Spaces:
Running
Running
File size: 16,758 Bytes
dbf7313 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 | """Import a historical Hugging Face dataset checkpoint into a clean snapshot.
This is mainly for legacy datasets that kept their richest data under
`_checkpoints/<snapshot_id>/...` instead of promoting the latest full snapshot
to the dataset root.
The importer:
1. selects a checkpoint from a source HF dataset repo
2. downloads the checkpoint parquet files
3. rewrites small tables to the current local schema when needed
4. regenerates derived artifacts (`links`, `issue_comments`, `pr_comments`)
5. writes a normal local snapshot directory that `analyze` / `pr-scope` can use
6. optionally republishes that clean snapshot
"""
from __future__ import annotations
import re
import shutil
from collections import defaultdict
from collections.abc import Mapping
from datetime import UTC, datetime
from pathlib import Path
from typing import Any
from huggingface_hub import HfApi, hf_hub_download
from slop_farmer.app.publish_dataset_snapshot import publish_dataset_snapshot
from slop_farmer.config import CheckpointImportOptions
from slop_farmer.data.dataset_card import build_hf_dataset_card
from slop_farmer.data.links import build_pr_duplicate_candidate_rows, build_text_link_rows
from slop_farmer.data.parquet_io import (
SCHEMAS,
read_json,
read_parquet_rows,
write_json,
write_parquet,
write_text,
)
CHECKPOINT_PATH_PATTERN = re.compile(
r"^(?P<root>_?checkpoints)/(?P<snapshot_id>[^/]+)/(?P<filename>[^/]+)$"
)
REQUIRED_CHECKPOINT_FILES = (
"issues.parquet",
"pull_requests.parquet",
"comments.parquet",
"reviews.parquet",
"review_comments.parquet",
"pr_files.parquet",
"pr_diffs.parquet",
"events.parquet",
)
SMALL_TABLES = (
"issues",
"pull_requests",
"comments",
"reviews",
"review_comments",
"events",
)
COPIED_TABLES = (
"pr_files",
"pr_diffs",
)
__all__ = [
"import_hf_checkpoint",
"materialize_checkpoint_snapshot",
]
def import_hf_checkpoint(options: CheckpointImportOptions) -> Path:
api = HfApi()
siblings = [
sibling.rfilename
for sibling in api.dataset_info(repo_id=options.source_repo_id).siblings or []
]
checkpoint_root, checkpoint_id = _select_checkpoint(
siblings,
checkpoint_id=options.checkpoint_id,
checkpoint_root=options.checkpoint_root,
)
checkpoint_files = {
filename: Path(
hf_hub_download(
repo_id=options.source_repo_id,
repo_type="dataset",
filename=f"{checkpoint_root}/{checkpoint_id}/{filename}",
)
)
for filename in REQUIRED_CHECKPOINT_FILES
}
source_manifest = _maybe_download_json(options.source_repo_id, "manifest.json")
source_progress = _maybe_download_json(
options.source_repo_id, f"{checkpoint_root}/{checkpoint_id}/progress.json"
)
snapshot_dir = materialize_checkpoint_snapshot(
checkpoint_files,
output_root=options.output_dir,
source_repo_id=options.source_repo_id,
checkpoint_root=checkpoint_root,
checkpoint_id=checkpoint_id,
source_manifest=source_manifest,
source_progress=source_progress,
force=options.force,
)
if options.publish_repo_id:
publish_dataset_snapshot(
snapshot_dir, options.publish_repo_id, private=options.private_hf_repo
)
return snapshot_dir
def materialize_checkpoint_snapshot(
checkpoint_files: Mapping[str, Path],
*,
output_root: Path,
source_repo_id: str,
checkpoint_root: str,
checkpoint_id: str,
source_manifest: Mapping[str, Any] | None = None,
source_progress: Mapping[str, Any] | None = None,
force: bool = False,
) -> Path:
source_manifest = dict(source_manifest or {})
source_progress = dict(source_progress or {})
output_root = output_root.resolve()
snapshot_dir = output_root / "snapshots" / _snapshot_dir_name(source_repo_id, checkpoint_id)
if snapshot_dir.exists():
if not force:
raise FileExistsError(f"Snapshot already exists: {snapshot_dir}")
shutil.rmtree(snapshot_dir)
snapshot_dir.mkdir(parents=True, exist_ok=True)
rows_by_table: dict[str, list[dict[str, Any]]] = {}
for table_name in SMALL_TABLES:
rows = read_parquet_rows(checkpoint_files[f"{table_name}.parquet"])
rows_by_table[table_name] = _project_rows(rows, table_name)
write_parquet(rows_by_table[table_name], snapshot_dir / f"{table_name}.parquet", table_name)
for table_name in COPIED_TABLES:
shutil.copy2(
checkpoint_files[f"{table_name}.parquet"], snapshot_dir / f"{table_name}.parquet"
)
repo_slug = _resolve_repo(rows_by_table, source_manifest, source_progress)
extracted_at = _resolve_extracted_at(rows_by_table, source_manifest)
link_rows = _derived_link_rows(
repo_slug=repo_slug,
snapshot_id=checkpoint_id,
extracted_at=extracted_at,
issues=rows_by_table["issues"],
pull_requests=rows_by_table["pull_requests"],
comments=rows_by_table["comments"],
reviews=rows_by_table["reviews"],
review_comments=rows_by_table["review_comments"],
events=rows_by_table["events"],
)
write_parquet(link_rows, snapshot_dir / "links.parquet", "links")
issue_comment_rows, pr_comment_rows = _viewer_comment_rows(
rows_by_table["comments"], rows_by_table["pull_requests"]
)
write_parquet(issue_comment_rows, snapshot_dir / "issue_comments.parquet", "comments")
write_parquet(pr_comment_rows, snapshot_dir / "pr_comments.parquet", "comments")
manifest = {
"repo": repo_slug,
"snapshot_id": checkpoint_id,
"extracted_at": extracted_at,
"imported_at": _iso_now(),
"source_type": "hf_checkpoint_import",
"source_hf_repo_id": source_repo_id,
"source_checkpoint_root": checkpoint_root,
"source_checkpoint_id": checkpoint_id,
"source_manifest": source_manifest,
"source_progress": source_progress,
"counts": {
"issues": len(rows_by_table["issues"]),
"pull_requests": len(rows_by_table["pull_requests"]),
"comments": len(rows_by_table["comments"]),
"reviews": len(rows_by_table["reviews"]),
"review_comments": len(rows_by_table["review_comments"]),
"pr_files": _parquet_row_count(snapshot_dir / "pr_files.parquet"),
"pr_diffs": _parquet_row_count(snapshot_dir / "pr_diffs.parquet"),
"timeline_events": len(rows_by_table["events"]),
"links": len(link_rows),
},
"notes": (
"Imported from a remote HF dataset checkpoint and regenerated locally derived link/comment views."
),
}
write_json(manifest, snapshot_dir / "manifest.json")
write_text(
_dataset_card(repo_slug, checkpoint_id, source_repo_id, checkpoint_root),
snapshot_dir / "README.md",
)
return snapshot_dir
def _select_checkpoint(
sibling_paths: list[str],
*,
checkpoint_id: str | None,
checkpoint_root: str | None,
) -> tuple[str, str]:
candidates: defaultdict[tuple[str, str], set[str]] = defaultdict(set)
for path in sibling_paths:
match = CHECKPOINT_PATH_PATTERN.match(path)
if match is None:
continue
root = match.group("root")
snapshot_id = match.group("snapshot_id")
filename = match.group("filename")
candidates[(root, snapshot_id)].add(filename)
viable = [
(root, snapshot_id)
for (root, snapshot_id), filenames in candidates.items()
if all(filename in filenames for filename in REQUIRED_CHECKPOINT_FILES)
]
if checkpoint_id is not None:
requested = [
candidate
for candidate in viable
if candidate[1] == checkpoint_id
and (checkpoint_root is None or candidate[0] == checkpoint_root)
]
if not requested:
raise ValueError(
f"Checkpoint {checkpoint_root or '*'}:{checkpoint_id} not found with required files."
)
return sorted(requested, key=_checkpoint_sort_key)[-1]
if checkpoint_root is not None:
viable = [candidate for candidate in viable if candidate[0] == checkpoint_root]
if not viable:
raise ValueError("No viable checkpoint directories were found in the source HF dataset.")
return sorted(viable, key=_checkpoint_sort_key)[-1]
def _checkpoint_sort_key(candidate: tuple[str, str]) -> tuple[str, int]:
root, snapshot_id = candidate
root_priority = 1 if root == "_checkpoints" else 0
return snapshot_id, root_priority
def _maybe_download_json(repo_id: str, filename: str) -> dict[str, Any]:
try:
path = hf_hub_download(repo_id=repo_id, repo_type="dataset", filename=filename)
except Exception:
return {}
return read_json(Path(path))
def _project_rows(rows: list[dict[str, Any]], table_name: str) -> list[dict[str, Any]]:
field_names = [field.name for field in SCHEMAS[table_name]]
return [{field_name: row.get(field_name) for field_name in field_names} for row in rows]
def _resolve_repo(
rows_by_table: Mapping[str, list[dict[str, Any]]],
source_manifest: Mapping[str, Any],
source_progress: Mapping[str, Any],
) -> str:
for table_name in ("issues", "pull_requests", "comments"):
rows = rows_by_table.get(table_name) or []
if rows and rows[0].get("repo"):
return str(rows[0]["repo"])
return str(source_manifest.get("repo") or source_progress.get("repo") or "")
def _resolve_extracted_at(
rows_by_table: Mapping[str, list[dict[str, Any]]],
source_manifest: Mapping[str, Any],
) -> str:
for table_name in (
"issues",
"pull_requests",
"comments",
"reviews",
"review_comments",
"events",
):
rows = rows_by_table.get(table_name) or []
if rows and rows[0].get("extracted_at"):
return str(rows[0]["extracted_at"])
return str(source_manifest.get("extracted_at") or _iso_now())
def _derived_link_rows(
*,
repo_slug: str,
snapshot_id: str,
extracted_at: str,
issues: list[dict[str, Any]],
pull_requests: list[dict[str, Any]],
comments: list[dict[str, Any]],
reviews: list[dict[str, Any]],
review_comments: list[dict[str, Any]],
events: list[dict[str, Any]],
) -> list[dict[str, Any]]:
owner, repo_name = repo_slug.split("/", 1)
link_rows: list[dict[str, Any]] = []
for issue_row in issues:
link_rows.extend(
build_text_link_rows(
repo=repo_slug,
owner=owner,
repo_name=repo_name,
source_type="issue",
source_number=int(issue_row["number"]),
source_id=issue_row.get("github_id"),
body=issue_row.get("body"),
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
for pr_row in pull_requests:
link_rows.extend(
build_text_link_rows(
repo=repo_slug,
owner=owner,
repo_name=repo_name,
source_type="pull_request",
source_number=int(pr_row["number"]),
source_id=pr_row.get("github_id"),
body=pr_row.get("body"),
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
for comment_row in comments:
parent_number = comment_row.get("parent_number")
if parent_number is None:
continue
link_rows.extend(
build_text_link_rows(
repo=repo_slug,
owner=owner,
repo_name=repo_name,
source_type="comment",
source_number=int(parent_number),
source_id=comment_row.get("github_id"),
body=comment_row.get("body"),
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
for review_row in reviews:
link_rows.extend(
build_text_link_rows(
repo=repo_slug,
owner=owner,
repo_name=repo_name,
source_type="review",
source_number=int(review_row["pull_request_number"]),
source_id=review_row.get("github_id"),
body=review_row.get("body"),
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
for review_comment_row in review_comments:
link_rows.extend(
build_text_link_rows(
repo=repo_slug,
owner=owner,
repo_name=repo_name,
source_type="review_comment",
source_number=int(review_comment_row["pull_request_number"]),
source_id=review_comment_row.get("github_id"),
body=review_comment_row.get("body"),
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
link_rows.extend(
build_pr_duplicate_candidate_rows(
repo=repo_slug,
pull_requests=pull_requests,
link_rows=link_rows,
snapshot_id=snapshot_id,
extracted_at=extracted_at,
)
)
for event in events:
source_issue_number = event.get("source_issue_number")
if not source_issue_number:
continue
link_rows.append(
{
"repo": repo_slug,
"source_type": event["parent_kind"],
"source_number": event["parent_number"],
"source_github_id": None,
"target_owner": owner,
"target_repo": repo_name,
"target_number": source_issue_number,
"link_type": f"timeline:{event['event']}",
"link_origin": "timeline",
"snapshot_id": snapshot_id,
"extracted_at": extracted_at,
}
)
return sorted(
_dedupe_link_rows(link_rows),
key=lambda row: (
str(row.get("source_type") or ""),
int(row.get("source_number") or 0),
int(row.get("source_github_id") or 0),
str(row.get("target_owner") or ""),
str(row.get("target_repo") or ""),
int(row.get("target_number") or 0),
str(row.get("link_type") or ""),
str(row.get("link_origin") or ""),
),
)
def _dedupe_link_rows(rows: list[dict[str, Any]]) -> list[dict[str, Any]]:
deduped: dict[tuple[Any, ...], dict[str, Any]] = {}
for row in rows:
key = (
row.get("repo"),
row.get("source_type"),
row.get("source_number"),
row.get("source_github_id"),
row.get("target_owner"),
row.get("target_repo"),
row.get("target_number"),
row.get("link_type"),
row.get("link_origin"),
)
deduped[key] = row
return list(deduped.values())
def _viewer_comment_rows(
comments: list[dict[str, Any]],
pull_requests: list[dict[str, Any]],
) -> tuple[list[dict[str, Any]], list[dict[str, Any]]]:
pr_numbers = {int(row["number"]) for row in pull_requests if row.get("number") is not None}
issue_comments: list[dict[str, Any]] = []
pr_comments: list[dict[str, Any]] = []
for row in comments:
parent_number = row.get("parent_number")
parent_kind = row.get("parent_kind")
if parent_kind == "pull_request" or parent_number in pr_numbers:
pr_comments.append(row)
else:
issue_comments.append(row)
return issue_comments, pr_comments
def _dataset_card(
repo_slug: str, snapshot_id: str, source_repo_id: str, checkpoint_root: str
) -> str:
return build_hf_dataset_card(
repo_slug,
snapshot_id,
notes=[
f"source HF dataset: `{source_repo_id}`",
f"source checkpoint root: `{checkpoint_root}`",
"links were regenerated locally from text references and timeline events",
],
)
def _snapshot_dir_name(source_repo_id: str, checkpoint_id: str) -> str:
return f"hf-{source_repo_id.replace('/', '--')}-{checkpoint_id}"
def _iso_now() -> str:
return datetime.now(tz=UTC).replace(microsecond=0).isoformat().replace("+00:00", "Z")
def _parquet_row_count(path: Path) -> int:
import pyarrow.parquet as pq
return pq.read_metadata(path).num_rows
|