| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341134213431344134513461347134813491350135113521353135413551356135713581359136013611362136313641365136613671368136913701371137213731374137513761377137813791380138113821383138413851386138713881389139013911392139313941395139613971398139914001401140214031404140514061407140814091410141114121413141414151416141714181419142014211422142314241425142614271428142914301431143214331434143514361437143814391440144114421443144414451446144714481449145014511452145314541455145614571458145914601461146214631464146514661467146814691470147114721473147414751476147714781479148014811482148314841485148614871488148914901491149214931494149514961497149814991500150115021503150415051506150715081509151015111512151315141515151615171518151915201521152215231524152515261527152815291530153115321533153415351536153715381539154015411542154315441545154615471548154915501551155215531554155515561557155815591560156115621563156415651566156715681569157015711572157315741575157615771578157915801581158215831584158515861587158815891590159115921593159415951596159715981599160016011602160316041605160616071608160916101611161216131614161516161617161816191620162116221623162416251626162716281629163016311632163316341635163616371638163916401641164216431644164516461647 |
- """HTTP orchestration boundary for data-research ingestion jobs."""
- from __future__ import annotations
- import io
- import logging
- from flask import current_app, g, jsonify, request, send_file
- from app import db
- from app.api.data_development import bp
- from app.core.data_research.errors import DataResearchError
- from app.models.result import failed, success
- logger = logging.getLogger(__name__)
- def get_ingestion_service():
- from app.core.data_research.ingestion import IngestionService
- from app.core.data_research.repository import SqlAlchemyIngestionJobRepository
- return IngestionService(
- SqlAlchemyIngestionJobRepository(db.session),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_database_source_registration_service():
- from app.core.data_research.repository import (
- SqlAlchemyIngestionSourceRepository,
- )
- from app.core.data_research.sources import (
- DatabaseSourceRegistrationService,
- )
- from app.core.data_source.runtime import get_data_source_manager
- manager = get_data_source_manager()
- return DatabaseSourceRegistrationService(
- SqlAlchemyIngestionSourceRepository(db.session),
- definition_resolver=manager.definitions.get,
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_catalog_snapshot_repository():
- from app.core.data_research.repository import (
- SqlAlchemyCatalogSnapshotRepository,
- )
- return SqlAlchemyCatalogSnapshotRepository(db.session)
- def get_catalog_ingestion_executor():
- from app.core.data_research.catalog.execution import (
- CatalogIngestionExecutor,
- )
- from app.core.data_research.catalog.mysql import MySqlCatalogCollector
- from app.core.data_research.catalog.postgresql import (
- PostgreSqlCatalogCollector,
- )
- from app.core.data_research.catalog.service import CatalogCollectionService
- from app.core.data_research.errors import IngestionSourceInvalid
- from app.core.data_source.runtime import get_data_source_manager
- manager = get_data_source_manager()
- def collector_resolver(database_type):
- if database_type == "postgresql":
- return PostgreSqlCatalogCollector()
- if database_type == "mysql":
- return MySqlCatalogCollector()
- raise IngestionSourceInvalid(
- f"database type {database_type} is not supported"
- )
- collector = CatalogCollectionService(
- manager,
- definition_resolver=manager.definitions.get,
- collector_resolver=collector_resolver,
- )
- return CatalogIngestionExecutor(
- get_ingestion_service(),
- collector,
- get_catalog_snapshot_repository(),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_data_element_service():
- from app.core.data_research.data_elements import DataElementService
- from app.core.data_research.repository import SqlAlchemyDataElementRepository
- from app.core.events.outbox import enqueue_outbox
- return DataElementService(
- SqlAlchemyDataElementRepository(db.session),
- outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
- )
- def get_device_asset_service():
- from app.core.data_research.device_asset_repository import (
- SqlAlchemyDeviceAssetRepository,
- )
- from app.core.data_research.device_assets import DeviceAssetService
- return DeviceAssetService(
- SqlAlchemyDeviceAssetRepository(db.session),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def _device_ontology_authorizer(ontology_repository):
- from app.core.data_research.device_semantics import (
- DeviceOntologyPublicationAuthorizer,
- )
- from app.core.governance.responsibilities import (
- ResponsibilityService,
- SqlAlchemyResponsibilityRepository,
- )
- responsibilities = ResponsibilityService(
- SqlAlchemyResponsibilityRepository(db.session)
- )
- return DeviceOntologyPublicationAuthorizer(
- ontology_repository,
- responsibility_lookup=responsibilities.get,
- )
- def get_device_semantic_service():
- from app.core.data_research.device_semantics import DeviceSemanticService
- from app.core.data_research.ontology.repository import (
- SqlAlchemyOntologyRepository,
- )
- return DeviceSemanticService(
- SqlAlchemyOntologyRepository(db.session),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_device_semantic_code_service():
- from app.core.data_research.device_semantic_repository import (
- SqlAlchemyDeviceSemanticCodeRepository,
- )
- from app.core.data_research.device_semantics import (
- DeviceSemanticCodeService,
- )
- from app.core.data_research.ontology.repository import (
- SqlAlchemyOntologyRepository,
- )
- ontology_repository = SqlAlchemyOntologyRepository(db.session)
- authorizer = _device_ontology_authorizer(ontology_repository)
- return DeviceSemanticCodeService(
- SqlAlchemyDeviceSemanticCodeRepository(db.session),
- review_authorizer=authorizer.assert_accountable,
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_device_entity_resolution_service():
- from app.core.data_research.device_entity_repository import (
- SqlAlchemyDeviceEntityResolutionRepository,
- )
- from app.core.data_research.device_entity_resolution import (
- DeviceEntityForbidden,
- DeviceEntityResolutionService,
- )
- from app.core.governance.responsibilities import (
- ResponsibilityService,
- SqlAlchemyResponsibilityRepository,
- )
- responsibilities = ResponsibilityService(
- SqlAlchemyResponsibilityRepository(db.session)
- )
- def assert_accountable(actor_uid):
- matrix = responsibilities.get(
- "device_mapping",
- "DEVICE_ENTITY_RESOLUTION",
- )
- accountable = [
- item
- for item in matrix.get("assignments", [])
- if item.get("responsibility_role") == "asset_manager"
- and item.get("raci_role") == "accountable"
- ]
- if len(accountable) != 1 or str(accountable[0].get("user_id")) != str(
- actor_uid
- ):
- raise DeviceEntityForbidden(
- "only the accountable device mapping asset manager may decide"
- )
- return DeviceEntityResolutionService(
- SqlAlchemyDeviceEntityResolutionRepository(db.session),
- review_authorizer=assert_accountable,
- auto_merge_enabled=current_app.config.get(
- "DEVICE_ENTITY_AUTO_MERGE_ENABLED",
- False,
- ),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_candidate_decision_service():
- from app.core.data_research.candidate_decisions import CandidateDecisionService
- from app.core.data_research.data_elements import DataElementService
- from app.core.data_research.repository import (
- SqlAlchemyCandidateDecisionRepository,
- SqlAlchemyDataElementRepository,
- )
- return CandidateDecisionService(
- SqlAlchemyCandidateDecisionRepository(db.session),
- data_elements=DataElementService(SqlAlchemyDataElementRepository(db.session)),
- )
- def _artifact_storage():
- from minio import Minio
- from app.core.data_research.artifacts import MinioArtifactStorage
- client = Minio(
- current_app.config["MINIO_HOST"],
- access_key=current_app.config["MINIO_USER"],
- secret_key=current_app.config["MINIO_PASSWORD"],
- secure=bool(current_app.config.get("MINIO_SECURE")),
- )
- return MinioArtifactStorage(client, current_app.config["MINIO_BUCKET"])
- def get_artifact_service():
- from app.core.common.identifiers import new_governance_uid
- from app.core.data_research.artifacts import (
- ArtifactService,
- SqlAlchemyArtifactRepository,
- )
- from app.core.data_research.file_policy import FilePolicy
- return ArtifactService(
- SqlAlchemyArtifactRepository(db.session),
- _artifact_storage(),
- FilePolicy(
- max_bytes=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_BYTES", 25 * 1024 * 1024)),
- max_pages=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_PAGES", 500)),
- ),
- uid_factory=new_governance_uid,
- )
- def get_evidence_service():
- from app.core.data_research.artifacts import EvidenceService
- return EvidenceService(db.session, _artifact_storage())
- def get_ontology_service():
- from app.core.data_research.ontology.publication import (
- OntologyApplicationService,
- OntologyPublicationService,
- )
- from app.core.data_research.ontology.repository import SqlAlchemyOntologyRepository
- from app.core.events.outbox import enqueue_outbox
- repository = SqlAlchemyOntologyRepository(db.session)
- authorizer = _device_ontology_authorizer(repository)
- publication = OntologyPublicationService(
- repository,
- outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
- publication_authorizer=authorizer,
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- return OntologyApplicationService(
- repository,
- publication=publication,
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_ontology_dynamic_service():
- from app.core.data_research.ontology.change_sets import (
- SqlAlchemyDynamicOntologyService,
- )
- return SqlAlchemyDynamicOntologyService(db.session)
- def get_ontology_exchange_service():
- from app.core.data_research.ontology.exchange import OntologyExchangeService
- from app.core.data_research.ontology.repository import SqlAlchemyOntologyRepository
- return OntologyExchangeService(
- SqlAlchemyOntologyRepository(db.session),
- commit=db.session.commit,
- rollback=db.session.rollback,
- )
- def get_semantic_query_service():
- from app.core.data_research.ontology.query import (
- Neo4jSemanticRepository,
- SemanticQueryService,
- )
- from app.services.neo4j_driver import neo4j_driver
- return SemanticQueryService(Neo4jSemanticRepository(neo4j_driver))
- def _identity():
- return getattr(g, "current_user", {}) or {}
- def _record(record):
- return {
- "uid": str(record.uid),
- "source_uid": str(record.source_uid),
- "artifact_uid": str(record.artifact_uid) if record.artifact_uid else None,
- "job_type": record.job_type,
- "parser_version": record.parser_version,
- "status": record.status,
- "parameters": dict(record.parameters or {}),
- "statistics": dict(record.statistics or {}),
- "last_error": record.last_error,
- "attempt_count": int(record.attempt_count or 0),
- "failure_stage": record.failure_stage,
- "actor_uid": record.actor_uid,
- "created_at": record.created_at.isoformat() if record.created_at else None,
- "updated_at": record.updated_at.isoformat() if record.updated_at else None,
- "started_at": record.started_at.isoformat() if record.started_at else None,
- "finished_at": record.finished_at.isoformat() if record.finished_at else None,
- }
- def _catalog_snapshot(record):
- return {
- "uid": str(record.uid),
- "job_uid": str(record.job_uid),
- "source_uid": str(record.source_uid),
- "attempt": int(record.attempt),
- "database_type": record.database_type,
- "content_hash": record.content_hash,
- "snapshot": dict(record.snapshot or {}),
- "evidence_count": int(record.evidence_count),
- "created_at": (
- record.created_at.isoformat()
- if record.created_at
- else None
- ),
- }
- def _element(record):
- return {
- "uid": str(record.uid),
- "code": record.code,
- "status": record.status,
- "current_version": int(record.current_version),
- "snapshot": dict(record.snapshot or {}),
- "created_by": record.created_by,
- "updated_by": record.updated_by,
- }
- def _decision(record):
- return {
- "uid": str(record.uid),
- "candidate_uid": str(record.candidate_uid),
- "action": record.action,
- "data_element_uid": (
- str(record.data_element_uid) if record.data_element_uid else None
- ),
- "evidence_uids": list(record.evidence_uids),
- "actor_uid": record.actor_uid,
- "reason": record.reason,
- }
- def _artifact(record):
- return {
- "uid": str(record.uid),
- "source_uid": str(record.source_uid),
- "filename": record.filename,
- "media_type": record.media_type,
- "size_bytes": int(record.size_bytes),
- "content_hash": record.content_hash,
- "parser_version": record.parser_version,
- }
- def _ontology(record):
- return {
- "uid": str(record.uid),
- "code": record.code,
- "name": record.name,
- "owner_uid": record.owner_uid,
- "status": record.status,
- "draft_revision": int(record.draft_revision),
- "active_version_uid": record.active_version_uid,
- "domain_links": [
- {"domain_uid": link.domain_uid, "role": link.role}
- for link in record.domain_links
- ],
- }
- def _ontology_version(record):
- return {
- "uid": str(record.uid),
- "ontology_uid": str(record.ontology_uid),
- "version": int(record.version),
- "parent_version_uid": record.parent_version_uid,
- "status": record.status,
- "content_hash": record.content_hash,
- "graph_document": record.graph_document.to_dict(),
- "created_by": record.created_by,
- }
- def _iso(value):
- return value.isoformat() if value else None
- def _device_asset(record):
- return {
- "uid": str(record.uid),
- "asset_type": record.asset_type,
- "name": record.name,
- "status": record.status,
- "current_version": int(record.current_version),
- "location": record.location,
- "organization": record.organization,
- "responsible_person": record.responsible_person,
- "attributes": dict(record.attributes or {}),
- "created_by": record.created_by,
- "updated_by": record.updated_by,
- "created_at": _iso(record.created_at),
- "updated_at": _iso(record.updated_at),
- }
- def _device_asset_mapping(record):
- return {
- "uid": str(record.uid),
- "asset_uid": str(record.asset_uid),
- "source_uid": str(record.source_uid),
- "source_entity": record.source_entity,
- "asset_type": record.asset_type,
- "source_code": record.source_code,
- "source_updated_at": _iso(record.source_updated_at),
- "first_seen_at": _iso(record.first_seen_at),
- "last_seen_at": _iso(record.last_seen_at),
- }
- def _device_asset_detail(detail):
- data = _device_asset(detail.asset)
- data["source_mappings"] = [
- _device_asset_mapping(mapping)
- for mapping in detail.mappings
- ]
- return data
- def _device_asset_version(record):
- return {
- "uid": str(record.uid),
- "asset_uid": str(record.asset_uid),
- "version": int(record.version),
- "snapshot": dict(record.snapshot or {}),
- "source_mapping_uid": str(record.source_mapping_uid),
- "actor_uid": record.actor_uid,
- "created_at": _iso(record.created_at),
- }
- def _device_asset_import_result(result):
- return {
- "records": [
- {
- "action": item.action,
- "asset": _device_asset(item.asset),
- "source_mapping": _device_asset_mapping(item.mapping),
- }
- for item in result.items
- ],
- "created_count": int(result.created_count),
- "updated_count": int(result.updated_count),
- "unchanged_count": int(result.unchanged_count),
- }
- def _device_semantic_profile(result):
- if result is None:
- return {
- "bootstrapped": False,
- "ontology": None,
- "version": None,
- "profile": {
- "ready_to_publish": False,
- "class_count": 0,
- "relation_count": 0,
- "mapping_count": 0,
- "missing_classes": [],
- "missing_relations": [],
- "missing_mappings": [],
- },
- }
- profile = result.profile
- return {
- "bootstrapped": True,
- "created": bool(result.created),
- "ontology": _ontology(result.ontology),
- "version": _ontology_version(result.version),
- "profile": {
- "ready_to_publish": bool(profile.ready_to_publish),
- "class_count": int(profile.class_count),
- "relation_count": int(profile.relation_count),
- "mapping_count": int(profile.mapping_count),
- "missing_classes": list(profile.missing_classes),
- "missing_relations": list(profile.missing_relations),
- "missing_mappings": list(profile.missing_mappings),
- },
- }
- def _device_semantic_code(record):
- return {
- "uid": str(record.uid),
- "ontology_uid": str(record.ontology_uid),
- "code_type": record.code_type,
- "canonical_code": record.canonical_code,
- "canonical_name": record.canonical_name,
- "definition": record.definition,
- "status": record.status,
- "current_version": int(record.current_version),
- "source_mappings": [
- dict(item) for item in record.source_mappings
- ],
- "evidence_uids": list(record.evidence_uids),
- "suggestion_source": record.suggestion_source,
- "confidence": record.confidence,
- "created_by": record.created_by,
- "updated_by": record.updated_by,
- "created_at": _iso(record.created_at),
- "updated_at": _iso(record.updated_at),
- }
- def _device_semantic_version(record):
- return {
- "uid": str(record.uid),
- "code_uid": str(record.code_uid),
- "version": int(record.version),
- "snapshot": dict(record.snapshot),
- "created_by": record.created_by,
- "created_at": _iso(record.created_at),
- }
- def _device_semantic_review(record):
- return {
- "uid": str(record.uid),
- "code_uid": str(record.code_uid),
- "version": int(record.version),
- "decision": record.decision,
- "reason": record.reason,
- "actor_uid": record.actor_uid,
- "created_at": _iso(record.created_at),
- }
- def _device_entity_candidate(record):
- return {
- "uid": str(record.uid),
- "left_asset_uid": str(record.left_asset_uid),
- "right_asset_uid": str(record.right_asset_uid),
- "canonical_asset_uid": (
- str(record.canonical_asset_uid)
- if record.canonical_asset_uid
- else None
- ),
- "status": record.status,
- "suggestion_source": record.suggestion_source,
- "confidence": float(record.confidence),
- "explanation": [
- dict(item) for item in record.explanation
- ],
- "evidence_uids": list(record.evidence_uids),
- "model_provider": record.model_provider,
- "model_name": record.model_name,
- "current_version": int(record.current_version),
- "created_by": record.created_by,
- "reviewed_by": record.reviewed_by,
- "created_at": _iso(record.created_at),
- "updated_at": _iso(record.updated_at),
- }
- def _device_entity_review(record):
- return {
- "uid": str(record.uid),
- "candidate_uid": str(record.candidate_uid),
- "version": int(record.version),
- "decision": record.decision,
- "reason": record.reason,
- "actor_uid": record.actor_uid,
- "created_at": _iso(record.created_at),
- }
- def _device_entity_merge(record):
- return {
- "uid": str(record.uid),
- "candidate_uid": str(record.candidate_uid),
- "canonical_asset_uid": str(record.canonical_asset_uid),
- "member_asset_uid": str(record.member_asset_uid),
- "review_uid": str(record.review_uid),
- "snapshot": dict(record.snapshot),
- "actor_uid": record.actor_uid,
- "created_at": _iso(record.created_at),
- }
- def _device_entity_rollback(record):
- return {
- "uid": str(record.uid),
- "merge_uid": str(record.merge_uid),
- "candidate_uid": str(record.candidate_uid),
- "reason": record.reason,
- "snapshot": dict(record.snapshot),
- "actor_uid": record.actor_uid,
- "created_at": _iso(record.created_at),
- }
- def _device_entity_generation(result):
- return {
- "records": [
- _device_entity_candidate(record)
- for record in result.records
- ],
- "created_count": int(result.created_count),
- "existing_count": int(result.existing_count),
- "evaluated_pair_count": int(result.evaluated_pair_count),
- "auto_merged_count": int(result.auto_merged_count),
- "auto_merge_enabled": bool(
- current_app.config.get(
- "DEVICE_ENTITY_AUTO_MERGE_ENABLED",
- False,
- )
- ),
- }
- def _device_asset_page(name, *, default, maximum):
- from app.core.data_research.errors import DeviceAssetInvalid
- raw = request.args.get(name)
- try:
- value = default if raw in (None, "") else int(raw)
- except (TypeError, ValueError) as error:
- raise DeviceAssetInvalid(f"{name} must be an integer") from error
- if value < 1 or value > maximum:
- raise DeviceAssetInvalid(
- f"{name} must be between 1 and {maximum}"
- )
- return value
- def _error(error):
- if isinstance(error, DataResearchError):
- return (
- jsonify(
- failed(
- str(error),
- code=error.http_status,
- error={"code": error.code},
- )
- ),
- error.http_status,
- )
- logger.exception("data-research ingestion request failed")
- return (
- jsonify(
- failed(
- "数据采集任务处理失败",
- code=500,
- error={"code": "DATA_RESEARCH_ERROR"},
- )
- ),
- 500,
- )
- @bp.route("/ingestion-jobs", methods=["POST"])
- def create_ingestion_job():
- payload = request.get_json(silent=True) or {}
- try:
- actor_uid = _identity().get("id") or _identity().get("sub")
- if payload.get("job_type") == "catalog_collect":
- get_database_source_registration_service().ensure(
- payload.get("source_uid"),
- actor_uid=actor_uid,
- )
- record, created = get_ingestion_service().create_job(
- payload,
- actor_uid=actor_uid,
- )
- return jsonify(success(_record(record))), 201 if created else 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs", methods=["GET"])
- def list_ingestion_jobs():
- filters = {
- name: request.args.get(name)
- for name in ("status", "source_uid")
- if request.args.get(name)
- }
- try:
- records = get_ingestion_service().list_jobs(filters)
- return jsonify(
- success({"records": [_record(item) for item in records], "total": len(records)})
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs/<job_uid>", methods=["GET"])
- def get_ingestion_job(job_uid):
- try:
- return jsonify(success(_record(get_ingestion_service().get_job(job_uid)))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs/<job_uid>/execute", methods=["POST"])
- def execute_ingestion_job(job_uid):
- try:
- record = get_catalog_ingestion_executor().execute(job_uid)
- return jsonify(success(_record(record))), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/ingestion-jobs/<job_uid>/catalog-snapshots",
- methods=["GET"],
- )
- def list_catalog_snapshots(job_uid):
- try:
- records = get_catalog_snapshot_repository().list(job_uid)
- return jsonify(
- success(
- {
- "records": [
- _catalog_snapshot(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs/<job_uid>/evidence", methods=["GET"])
- def list_ingestion_job_evidence(job_uid):
- try:
- records = get_evidence_service().list_for_job(job_uid)
- return jsonify(
- success({"records": records, "total": len(records)})
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs/<job_uid>/retry", methods=["POST"])
- def retry_ingestion_job(job_uid):
- try:
- return jsonify(success(_record(get_ingestion_service().retry(job_uid)))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ingestion-jobs/<job_uid>/cancel", methods=["POST"])
- def cancel_ingestion_job(job_uid):
- try:
- service = get_ingestion_service()
- record = service.get_job(job_uid)
- identity = _identity()
- permissions = set(identity.get("permissions") or [])
- actor_uid = identity.get("id") or identity.get("sub")
- if record.actor_uid != actor_uid and "ingestion:admin" not in permissions:
- return jsonify(failed("权限不足", code=403)), 403
- return jsonify(success(_record(service.cancel(job_uid)))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/data-elements", methods=["GET"])
- def list_data_elements():
- try:
- records = get_data_element_service().list(
- status=request.args.get("status"),
- business_domain_uid=request.args.get("business_domain_uid"),
- )
- return jsonify(
- success(
- {
- "records": [_element(item) for item in records],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-assets", methods=["GET"])
- def list_device_assets():
- filters = {
- name: request.args.get(name)
- for name in ("keyword", "asset_type", "status", "source_uid")
- if request.args.get(name)
- }
- try:
- page = _device_asset_page(
- "page",
- default=1,
- maximum=1_000_000,
- )
- page_size = _device_asset_page(
- "page_size",
- default=20,
- maximum=100,
- )
- records, total = get_device_asset_service().search(
- filters,
- page=page,
- page_size=page_size,
- )
- return jsonify(
- success(
- {
- "records": [
- _device_asset_detail(record)
- for record in records
- ],
- "total": int(total),
- "page": page,
- "page_size": page_size,
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-assets/import", methods=["POST"])
- def import_device_assets():
- try:
- result = get_device_asset_service().import_records(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_device_asset_import_result(result))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-assets/<asset_uid>", methods=["GET"])
- def get_device_asset(asset_uid):
- try:
- return jsonify(
- success(
- _device_asset_detail(
- get_device_asset_service().get(asset_uid)
- )
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-assets/<asset_uid>/versions", methods=["GET"])
- def list_device_asset_versions(asset_uid):
- try:
- records = get_device_asset_service().versions(asset_uid)
- return jsonify(
- success(
- {
- "records": [
- _device_asset_version(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-semantics/bootstrap", methods=["POST"])
- def bootstrap_device_semantics():
- try:
- result = get_device_semantic_service().bootstrap(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return (
- jsonify(success(_device_semantic_profile(result))),
- 201 if result.created else 200,
- )
- except Exception as error:
- return _error(error)
- @bp.route("/device-semantics/profile", methods=["GET"])
- def get_device_semantic_profile():
- try:
- result = get_device_semantic_service().profile()
- return jsonify(success(_device_semantic_profile(result))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-semantics/codes", methods=["GET"])
- def list_device_semantic_codes():
- filters = {
- name: request.args.get(name)
- for name in ("ontology_uid", "code_type", "status", "keyword")
- if request.args.get(name)
- }
- try:
- records, total = get_device_semantic_code_service().search(
- filters,
- page=request.args.get("page", 1),
- page_size=request.args.get("page_size", 20),
- )
- return jsonify(
- success(
- {
- "records": [
- _device_semantic_code(record)
- for record in records
- ],
- "total": int(total),
- "page": int(request.args.get("page", 1)),
- "page_size": int(request.args.get("page_size", 20)),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-semantics/codes", methods=["POST"])
- def create_device_semantic_code():
- try:
- record = get_device_semantic_code_service().create(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_device_semantic_code(record))), 201
- except Exception as error:
- return _error(error)
- @bp.route("/device-semantics/codes/<code_uid>", methods=["GET"])
- def get_device_semantic_code(code_uid):
- try:
- record = get_device_semantic_code_service().get(code_uid)
- return jsonify(success(_device_semantic_code(record))), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-semantics/codes/<code_uid>/revisions",
- methods=["POST"],
- )
- def revise_device_semantic_code(code_uid):
- payload = request.get_json(silent=True) or {}
- try:
- record = get_device_semantic_code_service().revise(
- code_uid,
- payload,
- expected_version=payload.get("expected_version"),
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_device_semantic_code(record))), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-semantics/codes/<code_uid>/submit",
- methods=["POST"],
- )
- def submit_device_semantic_code(code_uid):
- payload = request.get_json(silent=True) or {}
- try:
- record = get_device_semantic_code_service().submit(
- code_uid,
- expected_version=payload.get("expected_version"),
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_device_semantic_code(record))), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-semantics/codes/<code_uid>/review",
- methods=["POST"],
- )
- def review_device_semantic_code(code_uid):
- try:
- record, review = get_device_semantic_code_service().review(
- code_uid,
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(
- success(
- {
- "record": _device_semantic_code(record),
- "review": _device_semantic_review(review),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-semantics/codes/<code_uid>/versions",
- methods=["GET"],
- )
- def list_device_semantic_code_versions(code_uid):
- try:
- records = get_device_semantic_code_service().versions(code_uid)
- return jsonify(
- success(
- {
- "records": [
- _device_semantic_version(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-semantics/codes/<code_uid>/reviews",
- methods=["GET"],
- )
- def list_device_semantic_code_reviews(code_uid):
- try:
- records = get_device_semantic_code_service().reviews(code_uid)
- return jsonify(
- success(
- {
- "records": [
- _device_semantic_review(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-entities/candidates", methods=["GET"])
- def list_device_entity_candidates():
- filters = {
- name: request.args.get(name)
- for name in ("status", "suggestion_source")
- if request.args.get(name)
- }
- try:
- records, total = get_device_entity_resolution_service().search(
- filters,
- page=request.args.get("page", 1),
- page_size=request.args.get("page_size", 20),
- )
- return jsonify(
- success(
- {
- "records": [
- _device_entity_candidate(record)
- for record in records
- ],
- "total": int(total),
- "page": int(request.args.get("page", 1)),
- "page_size": int(request.args.get("page_size", 20)),
- "auto_merge_enabled": bool(
- current_app.config.get(
- "DEVICE_ENTITY_AUTO_MERGE_ENABLED",
- False,
- )
- ),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/device-entities/candidates/generate", methods=["POST"])
- def generate_device_entity_candidates():
- try:
- result = get_device_entity_resolution_service().generate(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- status = 201 if result.created_count else 200
- return jsonify(success(_device_entity_generation(result))), status
- except Exception as error:
- return _error(error)
- @bp.route("/device-entities/candidates", methods=["POST"])
- def submit_device_entity_candidate():
- try:
- record = (
- get_device_entity_resolution_service().submit_ai_candidate(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- )
- return jsonify(success(_device_entity_candidate(record))), 201
- except Exception as error:
- return _error(error)
- @bp.route("/device-entities/candidates/<candidate_uid>", methods=["GET"])
- def get_device_entity_candidate(candidate_uid):
- try:
- record = get_device_entity_resolution_service().get(candidate_uid)
- return jsonify(success(_device_entity_candidate(record))), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-entities/candidates/<candidate_uid>/review",
- methods=["POST"],
- )
- def review_device_entity_candidate(candidate_uid):
- try:
- candidate, review, merge = (
- get_device_entity_resolution_service().review(
- candidate_uid,
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- )
- return jsonify(
- success(
- {
- "candidate": _device_entity_candidate(candidate),
- "review": _device_entity_review(review),
- "merge": (
- _device_entity_merge(merge)
- if merge is not None
- else None
- ),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-entities/candidates/<candidate_uid>/reviews",
- methods=["GET"],
- )
- def list_device_entity_reviews(candidate_uid):
- try:
- records = get_device_entity_resolution_service().reviews(
- candidate_uid
- )
- return jsonify(
- success(
- {
- "records": [
- _device_entity_review(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-entities/candidates/<candidate_uid>/merges",
- methods=["GET"],
- )
- def list_device_entity_merges(candidate_uid):
- try:
- records = get_device_entity_resolution_service().merges(
- candidate_uid
- )
- return jsonify(
- success(
- {
- "records": [
- _device_entity_merge(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-entities/merges/<merge_uid>/rollback",
- methods=["POST"],
- )
- def rollback_device_entity_merge(merge_uid):
- try:
- candidate, rollback = (
- get_device_entity_resolution_service().rollback(
- merge_uid,
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- )
- return jsonify(
- success(
- {
- "candidate": _device_entity_candidate(candidate),
- "rollback": _device_entity_rollback(rollback),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route(
- "/device-entities/merges/<merge_uid>/rollbacks",
- methods=["GET"],
- )
- def list_device_entity_rollbacks(merge_uid):
- try:
- records = get_device_entity_resolution_service().rollbacks(
- merge_uid
- )
- return jsonify(
- success(
- {
- "records": [
- _device_entity_rollback(record)
- for record in records
- ],
- "total": len(records),
- }
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/data-elements", methods=["POST"])
- def create_data_element():
- try:
- record = get_data_element_service().create_draft(
- request.get_json(silent=True) or {},
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- db.session.commit()
- return jsonify(success(_element(record))), 201
- except Exception as error:
- db.session.rollback()
- return _error(error)
- @bp.route("/data-elements/<element_uid>/transitions", methods=["POST"])
- def transition_data_element(element_uid):
- payload = request.get_json(silent=True) or {}
- identity = _identity()
- target_status = str(payload.get("target_status") or "")
- if (
- target_status in {"published", "deprecated", "retired"}
- and "data-elements:publish" not in set(identity.get("permissions") or [])
- ):
- return jsonify(failed("权限不足", code=403)), 403
- try:
- record = get_data_element_service().transition(
- element_uid,
- target_status,
- expected_version=int(payload.get("expected_version")),
- actor_uid=identity.get("id") or identity.get("sub"),
- )
- db.session.commit()
- return jsonify(success(_element(record))), 200
- except Exception as error:
- db.session.rollback()
- return _error(error)
- @bp.route("/candidate-decisions", methods=["POST"])
- def decide_candidates():
- payload = request.get_json(silent=True) or {}
- try:
- records = get_candidate_decision_service().decide(
- payload.get("decisions") or [],
- actor_uid=_identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success([_decision(record) for record in records])), 200
- except Exception as error:
- return _error(error)
- @bp.route("/sources/files", methods=["POST"])
- def upload_source_file():
- uploaded = request.files.get("file")
- if uploaded is None:
- return jsonify(failed("缺少上传文件", code=400)), 400
- try:
- record, created = get_artifact_service().store(
- request.form.get("source_uid"),
- uploaded.filename,
- uploaded.mimetype,
- uploaded.read(),
- request.form.get("parser_version") or "auto-v1",
- )
- db.session.commit()
- return jsonify(success(_artifact(record))), 201 if created else 200
- except Exception as error:
- db.session.rollback()
- return _error(error)
- @bp.route("/evidence/<evidence_uid>", methods=["GET"])
- def get_evidence(evidence_uid):
- identity = _identity()
- try:
- service = get_evidence_service()
- if request.args.get("download") in {"1", "true", "yes"}:
- if "evidence:download" not in set(identity.get("permissions") or []):
- return jsonify(failed("权限不足", code=403)), 403
- content, filename, media_type = service.download(evidence_uid)
- return send_file(
- io.BytesIO(content),
- mimetype=media_type,
- as_attachment=True,
- download_name=filename,
- )
- preview = service.preview(evidence_uid)
- from app.core.data_research.artifacts import redact_excerpt
- preview["excerpt"] = redact_excerpt(preview.get("excerpt"))
- return jsonify(success(preview)), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies", methods=["GET"])
- def list_ontologies():
- try:
- return jsonify(success([_ontology(item) for item in get_ontology_service().list()])), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies", methods=["POST"])
- def create_ontology():
- try:
- record = get_ontology_service().create(
- request.get_json(silent=True) or {},
- _identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_ontology(record))), 201
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>", methods=["GET"])
- def get_ontology(ontology_uid):
- try:
- return jsonify(success(_ontology(get_ontology_service().get(ontology_uid)))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/versions", methods=["GET"])
- def list_ontology_versions(ontology_uid):
- try:
- return jsonify(
- success(
- [
- _ontology_version(item)
- for item in get_ontology_service().list_versions(ontology_uid)
- ]
- )
- ), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/graph", methods=["GET"])
- def get_ontology_graph(ontology_uid):
- try:
- service = get_ontology_service()
- ontology = service.get(ontology_uid)
- version = service.latest_version(ontology_uid)
- data = (
- _ontology_version(version)
- if version is not None
- else {
- "uid": None,
- "ontology_uid": str(ontology_uid),
- "version": 0,
- "parent_version_uid": None,
- "status": "draft",
- "content_hash": None,
- "graph_document": {
- name: []
- for name in (
- "classes",
- "properties",
- "relations",
- "constraints",
- "domain_links",
- "element_mappings",
- )
- },
- "created_by": None,
- }
- )
- response = jsonify(success(data))
- response.headers["ETag"] = f'"{int(ontology.draft_revision)}"'
- return response, 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/graph", methods=["PATCH"])
- def save_ontology_graph(ontology_uid):
- etag = str(request.headers.get("If-Match") or "").strip().strip('"')
- if not etag.isdigit():
- return jsonify(failed("缺少有效 If-Match 修订号", code=428)), 428
- try:
- version = get_ontology_service().save_draft(
- ontology_uid,
- request.get_json(silent=True) or {},
- int(etag),
- _identity().get("id") or _identity().get("sub"),
- )
- response = jsonify(success(_ontology_version(version)))
- response.headers["ETag"] = f'"{int(etag) + 1}"'
- return response, 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/validate", methods=["POST"])
- def validate_ontology(ontology_uid):
- try:
- issues = get_ontology_service().validate(ontology_uid)
- data = [
- {"code": item.code, "message": item.message, "path": item.path}
- for item in issues
- ]
- return jsonify(success(data)), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/publish", methods=["POST"])
- def publish_ontology(ontology_uid):
- try:
- version = get_ontology_service().publish(
- ontology_uid,
- request.headers.get("Idempotency-Key") or f"{ontology_uid}:publish",
- _identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_ontology_version(version))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/diff", methods=["GET"])
- def diff_ontology(ontology_uid):
- try:
- return jsonify(success(get_ontology_service().diff(
- ontology_uid, request.args.get("left"), request.args.get("right")
- ))), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/rollback", methods=["POST"])
- def rollback_ontology(ontology_uid):
- payload = request.get_json(silent=True) or {}
- try:
- version = get_ontology_service().rollback(
- ontology_uid,
- payload.get("target_version_uid"),
- int(payload.get("expected_revision")),
- _identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(_ontology_version(version))), 201
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/suggestions", methods=["POST"])
- def generate_ontology_suggestions(ontology_uid):
- try:
- result = get_ontology_dynamic_service().generate(
- ontology_uid,
- request.get_json(silent=True) or {},
- _identity().get("id") or _identity().get("sub"),
- )
- records = (
- result.get("suggestions") or []
- if isinstance(result, dict)
- else result
- )
- suggestions = [
- item
- if isinstance(item, dict)
- else {
- "uid": item.uid,
- "kind": item.kind,
- "payload": item.payload,
- "evidence_uids": list(item.evidence_uids),
- "confidence": item.confidence,
- "source": item.source,
- "model_version": item.model_version,
- "prompt_version": item.prompt_version,
- }
- for item in records
- ]
- data = {
- "change_set_uid": (
- result.get("change_set_uid")
- if isinstance(result, dict)
- else None
- ),
- "suggestions": suggestions,
- }
- return jsonify(success(data)), 201
- except Exception as error:
- return _error(error)
- @bp.route(
- "/ontologies/<ontology_uid>/change-sets/<change_set_uid>/decisions",
- methods=["POST"],
- )
- def decide_ontology_change_set(ontology_uid, change_set_uid):
- payload = request.get_json(silent=True) or {}
- try:
- result = get_ontology_dynamic_service().decide(
- ontology_uid,
- change_set_uid,
- payload.get("decisions") or [],
- _identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(result)), 200
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/<ontology_uid>/export", methods=["GET"])
- def export_ontology(ontology_uid):
- try:
- content, media_type, filename = get_ontology_exchange_service().export(
- ontology_uid, request.args.get("format") or "json"
- )
- return send_file(
- io.BytesIO(content),
- mimetype=media_type,
- as_attachment=True,
- download_name=filename,
- )
- except Exception as error:
- return _error(error)
- @bp.route("/ontologies/import", methods=["POST"])
- def import_ontology():
- try:
- uploaded = request.files.get("file")
- content = uploaded.read() if uploaded is not None else request.get_data(cache=False)
- result = get_ontology_exchange_service().import_document(
- content,
- request.args.get("format") or "json",
- _identity().get("id") or _identity().get("sub"),
- )
- return jsonify(success(result)), 201
- except Exception as error:
- return _error(error)
- @bp.route("/semantic/properties/<property_uid>", methods=["GET"])
- def query_semantic_property(property_uid):
- domain = str(request.args.get("business_domain_uid") or "").strip()
- identity = _identity()
- scoped = set(identity.get("business_domains") or [domain])
- if "*" not in scoped and domain not in scoped:
- return jsonify(failed("权限不足", code=403)), 403
- try:
- result = get_semantic_query_service().trace_property(
- property_uid,
- allowed_domains={domain},
- limit=int(request.args.get("limit") or 20),
- after_uid=request.args.get("after_uid"),
- )
- return jsonify(success(result)), 200
- except Exception as error:
- return _error(error)
|