".format(job_type)) ) env = job._get_environment() if job_type == "batch": context["job_options"] = batch_job_context context["use_fnapi"] = use_fnapi context["job_type"] = job_type if not use_fnapi: monkeypatch.setitem(context["pipeline_options"], "experiments", []) job._create_job_config(env, context, output_dir.strpath) expected_fixtures = os.path.join(FIXTURE_PATH, "expected") is_fnapi_dir = "fnapi" if use_fnapi else "no_fnapi" expected_fixtures = os.path.join(expected_fixtures, is_fnapi_dir) if job_type == "batch": fixture = os.path.join(expected_fixtures, "klio-job-batch.yaml") else: fixture = os.path.join(expected_fixtures, "klio-job.yaml") with open(fixture, "r") as f: expected = yaml.safe_load(f) ret_file = output_dir.join("klio-job.yaml") ret_contents = yaml.safe_load(ret_file.read()) assert expected == ret_contents @pytest.mark.parametrize("job_type", ("batch", "streaming")) def test_create_python_files(tmpdir, mocker, job, job_type): output_dir = tmpdir.mkdir("testing").mkdir("jobs").mkdir("test_job") env = job._get_environment() dt_patch = "klio_cli.commands.job.create.datetime.datetime" with mock.patch(dt_patch, MockDatetime): job._create_python_files(env, "test_job", job_type, output_dir.strpath) ret_init_file = output_dir.join("__init__.py") ret_init_contents = ret_init_file.read() ret_run_file = output_dir.join("run.py") ret_run_contents = ret_run_file.read() ret_transforms_file = output_dir.join("transforms.py") ret_transforms_contents = ret_transforms_file.read() expected_fixtures = os.path.join(FIXTURE_PATH, "expected") init_fixture = os.path.join(expected_fixtures, "__init__.py") run_fixture = os.path.join(expected_fixtures, "run.py") if job_type == "batch": transforms_fixture = os.path.join( expected_fixtures, "transforms-batch.py" ) else: transforms_fixture = os.path.join(expected_fixtures, "transforms.py") with open(init_fixture, "r") as f: expected_init = f.read() with open(run_fixture, "r") as f: expected_run = f.read() with open(transforms_fixture, "r") as f: expected_transforms = f.read() assert expected_init == ret_init_contents + "\n" assert expected_run == ret_run_contents + "\n" assert expected_transforms == ret_transforms_contents + "\n" def test_create_no_fnapi_files(tmpdir, job): output_dir = tmpdir.mkdir("testing").mkdir("jobs").mkdir("test_job") env = job._get_environment() context = {"use_fnapi": False, "package_name": "test-job"} job._create_no_fnapi_files(env, context, output_dir.strpath) manifest_file = output_dir.join("MANIFEST.in") ret_manifest_contents = manifest_file.read() setup_file = output_dir.join("setup.py") ret_setup_contents = setup_file.read() expected_fixtures = os.path.join(FIXTURE_PATH, "expected", "no_fnapi") manifest_fixture = os.path.join(expected_fixtures, "MANIFEST.in") setup_fixture = os.path.join(expected_fixtures, "setup.py") with open(manifest_fixture, "r") as f: expected_manifest = f.read() with open(setup_fixture, "r") as f: expected_setup = f.read() assert expected_manifest == ret_manifest_contents + "\n" assert expected_setup == ret_setup_contents + "\n" @pytest.mark.parametrize("use_fnapi", (True, False)) def test_create_reqs_file(use_fnapi, tmpdir, job): output_dir = tmpdir.mkdir("testing").mkdir("jobs").mkdir("test_job") env = job._get_environment() context = {"use_fnapi": use_fnapi} job._create_reqs_file(env, context, output_dir.strpath) ret_reqs_file = output_dir.join("job-requirements.txt") ret_reqs_contents = ret_reqs_file.read() expected_fixtures = os.path.join(FIXTURE_PATH, "expected") fixture = os.path.join(expected_fixtures, "job-requirements.txt") with open(fixture, "r") as f: expected = f.read() assert expected == ret_reqs_contents + "\n" @pytest.mark.parametrize("use_fnapi", (True, False)) def test_create_dockerfile(use_fnapi, tmpdir, job): output_dir = tmpdir.mkdir("testing").mkdir("jobs").mkdir("test_job") env = job._get_environment() context = { "pipeline_options": { "worker_harness_container_image": "gcr.io/foo/bar", "project": "test-gcp-project", }, "python_version": "3.6", "use_fnapi": use_fnapi, "create_resources": False, } job._create_dockerfile(env, context, output_dir.strpath) ret_dockerfile_file = output_dir.join("Dockerfile") ret_dockerfile_contents = ret_dockerfile_file.read() expected_fixtures = os.path.join(FIXTURE_PATH, "expected") is_fnapi_dir = "fnapi" if use_fnapi else "no_fnapi" expected_fixtures = os.path.join(expected_fixtures, is_fnapi_dir) fixture = os.path.join(expected_fixtures, "Dockerfile") with open(fixture, "r") as f: expected = f.read() assert expected == ret_dockerfile_contents def test_create_readme(tmpdir, job): output_dir = tmpdir.mkdir("testing").mkdir("test_job") env = job._get_environment() context = {"job_name": "test-job"} job._create_readme(env, context, output_dir.strpath) expected_fixtures = os.path.join(FIXTURE_PATH, "expected") ret_file = output_dir.join("README.md") ret_file_contents = ret_file.read() exp_file = os.path.join(expected_fixtures, "README.md") with open(exp_file, "r") as f: exp_file_contents = f.read() assert exp_file_contents == ret_file_contents def test_validate_worker_image(job): assert not job._validate_worker_image("foo") def test_validate_region(job): exp_region = "us-central1" ret_region = job._validate_region(exp_region) assert exp_region == ret_region def test_validate_region_raises(job): err_region = "not-a-region" with pytest.raises(click.BadParameter) as e: job._validate_region(err_region) assert e.match( '"{}" is not a valid region. Available: '.format(err_region) ) @pytest.mark.parametrize( "input_version,exp_output_version", ( ("3.6", "3.6"), ("3.6.1", "3.6"), ("3.7", "3.7"), ("3.7.1", "3.7"), ("3.8", "3.8"), ("3.8.1", "3.8"), ), ) def test_parse_python_version(input_version, exp_output_version, job): assert exp_output_version == job._parse_python_version(input_version) @pytest.mark.parametrize( "input_version,exp_msg", ( ("2", "Klio no longer supports Python 2.7"), ("2.7", "Klio no longer supports Python 2.7"), ("3", "Invalid Python version given"), ("3.3", "Invalid Python version given"), ("3.6.7.8", "Invalid Python version given"), ), ) def test_parse_python_version_raises(input_version, exp_msg, job): # only matching the start of the error message with pytest.raises(click.BadParameter, match=exp_msg): job._parse_python_version(input_version) def test_get_context_from_defaults(default_context, job): basic_context = { "job_name": "test-job", "gcp_project": "test-gcp-project", } ret_context, ret_create_dockerfile = job._get_context_from_defaults( basic_context ) default_context.pop("job_name") assert default_context == ret_context assert ret_create_dockerfile @pytest.fixture def context_overrides(): return { "job_name": "test-job", "job_type": "streaming", "gcp_project": "test-gcp-project", "worker_image": "gcr.io/foo/bar", "experiments": "beam_fn_api,another_experiment", "region": "us-central1", "staging_location": "gs://a-different/staging/location", "temp_location": "gs://a-different/temp/location", "num_workers": 1000000, "max_num_workers": 1000000, "autoscaling_algorithm": "THROUGHPUT_BASED", "disk_size_gb": 1000000, "input_topic": "projects/test-gcp-projects/topics/another-topic", "output_topic": "a-different-topic", "input_data_location": "gs://test-parent-gcp-project/test-parent-job", "output_data_location": "bq://somewhere/over/the/rainbow", "dependencies": [ {"job_name": "parent-job", "gcp_project": "parent-gcp-project"} ], "python_version": "3.7", "use_fnapi": "n", "create_resources": "n", } @pytest.fixture def expected_overrides(): return { "pipeline_options": { "project": "test-gcp-project", "worker_harness_container_image": "gcr.io/foo/bar", "experiments": ["beam_fn_api", "another_experiment"], "region": "us-central1", "staging_location": "gs://a-different/staging/location", "temp_location": "gs://a-different/temp/location", "num_workers": 1000000, "max_num_workers": 1000000, "autoscaling_algorithm": "THROUGHPUT_BASED", "disk_size_gb": 1000000, "worker_machine_type": "n4-highmem-l33t", }, "job_options": { "inputs": [ { "topic": "projects/test-gcp-projects/topics/another-topic", "subscription": ( "projects/test-gcp-project/subscriptions/" "another-topic-test-job" ), "data_location": ( "gs://test-parent-gcp-project/test-parent-job" ), } ], "outputs": [ { "topic": "a-different-topic", "data_location": "bq://somewhere/over/the/rainbow", } ], "dependencies": [ {"job_name": "parent-job", "gcp_project": "parent-gcp-project"} ], }, "python_version": "3.7", "use_fnapi": False, "create_resources": False, "job_type": "streaming", } def test_get_context_from_defaults_overrides( context_overrides, expected_overrides, job ): # FYI: Click will pass in kwargs as a flat dict context_overrides["worker_machine_type"] = "n4-highmem-l33t" ret_context, ret_create_dockerfile = job._get_context_from_defaults( context_overrides ) assert expected_overrides == ret_context assert not ret_create_dockerfile @pytest.mark.parametrize( "inputs,confirmation, regions,expected_dependencies", ( ( ["my-job", "my-proj", "my-topic", "europe-west1"], [False], ["europe-west1"], [ { "job_name": "my-job", "gcp_project": "my-proj", "input_topics": ["my-topic"], "region": "europe-west1", } ], ), # full input ( [ # mock confirm 1 "my-job", "my-proj", "my-topic", "europe-west1", # mock confirm 2 "my-job2", "my-proj", "my-topic2", "europe-west1", ], [True, False], ["europe-west1", "europe-west1"], [ { "job_name": "my-job", "gcp_project": "my-proj", "input_topics": ["my-topic"], "region": "europe-west1", }, { "job_name": "my-job2", "gcp_project": "my-proj", "input_topics": ["my-topic2"], "region": "europe-west1", }, ], ), # multiple full inputs ( ["my-job", "my-proj", "", ""], [False], [], [{"job_name": "my-job", "gcp_project": "my-proj"}], ), # use default inputs for input_topics and region ( ["my-job", "my-proj", "my-topic", ""], [False], [], [ { "job_name": "my-job", "gcp_project": "my-proj", "input_topics": ["my-topic"], } ], ), # use defaults for region ( ["my-job", "my-proj", "", "europe-west1"], [False], ["europe-west1"], [ { "job_name": "my-job", "gcp_project": "my-proj", "region": "europe-west1", } ], ), # use defaults for input topics ), ) def test_get_dependencies_from_user_inputs( mocker, mock_prompt, mock_confirm, inputs, confirmation, regions, expected_dependencies, job, ): mock_prompt.side_effect = inputs mock_confirm.side_effect = confirmation mock_validate_region = mocker.patch.object(job, "_validate_region") actual_dependencies = job._get_dependencies_from_user_inputs() assert expected_dependencies == actual_dependencies expected_validate_region_calls = [mocker.call(r) for r in regions] assert ( expected_validate_region_calls == mock_validate_region.call_args_list ) @pytest.fixture def mock_prompt(mocker): return mocker.patch.object(create.click, "prompt") @pytest.fixture def mock_confirm(mocker): return mocker.patch.object(create.click, "confirm") def test_get_context_from_user_inputs( context, mock_prompt, mock_confirm, mocker, job, ): # mimicking user inputs for each prompt prompt_side_effect = [ "streaming", "europe-west1", "Y", "n", ["beam_fn_api"], 2, 2, "NONE", 32, "n1-standard-2", "", "3.6", "gs://test-gcp-project-dataflow-tmp/test-job/staging", "gs://test-gcp-project-dataflow-tmp/test-job/temp", "projects/test-parent-gcp-project/topics/test-parent-job-output", "projects/test-gcp-project/topics/test-job-output", "gs://test-parent-gcp-project-output/test-parent-job", "gs://test-gcp-project-output/test-job", # <-- mock_confirm side effect 1 --> "test-parent-job", "test-parent-gcp-project", ( "projects/test-grandparent-gcp-project/topics/test-grandparent-" "job-output" ), "us-central1", # <-- mock_confirm side effect 2--> ] confirm_side_effect = [True, False] mock_prompt.side_effect = prompt_side_effect mock_confirm.side_effect = confirm_side_effect user_input_context = { "job_name": "test-job", "gcp_project": "test-gcp-project", } mock_validate_region = mocker.patch.object(job, "_validate_region") mock_validate_worker_image = mocker.patch.object( job, "_validate_worker_image" ) ret_context, ret_dockerfile = job._get_context_from_user_inputs( user_input_context ) assert len(prompt_side_effect) == mock_prompt.call_count assert len(confirm_side_effect) == mock_confirm.call_count exp_calls = [mocker.call("europe-west1"), mocker.call("us-central1")] assert exp_calls == mock_validate_region.call_args_list assert 2 == mock_validate_region.call_count # mock_validate_region.assert_called_once_with("europe-west1") gcr_url = "gcr.io/test-gcp-project/test-job-worker" mock_validate_worker_image.assert_called_once_with(gcr_url) context.pop("job_name") context["pipeline_options"].pop("project") context["use_fnapi"] = True context["create_resources"] = False assert context == ret_context assert ret_dockerfile def test_get_context_from_user_inputs_no_prompts( mocker, context_overrides, expected_overrides, mock_prompt, mock_confirm, job, ): context_overrides["machine_type"] = "n4-highmem-l33t" mock_validate_region = mocker.patch.object(job, "_validate_region") mock_validate_worker_image = mocker.patch.object( job, "_validate_worker_image" ) ret_context, ret_dockerfile = job._get_context_from_user_inputs( context_overrides ) expected_overrides["pipeline_options"].pop("project") expected_overrides["python_version"] = "3.6" assert not mock_prompt.call_count assert not mock_confirm.call_count mock_validate_region.assert_called_once_with("us-central1") mock_validate_worker_image.assert_called_once_with("gcr.io/foo/bar") assert not ret_dockerfile assert expected_overrides == ret_context @pytest.mark.parametrize( "users_provided_dependencies,confirmed_dependencies", ( (True, True), # (True, False), # not possible (False, True), (False, False), ), ) def test_get_context_from_user_inputs_dependency_settings( context_overrides, expected_overrides, mock_prompt, mock_confirm, mocker, monkeypatch, users_provided_dependencies, confirmed_dependencies, job, ): context_overrides["machine_type"] = "n4-highmem-l33t" saved_dependencies = context_overrides.pop("dependencies") mock_get_dependencies_from_user_inputs = mocker.Mock() if users_provided_dependencies: mock_get_dependencies_from_user_inputs.return_value = ( saved_dependencies ) else: mock_get_dependencies_from_user_inputs.return_value = None monkeypatch.setattr( job, "_get_dependencies_from_user_inputs", mock_get_dependencies_from_user_inputs, ) mock_confirm.side_effect = [confirmed_dependencies] mock_validate_region = mocker.patch.object(job, "_validate_region") mock_validate_worker_image = mocker.patch.object( job, "_validate_worker_image" ) ret_context, ret_dockerfile = job._get_context_from_user_inputs( context_overrides ) expected_overrides["pipeline_options"].pop("project") expected_overrides["python_version"] = "3.6" assert not mock_prompt.call_count assert 1 == mock_confirm.call_count mock_validate_region.assert_called_once_with("us-central1") mock_validate_worker_image.assert_called_once_with("gcr.io/foo/bar") assert not ret_dockerfile if users_provided_dependencies: assert expected_overrides == ret_context else: expected_overrides["job_options"]["dependencies"] = [] assert expected_overrides == ret_context if confirmed_dependencies: assert 1 == mock_get_dependencies_from_user_inputs.call_count else: assert 0 == mock_get_dependencies_from_user_inputs.call_count def test_get_context_from_user_inputs_no_prompts_image( mocker, context_overrides, expected_overrides, mock_prompt, mock_confirm, job, ): mock_prompt.side_effect = [""] context_overrides.pop("worker_image") context_overrides["machine_type"] = "n4-highmem-l33t" mock_validate_region = mocker.patch.object(job, "_validate_region") mock_validate_worker_image = mocker.patch.object( job, "_validate_worker_image" ) ret_context, ret_dockerfile = job._get_context_from_user_inputs( context_overrides ) gcr_url = "gcr.io/test-gcp-project/test-job-worker" exp_pipeline_opts = expected_overrides["pipeline_options"] exp_pipeline_opts.pop("project") exp_pipeline_opts["worker_harness_container_image"] = gcr_url assert 1 == mock_prompt.call_count assert not mock_confirm.call_count mock_validate_region.assert_called_once_with("us-central1") mock_validate_worker_image.assert_called_once_with(gcr_url) assert ret_dockerfile assert expected_overrides == ret_context @pytest.mark.parametrize("use_defaults", (True, False)) def test_get_user_input(use_defaults, mocker, job): ret_context = {"pipeline_options": {}} mock_get_context_defaults = mocker.patch.object( job, "_get_context_from_defaults" ) mock_get_context_defaults.return_value = (ret_context, True) mock_get_context_user = mocker.patch.object( job, "_get_context_from_user_inputs" ) mock_get_context_user.return_value = (ret_context, True) input_kwargs = { "use_defaults": use_defaults, "job_name": "test-job", "gcp_project": "test-gcp-project", } job._get_user_input(input_kwargs) if use_defaults: mock_get_context_defaults.assert_called_once_with(input_kwargs) else: mock_get_context_user.assert_called_once_with(input_kwargs) @pytest.mark.parametrize( "dependencies,expected_dependencies,num_warnings", ( ( ( "job_name=parent-job", "gcp_project=a-project", "region=a-region", "input_topic=d", "input_topics=a,b,c", ), { "job_name": "parent-job", "gcp_project": "a-project", "region": "a-region", "input_topic": "d", "input_topics": ["a", "b", "c"], }, 0, ), ( ( "job-name=parent-job", "gcp-project=a-project", "region=a-region", "input-topic=d", "input-topics=a,b,c", ), { "job_name": "parent-job", "gcp_project": "a-project", "region": "a-region", "input_topic": "d", "input_topics": ["a", "b", "c"], }, 0, ), ( ("job-name=parent-job", "gcp-project=a-project", "banana=1"), {"job_name": "parent-job", "gcp_project": "a-project"}, 1, ), ), ) def test_parse_dependency_args( dependencies, expected_dependencies, num_warnings, caplog, job, ): assert expected_dependencies == job._parse_dependency_args(dependencies) assert num_warnings == len(caplog.records) @pytest.mark.parametrize( "unknown_args,expected", ( (("--foo", "foobar"), {"foo": "foobar"}), ( ("--foo", "foobar", "--bar", "barfoo"), {"foo": "foobar", "bar": "barfoo"}, ), (("--foo", "bar", "baz"), {"foo": ("bar", "baz")}), ( ("--dependency", "job_name=parent-job", "gcp_project=a-project"), { "dependencies": [ {"job_name": "parent-job", "gcp_project": "a-project"} ] }, ), ( ("--dependency", "job_name=parent-job"), {"dependencies": [{"job_name": "parent-job"}]}, ), (("--dependency", "banana=phone"), {},), ), ) def test_parse_unknown_args(unknown_args, expected, job): ret_args = job._parse_unknown_args(unknown_args) assert expected == ret_args @pytest.mark.parametrize("create_dockerfile", (True, False)) @pytest.mark.parametrize("use_fnapi", (True, False)) @pytest.mark.parametrize("create_resources", (True, False)) def test_create( use_fnapi, create_dockerfile, create_resources, mocker, caplog, job ): context = { "job_name": "test-job", "use_fnapi": use_fnapi, "create_resources": create_resources, "job_type": "streaming", } mock_get_user_input = mocker.patch.object(job, "_get_user_input") mock_get_user_input.return_value = (context, create_dockerfile) mock_get_environment = mocker.patch.object(job, "_get_environment") mock_create_job_dir = mocker.patch.object(job, "_create_job_directory") mock_create_job_config = mocker.patch.object(job, "_create_job_config") mock_create_no_fnapi_files = mocker.patch.object( job, "_create_no_fnapi_files" ) mock_create_python_files = mocker.patch.object(job, "_create_python_files") mock_create_reqs_files = mocker.patch.object(job, "_create_reqs_file") mock_create_dockerfile = mocker.patch.object(job, "_create_dockerfile") mock_create_readme = mocker.patch.object(job, "_create_readme") mock_create_topics = mocker.patch.object(create.gcp_setup, "create_topics") mock_create_buckets = mocker.patch.object( create.gcp_setup, "create_buckets" ) mock_create_stackdriver = mocker.patch.object( create.gcp_setup, "create_stackdriver_dashboard" ) unknown_args = ("--foo", "bar") known_args = { "job_name": "test-job", "gcp-project": "test-gcp-project", } output_dir = "/testing/dir" job.create(unknown_args, known_args, output_dir) known_args["foo"] = "bar" mock_get_user_input.assert_called_once_with(known_args) mock_get_environment.assert_called_once_with() ret_env = mock_get_environment.return_value job_name = context["job_name"] package_name = job_name.replace("-", "_") mock_create_job_dir.assert_called_once_with(output_dir) mock_create_job_config.assert_called_once_with( ret_env, context, output_dir ) job_type = context.get("job_type") mock_create_python_files.assert_called_once_with( ret_env, package_name, job_type, output_dir ) if use_fnapi: mock_create_no_fnapi_files.assert_not_called() else: mock_create_no_fnapi_files.assert_called_once_with( ret_env, context, output_dir ) if create_resources: if job_type == "streaming": mock_create_topics.assert_called_once_with(context) mock_create_buckets.assert_called_once_with(context) mock_create_stackdriver.assert_called_once_with(context) else: mock_create_topics.assert_not_called() mock_create_buckets.assert_not_called() mock_create_stackdriver.assert_not_called() mock_create_reqs_files.assert_called_once_with( ret_env, context, output_dir ) if create_dockerfile: mock_create_dockerfile.assert_called_once_with( ret_env, context, output_dir ) mock_create_readme.assert_called_once_with(ret_env, context, output_dir) assert 1 == len(caplog.records) def test_get_batch_user_context(job, mock_prompt): kwargs = {"job_name": "test-job"} event_input = "events-in" data_input = "data-in" event_output = "events-out" data_output = "data-out" mock_prompt.side_effect = [ event_input, data_input, event_output, data_output, ] ret_context = job._get_batch_user_input_job_context(kwargs) expected_context = { "inputs": [ {"event_location": "events-in", "data_location": "data-in"} ], "outputs": [ {"event_location": "events-out", "data_location": "data-out"} ], } assert 4 == mock_prompt.call_count assert expected_context == ret_context def test_get_batch_user_context_no_prompt(job, mock_prompt): kwargs = { "job_name": "test-job", "batch_event_input": "input_ids.txt", "batch_event_output": "output_ids", "batch_data_input": "input-data", "batch_data_output": "output-data", } ret_context = job._get_batch_user_input_job_context(kwargs) mock_prompt.assert_not_called() expected_context = { "inputs": [ { "event_location": kwargs.get("batch_event_input"), "data_location": kwargs.get("batch_data_input"), } ], "outputs": [ { "event_location": kwargs.get("batch_event_output"), "data_location": kwargs.get("batch_data_output"), } ], } assert expected_context == ret_context def test_get_default_batch_job_context(job): kwargs = {"job_name": "test-job"} ret_context = job._get_default_batch_job_context(kwargs) expected_context = { "inputs": [ { "event_location": "test-job_input_elements.txt", "data_location": "test-job-input", } ], "outputs": [ { "event_location": "test-job_output_elements", "data_location": "test-job-output", } ], } assert expected_context == ret_context
gaybro8777/klio
core/tests/test_utils.py
<gh_stars>100-1000 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import os from unittest import mock import pytest from google.api_core import exceptions as gapi_exceptions from klio_core import utils def test_set_global(): utils.set_global("set-a-value", "a-value") actual = getattr(utils, "klio_global_state_set-a-value", None) assert "a-value" == actual delattr(utils, "klio_global_state_set-a-value") def test_get_global(): setattr(utils, "klio_global_state_get-a-value", "a-value") actual = utils.get_global("get-a-value") assert "a-value" == actual delattr(utils, "klio_global_state_get-a-value") @pytest.mark.parametrize("exists", (True, False)) def test_delete_global(exists): if exists: setattr(utils, "klio_global_state_delete-a-value", "a-value") utils.delete_global("delete-a-value") actual = getattr(utils, "klio_global_state_delete-a-value", None) assert not actual @pytest.mark.parametrize( "set_value,callable_init", ((True, False), (False, True), (False, False)) ) def test_get_or_initialize_global(set_value, callable_init, mocker): if set_value: setattr(utils, "klio_global_state_get-or-init-value", "a-value") if callable_init: initializer = mocker.Mock(return_value="a-value") else: initializer = "a-value" actual = utils.get_or_initialize_global("get-or-init-value", initializer) assert "a-value" == actual if not isinstance(initializer, str): initializer.assert_called_once_with() delattr(utils, "klio_global_state_get-or-init-value") @pytest.fixture def mock_publisher(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(utils.pubsub, "PublisherClient", mock) return mock def test_private_get_publisher(mock_publisher): ret_publisher = utils._get_publisher("a-topic") mock_publisher.assert_called_once_with() mock_publisher.return_value.create_topic.assert_called_once_with("a-topic") assert mock_publisher.return_value == ret_publisher def test_private_get_publisher_topic_exists(mock_publisher): client = mock_publisher.return_value client.create_topic.side_effect = gapi_exceptions.AlreadyExists("foo") ret_publisher = utils._get_publisher("a-topic") mock_publisher.assert_called_once_with() client.create_topic.assert_called_once_with("a-topic") assert client == ret_publisher def test_private_get_publisher_raises(mock_publisher): client = mock_publisher.return_value client.create_topic.side_effect = Exception("foo") with pytest.raises(Exception, match="foo"): utils._get_publisher("a-topic") mock_publisher.assert_called_once_with() client.create_topic.assert_called_once_with("a-topic") @pytest.mark.parametrize("in_globals", (True, False)) def test_get_publisher(in_globals, mock_publisher): client = mock_publisher.return_value if in_globals: setattr(utils, "klio_global_state_publisher_a-topic", client) ret_publisher = utils.get_publisher("a-topic") assert client == ret_publisher delattr(utils, "klio_global_state_publisher_a-topic") ######################### # Config Utils tests ######################### @pytest.fixture def patch_os_getcwd(monkeypatch, tmpdir): test_dir = str(tmpdir.mkdir("testing")) monkeypatch.setattr(os, "getcwd", lambda: test_dir) return test_dir def test_get_config_by_path(mocker, monkeypatch): m_open = mocker.mock_open() mock_open = mocker.patch("klio_core.utils.open", m_open) mock_safe_load = mocker.Mock() monkeypatch.setattr(utils.yaml, "safe_load", mock_safe_load) path = "path/to/a/file" utils.get_config_by_path(path) mock_open.assert_called_once_with(path) def test_get_config_by_path_error(mocker, monkeypatch, caplog): m_open = mocker.mock_open() mock_open = mocker.patch("klio_core.utils.open", m_open) mock_open.side_effect = IOError with pytest.raises(SystemExit): path = "path/to/a/file" utils.get_config_by_path(path) assert 1 == len(caplog.records) ######################### # Cli/exec command tests ######################### @pytest.mark.parametrize( "image", ( "dataflow.gcr.io/v1beta3/python", "dataflow.gcr.io/v1beta3/python-base", "dataflow.gcr.io/v1beta3/python-fnapi", ), ) def test_warn_if_py2_job(image, patch_os_getcwd, mocker): dockerfile = ( '## -*- docker-image-name: "gcr.io/foo/bar" -*-\n' "FROM {image}:1.2.3\n" 'LABEL maintainer "<EMAIL>"\n' ).format(image=image) m_open = mock.mock_open(read_data=dockerfile) mock_open = mocker.patch("klio_core.utils.open", m_open) m_is_file = mocker.Mock() m_is_file.return_value = True mock_is_file = mocker.patch("klio_core.utils.os.path.isfile", m_is_file) warn_msg = ( "Python 2 support in Klio is deprecated. " "Please upgrade to Python 3.5+" ) with pytest.warns(UserWarning, match=warn_msg): utils.warn_if_py2_job(patch_os_getcwd) exp_read_file = os.path.join(patch_os_getcwd, "Dockerfile") mock_is_file.assert_called_once_with(exp_read_file) mock_open.assert_called_once_with(exp_read_file, "r") @pytest.mark.parametrize("has_from_line", (True, False)) @pytest.mark.parametrize("file_exists", (True, False)) def test_warn_if_py2_job_no_warn( has_from_line, file_exists, patch_os_getcwd, mocker ): from_line = "\n" if has_from_line: from_line = "FROM dataflow.gcr.io/v1beta3/python36-fnapi:1.2.3\n" dockerfile = ( '## -*- docker-image-name: "gcr.io/foo/bar" -*-\n' + from_line + 'LABEL maintainer "<EMAIL>"\n' ) m_open = mock.mock_open(read_data=dockerfile) mock_open = mocker.patch("klio_core.utils.open", m_open) m_is_file = mocker.Mock() m_is_file.return_value = file_exists mock_is_file = mocker.patch("klio_core.utils.os.path.isfile", m_is_file) utils.warn_if_py2_job(patch_os_getcwd) exp_read_file = os.path.join(patch_os_getcwd, "Dockerfile") mock_is_file.assert_called_once_with(exp_read_file) if file_exists: mock_open.assert_called_once_with(exp_read_file, "r") @pytest.mark.parametrize( "job_dir,conf_file", ( (None, None), (None, "klio-job2.yaml"), ("foo/bar", None), ("foo/bar", "klio-job2.yaml"), ), ) def test_get_config_job_dir(job_dir, conf_file, patch_os_getcwd): exp_job_dir = patch_os_getcwd if job_dir: exp_job_dir = os.path.abspath(os.path.join(patch_os_getcwd, job_dir)) exp_conf_file = conf_file or os.path.join(exp_job_dir, "klio-job.yaml") if job_dir and conf_file: exp_conf_file = os.path.join(job_dir, conf_file) ret_job_dir, ret_conf_file = utils.get_config_job_dir(job_dir, conf_file) assert exp_job_dir == ret_job_dir assert exp_conf_file == ret_conf_file
gaybro8777/klio
exec/src/klio_exec/commands/audit_steps/multithreaded_tf.py
<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import sys from klio_exec.commands.audit_steps import base class MultithreadedTFUsage(base.BaseKlioAuditStep): """Use caution when running tensorflow in a multithreaded environment.""" AUDIT_STEP_NAME = "multithreaded_tf" @property def _is_tensorflow_loaded(self): return any(["tensorflow" in module for module in sys.modules]) @property def _is_job_single_threaded_per_container(self): exps = self.klio_config.pipeline_options.experiments return "worker_threads=1" in exps def after_tests(self): if not self._is_job_single_threaded_per_container: if self._is_tensorflow_loaded: self.emit_warning( "TensorFlow usage detected within job, but " "`worker_threads` is not explicitly set to 1 under " "`pipeline_options.experiments` in the job's configuration " "file! This can cause threading issues. Be careful." ) _init = MultithreadedTFUsage
gaybro8777/klio
exec/tests/unit/commands/test_audit.py
<reponame>gaybro8777/klio<filename>exec/tests/unit/commands/test_audit.py<gh_stars>100-1000 # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest from klio_exec.commands import audit @pytest.fixture def mock_pytest(mocker, monkeypatch): mock_pt = mocker.Mock() monkeypatch.setattr(audit, "pytest", mock_pt) return mock_pt @pytest.fixture def mock_get_audit_steps(mocker, monkeypatch): mock = mocker.Mock() monkeypatch.setattr(audit, "_get_audit_steps", mock) return mock @pytest.fixture def mock_run_pytest(mocker, monkeypatch): mock = mocker.Mock(return_value=False) monkeypatch.setattr(audit, "_run_pytest", mock) return mock @pytest.mark.parametrize( "exit_code,exp_pytest_failed,exp_msg,exp_kwargs", ( (1, True, "PyTest failed!\n", {"yellow": True}), (0, False, "PyTest passed!\n", {}), ), ) def test_run_pytest( exit_code, exp_pytest_failed, exp_msg, exp_kwargs, mock_pytest, mock_terminal_writer, mocker, ): mock_pytest.main.return_value = exit_code actual_pytest_failed = audit._run_pytest(mock_terminal_writer) assert exp_pytest_failed == actual_pytest_failed exp_calls = [ mocker.call("Running tests for audit validation...\n"), mocker.call(exp_msg, **exp_kwargs), ] mock_terminal_writer.write.assert_has_calls(exp_calls) def test_get_audit_steps(mock_terminal_writer, mocker, monkeypatch): mock_steps = [mocker.Mock() for i in range(3)] mock_get_plugins = mocker.Mock() mock_get_plugins.return_value = mock_steps monkeypatch.setattr( audit.plugin_utils, "load_plugins_by_namespace", mock_get_plugins ) ret = audit._get_audit_steps("job/dir", "config", mock_terminal_writer) mock_get_plugins.assert_called_once_with("klio.plugins.audit") assert ret == [s.return_value for s in mock_steps] for s in mock_steps: s.assert_called_once_with("job/dir", "config", mock_terminal_writer) @pytest.mark.parametrize("param_value", (None, "something")) @pytest.mark.parametrize("res_parsing", (True, False)) def test_list_audit_steps( res_parsing, param_value, mock_terminal_writer, mocker, monkeypatch ): mock_click_context = mocker.Mock() mock_click_context.resilient_parsing = res_parsing mock_print_plugins = mocker.Mock() monkeypatch.setattr( audit.plugin_utils, "print_plugins", mock_print_plugins ) audit.list_audit_steps(mock_click_context, None, param_value) if not param_value or res_parsing: mock_terminal_writer.sep.assert_not_called() mock_print_plugins.assert_not_called() mock_click_context.exit.assert_not_called() else: mock_terminal_writer.sep.assert_called_once_with( "=", "Installed audit steps" ) mock_print_plugins.assert_called_once_with( "klio.plugins.audit", mock_terminal_writer ) mock_click_context.exit.assert_called_once_with() @pytest.mark.parametrize("steps_passed", (0, 1)) @pytest.mark.parametrize("steps_warned", (0, 1)) def test_audit_steps( steps_passed, steps_warned, mock_terminal_writer, mock_get_audit_steps, mock_run_pytest, mocker, ): mock_passed = [] for i in range(steps_passed): name = "PassedStep{}".format(i + 1) mock_passed.append( mocker.Mock(AUDIT_STEP_NAME=name, warned=False, errored=False) ) mock_warned = [] for i in range(steps_warned): name = "WarnStep{}".format(i + 1) mock_warned.append( mocker.Mock(AUDIT_STEP_NAME=name, warned=True, errored=False) ) mock_audit_steps = mock_passed + mock_warned mock_get_audit_steps.return_value = mock_audit_steps audit.audit("job/dir", "config") mock_get_audit_steps.assert_called_once_with( "job/dir", "config", mock_terminal_writer ) for step in mock_audit_steps: step.before_tests.assert_called_once_with() mock_run_pytest.assert_called_once_with(mock_terminal_writer) # don't really care about the message strings, just that these funcs have # been called assert 2 == mock_terminal_writer.sep.call_count assert len(mock_passed) + 2 == mock_terminal_writer.write.call_count def test_audit_steps_errored( mock_terminal_writer, mock_get_audit_steps, mock_run_pytest, mocker ): mock_errored = [] for i in range(2): name = "WarnStep{}".format(i + 1) mock_errored.append( mocker.Mock(AUDIT_STEP_NAME=name, warned=False, errored=True) ) mock_get_audit_steps.return_value = mock_errored with pytest.raises(SystemExit): audit.audit("job/dir", "config") mock_get_audit_steps.assert_called_once_with( "job/dir", "config", mock_terminal_writer ) for step in mock_errored: step.before_tests.assert_called_once_with() mock_run_pytest.assert_called_once_with(mock_terminal_writer) # don't really care about the message strings, just that these funcs have # been called assert 3 == mock_terminal_writer.sep.call_count assert 1 == mock_terminal_writer.write.call_count def test_audit_raises(mock_get_audit_steps, mock_terminal_writer, caplog): mock_get_audit_steps.side_effect = Exception("Blah") with pytest.raises(SystemExit) as exc_info: audit.audit("job/dir", "config") assert 1 == exc_info.value.code # don't really care about the message strings, just that these funcs have # been called assert 1 == mock_terminal_writer.write.call_count assert 1 == len(caplog.records)
gaybro8777/klio
lib/src/klio/metrics/stackdriver.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ Klio ships with a Stackdriver Log-based Metrics relay client. The client creates metrics objects in Stackdriver Monitoring based off of filters in Stackdriver Logging. For more information on Stackdriver's Log-based Metrics, see `related documentation <https://cloud.google.com/logging/docs/ logs-based-metrics/>`_. When running on Dataflow, the client is on by default with no additional configuration needed. This must be actively turned off in ``klio-job.yaml`` if not wanted (see below). The log-based metrics client is not available for direct runner. To explicitly turn off log-based metrics, in ``klio-job.yaml``: .. code-block:: yaml job_config: metrics: stackdriver_logger: false """ import logging from googleapiclient import discovery from googleapiclient import errors as gapi_errors from klio.metrics import logger class StackdriverLogMetricsClient(logger.MetricsLoggerClient): """Stackdriver client for transform metrics. Intended to be instantiated by :class:`klio.metrics.client.MetricsRegistry` and not by itself. Args: klio_config (klio_core.config.KlioConfig): the job's configuration. """ RELAY_CLIENT_NAME = "stackdriver_logger" def __init__(self, klio_config): super(StackdriverLogMetricsClient, self).__init__(klio_config) self.job_name = klio_config.job_name @property def _stackdriver_client(self): client = getattr(self._thread_local, "stackdriver_client", None) if not client: # FYI this does not make a network call client = discovery.build("logging", "v2") self._thread_local.stackdriver_client = client return self._thread_local.stackdriver_client # get a list of metrics first and use any already-available metrics # to create metrics objects def counter(self, name, transform=None, tags=None, **kwargs): """Create a :class:`StackdriverLogMetricsCounter` object. .. note:: Stackdriver counts log lines so initializing a counter value is not supported . Args: name (str): name of counter transform (str): transform the counter is associated with tags (dict): any tags of additional contextual information to associate with the counter Returns: StackdriverLogMetricsCounter: a log-based counter """ # Since stackdriver literally counts loglines, initializing a # counter value is not supported if kwargs.get("value", 0) > 0: self.logger.log( logging.WARNING, "Initializing Stackdriver log-based counters with a value " "other than 0 is not supported. Defaulting to 0.", ) ctr = StackdriverLogMetricsCounter( name=name, job_name=self.job_name, project=self.klio_config.pipeline_options.project, transform=transform, tags=tags, ) ctr._init_metric(self._stackdriver_client) return ctr def gauge(self, *args, **kwargs): """Create a :class:`StackdriverLogMetricsGauge` object. .. warning:: Gauges for Stackdriver are not yet supported. This will default to standard logging. Args: name (str): name of gauge value (int): starting value of gauge; defaults to 0 transform (str): transform the gauge is associated with tags (dict): any tags of additional contextual information to associate with the gauge Returns: StackdriverLogMetricsGauge: a log-based gauge """ self.logger.log( logging.WARNING, "Gauge is not supported for Stackdriver log-based metrics, " "defaulting to standard logger.", ) return StackdriverLogMetricsGauge(*args, **kwargs) def timer(self, *args, **kwargs): """Create a :class:`StackdriverLogMetricsTimer` object. .. warning:: Timers for Stackdriver are not yet supported. This will default to standard logging. Args: name (str): name of gauge value (int): starting value of gauge; defaults to 0 transform (str): transform the gauge is associated with tags (dict): any tags of additional contextual information to associate with the gauge Returns: StackdriverLogMetricsTimer: a log-based timer """ self.logger.log( logging.WARNING, "Timer is not supported for Stackdriver log-based metrics, " "defaulting to standard logger.", ) return StackdriverLogMetricsTimer(*args, **kwargs) class StackdriverLogMetricsCounter(logger.LoggerCounter): """Stackdriver log-based counter metric. .. note:: Stackdriver counts log lines so initializing a counter value is not supported . Args: name (str): name of counter job_name (str): name of Dataflow job project (str): name of GCP project associated with Dataflow job transform (str): Name of transform associated with counter, if any. tags (dict): Tags to associate with counter. Note: ``{"metric_type": "counter"}`` will always be an included tag. """ # NOTE: The in-memory value (as kept track in the metric Dispatchers) may # be greater than 1, but it doesn't mean anything because: # 1. log-based counters in SD count the number log lines, and does not # extract a value from the log line, and # 2. state is not maintained within a pipeline, so the in-memory # value will eventually be wiped and re-initialized. # Therefore, we're hard-coding counters to the value of 1 DEFAULT_LOG_FORMAT = ( "[{name}] value: 1 transform: '{transform}' tags: {tags}" ) KLIO_TRANSFORM_LABEL_KEY = "klio_transform" def __init__(self, name, job_name, project, transform=None, tags=None): # Since stackdriver literally counts loglines, initializing a # counter value is not supported; defaulting to 0 super(StackdriverLogMetricsCounter, self).__init__( name, value=0, transform=transform, tags=tags ) self.job_name = job_name self.project = project self.parent = "projects/{}".format(project) self.desc = "Klio counter '{}' ".format(self.name) self.body = self._get_body() def _get_filter(self): _filter = ( 'resource.type="dataflow_step" ' 'logName="projects/{project}/logs/' 'dataflow.googleapis.com%2Fworker" ' # since stackdriver counters do not support regexes, we're # including the [] around the name to avoid name collisions like # my-counter & my-counter2 'jsonPayload.message:"[{name}]"'.format( project=self.project, name=self.name ) ) return _filter def _get_transform_label_extractor(self): # Grab the transform name from the log line, i.e.: # "[my-counter] value: 1 transform: 'HelloKlio' tags: {'metric_..."; # Needs to be able to follow valid Python chars for class names: # https://stackoverflow.com/a/10120327/1579977 label_extractor = "\"transform: '([a-zA-Z_][a-zA-Z0-9_]*)' tags:\"" # docs: https://cloud.google.com/logging/docs/ # logs-based-metrics/labels#create-label label_regex = "REGEXP_EXTRACT(jsonPayload.message, {})".format( label_extractor ) return {self.KLIO_TRANSFORM_LABEL_KEY: label_regex} def _get_body(self): labels = [ { "key": self.KLIO_TRANSFORM_LABEL_KEY, "valueType": "STRING", "description": "Name of Klio-based transform", } ] body = { "name": self.name, "description": self.desc, "filter": self._get_filter(), "metricDescriptor": { "metricKind": "DELTA", "valueType": "INT64", "unit": "1", "labels": labels, }, "labelExtractors": self._get_transform_label_extractor(), } return body def _init_metric(self, stackdriver_client): req = ( stackdriver_client.projects() .metrics() .create(parent=self.parent, body=self.body) ) try: req.execute() except gapi_errors.HttpError as e: # HTTP Conflict - metric already exists if e.resp.status == 409: logging.getLogger("klio.metrics").debug( "Metric {} already exists".format(self.name) ) return logging.getLogger("klio.metrics").error( "Error creating metric '{}': {}".format(self.name, e), exc_info=True, ) except Exception as e: logging.getLogger("klio.metrics").error( "Error creating metric '{}': {}".format(self.name, e), exc_info=True, ) # including method for documentation purposes class StackdriverLogMetricsGauge(logger.LoggerGauge): """Pass-thru object for naming only. Stackdriver log-based metrics does not support gauges. """ class StackdriverLogMetricsTimer(logger.LoggerTimer): """Pass-thru object for naming only. Stackdriver log-based metrics does not support timers. """
gaybro8777/klio
cli/src/klio_cli/commands/job/delete.py
<gh_stars>100-1000 # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import click from google.cloud import pubsub_v1 from google.cloud import storage from klio_cli.utils import stackdriver_utils as sd_utils class DeleteJob(object): resource_types = ["topic", "subscription", "location"] def __init__(self, config): self.config = config self._pipeline_options = config.pipeline_options self._job_config = config.job_config def _confirmation_dialog(self, resource_type, name): yes_no = click.confirm( "Do you want to delete {} {}? ".format(resource_type, name) ) if yes_no: confirmation = click.prompt( "Please confirm by typing out " "the name of the {}".format(resource_type) ) if confirmation != name: raise ValueError( "Invalid match. {} != {} " "Are you sure you meant to delete the resource? " "Please rerun this command and " "choose 'No' next time.".format(confirmation, name) ) else: return False return True def _get_resources(self): # Go through the list of resources set up # in input/output and ask for user input # on each one that we should delete. # The following are the types of resources we are considering: # 1. topics # 2. subscriptions # 3. buckets to_delete = { resource_type: [] for resource_type in self.resource_types } to_delete["stackdriver_group"] = False ev_inputs = self._job_config.events.inputs ev_outputs = self._job_config.events.outputs for resource in ev_inputs + ev_outputs: if "pubsub" == resource.name: if self._confirmation_dialog("topic", resource.topic): to_delete["topic"].append(resource.topic) if not hasattr(resource, "subscription"): continue if self._confirmation_dialog( "subscription", resource.subscription ): to_delete["subscription"].append(resource.subscription) data_inputs = self._job_config.data.inputs data_outputs = self._job_config.data.outputs for resource in data_inputs + data_outputs: if "gcs" == resource.name: if self._confirmation_dialog("location", resource.location): to_delete["location"].append(resource.location) # Now lets handle the stackdriver group _, dashboard_name = sd_utils.generate_group_meta( self._pipeline_options.project, self.config.job_name, self._pipeline_options.region, ) to_delete["stackdriver_group"] = self._confirmation_dialog( "stackdriver dashboard group", dashboard_name ) return to_delete def _delete_subscriptions(self, subscriptions): if not subscriptions: return client = pubsub_v1.SubscriberClient() for subscription in subscriptions: logging.info("Deleting subscription {}".format(subscription)) try: client.delete_subscription( request={"subscription": subscription} ) except Exception: logging.error( "Failed to delete subscription {}".format(subscription), exc_info=True, ) def _delete_topics(self, topics): if not topics: return client = pubsub_v1.PublisherClient() for topic in topics: logging.info("Deleting topic {}".format(topic)) try: client.delete_topic(request={"topic": topic}) except Exception: logging.error( "Failed to delete topic {}".format(topic), exc_info=True ) def _delete_buckets(self, project, data_locations): if not data_locations: return client = storage.Client(project=project) for data_location in data_locations: logging.info("Deleting data_location {}".format(data_location)) if data_location.startswith("gs://"): try: client.get_bucket(data_location.split("/")[2]).delete( force=True ) except Exception: logging.error( "Failed to delete bucket {}".format(data_location), exc_info=True, ) else: logging.info( "Skipping data location {}: Not a GCS location".format( data_location ) ) def delete(self): to_delete = self._get_resources() # Now we will go one by one through each resource and delete them all! self._delete_subscriptions(to_delete["subscription"]) self._delete_topics(to_delete["topic"]) self._delete_buckets( self._pipeline_options.project, to_delete["location"], ) if to_delete["stackdriver_group"]: sd_utils.delete_stackdriver_group( self._pipeline_options.project, self.config.job_name, self._pipeline_options.region, )
gaybro8777/klio
exec/tests/unit/commands/audit_steps/test_tempfile_usage.py
<filename>exec/tests/unit/commands/audit_steps/test_tempfile_usage.py # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import tempfile from klio_exec.commands.audit_steps import tempfile_usage def test_tempfile_usage(klio_config, mock_emit_error, mocker, monkeypatch): inst = tempfile_usage.TempFileUsage("job/dir", klio_config, "tw") assert "TemporaryFile" == tempfile.TemporaryFile.__name__ # sanity check inst.before_tests() assert "MockTemporaryFile" == tempfile.TemporaryFile.__name__ assert inst._tempfile_used is False inst.after_tests() mock_emit_error.assert_not_called() with tempfile.TemporaryFile(): pass assert inst._tempfile_used is True inst.after_tests() # don't really care about message, just that it was called assert 1 == mock_emit_error.call_count
gaybro8777/klio
exec/src/klio_exec/commands/utils/profile_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging try: import matplotlib.pyplot as plt except ImportError: # pragma: no cover # can be a problem on MacOS w certain Python & OS versions # https://stackoverflow.com/a/34583958/1579977 import matplotlib matplotlib.use("agg") import matplotlib.pyplot as plt import numpy as np def _get_profiling_data(filename): """Read a given file and parse its content for profiling data.""" data, timestamps = [], [] try: with open(filename, "r") as f: file_data = f.readlines() except Exception: logging.error("Could not read profiling data.", exc_info=True) raise SystemExit(1) for line in file_data: if line == "\n": continue line = line.strip() line_data = line.split(" ") if len(line_data) != 3: continue _, mem_usage, timestamp = line.split(" ") data.append(float(mem_usage)) timestamps.append(float(timestamp)) if not data: logging.error("No samples to parse in {}.".format(filename)) raise SystemExit(1) return {"data": data, "timestamp": timestamps} def plot(input_file, output_file, x_label, y_label, title): """Plot profiling data.""" profile_data = _get_profiling_data(input_file) data = np.asarray(profile_data["data"]) timestamp = np.asarray(profile_data["timestamp"]) global_start = float(timestamp[0]) t = timestamp - global_start # start at 0 rather than a specific time plt.figure(figsize=(14, 6), dpi=90) plt.plot(t, data, "+-c") # c is for `cyan` plt.xlabel(x_label) plt.ylabel(y_label) plt.title(title) plt.grid() plt.savefig(output_file)
gaybro8777/klio
lib/tests/unit/metrics/test_logger.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import threading import pytest from klio.metrics import logger @pytest.fixture def client(klio_config): return logger.MetricsLoggerClient(klio_config) @pytest.fixture def metric(): return logger.LoggerCounter(name="fixture-counter") @pytest.mark.parametrize( "metrics_config,exp_log_level", ( ({}, logging.DEBUG), ({"logger": True}, logging.DEBUG), ({"logger": {"key": "value"}}, logging.DEBUG), ({"logger": {"level": "info"}}, logging.INFO), ({"logger": {"level": "INFO"}}, logging.INFO), ({"logger": {"level": "notalevel"}}, logging.DEBUG), ), ) def test_client_set_log_level(metrics_config, exp_log_level, klio_config): klio_config.job_config.metrics = metrics_config client = logger.MetricsLoggerClient(klio_config) assert exp_log_level == client.log_level @pytest.mark.parametrize( "metrics_config,exp_timer_unit", ( ({}, "ns"), ({"logger": True}, "ns"), ({"logger": {"timer_unit": "seconds"}}, "s"), ({"logger": {"timer_unit": "notaunit"}}, "ns"), ), ) def test_client_set_timer_unit(metrics_config, exp_timer_unit, klio_config): klio_config.job_config.metrics = metrics_config client = logger.MetricsLoggerClient(klio_config) assert exp_timer_unit == client.timer_unit def test_client_logger(client): ret_logger = client.logger assert logging.getLogger("klio.metrics") == ret_logger def test_client_logger_local(client, monkeypatch): logger_patch = logging.getLogger("klio.metrics.patch") thread_local_patch = threading.local() thread_local_patch.klio_metrics_logger = logger_patch monkeypatch.setattr(client, "_thread_local", thread_local_patch) ret_logger = client.logger assert logger_patch == ret_logger def test_client_unmarshal(client, metric): exp_metric_dict = { "name": "fixture-counter", "value": 0, "transform": None, "tags": {"metric_type": "counter"}, } unmarshalled = client.unmarshal(metric) assert exp_metric_dict == unmarshalled def test_client_emit(client, metric, caplog): exp_log_record = ( "[fixture-counter] value: 0 transform: 'None' " "tags: {'metric_type': 'counter'}" ) client.emit(metric) assert 1 == len(caplog.records) assert exp_log_record == caplog.records[0].message def test_client_counter(client): counter = client.counter(name="my-counter") assert isinstance(counter, logger.LoggerCounter) def test_client_gauge(client): gauge = client.gauge(name="my-gauge") assert isinstance(gauge, logger.LoggerGauge) @pytest.mark.parametrize( "timer_unit,config_timer_unit,exp_timer_unit", ( (None, "ns", "ns"), (None, None, "ns"), ("s", None, "s"), ("microseconds", None, "us"), ("notaunit", None, "ns"), ), ) def test_client_timer( timer_unit, config_timer_unit, exp_timer_unit, client, monkeypatch ): if config_timer_unit: monkeypatch.setattr(client, "timer_unit", config_timer_unit) timer = client.timer(name="my-timer", timer_unit=timer_unit) assert isinstance(timer, logger.LoggerTimer) assert exp_timer_unit == timer.tags["unit"] def test_logger_counter(): expected_tags = {"metric_type": "counter"} counter = logger.LoggerCounter(name="my-counter") assert expected_tags == counter.tags def test_logger_gauge(): expected_tags = {"metric_type": "gauge"} gauge = logger.LoggerGauge(name="my-gauge") assert expected_tags == gauge.tags def test_logger_timer(): expected_tags = {"metric_type": "timer", "unit": "ns"} timer = logger.LoggerTimer(name="my-timer") assert expected_tags == timer.tags expected_tags["unit"] = "s" timer = logger.LoggerTimer(name="my-timer", timer_unit="s") assert expected_tags == timer.tags # pseudo-integration test @pytest.mark.parametrize("disabled,exp", ((True, 0), (False, 1))) def test_logger_disabled(disabled, exp, client, metric, caplog, monkeypatch): caplog.set_level(logging.DEBUG, logger="klio.metrics") # sanity check / clear out thread local client._thread_local.klio_metrics_logger = None monkeypatch.setattr(client, "disabled", disabled) client.emit(metric) assert exp == len(caplog.records)
gaybro8777/klio
exec/src/klio_exec/commands/utils/memory_utils.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import functools import inspect import memory_profiler as mp from klio_exec.commands.utils import wrappers class KMemoryLineProfiler(wrappers.KLineProfilerMixin, mp.LineProfiler): @classmethod def _wrap_per_element_func(cls, func, stream=None): @functools.wraps(func) def wrapper(*args, **kwargs): prof = cls(backend="psutil") try: return prof(func)(*args, **kwargs) finally: mp.show_results(prof, stream=stream) return wrapper @classmethod def _wrap_per_element_gen(cls, func, stream=None): @functools.wraps(func) def wrapper(*args, **kwargs): prof = cls(backend="psutil") try: yield from prof(func)(*args, **kwargs) finally: mp.show_results(prof, stream=stream) return wrapper @classmethod def wrap_per_element(cls, func, **kwargs): if inspect.isgeneratorfunction(func): return cls._wrap_per_element_gen(func, **kwargs) return cls._wrap_per_element_func(func, **kwargs) @staticmethod def _wrap_maximum_func(prof, func): @functools.wraps(func) def wrapper(*args, **kwargs): return prof(func)(*args, **kwargs) return wrapper @staticmethod def _wrap_maximum_gen(prof, func): @functools.wraps(func) def wrapper(*args, **kwargs): yield from prof(func)(*args, **kwargs) return wrapper @classmethod def wrap_maximum(cls, prof, func, **_): if inspect.isgeneratorfunction(func): return cls._wrap_maximum_gen(prof, func) return cls._wrap_maximum_func(prof, func)
gaybro8777/klio
audio/src/klio_audio/decorators.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import collections import functools import inspect import io import pickle import types import apache_beam as beam import numpy as np from apache_beam import pvalue from klio.transforms import core # Actual KlioMessage will not take payload that's not bytes (i.e. numpy # arrays), so we just re-wrap it into a faked klio-like message (just the # data part) before handing it off to the user FakeKlioMsg = collections.namedtuple("FakeKlioMsg", ["element", "payload"]) def __is_method(obj): # we have to do this instead of inspect.ismethod since # it's not a "bounded" method, ie Foo.process (unbounded) # vs Foo().process (bounded) args = inspect.getfullargspec(obj).args if args: return "self" == args[0] return False def __is_dofn_process_method(self, meth): if meth.__name__ == "process": if issubclass(self.__class__, beam.DoFn): return True return False class _BinarySerializer(object): def __init__(self, skip_load, skip_dump, load_with_numpy, save_with_numpy): self.skip_load = skip_load self.skip_dump = skip_dump self.load_with_numpy = load_with_numpy self.save_with_numpy = save_with_numpy @staticmethod def _pickle_load(ctx, kmsg): try: payload = pickle.loads(kmsg.payload) return FakeKlioMsg(element=kmsg.element, payload=payload) except Exception as err: ctx.logger.error( "Exception occurred when unpickling payload for '%s'.\n" "Error: %s" % (kmsg.element, err) ) raise err @staticmethod def _numpy_load(ctx, kmsg): try: in_ = io.BytesIO(kmsg.payload) in_.seek(0) # push the read pointer to the beginning payload = np.load(in_, allow_pickle=True) return FakeKlioMsg(element=kmsg.element, payload=payload) except Exception as err: ctx.logger.error( "Exception occurred when loading numpy payload for '%s'.\n" "Error: %s" % (kmsg.element, err) ) raise err @staticmethod def _pickle_dump(ctx, kmsg, ret): tagged, tag = False, None if isinstance(ret, pvalue.TaggedOutput): tagged = True tag = ret.tag ret = ret.value try: dumped = pickle.dumps(ret) if tagged: return pvalue.TaggedOutput(tag, dumped) return dumped except Exception as err: ctx.logger.error( "Exception occurred when pickling payload for '%s'.\nError: %s" % (kmsg.element, err) ) raise err @staticmethod def _numpy_dump(ctx, kmsg, ret): tagged, tag = False, None if isinstance(ret, pvalue.TaggedOutput): tagged = True tag = ret.tag ret = ret.value try: out = io.BytesIO() np.save(out, ret) dumped = out.getvalue() # returns the data in `out` in bytes if tagged: return pvalue.TaggedOutput(tag, dumped) return dumped except Exception as err: ctx.logger.error( "Exception occurred when dumping numpy payload for '%s'.\n" "Error: %s" % (kmsg.element, err) ) raise err def load(self, ctx, kmsg): if self.skip_load: return kmsg if self.load_with_numpy: return self._numpy_load(ctx, kmsg) return self._pickle_load(ctx, kmsg) def dump(self, ctx, kmsg, ret_val): if self.skip_dump: return ret_val if self.save_with_numpy: return self._numpy_dump(ctx, kmsg, ret_val) return self._pickle_dump(ctx, kmsg, ret_val) # A separate function from __handle_binary_generator so we can specifically # `yield from` it (and exhaust transforms that have multiple yields) def __yield_from_handle_binary(serializer, self, new_msg, ret_generator): for ret in ret_generator: ret = serializer.dump(self._klio, new_msg, ret) yield ret def __handle_binary_generator(self, meth, kmsg, serializer, *args, **kwargs): new_msg = serializer.load(self._klio, kmsg) # any error should be caught further up (i.e. by `@handle_klio`) ret = meth(self, new_msg, *args, **kwargs) if not isinstance(ret, types.GeneratorType): ret = serializer.dump(self._klio, new_msg, ret) yield ret else: yield from __yield_from_handle_binary(serializer, self, new_msg, ret) def __handle_binary(ctx, func, kmsg, serializer, *args, **kwargs): _self = ctx if not isinstance(ctx, core.KlioContext): ctx = _self._klio new_msg = serializer.load(ctx, kmsg) # any error should be caught further up (i.e. by `@handle_klio`) ret = func(ctx, new_msg, *args, **kwargs) ret = serializer.dump(ctx, new_msg, ret) return ret def handle_binary(*decorator_args, **decorator_kwargs): """Decorator to handle the required loading/unloading of binary data. .. caution:: The ``@handle_binary`` decorator **must** be used in conjunction with the :func:`@handle_klio <klio.transforms.decorators.handle_klio>` decorator. As well, ``@handle_binary`` **must** also be applied **after** ``@handle_klio``. Example usage: .. code-block:: python class MelSpectrogram(beam.DoFn): @handle_klio @handle_binary def process(self, item): self._klio.logger.info( f"Generating specgram for {item.element}" ) audio = item.payload yield librosa.feature.melspectrogram(y=audio, sr=22050) @handle_klio @handle_binary(skip_dump=True) def save_plt_as_png(ctx, item): fig = item.payload output = os.path.join(".", item.element.decode("utf-8") + ".png") plt.savefig(output, format="png", transparent=True, pad_inches=0) ctx.logger.info(f"Saved spectrogram: {output}") return output class DownloadAudio(beam.DoFn): def setup(self): self.client = SomeClient() @handle_klio @handle_binary(skip_load=True, save_with_numpy=True) def process(self, item): self._klio.logger.info(f"Downloading {item.element}") filename = item.payload.decode("utf-8") location = self._klio.config.job_config.data.inputs[0].location source_path = os.path.join(location, filename) with self.client.open(source_path, "rb") as source: out = io.BytesIO(source.read()) self._klio.logger.info(f"Downloaded {item.element} to memory") yield out Args: skip_load (bool): Skip loading the ``KlioMessage`` payload via pickle. Set this to ``True`` if the incoming ``KlioMessage`` payload is not binary data, or otherwise has not been pickled to bytes. Default: ``False`` skip_dump (bool): Skip dumping the ``KlioMessage`` payload via pickle. Set this to ``True`` if the outgoing ``KlioMessage`` payload is not binary data, or otherwise should not be pickled to bytes. Default: ``False`` load_with_numpy (bool): Use :func:`numpy.load` instead of :func:`pickle.loads` to load arrays or pickled numpy objects. This is less performant than standard pickling, but uses less memory. Default: ``False``. save_with_numpy (bool): Use :func:`numpy.save` instead of :func:`pickle.dumps` to save arrays as ``bytes``. This is less performant than standard pickling, but uses less memory. Default: ``False`` """ skip_load = decorator_kwargs.pop("skip_load", False) skip_dump = decorator_kwargs.pop("skip_dump", False) load_with_numpy = decorator_kwargs.pop("load_with_numpy", False) save_with_numpy = decorator_kwargs.pop("save_with_numpy", False) serializer = _BinarySerializer( skip_load, skip_dump, load_with_numpy, save_with_numpy ) def inner(func_or_meth): @functools.wraps(func_or_meth) def method_wrapper(self, kmsg, *args, **kwargs): wrapper = __handle_binary # Only the process method of a DoFn is a generator - otherwise # beam can't pickle a generator if __is_dofn_process_method(self, func_or_meth): wrapper = __handle_binary_generator return wrapper( self, func_or_meth, kmsg, serializer, *args, **kwargs ) @functools.wraps(func_or_meth) def func_wrapper(ctx, kmsg, *args, **kwargs): return __handle_binary( ctx, func_or_meth, kmsg, serializer, *args, **kwargs ) if __is_method(func_or_meth): return method_wrapper return func_wrapper # allows @handle_binary to be used without parens (i.e. no need to do # `@handle_binary()`) when there are no args/kwargs provided if decorator_args: return inner(decorator_args[0]) return inner
gaybro8777/klio
lib/tests/unit/conftest.py
<filename>lib/tests/unit/conftest.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import pytest from klio_core import config @pytest.fixture def caplog(caplog): """Set global test logging levels.""" caplog.set_level(logging.DEBUG) return caplog @pytest.fixture def job_config_dict(): return { "metrics": {"logger": {}}, "allow_non_klio_messages": False, "number_of_retries": 3, "inputs": [ { "topic": "test-parent-job-out", "subscription": "test-parent-job-out-sub", "data_location": "gs://sigint-output/test-parent-job-out", } ], "outputs": [ { "topic": "test-job-out", "data_location": "gs://sigint-output/test-job-out", } ], "dependencies": [ {"gcp_project": "sigint", "job_name": "test-parent-job"} ], "more": "config", "that": {"the": "user"}, "might": ["include"], } @pytest.fixture def pipeline_config_dict(): return { "project": "test-project", "staging_location": "gs://some/stage", "temp_location": "gs://some/temp", "worker_harness_container_image": "gcr.io/sigint/foo", "streaming": True, "experiments": ["beam_fn_api"], "region": "us-central1", "num_workers": 3, "max_num_workers": 5, "disk_size_gb": 50, "worker_machine_type": "n1-standard-4", "runner": "direct", "autoscaling_algorithm": "THROUGHPUT_BASED", "update": False, } @pytest.fixture def config_dict(job_config_dict, pipeline_config_dict): return { "job_config": job_config_dict, "pipeline_options": pipeline_config_dict, "job_name": "test-job", "version": 1, } @pytest.fixture def klio_config(config_dict): return config.KlioConfig(config_dict) @pytest.fixture def mock_config(mocker, monkeypatch): mconfig = mocker.Mock(name="MockKlioConfig") mconfig.job_name = "a-job" mconfig.pipeline_options.streaming = True mconfig.pipeline_options.project = "not-a-real-project" mock_data_input = mocker.Mock(name="MockDataGcsInput") mock_data_input.type = "gcs" mock_data_input.location = "gs://hopefully-this-bucket-doesnt-exist" mock_data_input.file_suffix = "" mock_data_input.skip_klio_existence_check = True mconfig.job_config.data.inputs = [mock_data_input] monkeypatch.setattr( "klio.transforms.core.RunConfig.get", lambda: mconfig, ) return mconfig
gaybro8777/klio
cli/src/klio_cli/utils/stackdriver_utils.py
<reponame>gaybro8777/klio<filename>cli/src/klio_cli/utils/stackdriver_utils.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging from google.cloud import monitoring STACKDRIVER_GROUP_BASE_URL = "https://app.google.stackdriver.com/groups" STACKDRIVER_GROUP_TPL = ( "{base_url}/{group_id}/{display_name}?project={project}" ) DASHBOARD_NAME_TPL = "{job_name}-{region}-klio-dashboard" def generate_group_meta(project, job_name, region): name = "projects/{}".format(project) dashboard_name = DASHBOARD_NAME_TPL.format( job_name=job_name, region=region ) return name, dashboard_name # callees of this function should wrap in try/except since the call to # client.list_groups does not def get_stackdriver_group_url(project, job_name, region): client = monitoring.GroupServiceClient() name, dashboard_name = generate_group_meta(project, job_name, region) groups = client.list_groups(request={"name": name}) for group in groups: if group.display_name == dashboard_name: group_id = group.name.split("/")[-1] return STACKDRIVER_GROUP_TPL.format( base_url=STACKDRIVER_GROUP_BASE_URL, group_id=group_id, display_name=dashboard_name, project=project, ) # NOTE: this will create a new group with the same name, rather than raise # an "Already Exists" error. def create_stackdriver_group(project, job_name, region): client = monitoring.GroupServiceClient() name, dashboard_name = generate_group_meta(project, job_name, region) group = { "display_name": dashboard_name, "filter": "resource.metadata.name=starts_with({})".format(job_name), } try: group = client.create_group(request={"name": name, "group": group}) except Exception as e: msg = ( "Could not create a Stackdriver for job '{}': {}. " "Skipping...".format(job_name, e) ) logging.error(msg) return group_id = group.name.split("/")[-1] url = STACKDRIVER_GROUP_TPL.format( base_url=STACKDRIVER_GROUP_BASE_URL, group_id=group_id, display_name=dashboard_name, project=project, ) msg = "Created dashboard '{}' for job '{}': {}".format( dashboard_name, job_name, url ) logging.info(msg) return url # Note: This will attempt to delete a group without retries. def delete_stackdriver_group(project, job_name, region): client = monitoring.GroupServiceClient() name, dashboard_name = generate_group_meta(project, job_name, region) try: for group in client.list_groups(request={"name": name}): if group.display_name == dashboard_name: client.delete_group( request={"name": group.name, "recursive": True} ) msg = "Deleted dashboard '{}' for job '{}'".format( dashboard_name, job_name ) logging.info(msg) return except Exception as e: msg = ( "Could not delete a Stackdriver for job '{}': {}. " "Skipping...".format(job_name, e) ) logging.error(msg) return logging.warning( "No dashboard for job '{}' could be found. Nothing deleted".format( job_name ) )
gaybro8777/klio
lib/tests/unit/metrics/test_base.py
<filename>lib/tests/unit/metrics/test_base.py # Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import pytest import six from klio.metrics import base as base_metrics class FakeObject(object): pass @pytest.mark.parametrize("obj", [None, FakeObject()]) def test_abstract_attr(obj): ret_obj = base_metrics.abstract_attr(obj) if obj: assert isinstance(ret_obj, FakeObject) else: assert isinstance(ret_obj, base_metrics._DummyAttribute) assert hasattr(ret_obj, "__isabstractattr__") def test__has_abstract_attributes_implemented_decorator(): class TestMeta(six.with_metaclass(base_metrics._ABCBaseMeta)): @property @base_metrics.abstract_attr def foo(self): pass class TestABC(TestMeta): foo = "bar" rc = TestABC() assert hasattr(rc, "foo") def test_has_abstract_attributes_implemented_cls_attr(): class TestMeta(six.with_metaclass(base_metrics._ABCBaseMeta)): foo = base_metrics.abstract_attr() class TestABC(TestMeta): foo = "bar" rc = TestABC() assert hasattr(rc, "foo") def test_has_abstract_attributes_implemented_raises(): class TestMeta(six.with_metaclass(base_metrics._ABCBaseMeta)): foo = base_metrics.abstract_attr() with pytest.raises(NotImplementedError): class TestABC(TestMeta): not_foo = "nope" def test_abstract_relay_client(): class RelayClient(base_metrics.AbstractRelayClient): RELAY_CLIENT_NAME = "test-relay-client" def unmarshal(self, metric): return {} def emit(self, metric): return {} def counter(self, name, value=0, transform=None): return {} def gauge(self, name, value=0, transform=None): return {} def timer(self, name, value=0, transform=None): return {} rc = RelayClient({}) assert hasattr(rc, "RELAY_CLIENT_NAME") assert hasattr(rc, "unmarshal") assert hasattr(rc, "emit") assert hasattr(rc, "counter") assert hasattr(rc, "gauge") assert hasattr(rc, "timer") def test_abstract_relay_client_raises(): with pytest.raises(NotImplementedError): class RelayClient(base_metrics.AbstractRelayClient): pass def test_base_metric(): class MyMetric(base_metrics.BaseMetric): def __init__(self, name, value=0, transform=None, tags=None, **kwargs): super(MyMetric, self).__init__( name, value=value, transform=transform ) self.tags = tags my_metric_inst = MyMetric(name="foo") assert isinstance(my_metric_inst, base_metrics.BaseMetric) assert hasattr(my_metric_inst, "name") assert hasattr(my_metric_inst, "value") assert hasattr(my_metric_inst, "transform") assert hasattr(my_metric_inst, "update") assert 0 == my_metric_inst.value my_metric_inst.update(1) assert 1 == my_metric_inst.value
gaybro8777/klio
integration/multi-event-input-batch/tests/test_run.py
<reponame>gaybro8777/klio # Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import logging import pytest from klio_core.proto import klio_pb2 import run @pytest.fixture def caplog(caplog): """Set global test logging levels.""" caplog.set_level(logging.DEBUG) return caplog @pytest.fixture def klio_msg(): element = b"s0m3_tr4ck_1d" msg = klio_pb2.KlioMessage() msg.data.element = element msg.version = klio_pb2.Version.V2 return msg @pytest.mark.parametrize( "func_to_test,exp_log_prefix", ( (run.first_func, "[first_func]:"), (run.second_func, "[second_func]:"), (run.combined_func, "[combined_func]:"), ) ) def test_process_funcs(func_to_test, exp_log_prefix, klio_msg, caplog): ret = func_to_test(klio_msg.SerializeToString()) assert klio_msg.SerializeToString() == ret assert 1 == len(caplog.records) exp_log_msg = f"{exp_log_prefix} {klio_msg.data.element}" assert exp_log_msg == caplog.records[0].message assert "INFO" == caplog.records[0].levelname
gaybro8777/klio
devtools/src/klio_devtools/cli.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import os import click from klio_cli import cli as main_cli from klio_cli import options from klio_cli.cli import main from klio_cli.utils import cli_utils from klio_core import config from klio_core import options as core_options from klio_core import utils as core_utils from klio_devtools.commands import develop @main.command( "develop", short_help="Develop on the klio ecosystem in a job's container.", help=( "Builds & runs a job's container, mounts the job's code in " "`/usr/src/app`, installs klio packages as 'editable' packages " "that will automatically pick up local changes, and attaches to " "the container with an interactive terminal to enable manual " "runs of `klioexec`.\n\nNOTE: It's probably a good idea to locally " "bump the versions of the libraries to ensure proper installation." ), ) @core_options.job_dir @core_options.config_file @core_options.image_tag(default=None, show_default="``git-sha[dirty?]``") @options.runtime @click.option( "--klio-path", type=click.Path( exists=True, dir_okay=True, file_okay=False, readable=True, writable=True, resolve_path=True, ), help="Path to klio repo", required=True, ) @click.option( "--exclude", help="exclude installing a particular package", multiple=True, ) def develop_job(job_dir, config_file, **kwargs): job_dir, config_path = core_utils.get_config_job_dir(job_dir, config_file) config_data = core_utils.get_config_by_path(config_path) conf = config.KlioConfig(config_data) git_sha = cli_utils.get_git_sha(job_dir, kwargs.get("image_tag")) image_tag = kwargs.get("image_tag") or git_sha if config_file: basename = os.path.basename(config_file) image_tag = "{}-{}".format(image_tag, basename) runtime_config = main_cli.DockerRuntimeConfig( image_tag=image_tag, force_build=kwargs.get("force_build"), config_file_override=config_file, ) klio_pipeline = develop.DevelopKlioContainer( job_dir, conf, runtime_config, kwargs["klio_path"], kwargs["exclude"] ) klio_pipeline.run()
gaybro8777/klio
lib/src/klio/transforms/_timeout.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # """ This module is heavily inspired by https://github.com/pnpnpn/timeout-decorator and https://github.com/bitranox/wrapt_timeout_decorator. It's been simplified for what we need, and adapted to work with generators. """ import multiprocessing import sys import time import traceback import types import psutil TRACEBACK_HEADER = "Traceback (most recent call last):\n" class KlioTimeoutError(Exception): """Transform timed out while processing a KlioMessage.""" def _target(queue, function, *args, **kwargs): """Run user's function with args/kwargs and return output via a queue.""" try: ret = function(*args, **kwargs) if isinstance(ret, types.GeneratorType): ret = next(ret) queue.put((True, ret)) except Exception: # Note: here we're trying to get the traceback context that's relevant # to the user. Without it, the user would see the timeout wrapper # context, which won't help the user. So we grab the previous traceback # (accessed by tb_next) and attach it to the exception as a new # attribute. Then when we catch the exception in the caller code, we # can check for that attribute and log it. exc_infos = sys.exc_info() user_exc = exc_infos[1] user_exc_tb = "".join(traceback.format_tb(exc_infos[2].tb_next)) user_exc_tb = TRACEBACK_HEADER + user_exc_tb setattr(user_exc, "_klio_traceback", user_exc_tb) queue.put((False, user_exc)) class KlioTimeoutWrapper(object): """Wrap a function to execute in a separate process with a timeout.""" DEFAULT_EXC_MSG = "Function '{}' timed out after {} seconds." def __init__( self, function, seconds, timeout_exception=None, exception_message=None ): self._function = function self._func_name = getattr(function, "__qualname__", function.__name__) self._seconds = seconds self._timeout_exception = timeout_exception or KlioTimeoutError self._exception_message = exception_message def __call__(self, *args, **kwargs): self._queue = multiprocessing.Queue(maxsize=1) args = (self._queue, self._function) + args proc_name = "KlioTimeoutProcess-{}".format(self._func_name) self._process = multiprocessing.Process( target=_target, name=proc_name, args=args, kwargs=kwargs ) # can't daemonize process in case users nest decorators self._process.daemon = False self._process.start() self._timeout = self._seconds + time.monotonic() while not self.ready: time.sleep(0.01) return self.value def _raise_exception(self): if self._exception_message is None: self._exception_message = self.DEFAULT_EXC_MSG.format( self._func_name, self._seconds ) raise self._timeout_exception(self._exception_message) def _terminate_child_processes(self): # This is to make sure we clean up any spawned child processes, # particularly if the user uses `subprocess` in their timeout- # decorated code proc = psutil.Process(pid=self._process.pid) children = proc.children(recursive=True) while children: # top-down/handle parents first; if we terminate children first, # the parents could just respawn child = children.pop(0) try: child.terminate() except psutil.NoSuchProcess: pass def cancel(self): """Terminate any possible execution of the process running the func.""" if self._process.is_alive(): self._terminate_child_processes() self._process.terminate() # join process so the parent process can reap it - no zombies self._process.join(timeout=0.1) self._raise_exception() @property def ready(self): """Manage the status of "value" property.""" if self._timeout < time.monotonic(): self.cancel() return self._queue.full() and not self._queue.empty() @property def value(self): """Get data from the processing queue that was added by the func.""" if self.ready is True: is_successful, ret_value = self._queue.get() if is_successful: return ret_value raise ret_value
gaybro8777/klio
integration/read-file/tests/test_transforms.py
# Copyright 2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import json import logging import os import pytest from klio_core.proto import klio_pb2 import transforms @pytest.fixture def caplog(caplog): """Set global test logging levels.""" caplog.set_level(logging.DEBUG) return caplog @pytest.fixture def klio_msg(): msg = klio_pb2.KlioMessage() msg.data.element = b"s0m3_tr4ck_1d" msg.version = klio_pb2.Version.V2 return msg @pytest.fixture def expected_log_messages(klio_msg): return [ ( "KlioThreadLimiter(name=LogKlioMessage.process) Blocked – " "waiting on semaphore for an available thread (available threads:" ), ( "KlioThreadLimiter(name=LogKlioMessage.process) Released " "semaphore (available threads:" ), "Hello, Klio!", "Received element {}".format(klio_msg.data.element), "Received payload {}".format(klio_msg.data.payload), ] def test_process(klio_msg, expected_log_messages, caplog): helloklio_fn = transforms.LogKlioMessage() output = helloklio_fn.process(klio_msg.SerializeToString()) assert klio_msg.SerializeToString() == list(output)[0] assert len(caplog.records) == len(expected_log_messages) for index, record in enumerate(caplog.records): assert "INFO" == record.levelname assert expected_log_messages[index] in record.message
gaybro8777/klio
cli/src/klio_cli/commands/job/stop.py
# Copyright 2019-2020 Spotify AB # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. # You may obtain a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. # import datetime import logging import time import emoji from googleapiclient import discovery JOB_STATE_MAP = {"cancel": "JOB_STATE_CANCELLED", "drain": "JOB_STATE_DRAINED"} class StopJob(object): def __init__(self, api_version=None): self._set_dataflow_client(api_version) def _set_dataflow_client(self, api_version): if not api_version: api_version = "v1b3" self._client = discovery.build("dataflow", api_version) def _check_job_running(self, job_name, project, region): request = ( self._client.projects() .locations() .jobs() .list(projectId=project, location=region, filter="ACTIVE",) ) try: response = request.execute() except Exception as e: logging.warning( "Could not find running job '{}' in project '{}': {}".format( job_name, project, e ) ) logging.warning( "Continuing to attempt deploying '{}'".format(job_name) ) return job_results = response.get("jobs", []) if job_results: for result in job_results: if result["name"] == job_name: return result def _update_job_state(self, job, req_state=None, retries=None): if retries is None: retries = 0 _req_state = JOB_STATE_MAP.get(req_state, JOB_STATE_MAP["cancel"]) if job.get("requestedState") is not _req_state: job["requestedState"] = _req_state request = ( self._client.projects() .locations() .jobs() .update( jobId=job["id"], projectId=job["projectId"], location=job["location"], body=job, ) ) try: request.execute() except Exception as e: # generic catch if 4xx error - probably shouldn't retry if getattr(e, "resp", None): if e.resp.status < 500: msg = "Failed to {} job '{}': {}".format( req_state, job["name"], e ) logging.error(msg) raise SystemExit(1) if retries > 2: msg = "Max retries reached: could not {} job '{}': {}".format( req_state, job["name"], e ) logging.error(msg) raise SystemExit(1) logging.info( "Failed to {} job '{}'. Trying again after 30s...".format( req_state, job["name"] ) ) retries += 1 time.sleep(30) self._update_job_state(job, req_state, retries) def _watch_job_state(self, job, timeout=600): timeout = datetime.datetime.now() + datetime.timedelta(seconds=timeout) request = ( self._client.projects() .locations() .jobs() .get( jobId=job["id"], projectId=job["projectId"], location=job["location"], ) ) while datetime.datetime.now() < timeout: try: resp = request.execute() except Exception as e: msg = ( "Failed to get current status for job '{}'. Error: {}.\n" "Trying again after 5s...".format(job["name"], e) ) logging.info(msg) time.sleep(5) continue if resp["currentState"] in JOB_STATE_MAP.values(): return else: msg = "Waiting for job '{}' to reach terminal state...".format( job["name"] ) logging.info(msg) time.sleep(5) msg = "Job '{}' did not reach terminal state after '{}' secs.".format( job["name"], timeout ) logging.error(msg) raise SystemExit(1) def stop(self, job_name, project, region, strategy, api_version=None): self._set_dataflow_client(api_version) current_running_job = self._check_job_running( job_name, project, region ) if not current_running_job: return self._update_job_state(current_running_job, req_state=strategy) self._watch_job_state(current_running_job) verb = "cancelled" if strategy == "cancel" else "drained" msg = "Successfully {} job '{}' :smile_cat:".format(verb, job_name) logging.info(emoji.emojize(msg, use_aliases=True))
harwiltz/mltools
mltools/functional.py
<filename>mltools/functional.py Id = lambda x: x def cmp(f, g): return lambda x: f(g(x)) def mul(f, g): return lambda x: f(x) * g(x) def mulk(f, k): return mul(f, const(k)) def add(f, g): return lambda x: f(x) + g(x) def addk(f, k): return add(f, const(k)) def negate(f): return lambda x: -f(x) def const(k): return lambda _: k
harwiltz/mltools
examples/memtest.py
<filename>examples/memtest.py import jax import jax.numpy as jnp from haiku import PRNGSequence from mltools.logging.html import HTMLExperimentLogger @profile def main(): monitor = HTMLExperimentLogger("memtest", template_path="/home/harwiltz/lab/mltools/examples/", template_name="html-logger-template.html") rngs = PRNGSequence(jax.random.PRNGKey(0)) i = 0 monitor.log_metric("dummy", jax.random.uniform(next(rngs), shape=(4,)).sum(), i) i += 1 monitor.log_metric("dummy", jax.random.uniform(next(rngs), shape=(4,)).sum(), i) i += 1 monitor.log_metric("dummy", jax.random.uniform(next(rngs), shape=(4,)).sum(), i) if __name__ == "__main__": main()
harwiltz/mltools
mltools/logging/__init__.py
from .experiment_logger import ExperimentLogger from .experiment_logger import CLIExperimentLogger
harwiltz/mltools
mltools/logging/experiment_logger.py
import logging class ExperimentLogger(object): def __init__(self, exp_name, tags=[], **kwargs): self.exp_name = exp_name self.epoch = 0 self.tags = tags def instantiate(logger_type, exp_name, **kwargs): logger_type = logger_type.lower() if (logger_type == 'default') or (logger_type == 'cli'): return CLIExperimentLogger(exp_name, **kwargs) if logger_type == 'comet': return CometExperimentLogger(exp_name, **kwargs) def log_metric(self, tag, value, step, **kwargs): raise NotImplementedError def log_image(self, tag, img, step, **kwargs): raise NotImplementedError def log_plt(self, tag, plt, step, **kwargs): raise NotImplementedError def log_text(self, tag, text, **kwargs): raise NotImplementedError def log_parameters(self, params, **kwargs): raise NotImplementedError def start_epoch(self, **kwargs): pass def end_epoch(self, **kwargs): self.epoch += 1 def end_experiment(self): raise NotImplementedError class CLIExperimentLogger(ExperimentLogger): def __init__(self, exp_name, log_level=logging.INFO, **kwargs): super(CLIExperimentLogger, self).__init__(exp_name, **kwargs) self._has_warned_img = False self._has_warned_plt = False self._logger = logging.getLogger("Experiment Logger") self._logger.setLevel(log_level) self._logger.debug(f"Instantiated logger for experiment \"{exp_name}\"") def log_metric(self, tag, value, step, **kwargs): self._logger.info("[METRIC] {:>32} ({:>5}): {:>16.11f}".format(tag, step, value)) def log_image(self, tag, img, step, **kwargs): if self._has_warned_img: return else: warn_message_pre = f"[ WARN] {self.__class__} does not support image logging" warn_message_suf = f"skipping log of {tag} ({step})" self._logger.warn(f"{warn_message_pre} -- {warn_message_suf}") self._has_warned_img = True def log_plt(self, tag, plt, step, **kwargs): if self._has_warned_plt: return else: warn_message_pre = f"[ WARN] {self.__class__} does not support plt logging" warn_message_suf = f"skipping log of {tag} ({step})" self._logger.warn(f"{warn_message_pre} -- {warn_message_suf}") self._has_warned_plt = True def log_text(self, tag, text, **kwargs): tag = tag[:min(6, len(tag))] self._logger.info("[{:>6}] {}".format(tag, text)) def log_parameters(self, params, **kwargs): self._logger.info("<PARAMS>") self._logger.info(params) self._logger.info("</PARAMS>") def start_epoch(self, **kwargs): super(CLIExperimentLogger, self).start_epoch() self._logger.debug(f"\n{64 * '='}\n") self._logger.debug(f"Starting epoch {self.epoch + 1}...") def end_experiment(self): self._logger.debug(f"\n[ DONE] Experiment \"{self.exp_name}\" has finished.")
harwiltz/mltools
mltools/logging/comet.py
from comet_ml import Experiment, OfflineExperiment from mltools.logging import ExperimentLogger class CometExperimentLogger(ExperimentLogger): def __init__(self, project_name, online=True, offline_directory=None, tags=[], **kwargs): super(CometExperimentLogger, self).__init__(project_name, tags=tags, **kwargs) if online: self.comet = Experiment(project_name=project_name, **kwargs) else: self.comet = OfflineExperiment(project_name=project_name, offline_directory=offline_directory, **kwargs) for tag in self.tags: self.comet.add_tag(tag) def log_metric(self, tag, value, step, **kwargs): self.comet.log_metric(tag, value, step=step, **kwargs) def log_image(self, tag, img, step, **kwargs): self.comet.log_image(img, name=tag, step=step, **kwargs) def log_plt(self, tag, plt, step, **kwargs): self.comet.log_figure(figure=plt, figure_name=tag, step=step, **kwargs) def log_text(self, tag, text, **kwargs): self.comet.log_text(text, **kwargs) def log_parameters(self, params, **kwargs): self.comet.log_parameters(params, **kwargs) def start_epoch(self, **kwargs): super(CometExperimentLogger, self).start_epoch() def end_epoch(self, **kwargs): super(CometExperimentLogger, self).end_epoch() self.comet.log_epoch_end(self.epoch, **kwargs) def end_experiment(self): self.comet.end()
harwiltz/mltools
test/mltools/test_ops.py
import jax import jax.numpy as jnp import unittest import mltools.ops as ops from functools import partial class TestOps(unittest.TestCase): def setUp(self): self.bound = 10. self.small_vector = jnp.array([3.,4.]) self.big_vector = jnp.array([12, 16.]) def test_confine_vector(self): v = ops.confine(self.bound, self.small_vector) self.assertTrue(jnp.allclose(v, self.small_vector)) V = ops.confine(self.bound, self.big_vector) self.assertFalse(jnp.allclose(V, self.big_vector)) self.assertEqual(jnp.sqrt(jnp.square(V).sum()), self.bound) self.assertEqual(V[1] / V[0], self.big_vector[1] / self.big_vector[0]) def test_confine_matrix(self): matrix = jnp.array([self.small_vector, self.big_vector]) m = ops.confine(self.bound, matrix) self.assertTrue(norm(m) <= self.bound) self.assertFalse(jnp.allclose(m[0], self.small_vector)) self.assertTrue(same_ratio(m[0], self.small_vector)) self.assertFalse(jnp.allclose(m[1], self.big_vector)) self.assertTrue(same_ratio(m[1], self.big_vector)) self.assertEqual(norm(m[1]), self.bound) def test_confine_batch_vectors(self): batch = jnp.array([self.small_vector, self.big_vector]) b = jax.vmap(partial(ops.confine, self.bound))(batch) self.assertTrue(jnp.allclose(b[0], self.small_vector)) self.assertFalse(jnp.allclose(b[1], self.big_vector)) self.assertEqual(norm(b[1]), self.bound) self.assertTrue(same_ratio(b[1], self.big_vector)) def same_ratio(v1, v2, eps=1e-6): return jnp.abs((v1[1] / v1[0]) - (v2[1] / v2[0])) < eps def norm(v): return jnp.max(jnp.sqrt(jnp.square(v).sum(axis=-1))) if __name__ == "__main__": unittest.main()
harwiltz/mltools
mltools/ops.py
<filename>mltools/ops.py import jax import jax.numpy as jnp @jax.custom_vjp def clip_gradient(lo, hi, x): r""" Differentiable gradient clipping function. Taken from https://jax.readthedocs.io/en/latest/notebooks/Custom_derivative_rules_for_Python_code.html """ return x def clip_gradient_fwd(lo, hi, x): return x, (lo, hi) def clip_gradient_bwd(res, g): lo, hi = res return (None, None, jnp.clip(g, lo, hi)) clip_gradient.defvjp(clip_gradient_fwd, clip_gradient_bwd) def confine(bound: float, x : jnp.ndarray) -> jnp.ndarray: r""" Rescale a vector if its norm is too large. Parameters ---------- bound : float The maximum allowable norm x : jnp.ndarray The vector to rescale """ norm = jnp.max(jnp.sqrt(jnp.square(x).sum(axis=-1))) return jnp.where(norm > bound, x * bound / norm, x)
harwiltz/mltools
mltools/distribution/normal.py
<reponame>harwiltz/mltools import jax import jax.numpy as jnp import jax.scipy.stats.norm as norm from mltools.distribution import Distribution, Probability from typing import Any, List, Optional, Union class Normal(Distribution): """ Univariate normal distribution """ def __init__(self, loc: float, scale: float) -> Distribution: self.loc = loc self.scale = scale def sample(self, rng: jax.random.PRNGKey) -> float: return self.loc + self.scale * jax.random.normal(rng) def pdf(self, x: float) -> Probability: return norm.pdf(x, loc=self.loc, scale=self.scale) def instantiate(params: str) -> Distribution: tokens = params.split() assert len(tokens) >= 2, \ f"{__class__.__name__} requires 2 parameters: loc and scale" loc = float(tokens[0]) scale = float(tokens[1]) return Normal(loc, scale) class MixedNormal(Distribution): """ Univariate mixture of gaussians """ def __init__(self, dists: List[Normal], weights: Optional[List[float]]=None) -> Distribution: self.dists = dists self.n = len(dists) if weights is None: self.weights = jnp.ones(self.n, dtype=jnp.float32) / self.n else: weights = jnp.array(weights) assert jnp.all(weights >= 0), "Mixture weights must be nonnegative" assert len(weights) == self.n, \ "Number of weights does not match number of distributions:\n" + \ f"Expect {self.n}, got {len(weights)}" self.weights = weights / jnp.sum(weights) def sample(self, rng: jax.random.PRNGKey) -> float: rng, sub = jax.random.split(rng) i = jax.random.choice(rng, self.n, p=self.weights) return self.dists[i].sample(sub) def pdf(self, x: float) -> Probability: densities = jnp.array([p.pdf(x) for p in self.dists]) return densities.dot(self.weights) def instantiate(params: str) -> Distribution: args = params.split(",") dists = [] weights = [] for arg in args: tokens = arg.split() dists.append(Normal.instantiate(" ".join(tokens[:2]))) if len(tokens) >= 3: weights.append(float(tokens[2])) else: weights.append(None) assert jnp.all(weights is None) or jnp.all(weights is not None), \ f"Error instantiating {__class__.__name__} with weights {weights}:\n" + \ "Cannot leave a subset of weights unspecified" if jnp.all(weights is None): weights = None return MixedNormal(dists, weights)
harwiltz/mltools
mltools/util.py
import functools from typing import Mapping def flatten_hparams(hparams: dict, base_key: str='') -> dict: """ Flattens nested dicts for easier viewing Some tools (like comet.ml) require hyperparameters to be in this format for logging Parameters ---------- hparams: dict A dictionary base_key: str (optional) A prefix to prepend to hyperparameter keys Returns ------- dict: A dict whose elements are not dicts """ def parse_object(d: dict, key: str, base_key: str) -> dict: if len(base_key) > 0 : base_key = base_key + '.' if not isinstance(d[key], Mapping): return {f"{base_key}{key}": d[key]} return flatten_hparams(d[key], f"{base_key}{key}") return functools.reduce( lambda acc, key: {**acc, **parse_object(hparams, key, base_key)}, hparams.keys(), {} )
harwiltz/mltools
mltools/distribution/__init__.py
from .base import Distribution, Probability from .normal import ( Normal, MixedNormal ) def parse_distribution(dist_type, dist_params): dist_type = dist_type.lower() if dist_type == "normal": return Normal.instantiate(dist_params) if dist_type == "mixed_normal": return MixedNormal.instantiate(dist_params) raise NotImplementedError(f"Distribution type \"{dist_type}\" not recognized")
harwiltz/mltools
mltools/distribution/base.py
<reponame>harwiltz/mltools import jax import jax.numpy as jnp from typing import Any Probability = float class Distribution: """ Distribution base class """ def sample(self, rng : jax.random.PRNGKey) -> Any: raise NotImplementedError def pdf(self, x: Any) -> Probability: raise NotImplementedError
harwiltz/mltools
mltools/logging/html.py
import hashlib import jinja2 import logging import matplotlib.pyplot as plt import numpy as np import os from jinja2 import Environment, FileSystemLoader, select_autoescape from mltools.logging import ExperimentLogger class HTMLExperimentLogger(ExperimentLogger): FILENAME = "index.html" def __init__(self, exp_name, root_path=".htmllogs", template_path="./templates", template_name="index.html", metric_smoothing=0.7, metric_figsize=(4, 3), plt_style="seaborn", **kwargs): super(HTMLExperimentLogger, self).__init__(exp_name, **kwargs) self.data = { "metrics": {}, "metrics_rendered": {}, "images": {}, "plots": {}, "text": {}, "parameters": {}, } self.metric_smoothing = metric_smoothing self.metric_figsize = metric_figsize self.plt_style = plt_style plt.style.use(plt_style) self.rootdir = os.path.join(root_path, self.exp_name) os.makedirs(self.rootdir, exist_ok=True) self.filename = os.path.join(self.rootdir, HTMLExperimentLogger.FILENAME) self._env = Environment( loader=FileSystemLoader(os.path.abspath(template_path)), autoescape=select_autoescape(['html']) ) self._template = self._env.get_template(template_name) self._first_write = True def build_page(self): if self._first_write: (logging .getLogger("HTMLExperimentLogger") .info(f"Displaying logs at file://{os.path.abspath(self.filename)}")) self._first_write = False with open(self.filename, 'w') as f: f.write(self._template.render(title=self.exp_name, epoch=self.epoch, data=self.data)) def log_metric(self, tag, value, step, **kwargs): if tag not in self.data['metrics'].keys(): self.data['metrics'][tag] = [value] else: prev = self.data['metrics'][tag][-1] v = (value - prev) * (1 - self.metric_smoothing) + prev self.data['metrics'][tag].append(v) metrics = self.data['metrics'][tag] fig = plt.figure(figsize=self.metric_figsize) plt.plot(metrics) fname = self._plot_fname(tag) plt.savefig(fname) plt.close(fig) del(fig) self.data['metrics_rendered'][tag] = fname self.build_page() def log_image(self, tag: str, img: str, step: int, **kwargs): self.data['images'][tag] = (img, step) self.build_page() def log_plt(self, tag, fig, step, **kwargs): fname = self._plot_fname(tag) fig.savefig(fname) self.data['plots'][tag] = (fname, step) self.build_page() def log_text(self, tag, text, **kwargs): if tag not in self.data['text'].keys(): self.data['text'][tag] = [] self.data['text'][tag].append(text) self.build_page() def log_parameters(self, params, **kwargs): self.data['parameters'] = params def _plot_fname(self, tag): h = str(hash(tag)).encode() fname = hashlib.md5(h).hexdigest() return os.path.abspath(f"{os.path.join(self.rootdir, fname)}.png")
schreven/OMPEval
hand_evaluators.py
#!/usr/bin/env python3 # -*- coding: utf-8 -*- """ Created on Fri Mar 22 14:37:11 2019 @author: cyril """ """ This file compares the speed of three different methods that compute the equity of a hand: OMPEval, Deuces, and PyPokerEngine""" import sys sys.path.append("..") import time ##### OMPEVAL ##### print("## OMPEVAL ##") # Import ctypes, it is native to python import numpy.ctypeslib as ctl import ctypes libname = 'libhandequity.so' # The path may have to be changed libdir = '../OMPEval-fork/lib/' lib = ctl.load_library(libname, libdir) # Defining the python function from the library omp_hand_equity = lib.hand_equity # Determining its arguments and return types omp_hand_equity.argtypes = [ctypes.c_char_p, ctypes.c_char_p, ctypes.c_int, ctypes.c_int, ctypes.c_double, ctypes.c_bool] omp_hand_equity.restype = ctypes.c_double # Defining the arguments hole_card = "9sKs" # 9 of spades and ten of spades community_card = "8dAhKh" # Number of cards defined here may very between 0 and 5 nb_players = 2 # Number of players may vary between 2 and 6 nb_board_cards = 5 # Default is 5. If = 3, showdown is at flop std_err_tol = 10**-3 # Default is 10**-5. This is the std in % at which the hand equity will be returned verbose = True # Default is False time_1 = time.time() # Calling the function (goes from python to C++ and back) hand_equity = omp_hand_equity(hole_card.encode(), community_card.encode(), nb_players, nb_board_cards, std_err_tol, verbose) print ("The equity is: " + str(hand_equity*100)+"%") time_2 = time.time() print("The total time taken is: "+ str((time_2-time_1)*1000)+ " [ms]") #print(results) ##### DEUCES ##### print("## DEUCES ##") from deuces.card import Card from deuces.evaluator import Evaluator from deuces.deck import Deck def deuces_estimate_win_rate(nb_simulation, nb_player, hole_card, board): # Estimate the win count by doing a Monte Carlo simulation win_count = sum([deuces_montecarlo_simulation(nb_player, hole_card, board) for _ in range(nb_simulation)]) return 1.0 * win_count / nb_simulation def deuces_montecarlo_simulation(nb_player, hole_card, board): deck = Deck() for card in hole_card+board: deck.remove_card(card) board_full = board + deck.draw(5-len(board)) opponents_hole = [deck.draw(2) for i in range(nb_player - 1)]#[unused_cards[2 * i:2 * i + 2] opponents_score = [evaluator.evaluate(board_full, hole) for hole in opponents_hole]#[HandEvaluator.eval_hand(hole, community_card) for hole in opponents_hole] my_score = evaluator.evaluate(board_full, hole_card)#HandEvaluator.eval_hand(hole_card, community_card) return 1 if my_score < min(opponents_score) else 0 evaluator = Evaluator() hole_card = [Card.new('9s'),Card.new('Ks')] board = [Card.new('8d'),Card.new('Ah'),Card.new('Kh')] nb_players = 6 nb_simulations = 10000 time_1 = time.time() hand_equity = deuces_estimate_win_rate(nb_simulations,nb_players,hole_card, board) print ("The equity is: " + str(hand_equity*100)+"%") time_2 = time.time() print("The total time taken is: "+ str((time_2-time_1)*1000)+ " [ms]") ### Getting evals/s evaluator = Evaluator() hole_card = [Card.new('9s'),Card.new('Ks')] board_full = [Card.new('8d'),Card.new('Ah'),Card.new('Kh'),Card.new('2s'),Card.new('6h')] nb_simulations = 100000 time_1 = time.time() for _ in range (nb_simulations): evaluator.evaluate(board_full, hole_card) time_2 = time.time() eval_per_sec = nb_simulations / (time_2-time_1) print("Speed: "+ str(eval_per_sec*10**-3)+" [kEval/s]") ##### PYPOKERENGINE ##### print('### PYPOKERENGINE ###') from pypokerengine.engine.hand_evaluator import HandEvaluator from pypokerengine.utils.card_utils import _pick_unused_card, _fill_community_card, gen_cards # Estimate the ratio of winning games given the current state of the game def estimate_win_rate(nb_simulation, nb_player, hole_card, community_card=None): if not community_card: community_card = [] # Make lists of Card objects out of the list of cards community_card = gen_cards(community_card) hole_card = gen_cards(hole_card) # Estimate the win count by doing a Monte Carlo simulation win_count = sum([montecarlo_simulation(nb_player, hole_card, community_card) for _ in range(nb_simulation)]) return 1.0 * win_count / nb_simulation def montecarlo_simulation(nb_player, hole_card, community_card): # Do a Monte Carlo simulation given the current state of the game by evaluating the hands community_card = _fill_community_card(community_card, used_card=hole_card + community_card) unused_cards = _pick_unused_card((nb_player - 1) * 2, hole_card + community_card) opponents_hole = [unused_cards[2 * i:2 * i + 2] for i in range(nb_player - 1)] opponents_score = [HandEvaluator.eval_hand(hole, community_card) for hole in opponents_hole] my_score = HandEvaluator.eval_hand(hole_card, community_card) return 1 if my_score > max(opponents_score) else 0 hole_card = ['S9','SK'] board = ['D8','HA','HK'] nb_players = 6 nb_simulations = 10000 time_1 = time.time() hand_equity = estimate_win_rate(nb_simulations,nb_players,hole_card, board) print ("The equity is: " + str(hand_equity*100)+"%") time_2 = time.time() print("The total time taken is: "+ str((time_2-time_1)*1000)+ " [ms]") ### Getting evals/s hole_card = ['S9','SK'] board_full = ['D8','HA','HK','S2','H6'] hole_card = gen_cards(hole_card) board_full = gen_cards(board_full) nb_simulations = 100000 time_1 = time.time() for _ in range (nb_simulations): HandEvaluator.eval_hand(hole_card, board_full) time_2 = time.time() eval_per_sec = nb_simulations / (time_2-time_1) print("Speed: "+ str(eval_per_sec*10**-3)+" [kEval/s]")
retroxsky06/Final_Project
webpage/app.py
<filename>webpage/app.py from flask import Flask, render_template, request from run_ml import predictions app = Flask (__name__) @app.route("/") def index(): return render_template("index.html") @app.route('/predict',methods=['POST']) def predict(): # Get the data from the POST request. if request.method == "POST": print(request.form["age"]) age = float(request.form["age"]) gender = request.form["gender"] hypertension = request.form ["hypertension"] heart_disease = request.form["heart_disease"] ever_married = request.form["ever_married"] smoking_status = request.form["smoking_status"] model_age = 0 model_gender = 0 model_hyper = 0 model_heart = 0 model_married = 0 model_smoking = 0 #gender if(gender == "Male"): model_gender = 1 elif(gender == "Female"): model_gender = 0 elif(gender == "Other"): model_gender = 2 #hypertension if(hypertension == "No"): model_hyper = 0 elif(hypertension == "Yes"): model_hyper = 1 #heart if(heart_disease == "No"): model_heart = 0 elif(heart_disease == "Yes"): model_heart = 1 #married if(ever_married == "No"): model_married = 0 elif(ever_married == "Yes"): model_married = 1 #smoking if(smoking_status == "former"): model_smoking = 0 elif(smoking_status == "never smoked"): model_smoking = 1 elif(smoking_status == "smokes"): model_smoking = 2 elif(smoking_status == "unknown"): model_smoking = 3 # data = float(request.form['exp']) # print("Data", model.predict([[data]])) prediction = predictions(model_age, model_gender, model_hyper, model_heart, model_married, model_smoking) output = prediction[0] results = "" if(output == 0): results = "No Stroke - keep up the healthy habits!" elif(output == 1): results = "Stroke - sorry buddy." print(output) return render_template("results.html", results=results)