repo_name string | dataset string | owner string | lang string | func_name string | code string | docstring string | url string | sha string |
|---|---|---|---|---|---|---|---|---|
project-lakechain | github_2023 | awslabs | typescript | ExtractPagesTask.toJSON | public toJSON() {
return (this.props);
} | /**
* @returns the JSON representation of the task.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/pdf-text-converter/src/definitions/tasks/extract-pages-task.ts#L191-L193 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitterBuilder.withChunkSize | public withChunkSize(chunkSize: number) {
this.providerProps.chunkSize = chunkSize;
return (this);
} | /**
* Sets the chunk size.
* @param chunkSize the chunk size to assign.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L90-L93 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitterBuilder.withChunkOverlap | public withChunkOverlap(chunkOverlap: number) {
this.providerProps.chunkOverlap = chunkOverlap;
return (this);
} | /**
* Sets the chunk overlap.
* @param chunkOverlap the chunk overlap to assign.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L99-L102 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitterBuilder.withSeparators | public withSeparators(separators: string[]) {
this.providerProps.separators = separators;
return (this);
} | /**
* Sets the separators to use between chunks.
* @param separators the separators to use.
* @returns the builder itself.
* @default ["\n\n", "\n", " ", ""]
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L110-L113 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitterBuilder.build | public build(): RecursiveCharacterTextSplitter {
return (new RecursiveCharacterTextSplitter(
this.scope,
this.identifier, {
...this.providerProps as RecursiveCharacterTextSplitterProps,
...this.props
}
));
} | /**
* @returns a new instance of the `RecursiveCharacterTextSplitter`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L119-L127 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.constructor | constructor(scope: Construct, id: string, props: RecursiveCharacterTextSplitterProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
6 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
props = this.parse(RecursiveCharacterTextSplitterPropsSchema, props);
///////////////////////////////////////////
///////// Processing Storage //////
///////////////////////////////////////////
this.storage = new CacheStorage(this, 'Storage', {
encryptionKey: props.kmsKey
});
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.processor = new node.NodejsFunction(this, 'Compute', {
description: 'Middleware splitting text documents in chunks.',
entry: path.resolve(__dirname, 'lambdas', 'text-splitter', 'index.js'),
vpc: props.vpc,
memorySize: props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
runtime: EXECUTION_RUNTIME,
architecture: lambda.Architecture.ARM_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: props.kmsKey,
logGroup: this.logGroup,
insightsVersion: props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
PROCESSED_FILES_BUCKET: this.storage.id(),
SEPARATORS: JSON.stringify(props.separators ?? DEFAULT_SEPARATORS),
CHUNK_SIZE: `${props.chunkSize ?? DEFAULT_CHUNK_SIZE}`,
CHUNK_OVERLAP: `${props.chunkOverlap ?? DEFAULT_CHUNK_OVERLAP}`
},
bundling: {
minify: true,
externalModules: [
'@aws-sdk/client-s3',
'@aws-sdk/client-sns'
]
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.processor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.processor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: props.batchSize ?? 5,
reportBatchItemFailures: true
}));
// Function permissions.
this.storage.grantReadWrite(this.processor);
this.eventBus.grantPublish(this.processor);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L154-L225 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
return (this.storage.grantRead(grantee));
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L231-L233 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.supportedInputTypes | supportedInputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L239-L243 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.supportedOutputTypes | supportedOutputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L249-L253 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.supportedComputeTypes | supportedComputeTypes(): ComputeType[] {
return ([
ComputeType.CPU
]);
} | /**
* @returns the supported compute types by a given
* middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L259-L263 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RecursiveCharacterTextSplitter.conditional | conditional() {
return (super
.conditional()
.and(when('type').equals('document-created'))
);
} | /**
* @returns the middleware conditional statement defining
* in which conditions this middleware should be executed.
* In this case, we want the middleware to only be invoked
* when the document mime-type is supported, and the event
* type is `document-created`.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/index.ts#L272-L277 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.getMetadata | getMetadata(chunk: string, order: number, total: number): DocumentMetadata {
return ({
properties: {
kind: 'text',
attrs: {
chunk: {
id: crypto
.createHash('sha256')
.update(chunk)
.digest('hex'),
order,
total
}
}
}
});
} | /**
* @param chunk the chunk of text to return metadata for.
* @param order the order of the chunk.
* @param total the total number of chunks in the document.
* @returns the metadata for the chunk.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/lambdas/text-splitter/index.ts#L67-L83 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.processEvent | async processEvent(event: CloudEvent): Promise<CloudEvent> {
const document = event.data().document();
// We load the text file in memory.
const text = (await document.data().asBuffer()).toString('utf-8');
// Create the text splitter.
const textSplitter = new RecursiveCharacterTextSplitter({
chunkSize: CHUNK_SIZE,
chunkOverlap: CHUNK_OVERLAP,
separators: SEPARATORS
});
// Split the text into chunks.
const chunks = await textSplitter.splitText(text);
const total = chunks.length;
// Publish each chunk as a separate document.
for (const [idx, chunk] of chunks.entries()) {
await this.onChunk(chunk, idx, total, event);
}
return (event);
} | /**
* @param event the event to process.
* @note the next decorator will automatically forward the
* returned cloud event to the next middlewares
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/lambdas/text-splitter/index.ts#L124-L147 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.handler | async handler(event: SQSEvent, _: Context): Promise<SQSBatchResponse> {
return (await processPartialResponse(
event,
(record: SQSRecord) => this.processEvent(
CloudEvent.from(JSON.parse(record.body))
),
processor
));
} | // eslint-disable-next-line @typescript-eslint/no-unused-vars | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/recursive-character-text-splitter/src/lambdas/text-splitter/index.ts#L156-L164 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitterBuilder.withSeparator | public withSeparator(separator: string | RegExp) {
this.providerProps.separator = separator;
return (this);
} | /**
* Sets the separator.
* @param separator a string or regular expression
* to use as a separator.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L75-L78 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitterBuilder.build | public build(): RegexpTextSplitter {
return (new RegexpTextSplitter(
this.scope,
this.identifier, {
...this.providerProps as RegexpTextSplitterProps,
...this.props
}
));
} | /**
* @returns a new instance of the `RegexpTextSplitter`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L84-L92 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.constructor | constructor(scope: Construct, id: string, props: RegexpTextSplitterProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
6 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
props = this.parse(RegexpTextSplitterPropsSchema, props);
///////////////////////////////////////////
///////// Processing Storage //////
///////////////////////////////////////////
this.storage = new CacheStorage(this, 'Storage', {
encryptionKey: props.kmsKey
});
///////////////////////////////////////////
////// Separator serialization /////
///////////////////////////////////////////
const separator = props.separator instanceof RegExp ?
props.separator.toString() :
props.separator;
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.processor = new node.NodejsFunction(this, 'Compute', {
description: 'Middleware splitting text documents in chunks.',
entry: path.resolve(__dirname, 'lambdas', 'text-splitter', 'index.js'),
vpc: props.vpc,
memorySize: props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
runtime: EXECUTION_RUNTIME,
architecture: lambda.Architecture.ARM_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: props.kmsKey,
logGroup: this.logGroup,
insightsVersion: props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
PROCESSED_FILES_BUCKET: this.storage.id(),
SEPARATOR_TYPE: props.separator instanceof RegExp ?
'regexp' :
'string',
SEPARATOR: separator
},
bundling: {
minify: true,
externalModules: [
'@aws-sdk/client-s3',
'@aws-sdk/client-sns'
]
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.processor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.processor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: props.batchSize ?? 5,
reportBatchItemFailures: true
}));
// Function permissions.
this.storage.grantWrite(this.processor);
this.eventBus.grantPublish(this.processor);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L119-L199 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
return (this.storage.grantRead(grantee));
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L205-L207 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.supportedInputTypes | supportedInputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L213-L217 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.supportedOutputTypes | supportedOutputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L223-L227 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.supportedComputeTypes | supportedComputeTypes(): ComputeType[] {
return ([
ComputeType.CPU
]);
} | /**
* @returns the supported compute types by a given
* middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L233-L237 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | RegexpTextSplitter.conditional | conditional() {
return (super
.conditional()
.and(when('type').equals('document-created'))
);
} | /**
* @returns the middleware conditional statement defining
* in which conditions this middleware should be executed.
* In this case, we want the middleware to only be invoked
* when the document mime-type is supported, and the event
* type is `document-created`.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/index.ts#L246-L251 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | splitText | const splitText = (text: string): string[] => {
let separator: any = SEPARATOR;
// If the separator is a regexp, we deserialize the regexp expression.
if (SEPARATOR_TYPE === 'regexp') {
separator = new RegExp(
SEPARATOR.slice(1, SEPARATOR.lastIndexOf('/')),
SEPARATOR.slice(SEPARATOR.lastIndexOf('/') + 1)
);
}
return (text
.split(separator)
.map((t) => t.trim())
);
}; | /**
* Splits the given text into an array of substrings
* based on the specified separator.
* @param text the text to split.
* @returns an array of substrings.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/lambdas/text-splitter/index.ts#L57-L72 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.getMetadata | getMetadata(chunk: string, order: number, total: number): DocumentMetadata {
return ({
properties: {
kind: 'text',
attrs: {
chunk: {
id: crypto
.createHash('sha256')
.update(chunk)
.digest('hex'),
order,
total
}
}
}
});
} | /**
* @param chunk the chunk of text to return metadata for.
* @param order the order of the chunk.
* @param total the total number of chunks in the document.
* @returns the metadata for the chunk.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/lambdas/text-splitter/index.ts#L88-L104 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.processEvent | async processEvent(event: CloudEvent): Promise<CloudEvent> {
const document = event.data().document();
// We load the text file in memory.
const text = (await document.data().asBuffer()).toString('utf-8');
// Split the text into chunks.
const chunks = splitText(text);
const total = chunks.length;
// Publish each chunk as a separate document.
for (const [idx, chunk] of chunks.entries()) {
await this.onChunk(chunk, idx, total, event);
}
return (event);
} | /**
* @param event the event to process.
* @note the next decorator will automatically forward the
* returned cloud event to the next middlewares
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/lambdas/text-splitter/index.ts#L145-L161 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.handler | async handler(event: SQSEvent, _: Context): Promise<SQSBatchResponse> {
return (await processPartialResponse(
event,
(record: SQSRecord) => this.processEvent(
CloudEvent.from(JSON.parse(record.body))
),
processor
));
} | // eslint-disable-next-line @typescript-eslint/no-unused-vars | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/regexp-text-splitter/src/lambdas/text-splitter/index.ts#L170-L178 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractorBuilder.withRegion | public withRegion(region: string) {
this.middlewareProps.region = region;
return (this);
} | /**
* Sets the AWS region in which the model
* will be invoked.
* @param region the AWS region in which the model
* will be invoked.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L90-L93 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractorBuilder.withOntologyClassifier | public withOntologyClassifier(ontologyClassifier: DefaultOntologyClassifier | CustomOntologyClassifier) {
this.middlewareProps.ontologyClassifier = ontologyClassifier;
return (this);
} | /**
* Sets the ontology classifier to use to extract
* semantic ontology from documents.
* @param ontologyClassifier the ontology classifier to use.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L101-L104 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractorBuilder.build | public build(): SemanticOntologyExtractor {
return (new SemanticOntologyExtractor(
this.scope,
this.identifier, {
...this.middlewareProps as SemanticOntologyExtractorProps,
...this.props
}
));
} | /**
* @returns a new instance of the `SemanticOntologyExtractor`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L110-L118 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.constructor | constructor(scope: Construct, id: string, private props: SemanticOntologyExtractorProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
2 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
this.props = this.parse(SemanticOntologyExtractorPropsSchema, props);
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.eventProcessor = new node.NodejsFunction(this, 'Compute', {
description: 'Extracts semantic ontology from processed documents.',
entry: path.resolve(__dirname, 'lambdas', 'handler', 'index.js'),
vpc: this.props.vpc,
memorySize: this.props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
runtime: EXECUTION_RUNTIME,
architecture: lambda.Architecture.ARM_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: this.props.kmsKey,
logGroup: this.logGroup,
insightsVersion: this.props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
LAKECHAIN_CACHE_STORAGE: this.props.cacheStorage.id(),
BEDROCK_REGION: this.props.region ?? cdk.Aws.REGION,
MODEL_ID: DEFAULT_MODEL_ID,
ONTOLOGY_CLASSIFIER_PROPS: JSON.stringify(
this.props.ontologyClassifier
)
},
bundling: {
minify: true,
externalModules: [
'@aws-sdk/client-s3',
'@aws-sdk/client-sns'
]
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.eventProcessor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.eventProcessor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: this.props.batchSize ?? 1,
maxConcurrency: 3,
reportBatchItemFailures: true
}));
// Allow access to the Bedrock API.
this.eventProcessor.addToRolePolicy(new iam.PolicyStatement({
effect: iam.Effect.ALLOW,
actions: [
'bedrock:InvokeModel'
],
resources: [
`arn:${cdk.Aws.PARTITION}:bedrock:${this.props.region ?? cdk.Aws.REGION}::foundation-model/*`,
]
}));
// Grant the compute type permissions to
// publish to the SNS topic.
this.eventBus.grantPublish(this.grantPrincipal);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L139-L216 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
// Since this middleware simply passes through the data
// from the previous middleware, we grant any subsequent
// middlewares in the pipeline read access to the
// data of all source middlewares.
for (const source of this.sources) {
source.grantReadProcessedDocuments(grantee);
}
return ({} as iam.Grant);
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L222-L231 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.supportedInputTypes | supportedInputTypes(): string[] {
return ([
...BASE_TEXT_INPUTS,
...BASE_IMAGE_INPUTS
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L237-L242 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.supportedOutputTypes | supportedOutputTypes(): string[] {
return (this.supportedInputTypes());
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L248-L250 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.supportedComputeTypes | supportedComputeTypes(): ComputeType[] {
return ([
ComputeType.CPU
]);
} | /**
* @returns the supported compute types by a given
* middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L256-L260 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SemanticOntologyExtractor.conditional | conditional() {
return (super
.conditional()
.and(when('type').equals('document-created'))
);
} | /**
* @returns the middleware conditional statement defining
* in which conditions this middleware should be executed.
* In this case, we want the middleware to only be invoked
* when the document mime-type is supported, and the event
* type is `document-created`.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/index.ts#L269-L274 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | parse | const parse = (schema: z.ZodSchema, opts: CustomOntologyClassifierProps): any => {
const options: ErrorMessageOptions = {
delimiter: {
component: ' - ',
},
path: {
enabled: true,
type: 'zodPathArray',
label: 'Path: ',
},
code: {
enabled: false
},
message: {
enabled: true,
label: '',
},
transform: ({ errorMessage, index }) => {
return (`❗ Custom Ontology Classifier - Error #${index + 1}: ${errorMessage}`);
}
};
const result = schema.safeParse(opts);
if (!result.success) {
throw new Error(generateErrorMessage(result.error.issues, options));
}
return (result.data);
}; | /**
* An internal helper allowing to validate schemas using a
* predefined visualization for error messages.
* @param schema the Zod schema to use to validate the properties.
* @param opts the options to validate.
* @returns the parsed properties.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L80-L107 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifierBuilder.withNodes | public withNodes(nodes: Node[] | 'automatic'): CustomOntologyClassifierBuilder {
this.props.nodes = nodes;
return (this);
} | /**
* Sets the nodes defined in the ontology.
* @param nodes the nodes defined in the ontology.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L171-L174 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifierBuilder.withEdges | public withEdges(edges: Edge[] | 'automatic'): CustomOntologyClassifierBuilder {
this.props.edges = edges;
return (this);
} | /**
* Sets the edges defined in the ontology.
* @param edges the edges defined in the ontology.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L181-L184 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifierBuilder.build | public build(): CustomOntologyClassifier {
return (CustomOntologyClassifier.from(this.props));
} | /**
* @returns a new instance of the `CustomOntologyClassifier`
* classifier constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L190-L192 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.constructor | constructor(public props: CustomOntologyClassifierProps) {
this.defaultClassifier = new DefaultOntologyClassifier.Builder().build();
} | /**
* Creates a new instance of the `CustomOntologyClassifier` class.
* @param props the task properties.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L214-L216 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.classifierType | public classifierType(): string {
return (this.props.classifierType);
} | /**
* @returns the type of the classifier.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L221-L223 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.nodes | public nodes(): Node[] | 'automatic' {
return (this.props.nodes);
} | /**
* @returns the nodes defined in the ontology.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L228-L230 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.edges | public edges(): Edge[] | 'automatic' {
return (this.props.edges);
} | /**
* @returns the edges defined in the ontology.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L235-L237 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.from | public static from(props: any) {
return (new CustomOntologyClassifier(
parse(CustomOntologyClassifierPropsSchema, props)
));
} | /**
* Creates a new instance of the `CustomOntologyClassifier` class.
* @param props the task properties.
* @returns a new instance of the `CustomOntologyClassifier` class.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L244-L248 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.getPrompts | public getPrompts(): Prompt[] {
const prompts = this.defaultClassifier.getPrompts();
let userPrompt = USER_PROMPT;
// Nodes definition.
if (this.nodes() === 'automatic') {
userPrompt += `
Define the nodes automatically based on your analysis of the documents.
The nodes cannot be of type "Document", "Author", "Publisher", "Topic", "Class", or "Kind" as they are reserved types.
`;
} else {
userPrompt += `
Attempt to extract the following nodes from the documents.
Only include nodes for which the properties are found in the document; do not include nodes without properties.
<json>
${JSON.stringify(this.nodes(), null, 2)}
</json>
`;
}
if (this.edges() === 'automatic') {
userPrompt += `
Define the edges in the nodes based on your analysis of the documents.
Create edges using capital letters and underscores such as "IS_ASSOCIATED_WITH" or "PARTICIPATED_IN".
`;
} else {
userPrompt += `
Attempt to extract the following edges from the documents.
Only include edges for which the source and target nodes are found in the document; do not include edges without source or target nodes.
<json>
${JSON.stringify(this.edges(), null, 2)}
</json>
`;
}
// Add custom prompts.
prompts.push({
type: 'custom',
systemPrompt: SYSTEM_PROMPT,
userPrompt
});
return (prompts);
} | /**
* @returns the prompts to execute to extract the
* semantic ontology from documents.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L254-L297 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.isNodeValid | private isNodeValid(node: any): boolean {
return (node
&& node.id
&& node.type
&& node.description
&& Array.isArray(node.props)
);
} | /**
* Validates the given node.
* @param node the node to validate.
* @returns `true` if the node is valid; `false` otherwise.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L304-L311 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.getId | private getId(node: any): string {
const attribute = node.props.find((prop: any) => prop.uniqueIdentifier);
return (attribute ? `${node.type}-${attribute.value}` : `${node.type}-${randomUUID()}`);
} | /**
* A helper defining the unique identifier of node.
* This makes it possible to namespace nodes, and honors
* the `uniqueIdentifier` attribute.
* @param node the node to get the id for.
* @returns the unique identifier of the node.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L320-L323 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.makeGraph | private async makeGraph(event: CloudEvent, json: any): Promise<CloudEvent> {
const metadata = event.data().metadata();
const cache = new CacheStorage();
let graph: DirectedGraph;
// If the graph does not exist, we create it.
if (metadata.ontology) {
graph = await metadata.ontology.resolve();
} else {
graph = new DirectedGraph();
}
// Add the nodes to the graph.
for (const node of json.nodes) {
if (this.isNodeValid(node)) {
const props: Record<string, any> = {
type: node.type,
attrs: {
description: node.description,
isHead: node.isHead
}
};
// Set a unique identifier for the node.
const id = this.getId(node);
// Add the properties to the node.
node.props
.filter((prop: any) => prop.name && prop.value)
.forEach((prop: any) => props.attrs[prop.name] = prop.value);
// Add the node to the graph.
graph.addNode(id, props);
// Update all edges that have the node as a source or target
// in `json.edges` to match the new node id.
for (const edge of json.edges) {
if (edge.source === node.id) {
edge.source = id;
}
if (edge.target === node.id) {
edge.target = id;
}
}
}
}
// Add the edges to the graph.
for (const edge of json.edges) {
if (graph.hasNode(edge.source) && graph.hasNode(edge.target)) {
const props: Record<string, any> = {
type: edge.type,
attrs: {
description: edge.description
}
};
// Add the properties to the edge.
(edge.props ?? [])
.filter((prop: any) => prop.name && prop.value)
.forEach((prop: any) => props.attrs[prop.name] = prop.value);
// Add the edge to the graph.
graph.addEdge(edge.source, edge.target, props);
}
}
// Save the graph in the cache storage.
metadata.ontology = await cache.put('ontology', graph);
return (Promise.resolve(event));
} | /**
* Creates a graph from the given JSON object.
* @param event the cloud event to update.
* @param json the JSON object extracted from the LLM.
* @returns the updated cloud event.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L331-L402 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.update | public async update(json: any, prompt: Prompt, event: CloudEvent): Promise<CloudEvent> {
if (prompt.type === 'default') {
return (Promise.resolve(
this.defaultClassifier.update(json, prompt, event)
));
} else if (prompt.type === 'custom') {
return (this.makeGraph(event, json));
} else {
throw new Error(`Unknown prompt type: ${prompt.type}`);
}
} | /**
* Updates the given cloud event based on the answer generated by the LLM
* and the prompt.
* @param json the JSON object extracted from the LLM.
* @param prompt the prompt used to generate the JSON object.
* @param event the cloud event to update.
* @returns the updated cloud event.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L412-L422 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | CustomOntologyClassifier.toJSON | public toJSON() {
return (this.props);
} | /**
* @returns the JSON representation of the classifier.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/custom-ontology-classifier.ts#L427-L429 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.constructor | constructor(public props: DefaultOntologyClassifierProps) {} | /**
* Creates a new instance of the `DefaultOntologyClassifier` class.
* @param props the task properties.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L108-L108 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.classifierType | public classifierType(): string {
return (this.props.classifierType);
} | /**
* @returns the type of the classifier.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L113-L115 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.from | public static from(props: any) {
return (new DefaultOntologyClassifier(
DefaultOntologyClassifierPropsSchema.parse(props)
));
} | /**
* Creates a new instance of the `DefaultOntologyClassifier` class.
* @param props the task properties.
* @returns a new instance of the `DefaultOntologyClassifier` class.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L122-L126 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.getPrompts | public getPrompts(): Prompt[] {
return ([{
type: 'default',
userPrompt: USER_PROMPT,
systemPrompt: SYSTEM_PROMPT
}]);
} | /**
* @returns the prompts to execute to extract the
* semantic ontology from documents.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L132-L138 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.update | public update(json: any, _: Prompt, event: CloudEvent): Promise<CloudEvent> {
const metadata = event.data().metadata();
// Document type.
if (json.type) {
metadata.type = json.type;
}
// Dates.
if (json.createdAt) {
metadata.createdAt = json.createdAt;
}
if (json.updatedAt) {
metadata.updatedAt = json.updatedAt;
}
// Title.
if (json.title) {
metadata.title = json.title;
}
// Description.
if (json.description) {
metadata.description = json.description;
}
// Keywords.
if (Array.isArray(json.keywords)) {
if (!metadata.keywords?.length) {
metadata.keywords = json.keywords;
} else {
metadata.keywords.push(...json.keywords);
}
}
// Topics.
if (Array.isArray(json.topics) && json.topics.length) {
if (!metadata.topics?.length) {
metadata.topics = json.topics;
} else {
metadata.topics.push(...json.topics);
}
}
// Authors.
if (Array.isArray(json.authors) && json.authors.length) {
metadata.authors = json.authors;
}
// Language.
if (json.language) {
metadata.language = json.language;
}
return (Promise.resolve(event));
} | /**
* Updates the given cloud event based on the answer generated by the LLM
* and the prompt.
* @param json the JSON object extracted from the LLM.
* @param _ the prompt used to generate the JSON object.
* @param event the cloud event to update.
* @returns the updated cloud event.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L148-L203 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | DefaultOntologyClassifier.toJSON | public toJSON() {
return (this.props);
} | /**
* @returns the JSON representation of the classifier.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/definitions/classifiers/default-ontology-classifier.ts#L208-L210 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.resolveEvents | private async resolveEvents(event: CloudEvent): Promise<CloudEvent[]> {
const document = event.data().document();
if (document.mimeType() === 'application/cloudevents+json') {
const data = JSON.parse((await document
.data()
.asBuffer())
.toString('utf-8')
);
return (data.map((event: string) => CloudEvent.from(event)));
} else {
return ([event]);
}
} | /**
* Resolves the input documents to process.
* This function supports both single and composite events.
* @param event the received event.
* @returns a promise to an array of CloudEvents to process.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/index.ts#L66-L79 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.recordHandler | async recordHandler(record: SQSRecord): Promise<CloudEvent> {
return (await this.processEvent(
CloudEvent.from(record.body)
));
} | /**
* @param record the SQS record to process that contains
* the EventBridge event providing information about the
* S3 event.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/index.ts#L113-L117 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.handler | public async handler(event: SQSEvent, _: Context) {
return (await processPartialResponse(
event, this.recordHandler.bind(this), processor
));
} | // eslint-disable-next-line @typescript-eslint/no-unused-vars | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/index.ts#L127-L131 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | groupBy | const groupBy = <T, K extends keyof any>(arr: T[], key: (i: T) => K) =>
arr.reduce((groups, item) => {
(groups[key(item)] ||= []).push(item);
return groups;
}, {} as Record<K, T[]>
); | /**
* Group an array of elements by a given key.
* @param arr the array to group.
* @param key the key to group by.
* @returns a record mapping the given key to the
* elements of the array.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/invoke.ts#L63-L68 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | getMessages | const getMessages = async (events: CloudEvent[], prompt: Prompt) => {
const messages = [{ role: 'user', content: [
{ type: 'text', text: prompt.userPrompt }
] as any[] }];
// Get the input documents.
events = events.filter((e: CloudEvent) => {
const type = e.data().document().mimeType();
return (BASE_TEXT_INPUTS.includes(type) || BASE_IMAGE_INPUTS.includes(type));
});
// If no documents match the supported types, throw an error.
if (events.length === 0) {
throw new Error('No valid input documents found.');
}
// Group documents by either `text` or `image` type.
const group = groupBy(events, e => {
const type = e.data().document().mimeType();
return (BASE_TEXT_INPUTS.includes(type) ? 'text' : 'image');
});
// Text documents are concatenated into a single message
// in the case where we're handling a composite event.
if (group.text) {
let text = '';
if (group.text.length > 1) {
for (const [idx, e] of group.text.entries()) {
const document = e.data().document();
text += `Document ${idx + 1}\n${(await document.data().asBuffer()).toString('utf-8')}\n\n`;
}
} else {
const document = group.text[0].data().document();
text = (await document.data().asBuffer()).toString('utf-8');
}
messages[0].content.push({ type: 'text', text });
}
// Image documents are added as separate messages.
if (group.image) {
for (const e of group.image) {
const document = e.data().document();
const type = document.mimeType();
messages[0].content.push({
type: 'image',
source: {
type: 'base64',
media_type: type,
data: (await document.data().asBuffer()).toString('base64')
}
});
}
}
// Prepend the assistant prompt with a `{` character
// to guide the LLM to generate a JSON object.
messages.push({ role: 'assistant', content: [
{ type: 'text', text: '{' }
] as any[] });
return (messages);
}; | /**
* Computes a list of messages constructed from the input documents.
* @param event the received event.
* @returns an array of messages to use for generating text.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/invoke.ts#L75-L139 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | getBody | const getBody = async (events: CloudEvent[], prompt: Prompt) => {
return ({
anthropic_version: 'bedrock-2023-05-31',
system: prompt.systemPrompt,
messages: await getMessages(events, prompt),
max_tokens: 4096,
temperature: 0.3
});
}; | /**
* Constructs the body to pass to the Bedrock API.
* @param event the cloud event associated with the
* input document(s).
* @returns the body to pass to the Bedrock API.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/invoke.ts#L147-L155 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | parseJson | const parseJson = (data: string) => {
try {
return (JSON.parse(data));
} catch (err) {
// If the document is not a valid JSON, we attempt to extract
// the JSON object embedded within the string.
const firstBracket = data.indexOf('{');
const lastBracket = data.lastIndexOf('}');
return (JSON.parse(
data.slice(firstBracket, lastBracket + 1)
));
}
}; | /**
* A helper function used to parse a JSON document
* embedded within the given string.
* @param data the string to parse.
* @returns a JSON object.
* @throws an error if the JSON document cannot be extracted.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/semantic-ontology-extractor/src/lambdas/handler/invoke.ts#L164-L176 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitterBuilder.withMaxBytesLength | public withMaxBytesLength(maxBytesLength: number) {
this.providerProps.maxBytesLength = maxBytesLength;
return (this);
} | /**
* Sets the maximum number of bytes to keep
* for a chunk.
* @param maxBytesLength the maximum number of bytes
* to keep for a chunk.
* @returns the builder itself.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L72-L75 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitterBuilder.build | public build(): SentenceTextSplitter {
return (new SentenceTextSplitter(
this.scope,
this.identifier, {
...this.providerProps as SentenceTextSplitterProps,
...this.props
}
));
} | /**
* @returns a new instance of the `SentenceTextSplitter`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L81-L89 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.constructor | constructor(scope: Construct, id: string, props: SentenceTextSplitterProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
6 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
props = this.parse(SentenceTextSplitterPropsSchema, props);
///////////////////////////////////////////
///////// Processing Storage //////
///////////////////////////////////////////
this.storage = new CacheStorage(this, 'Storage', {
encryptionKey: props.kmsKey
});
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.processor = new lambda.DockerImageFunction(this, 'Compute', {
description: 'A function splitting text documents in chunks.',
code: lambda.DockerImageCode.fromImageAsset(
path.resolve(__dirname, 'lambdas', 'text-splitter')
),
vpc: props.vpc,
memorySize: props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
architecture: lambda.Architecture.X86_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: props.kmsKey,
logGroup: this.logGroup,
insightsVersion: props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
PROCESSED_FILES_BUCKET: this.storage.id(),
MAX_BYTES_LENGTH: `${props.maxBytesLength}`,
CACHE_DIR: '/tmp'
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.processor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.processor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: props.batchSize ?? 5,
reportBatchItemFailures: true
}));
// Function permissions.
this.storage.grantWrite(this.processor);
this.eventBus.grantPublish(this.processor);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L116-L180 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
return (this.storage.grantRead(grantee));
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L186-L188 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.supportedInputTypes | supportedInputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L194-L198 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.supportedOutputTypes | supportedOutputTypes(): string[] {
return ([
'text/plain'
]);
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L204-L208 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.supportedComputeTypes | supportedComputeTypes(): ComputeType[] {
return ([
ComputeType.CPU
]);
} | /**
* @returns the supported compute types by a given
* middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L214-L218 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SentenceTextSplitter.conditional | conditional() {
return (super
.conditional()
.and(when('type').equals('document-created'))
);
} | /**
* @returns the middleware conditional statement defining
* in which conditions this middleware should be executed.
* In this case, we want the middleware to only be invoked
* when the document mime-type is supported, and the event
* type is `document-created`.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/sentence-text-splitter/src/index.ts#L227-L232 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withRegion | public withRegion(region: string) {
this.middlewareProps.region = region;
return (this);
} | /**
* Sets the AWS region in which the model
* will be invoked.
* @param region the AWS region in which the model
* will be invoked.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L87-L90 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withSchema | public withSchema(schema: z.ZodSchema<any>) {
this.middlewareProps.schema = schema;
return (this);
} | /**
* Sets the schema to use to extract structured
* entities from documents.
* @param schema the schema to use to extract structured
* entities from documents.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L99-L102 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withOutputType | public withOutputType(outputType: 'json' | 'metadata') {
this.middlewareProps.outputType = outputType;
return (this);
} | /**
* Sets the output type of the structured entities.
* @param outputType the output type of the structured entities.
* @returns the current builder instance.
* @default 'json'
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L110-L113 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withModel | public withModel(model: Model) {
this.middlewareProps.model = model;
return (this);
} | /**
* Sets the model to use for structured entity extraction.
* @param model the model to use for structured entity extraction.
* @returns the current builder instance.
* @default 'anthropic.claude-3-sonnet-20240229-v1:0'
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L121-L124 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withInstructions | public withInstructions(instructions: string) {
this.middlewareProps.instructions = instructions;
return (this);
} | /**
* Sets optional instruction to pass to the model to guide it
* in extracting structured entities.
* @param instructions optional instruction to pass to the model to guide it
* in extracting structured entities.
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L133-L136 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.withModelParameters | public withModelParameters(parameters: ModelParameters) {
this.middlewareProps.modelParameters = parameters;
return (this);
} | /**
* Sets the parameters to pass to the text model.
* @param parameters the parameters to pass to the text model.
* @default {}
* @returns the current builder instance.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L144-L147 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractorBuilder.build | public build(): StructuredEntityExtractor {
return (new StructuredEntityExtractor(
this.scope,
this.identifier, {
...this.middlewareProps as StructuredEntityExtractorProps,
...this.props
}
));
} | /**
* @returns a new instance of the `StructuredEntityExtractor`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L153-L161 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.constructor | constructor(scope: Construct, id: string, private props: StructuredEntityExtractorProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
2 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
this.props = this.parse(StructuredEntityExtractorPropsSchema, props);
///////////////////////////////////////////
//////// Processing Storage ///////
///////////////////////////////////////////
this.storage = new CacheStorage(this, 'Storage', {
encryptionKey: props.kmsKey
});
///////////////////////////////////////////
////////// Schema Handler /////////
///////////////////////////////////////////
// Convert the Zod schema to a JSON schema.
const jsonSchema = zodToJsonSchema(this.props.schema) as any;
// Stringified JSON schema.
const schema = JSON.stringify(jsonSchema);
// Ensure the root schema is an object.
if (jsonSchema.type !== 'object') {
throw new Error(`
The root schema you provide must be of type 'object'.
`);
}
// Create a reference to the schema.
let reference: r.IReference<any> = r.reference(r.value(schema));
// If the schema is bigger than a certain threshold, we upload the schema
// to the internal storage and reference it in the lambda environment.
if (schema.length > 3072) {
new s3deploy.BucketDeployment(this, 'Schema', {
sources: [s3deploy.Source.data('schema', schema)],
destinationBucket: this.storage.getBucket()
});
reference = r.reference(r.url(`s3://${this.storage.getBucket().bucketName}/schema`));
}
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.eventProcessor = new node.NodejsFunction(this, 'Compute', {
description: 'Extracts structured entities from processed documents.',
entry: path.resolve(__dirname, 'lambdas', 'handler', 'index.js'),
vpc: this.props.vpc,
memorySize: this.props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
runtime: EXECUTION_RUNTIME,
architecture: lambda.Architecture.ARM_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: this.props.kmsKey,
logGroup: this.logGroup,
insightsVersion: this.props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
PROCESSED_FILES_BUCKET: this.storage.id(),
LAKECHAIN_CACHE_STORAGE: this.props.cacheStorage.id(),
BEDROCK_REGION: this.props.region ?? cdk.Aws.REGION,
MODEL_ID: this.props.model,
SCHEMA: JSON.stringify(reference),
OUTPUT_TYPE: this.props.outputType,
INSTRUCTIONS: this.props.instructions ?? '',
MODEL_PARAMETERS: JSON.stringify(this.props.modelParameters)
},
bundling: {
minify: true,
externalModules: [
'@aws-sdk/client-s3',
'@aws-sdk/client-sns'
]
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.eventProcessor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.eventProcessor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: this.props.batchSize ?? 1,
maxConcurrency: this.props.maxConcurrency ?? 3,
reportBatchItemFailures: true
}));
// Allow access to the Bedrock API.
this.eventProcessor.addToRolePolicy(new iam.PolicyStatement({
effect: iam.Effect.ALLOW,
actions: [
'bedrock:InvokeModel'
],
resources: [
`arn:${cdk.Aws.PARTITION}:bedrock:${this.props.region ?? cdk.Aws.REGION}::foundation-model/*`,
]
}));
// Grant the compute type permissions to
// publish to the SNS topic.
this.eventBus.grantPublish(this.grantPrincipal);
// Grant the compute type permissions to
// write to the storage.
this.storage.grantWrite(this.grantPrincipal);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L187-L308 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
if (this.props.outputType === 'json') {
return (this.storage.grantRead(grantee));
} else {
// Since this middleware simply passes through the data
// from the previous middleware, we grant any subsequent
// middlewares in the pipeline read access to the
// data of all source middlewares.
for (const source of this.sources) {
source.grantReadProcessedDocuments(grantee);
}
return ({} as iam.Grant);
}
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L314-L327 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.supportedInputTypes | supportedInputTypes(): string[] {
return ([
'text/plain',
'text/markdown',
'text/csv',
'text/html',
'application/x-subrip',
'text/vtt',
'application/json',
'application/xml',
'application/cloudevents+json'
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L333-L345 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.supportedOutputTypes | supportedOutputTypes(): string[] {
return (this.props.outputType === 'json' ?
['application/json'] :
this.supportedInputTypes()
);
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L351-L356 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.supportedComputeTypes | supportedComputeTypes(): ComputeType[] {
return ([
ComputeType.CPU
]);
} | /**
* @returns the supported compute types by a given
* middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L362-L366 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | StructuredEntityExtractor.conditional | conditional() {
return (super
.conditional()
.and(when('type').equals('document-created'))
);
} | /**
* @returns the middleware conditional statement defining
* in which conditions this middleware should be executed.
* In this case, we want the middleware to only be invoked
* when the document mime-type is supported, and the event
* type is `document-created`.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/index.ts#L375-L380 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.resolveEvents | private async resolveEvents(event: CloudEvent): Promise<CloudEvent[]> {
const document = event.data().document();
if (document.mimeType() === 'application/cloudevents+json') {
const data = JSON.parse((await document
.data()
.asBuffer())
.toString('utf-8')
);
return (data.map((event: string) => CloudEvent.from(event)));
} else {
return ([event]);
}
} | /**
* Resolves the input documents to process.
* This function supports both single and composite events.
* @param event the received event.
* @returns a promise to an array of CloudEvents to process.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L123-L136 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.getContent | private async getContent(events: CloudEvent[]): Promise<ContentBlock[]> {
const content = [{
text: USER_PROMPT
}];
// Add additional instructions if available.
if (INSTRUCTIONS) {
content.push({
text: `Here are additional instructions you must follow when extracting structured data:
<instructions>
${INSTRUCTIONS}
</instructions>`
});
}
// Add the documents to the prompt.
for (const event of events) {
const document = event.data().document();
const text = (await document.data().asBuffer()).toString('utf-8');
content.push({
text: `<document>\n${text}\n</document>`
});
}
return (content);
} | /**
* Creates the content array to pass to the model.
* @param events the events to create a prompt for.
* @returns a promise to an array of messages to pass to the model.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L143-L168 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.parseOutput | private parseOutput(result: any, validateFn: any) {
const output = result.output?.message;
const content = output?.content;
const response = content.find((item: any) => 'toolUse' in item);
if (result.stopReason !== 'tool_use') {
throw new InvalidResponseException('The data extraction tool was not invoked.');
}
if (!response?.toolUse) {
throw new InvalidResponseException('The data extraction tool did not return any results.');
}
// Validate the extracted data.
const isValid = validateFn(response.toolUse.input);
if (!isValid) {
throw new InvalidSchemaException(validateFn.errors, response);
}
return (response.toolUse.input);
} | /**
* Parses the output of the data extraction tool.
* @param result the result of the data extraction.
* @param validateFn the validation function to use.
* @returns the structured data extracted from the documents.
* @throws an error if the data extraction fails.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L177-L196 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.extract | private async extract(events: CloudEvent[], maxRetries = DEFAULT_MAX_RETRIES): Promise<any> {
const messages: Message[] = [{
role: 'user',
content: await this.getContent(events)
}];
// The JSON schema to extract structured data from
// the provided documents.
const schema = JSON.parse(
await events[0].resolve(SCHEMA_REFERENCE)
);
// The validation function to use.
const validate = ajv.compile(schema);
// For up to `maxRetries` attempts, we will try to extract
// structured data from the documents.
while (maxRetries-- > 0) {
const result = await bedrock.send(new ConverseCommand({
modelId: MODEL_ID,
system: [{
text: SYSTEM_PROMPT
}],
messages,
toolConfig: {
tools: [{
toolSpec: {
name: TOOL_NAME,
description: 'A function receiving the results of the data extraction.',
inputSchema: {
json: schema
}
}
}],
toolChoice: {
tool: {
name: TOOL_NAME
}
}
},
inferenceConfig: MODEL_PARAMETERS
}));
// We add the assistant response to the messages.
messages.push(result.output!.message!);
// Parse and validate the extracted data.
try {
return (this.parseOutput(result, validate));
} catch (error) {
if (error instanceof InvalidSchemaException) {
// The extracted data are invalid, we want to prompt the model
// with the errors to correct the generated data.
messages.push({
role: 'user',
content: [{
toolResult: {
toolUseId: error.getResponse().toolUse.toolUseId,
content: [{
text: `${ERROR_PROMPT}\n${JSON.stringify(error.getErrors())}`
}],
status: 'error'
}
}]
});
}
}
}
throw new Error('Failed to extract structured data.');
} | /**
* Extracts structured data from the provided documents.
* @param events the CloudEvents to process.
* @param maxRetries the maximum number of retries to attempt.
* @returns a promise to an object containing the structured data
* matching the given schema.
* @throws an error if the data extraction fails.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L206-L276 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.recordHandler | async recordHandler(record: SQSRecord): Promise<CloudEvent> {
return (await this.processEvent(
CloudEvent.from(record.body)
));
} | /**
* @param record the SQS record to process that contains
* the EventBridge event providing information about the
* S3 event.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L321-L325 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | Lambda.handler | public async handler(event: SQSEvent, _: Context) {
return (await processPartialResponse(
event, this.recordHandler.bind(this), processor
));
} | // eslint-disable-next-line @typescript-eslint/no-unused-vars | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/index.ts#L335-L339 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | InvalidResponseException.constructor | constructor(description: string) {
super(description);
} | /**
* `InvalidResponseException` constructor.
* @param description the description of the exception.
* @param errors the errors that caused the response to be invalid.
* @param response the response that is not valid.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/exceptions/invalid-response.ts#L29-L31 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | InvalidSchemaException.constructor | constructor(private errors: ErrorObject[], private response: any) {
super(`The extracted data are invalid : ${JSON.stringify(errors)}`);
} | /**
* `InvalidSchemaException` constructor.
* @param errors the errors that caused the response to be invalid.
* @param response the response that is not valid.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/exceptions/invalid-schema.ts#L30-L32 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | InvalidSchemaException.getResponse | public getResponse(): any {
return (this.response);
} | /**
* @returns the response that is not valid.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/exceptions/invalid-schema.ts#L37-L39 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | InvalidSchemaException.getErrors | public getErrors(): ErrorObject[] {
return (this.errors);
} | /**
* @returns the errors that caused the response to be invalid.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/structured-entity-extractor/src/lambdas/handler/exceptions/invalid-schema.ts#L44-L46 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessorBuilder.withOutputFormats | public withOutputFormats(...outputFormats: OutputFormat[]) {
this.providerProps.outputFormats = outputFormats;
return (this);
} | /**
* Sets the chunk size.
* @param chunkSize the chunk size to assign.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L75-L78 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessorBuilder.build | public build(): SubtitleProcessor {
return (new SubtitleProcessor(
this.scope,
this.identifier, {
...this.providerProps as SubtitleProcessorProps,
...this.props
}
));
} | /**
* @returns a new instance of the `SubtitleProcessor`
* service constructed with the given parameters.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L84-L92 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessor.constructor | constructor(scope: Construct, id: string, private props: SubtitleProcessorProps) {
super(scope, id, description, {
...props,
queueVisibilityTimeout: cdk.Duration.seconds(
3 * PROCESSING_TIMEOUT.toSeconds()
)
});
// Validate the properties.
this.props = this.parse(SubtitleProcessorPropsSchema, props);
///////////////////////////////////////////
///////// Processing Storage //////
///////////////////////////////////////////
this.storage = new CacheStorage(this, 'Storage', {
encryptionKey: this.props.kmsKey
});
///////////////////////////////////////////
////// Middleware Event Handler ////
///////////////////////////////////////////
this.processor = new node.NodejsFunction(this, 'Compute', {
description: 'A middleware parsing subtitle documents.',
entry: path.resolve(__dirname, 'lambdas', 'parser', 'index.js'),
vpc: this.props.vpc,
memorySize: this.props.maxMemorySize ?? DEFAULT_MEMORY_SIZE,
timeout: PROCESSING_TIMEOUT,
runtime: EXECUTION_RUNTIME,
architecture: lambda.Architecture.ARM_64,
tracing: lambda.Tracing.ACTIVE,
environmentEncryption: this.props.kmsKey,
logGroup: this.logGroup,
insightsVersion: this.props.cloudWatchInsights ?
LAMBDA_INSIGHTS_VERSION :
undefined,
environment: {
POWERTOOLS_SERVICE_NAME: description.name,
POWERTOOLS_METRICS_NAMESPACE: NAMESPACE,
SNS_TARGET_TOPIC: this.eventBus.topicArn,
PROCESSED_FILES_BUCKET: this.storage.id(),
OUTPUT_FORMATS: JSON.stringify(this.props.outputFormats)
},
bundling: {
minify: true,
externalModules: [
'@aws-sdk/client-s3',
'@aws-sdk/client-sns'
]
}
});
// Allows this construct to act as a `IGrantable`
// for other middlewares to grant the processing
// lambda permissions to access their resources.
this.grantPrincipal = this.processor.grantPrincipal;
// Plug the SQS queue into the lambda function.
this.processor.addEventSource(new sources.SqsEventSource(this.eventQueue, {
batchSize: this.props.batchSize ?? 5,
reportBatchItemFailures: true
}));
// Function permissions.
this.storage.grantWrite(this.processor);
this.eventBus.grantPublish(this.processor);
super.bind();
} | /**
* Construct constructor.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L119-L188 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessor.grantReadProcessedDocuments | grantReadProcessedDocuments(grantee: iam.IGrantable): iam.Grant {
return (this.storage.grantRead(grantee));
} | /**
* Allows a grantee to read from the processed documents
* generated by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L194-L196 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessor.supportedInputTypes | supportedInputTypes(): string[] {
return ([
'text/vtt',
'application/x-subrip'
]);
} | /**
* @returns an array of mime-types supported as input
* type by this middleware.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L202-L207 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
project-lakechain | github_2023 | awslabs | typescript | SubtitleProcessor.supportedOutputTypes | supportedOutputTypes(): string[] {
const outputs = [];
for (const format of this.props.outputFormats) {
if (format === 'text') {
outputs.push('text/plain');
} else if (format === 'json') {
outputs.push('application/json');
}
}
return (outputs);
} | /**
* @returns an array of mime-types supported as output
* type by the data producer.
*/ | https://github.com/awslabs/project-lakechain/blob/4285173e80584eedfc1a8424d3d1b6c1a7038088/packages/middlewares/text-processors/subtitle-processor/src/index.ts#L213-L224 | 4285173e80584eedfc1a8424d3d1b6c1a7038088 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.