routes.py 41 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319
  1. """HTTP orchestration boundary for data-research ingestion jobs."""
  2. from __future__ import annotations
  3. import io
  4. import logging
  5. from flask import current_app, g, jsonify, request, send_file
  6. from app import db
  7. from app.api.data_development import bp
  8. from app.core.data_research.errors import DataResearchError
  9. from app.models.result import failed, success
  10. logger = logging.getLogger(__name__)
  11. def get_ingestion_service():
  12. from app.core.data_research.ingestion import IngestionService
  13. from app.core.data_research.repository import SqlAlchemyIngestionJobRepository
  14. return IngestionService(
  15. SqlAlchemyIngestionJobRepository(db.session),
  16. commit=db.session.commit,
  17. rollback=db.session.rollback,
  18. )
  19. def get_database_source_registration_service():
  20. from app.core.data_research.repository import (
  21. SqlAlchemyIngestionSourceRepository,
  22. )
  23. from app.core.data_research.sources import (
  24. DatabaseSourceRegistrationService,
  25. )
  26. from app.core.data_source.runtime import get_data_source_manager
  27. manager = get_data_source_manager()
  28. return DatabaseSourceRegistrationService(
  29. SqlAlchemyIngestionSourceRepository(db.session),
  30. definition_resolver=manager.definitions.get,
  31. commit=db.session.commit,
  32. rollback=db.session.rollback,
  33. )
  34. def get_catalog_snapshot_repository():
  35. from app.core.data_research.repository import (
  36. SqlAlchemyCatalogSnapshotRepository,
  37. )
  38. return SqlAlchemyCatalogSnapshotRepository(db.session)
  39. def get_catalog_ingestion_executor():
  40. from app.core.data_research.catalog.execution import (
  41. CatalogIngestionExecutor,
  42. )
  43. from app.core.data_research.catalog.mysql import MySqlCatalogCollector
  44. from app.core.data_research.catalog.postgresql import (
  45. PostgreSqlCatalogCollector,
  46. )
  47. from app.core.data_research.catalog.service import CatalogCollectionService
  48. from app.core.data_research.errors import IngestionSourceInvalid
  49. from app.core.data_source.runtime import get_data_source_manager
  50. manager = get_data_source_manager()
  51. def collector_resolver(database_type):
  52. if database_type == "postgresql":
  53. return PostgreSqlCatalogCollector()
  54. if database_type == "mysql":
  55. return MySqlCatalogCollector()
  56. raise IngestionSourceInvalid(
  57. f"database type {database_type} is not supported"
  58. )
  59. collector = CatalogCollectionService(
  60. manager,
  61. definition_resolver=manager.definitions.get,
  62. collector_resolver=collector_resolver,
  63. )
  64. return CatalogIngestionExecutor(
  65. get_ingestion_service(),
  66. collector,
  67. get_catalog_snapshot_repository(),
  68. commit=db.session.commit,
  69. rollback=db.session.rollback,
  70. )
  71. def get_data_element_service():
  72. from app.core.data_research.data_elements import DataElementService
  73. from app.core.data_research.repository import SqlAlchemyDataElementRepository
  74. from app.core.events.outbox import enqueue_outbox
  75. return DataElementService(
  76. SqlAlchemyDataElementRepository(db.session),
  77. outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
  78. )
  79. def get_device_asset_service():
  80. from app.core.data_research.device_asset_repository import (
  81. SqlAlchemyDeviceAssetRepository,
  82. )
  83. from app.core.data_research.device_assets import DeviceAssetService
  84. return DeviceAssetService(
  85. SqlAlchemyDeviceAssetRepository(db.session),
  86. commit=db.session.commit,
  87. rollback=db.session.rollback,
  88. )
  89. def _device_ontology_authorizer(ontology_repository):
  90. from app.core.data_research.device_semantics import (
  91. DeviceOntologyPublicationAuthorizer,
  92. )
  93. from app.core.governance.responsibilities import (
  94. ResponsibilityService,
  95. SqlAlchemyResponsibilityRepository,
  96. )
  97. responsibilities = ResponsibilityService(
  98. SqlAlchemyResponsibilityRepository(db.session)
  99. )
  100. return DeviceOntologyPublicationAuthorizer(
  101. ontology_repository,
  102. responsibility_lookup=responsibilities.get,
  103. )
  104. def get_device_semantic_service():
  105. from app.core.data_research.device_semantics import DeviceSemanticService
  106. from app.core.data_research.ontology.repository import (
  107. SqlAlchemyOntologyRepository,
  108. )
  109. return DeviceSemanticService(
  110. SqlAlchemyOntologyRepository(db.session),
  111. commit=db.session.commit,
  112. rollback=db.session.rollback,
  113. )
  114. def get_device_semantic_code_service():
  115. from app.core.data_research.device_semantic_repository import (
  116. SqlAlchemyDeviceSemanticCodeRepository,
  117. )
  118. from app.core.data_research.device_semantics import (
  119. DeviceSemanticCodeService,
  120. )
  121. from app.core.data_research.ontology.repository import (
  122. SqlAlchemyOntologyRepository,
  123. )
  124. ontology_repository = SqlAlchemyOntologyRepository(db.session)
  125. authorizer = _device_ontology_authorizer(ontology_repository)
  126. return DeviceSemanticCodeService(
  127. SqlAlchemyDeviceSemanticCodeRepository(db.session),
  128. review_authorizer=authorizer.assert_accountable,
  129. commit=db.session.commit,
  130. rollback=db.session.rollback,
  131. )
  132. def get_candidate_decision_service():
  133. from app.core.data_research.candidate_decisions import CandidateDecisionService
  134. from app.core.data_research.data_elements import DataElementService
  135. from app.core.data_research.repository import (
  136. SqlAlchemyCandidateDecisionRepository,
  137. SqlAlchemyDataElementRepository,
  138. )
  139. return CandidateDecisionService(
  140. SqlAlchemyCandidateDecisionRepository(db.session),
  141. data_elements=DataElementService(SqlAlchemyDataElementRepository(db.session)),
  142. )
  143. def _artifact_storage():
  144. from minio import Minio
  145. from app.core.data_research.artifacts import MinioArtifactStorage
  146. client = Minio(
  147. current_app.config["MINIO_HOST"],
  148. access_key=current_app.config["MINIO_USER"],
  149. secret_key=current_app.config["MINIO_PASSWORD"],
  150. secure=bool(current_app.config.get("MINIO_SECURE")),
  151. )
  152. return MinioArtifactStorage(client, current_app.config["MINIO_BUCKET"])
  153. def get_artifact_service():
  154. from app.core.common.identifiers import new_governance_uid
  155. from app.core.data_research.artifacts import (
  156. ArtifactService,
  157. SqlAlchemyArtifactRepository,
  158. )
  159. from app.core.data_research.file_policy import FilePolicy
  160. return ArtifactService(
  161. SqlAlchemyArtifactRepository(db.session),
  162. _artifact_storage(),
  163. FilePolicy(
  164. max_bytes=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_BYTES", 25 * 1024 * 1024)),
  165. max_pages=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_PAGES", 500)),
  166. ),
  167. uid_factory=new_governance_uid,
  168. )
  169. def get_evidence_service():
  170. from app.core.data_research.artifacts import EvidenceService
  171. return EvidenceService(db.session, _artifact_storage())
  172. def get_ontology_service():
  173. from app.core.data_research.ontology.publication import (
  174. OntologyApplicationService,
  175. OntologyPublicationService,
  176. )
  177. from app.core.data_research.ontology.repository import SqlAlchemyOntologyRepository
  178. from app.core.events.outbox import enqueue_outbox
  179. repository = SqlAlchemyOntologyRepository(db.session)
  180. authorizer = _device_ontology_authorizer(repository)
  181. publication = OntologyPublicationService(
  182. repository,
  183. outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
  184. publication_authorizer=authorizer,
  185. commit=db.session.commit,
  186. rollback=db.session.rollback,
  187. )
  188. return OntologyApplicationService(
  189. repository,
  190. publication=publication,
  191. commit=db.session.commit,
  192. rollback=db.session.rollback,
  193. )
  194. def get_ontology_dynamic_service():
  195. from app.core.data_research.ontology.change_sets import (
  196. SqlAlchemyDynamicOntologyService,
  197. )
  198. return SqlAlchemyDynamicOntologyService(db.session)
  199. def get_ontology_exchange_service():
  200. from app.core.data_research.ontology.exchange import OntologyExchangeService
  201. from app.core.data_research.ontology.repository import SqlAlchemyOntologyRepository
  202. return OntologyExchangeService(
  203. SqlAlchemyOntologyRepository(db.session),
  204. commit=db.session.commit,
  205. rollback=db.session.rollback,
  206. )
  207. def get_semantic_query_service():
  208. from app.core.data_research.ontology.query import (
  209. Neo4jSemanticRepository,
  210. SemanticQueryService,
  211. )
  212. from app.services.neo4j_driver import neo4j_driver
  213. return SemanticQueryService(Neo4jSemanticRepository(neo4j_driver))
  214. def _identity():
  215. return getattr(g, "current_user", {}) or {}
  216. def _record(record):
  217. return {
  218. "uid": str(record.uid),
  219. "source_uid": str(record.source_uid),
  220. "artifact_uid": str(record.artifact_uid) if record.artifact_uid else None,
  221. "job_type": record.job_type,
  222. "parser_version": record.parser_version,
  223. "status": record.status,
  224. "parameters": dict(record.parameters or {}),
  225. "statistics": dict(record.statistics or {}),
  226. "last_error": record.last_error,
  227. "attempt_count": int(record.attempt_count or 0),
  228. "failure_stage": record.failure_stage,
  229. "actor_uid": record.actor_uid,
  230. "created_at": record.created_at.isoformat() if record.created_at else None,
  231. "updated_at": record.updated_at.isoformat() if record.updated_at else None,
  232. "started_at": record.started_at.isoformat() if record.started_at else None,
  233. "finished_at": record.finished_at.isoformat() if record.finished_at else None,
  234. }
  235. def _catalog_snapshot(record):
  236. return {
  237. "uid": str(record.uid),
  238. "job_uid": str(record.job_uid),
  239. "source_uid": str(record.source_uid),
  240. "attempt": int(record.attempt),
  241. "database_type": record.database_type,
  242. "content_hash": record.content_hash,
  243. "snapshot": dict(record.snapshot or {}),
  244. "evidence_count": int(record.evidence_count),
  245. "created_at": (
  246. record.created_at.isoformat()
  247. if record.created_at
  248. else None
  249. ),
  250. }
  251. def _element(record):
  252. return {
  253. "uid": str(record.uid),
  254. "code": record.code,
  255. "status": record.status,
  256. "current_version": int(record.current_version),
  257. "snapshot": dict(record.snapshot or {}),
  258. "created_by": record.created_by,
  259. "updated_by": record.updated_by,
  260. }
  261. def _decision(record):
  262. return {
  263. "uid": str(record.uid),
  264. "candidate_uid": str(record.candidate_uid),
  265. "action": record.action,
  266. "data_element_uid": (
  267. str(record.data_element_uid) if record.data_element_uid else None
  268. ),
  269. "evidence_uids": list(record.evidence_uids),
  270. "actor_uid": record.actor_uid,
  271. "reason": record.reason,
  272. }
  273. def _artifact(record):
  274. return {
  275. "uid": str(record.uid),
  276. "source_uid": str(record.source_uid),
  277. "filename": record.filename,
  278. "media_type": record.media_type,
  279. "size_bytes": int(record.size_bytes),
  280. "content_hash": record.content_hash,
  281. "parser_version": record.parser_version,
  282. }
  283. def _ontology(record):
  284. return {
  285. "uid": str(record.uid),
  286. "code": record.code,
  287. "name": record.name,
  288. "owner_uid": record.owner_uid,
  289. "status": record.status,
  290. "draft_revision": int(record.draft_revision),
  291. "active_version_uid": record.active_version_uid,
  292. "domain_links": [
  293. {"domain_uid": link.domain_uid, "role": link.role}
  294. for link in record.domain_links
  295. ],
  296. }
  297. def _ontology_version(record):
  298. return {
  299. "uid": str(record.uid),
  300. "ontology_uid": str(record.ontology_uid),
  301. "version": int(record.version),
  302. "parent_version_uid": record.parent_version_uid,
  303. "status": record.status,
  304. "content_hash": record.content_hash,
  305. "graph_document": record.graph_document.to_dict(),
  306. "created_by": record.created_by,
  307. }
  308. def _iso(value):
  309. return value.isoformat() if value else None
  310. def _device_asset(record):
  311. return {
  312. "uid": str(record.uid),
  313. "asset_type": record.asset_type,
  314. "name": record.name,
  315. "status": record.status,
  316. "current_version": int(record.current_version),
  317. "location": record.location,
  318. "organization": record.organization,
  319. "responsible_person": record.responsible_person,
  320. "attributes": dict(record.attributes or {}),
  321. "created_by": record.created_by,
  322. "updated_by": record.updated_by,
  323. "created_at": _iso(record.created_at),
  324. "updated_at": _iso(record.updated_at),
  325. }
  326. def _device_asset_mapping(record):
  327. return {
  328. "uid": str(record.uid),
  329. "asset_uid": str(record.asset_uid),
  330. "source_uid": str(record.source_uid),
  331. "source_entity": record.source_entity,
  332. "asset_type": record.asset_type,
  333. "source_code": record.source_code,
  334. "source_updated_at": _iso(record.source_updated_at),
  335. "first_seen_at": _iso(record.first_seen_at),
  336. "last_seen_at": _iso(record.last_seen_at),
  337. }
  338. def _device_asset_detail(detail):
  339. data = _device_asset(detail.asset)
  340. data["source_mappings"] = [
  341. _device_asset_mapping(mapping)
  342. for mapping in detail.mappings
  343. ]
  344. return data
  345. def _device_asset_version(record):
  346. return {
  347. "uid": str(record.uid),
  348. "asset_uid": str(record.asset_uid),
  349. "version": int(record.version),
  350. "snapshot": dict(record.snapshot or {}),
  351. "source_mapping_uid": str(record.source_mapping_uid),
  352. "actor_uid": record.actor_uid,
  353. "created_at": _iso(record.created_at),
  354. }
  355. def _device_asset_import_result(result):
  356. return {
  357. "records": [
  358. {
  359. "action": item.action,
  360. "asset": _device_asset(item.asset),
  361. "source_mapping": _device_asset_mapping(item.mapping),
  362. }
  363. for item in result.items
  364. ],
  365. "created_count": int(result.created_count),
  366. "updated_count": int(result.updated_count),
  367. "unchanged_count": int(result.unchanged_count),
  368. }
  369. def _device_semantic_profile(result):
  370. if result is None:
  371. return {
  372. "bootstrapped": False,
  373. "ontology": None,
  374. "version": None,
  375. "profile": {
  376. "ready_to_publish": False,
  377. "class_count": 0,
  378. "relation_count": 0,
  379. "mapping_count": 0,
  380. "missing_classes": [],
  381. "missing_relations": [],
  382. "missing_mappings": [],
  383. },
  384. }
  385. profile = result.profile
  386. return {
  387. "bootstrapped": True,
  388. "created": bool(result.created),
  389. "ontology": _ontology(result.ontology),
  390. "version": _ontology_version(result.version),
  391. "profile": {
  392. "ready_to_publish": bool(profile.ready_to_publish),
  393. "class_count": int(profile.class_count),
  394. "relation_count": int(profile.relation_count),
  395. "mapping_count": int(profile.mapping_count),
  396. "missing_classes": list(profile.missing_classes),
  397. "missing_relations": list(profile.missing_relations),
  398. "missing_mappings": list(profile.missing_mappings),
  399. },
  400. }
  401. def _device_semantic_code(record):
  402. return {
  403. "uid": str(record.uid),
  404. "ontology_uid": str(record.ontology_uid),
  405. "code_type": record.code_type,
  406. "canonical_code": record.canonical_code,
  407. "canonical_name": record.canonical_name,
  408. "definition": record.definition,
  409. "status": record.status,
  410. "current_version": int(record.current_version),
  411. "source_mappings": [
  412. dict(item) for item in record.source_mappings
  413. ],
  414. "evidence_uids": list(record.evidence_uids),
  415. "suggestion_source": record.suggestion_source,
  416. "confidence": record.confidence,
  417. "created_by": record.created_by,
  418. "updated_by": record.updated_by,
  419. "created_at": _iso(record.created_at),
  420. "updated_at": _iso(record.updated_at),
  421. }
  422. def _device_semantic_version(record):
  423. return {
  424. "uid": str(record.uid),
  425. "code_uid": str(record.code_uid),
  426. "version": int(record.version),
  427. "snapshot": dict(record.snapshot),
  428. "created_by": record.created_by,
  429. "created_at": _iso(record.created_at),
  430. }
  431. def _device_semantic_review(record):
  432. return {
  433. "uid": str(record.uid),
  434. "code_uid": str(record.code_uid),
  435. "version": int(record.version),
  436. "decision": record.decision,
  437. "reason": record.reason,
  438. "actor_uid": record.actor_uid,
  439. "created_at": _iso(record.created_at),
  440. }
  441. def _device_asset_page(name, *, default, maximum):
  442. from app.core.data_research.errors import DeviceAssetInvalid
  443. raw = request.args.get(name)
  444. try:
  445. value = default if raw in (None, "") else int(raw)
  446. except (TypeError, ValueError) as error:
  447. raise DeviceAssetInvalid(f"{name} must be an integer") from error
  448. if value < 1 or value > maximum:
  449. raise DeviceAssetInvalid(
  450. f"{name} must be between 1 and {maximum}"
  451. )
  452. return value
  453. def _error(error):
  454. if isinstance(error, DataResearchError):
  455. return (
  456. jsonify(
  457. failed(
  458. str(error),
  459. code=error.http_status,
  460. error={"code": error.code},
  461. )
  462. ),
  463. error.http_status,
  464. )
  465. logger.exception("data-research ingestion request failed")
  466. return (
  467. jsonify(
  468. failed(
  469. "数据采集任务处理失败",
  470. code=500,
  471. error={"code": "DATA_RESEARCH_ERROR"},
  472. )
  473. ),
  474. 500,
  475. )
  476. @bp.route("/ingestion-jobs", methods=["POST"])
  477. def create_ingestion_job():
  478. payload = request.get_json(silent=True) or {}
  479. try:
  480. actor_uid = _identity().get("id") or _identity().get("sub")
  481. if payload.get("job_type") == "catalog_collect":
  482. get_database_source_registration_service().ensure(
  483. payload.get("source_uid"),
  484. actor_uid=actor_uid,
  485. )
  486. record, created = get_ingestion_service().create_job(
  487. payload,
  488. actor_uid=actor_uid,
  489. )
  490. return jsonify(success(_record(record))), 201 if created else 200
  491. except Exception as error:
  492. return _error(error)
  493. @bp.route("/ingestion-jobs", methods=["GET"])
  494. def list_ingestion_jobs():
  495. filters = {
  496. name: request.args.get(name)
  497. for name in ("status", "source_uid")
  498. if request.args.get(name)
  499. }
  500. try:
  501. records = get_ingestion_service().list_jobs(filters)
  502. return jsonify(
  503. success({"records": [_record(item) for item in records], "total": len(records)})
  504. ), 200
  505. except Exception as error:
  506. return _error(error)
  507. @bp.route("/ingestion-jobs/<job_uid>", methods=["GET"])
  508. def get_ingestion_job(job_uid):
  509. try:
  510. return jsonify(success(_record(get_ingestion_service().get_job(job_uid)))), 200
  511. except Exception as error:
  512. return _error(error)
  513. @bp.route("/ingestion-jobs/<job_uid>/execute", methods=["POST"])
  514. def execute_ingestion_job(job_uid):
  515. try:
  516. record = get_catalog_ingestion_executor().execute(job_uid)
  517. return jsonify(success(_record(record))), 200
  518. except Exception as error:
  519. return _error(error)
  520. @bp.route(
  521. "/ingestion-jobs/<job_uid>/catalog-snapshots",
  522. methods=["GET"],
  523. )
  524. def list_catalog_snapshots(job_uid):
  525. try:
  526. records = get_catalog_snapshot_repository().list(job_uid)
  527. return jsonify(
  528. success(
  529. {
  530. "records": [
  531. _catalog_snapshot(record)
  532. for record in records
  533. ],
  534. "total": len(records),
  535. }
  536. )
  537. ), 200
  538. except Exception as error:
  539. return _error(error)
  540. @bp.route("/ingestion-jobs/<job_uid>/evidence", methods=["GET"])
  541. def list_ingestion_job_evidence(job_uid):
  542. try:
  543. records = get_evidence_service().list_for_job(job_uid)
  544. return jsonify(
  545. success({"records": records, "total": len(records)})
  546. ), 200
  547. except Exception as error:
  548. return _error(error)
  549. @bp.route("/ingestion-jobs/<job_uid>/retry", methods=["POST"])
  550. def retry_ingestion_job(job_uid):
  551. try:
  552. return jsonify(success(_record(get_ingestion_service().retry(job_uid)))), 200
  553. except Exception as error:
  554. return _error(error)
  555. @bp.route("/ingestion-jobs/<job_uid>/cancel", methods=["POST"])
  556. def cancel_ingestion_job(job_uid):
  557. try:
  558. service = get_ingestion_service()
  559. record = service.get_job(job_uid)
  560. identity = _identity()
  561. permissions = set(identity.get("permissions") or [])
  562. actor_uid = identity.get("id") or identity.get("sub")
  563. if record.actor_uid != actor_uid and "ingestion:admin" not in permissions:
  564. return jsonify(failed("权限不足", code=403)), 403
  565. return jsonify(success(_record(service.cancel(job_uid)))), 200
  566. except Exception as error:
  567. return _error(error)
  568. @bp.route("/data-elements", methods=["GET"])
  569. def list_data_elements():
  570. try:
  571. records = get_data_element_service().list(
  572. status=request.args.get("status"),
  573. business_domain_uid=request.args.get("business_domain_uid"),
  574. )
  575. return jsonify(
  576. success(
  577. {
  578. "records": [_element(item) for item in records],
  579. "total": len(records),
  580. }
  581. )
  582. ), 200
  583. except Exception as error:
  584. return _error(error)
  585. @bp.route("/device-assets", methods=["GET"])
  586. def list_device_assets():
  587. filters = {
  588. name: request.args.get(name)
  589. for name in ("keyword", "asset_type", "status", "source_uid")
  590. if request.args.get(name)
  591. }
  592. try:
  593. page = _device_asset_page(
  594. "page",
  595. default=1,
  596. maximum=1_000_000,
  597. )
  598. page_size = _device_asset_page(
  599. "page_size",
  600. default=20,
  601. maximum=100,
  602. )
  603. records, total = get_device_asset_service().search(
  604. filters,
  605. page=page,
  606. page_size=page_size,
  607. )
  608. return jsonify(
  609. success(
  610. {
  611. "records": [
  612. _device_asset_detail(record)
  613. for record in records
  614. ],
  615. "total": int(total),
  616. "page": page,
  617. "page_size": page_size,
  618. }
  619. )
  620. ), 200
  621. except Exception as error:
  622. return _error(error)
  623. @bp.route("/device-assets/import", methods=["POST"])
  624. def import_device_assets():
  625. try:
  626. result = get_device_asset_service().import_records(
  627. request.get_json(silent=True) or {},
  628. actor_uid=_identity().get("id") or _identity().get("sub"),
  629. )
  630. return jsonify(success(_device_asset_import_result(result))), 200
  631. except Exception as error:
  632. return _error(error)
  633. @bp.route("/device-assets/<asset_uid>", methods=["GET"])
  634. def get_device_asset(asset_uid):
  635. try:
  636. return jsonify(
  637. success(
  638. _device_asset_detail(
  639. get_device_asset_service().get(asset_uid)
  640. )
  641. )
  642. ), 200
  643. except Exception as error:
  644. return _error(error)
  645. @bp.route("/device-assets/<asset_uid>/versions", methods=["GET"])
  646. def list_device_asset_versions(asset_uid):
  647. try:
  648. records = get_device_asset_service().versions(asset_uid)
  649. return jsonify(
  650. success(
  651. {
  652. "records": [
  653. _device_asset_version(record)
  654. for record in records
  655. ],
  656. "total": len(records),
  657. }
  658. )
  659. ), 200
  660. except Exception as error:
  661. return _error(error)
  662. @bp.route("/device-semantics/bootstrap", methods=["POST"])
  663. def bootstrap_device_semantics():
  664. try:
  665. result = get_device_semantic_service().bootstrap(
  666. request.get_json(silent=True) or {},
  667. actor_uid=_identity().get("id") or _identity().get("sub"),
  668. )
  669. return (
  670. jsonify(success(_device_semantic_profile(result))),
  671. 201 if result.created else 200,
  672. )
  673. except Exception as error:
  674. return _error(error)
  675. @bp.route("/device-semantics/profile", methods=["GET"])
  676. def get_device_semantic_profile():
  677. try:
  678. result = get_device_semantic_service().profile()
  679. return jsonify(success(_device_semantic_profile(result))), 200
  680. except Exception as error:
  681. return _error(error)
  682. @bp.route("/device-semantics/codes", methods=["GET"])
  683. def list_device_semantic_codes():
  684. filters = {
  685. name: request.args.get(name)
  686. for name in ("ontology_uid", "code_type", "status", "keyword")
  687. if request.args.get(name)
  688. }
  689. try:
  690. records, total = get_device_semantic_code_service().search(
  691. filters,
  692. page=request.args.get("page", 1),
  693. page_size=request.args.get("page_size", 20),
  694. )
  695. return jsonify(
  696. success(
  697. {
  698. "records": [
  699. _device_semantic_code(record)
  700. for record in records
  701. ],
  702. "total": int(total),
  703. "page": int(request.args.get("page", 1)),
  704. "page_size": int(request.args.get("page_size", 20)),
  705. }
  706. )
  707. ), 200
  708. except Exception as error:
  709. return _error(error)
  710. @bp.route("/device-semantics/codes", methods=["POST"])
  711. def create_device_semantic_code():
  712. try:
  713. record = get_device_semantic_code_service().create(
  714. request.get_json(silent=True) or {},
  715. actor_uid=_identity().get("id") or _identity().get("sub"),
  716. )
  717. return jsonify(success(_device_semantic_code(record))), 201
  718. except Exception as error:
  719. return _error(error)
  720. @bp.route("/device-semantics/codes/<code_uid>", methods=["GET"])
  721. def get_device_semantic_code(code_uid):
  722. try:
  723. record = get_device_semantic_code_service().get(code_uid)
  724. return jsonify(success(_device_semantic_code(record))), 200
  725. except Exception as error:
  726. return _error(error)
  727. @bp.route(
  728. "/device-semantics/codes/<code_uid>/revisions",
  729. methods=["POST"],
  730. )
  731. def revise_device_semantic_code(code_uid):
  732. payload = request.get_json(silent=True) or {}
  733. try:
  734. record = get_device_semantic_code_service().revise(
  735. code_uid,
  736. payload,
  737. expected_version=payload.get("expected_version"),
  738. actor_uid=_identity().get("id") or _identity().get("sub"),
  739. )
  740. return jsonify(success(_device_semantic_code(record))), 200
  741. except Exception as error:
  742. return _error(error)
  743. @bp.route(
  744. "/device-semantics/codes/<code_uid>/submit",
  745. methods=["POST"],
  746. )
  747. def submit_device_semantic_code(code_uid):
  748. payload = request.get_json(silent=True) or {}
  749. try:
  750. record = get_device_semantic_code_service().submit(
  751. code_uid,
  752. expected_version=payload.get("expected_version"),
  753. actor_uid=_identity().get("id") or _identity().get("sub"),
  754. )
  755. return jsonify(success(_device_semantic_code(record))), 200
  756. except Exception as error:
  757. return _error(error)
  758. @bp.route(
  759. "/device-semantics/codes/<code_uid>/review",
  760. methods=["POST"],
  761. )
  762. def review_device_semantic_code(code_uid):
  763. try:
  764. record, review = get_device_semantic_code_service().review(
  765. code_uid,
  766. request.get_json(silent=True) or {},
  767. actor_uid=_identity().get("id") or _identity().get("sub"),
  768. )
  769. return jsonify(
  770. success(
  771. {
  772. "record": _device_semantic_code(record),
  773. "review": _device_semantic_review(review),
  774. }
  775. )
  776. ), 200
  777. except Exception as error:
  778. return _error(error)
  779. @bp.route(
  780. "/device-semantics/codes/<code_uid>/versions",
  781. methods=["GET"],
  782. )
  783. def list_device_semantic_code_versions(code_uid):
  784. try:
  785. records = get_device_semantic_code_service().versions(code_uid)
  786. return jsonify(
  787. success(
  788. {
  789. "records": [
  790. _device_semantic_version(record)
  791. for record in records
  792. ],
  793. "total": len(records),
  794. }
  795. )
  796. ), 200
  797. except Exception as error:
  798. return _error(error)
  799. @bp.route(
  800. "/device-semantics/codes/<code_uid>/reviews",
  801. methods=["GET"],
  802. )
  803. def list_device_semantic_code_reviews(code_uid):
  804. try:
  805. records = get_device_semantic_code_service().reviews(code_uid)
  806. return jsonify(
  807. success(
  808. {
  809. "records": [
  810. _device_semantic_review(record)
  811. for record in records
  812. ],
  813. "total": len(records),
  814. }
  815. )
  816. ), 200
  817. except Exception as error:
  818. return _error(error)
  819. @bp.route("/data-elements", methods=["POST"])
  820. def create_data_element():
  821. try:
  822. record = get_data_element_service().create_draft(
  823. request.get_json(silent=True) or {},
  824. actor_uid=_identity().get("id") or _identity().get("sub"),
  825. )
  826. db.session.commit()
  827. return jsonify(success(_element(record))), 201
  828. except Exception as error:
  829. db.session.rollback()
  830. return _error(error)
  831. @bp.route("/data-elements/<element_uid>/transitions", methods=["POST"])
  832. def transition_data_element(element_uid):
  833. payload = request.get_json(silent=True) or {}
  834. identity = _identity()
  835. target_status = str(payload.get("target_status") or "")
  836. if (
  837. target_status in {"published", "deprecated", "retired"}
  838. and "data-elements:publish" not in set(identity.get("permissions") or [])
  839. ):
  840. return jsonify(failed("权限不足", code=403)), 403
  841. try:
  842. record = get_data_element_service().transition(
  843. element_uid,
  844. target_status,
  845. expected_version=int(payload.get("expected_version")),
  846. actor_uid=identity.get("id") or identity.get("sub"),
  847. )
  848. db.session.commit()
  849. return jsonify(success(_element(record))), 200
  850. except Exception as error:
  851. db.session.rollback()
  852. return _error(error)
  853. @bp.route("/candidate-decisions", methods=["POST"])
  854. def decide_candidates():
  855. payload = request.get_json(silent=True) or {}
  856. try:
  857. records = get_candidate_decision_service().decide(
  858. payload.get("decisions") or [],
  859. actor_uid=_identity().get("id") or _identity().get("sub"),
  860. )
  861. return jsonify(success([_decision(record) for record in records])), 200
  862. except Exception as error:
  863. return _error(error)
  864. @bp.route("/sources/files", methods=["POST"])
  865. def upload_source_file():
  866. uploaded = request.files.get("file")
  867. if uploaded is None:
  868. return jsonify(failed("缺少上传文件", code=400)), 400
  869. try:
  870. record, created = get_artifact_service().store(
  871. request.form.get("source_uid"),
  872. uploaded.filename,
  873. uploaded.mimetype,
  874. uploaded.read(),
  875. request.form.get("parser_version") or "auto-v1",
  876. )
  877. db.session.commit()
  878. return jsonify(success(_artifact(record))), 201 if created else 200
  879. except Exception as error:
  880. db.session.rollback()
  881. return _error(error)
  882. @bp.route("/evidence/<evidence_uid>", methods=["GET"])
  883. def get_evidence(evidence_uid):
  884. identity = _identity()
  885. try:
  886. service = get_evidence_service()
  887. if request.args.get("download") in {"1", "true", "yes"}:
  888. if "evidence:download" not in set(identity.get("permissions") or []):
  889. return jsonify(failed("权限不足", code=403)), 403
  890. content, filename, media_type = service.download(evidence_uid)
  891. return send_file(
  892. io.BytesIO(content),
  893. mimetype=media_type,
  894. as_attachment=True,
  895. download_name=filename,
  896. )
  897. preview = service.preview(evidence_uid)
  898. from app.core.data_research.artifacts import redact_excerpt
  899. preview["excerpt"] = redact_excerpt(preview.get("excerpt"))
  900. return jsonify(success(preview)), 200
  901. except Exception as error:
  902. return _error(error)
  903. @bp.route("/ontologies", methods=["GET"])
  904. def list_ontologies():
  905. try:
  906. return jsonify(success([_ontology(item) for item in get_ontology_service().list()])), 200
  907. except Exception as error:
  908. return _error(error)
  909. @bp.route("/ontologies", methods=["POST"])
  910. def create_ontology():
  911. try:
  912. record = get_ontology_service().create(
  913. request.get_json(silent=True) or {},
  914. _identity().get("id") or _identity().get("sub"),
  915. )
  916. return jsonify(success(_ontology(record))), 201
  917. except Exception as error:
  918. return _error(error)
  919. @bp.route("/ontologies/<ontology_uid>", methods=["GET"])
  920. def get_ontology(ontology_uid):
  921. try:
  922. return jsonify(success(_ontology(get_ontology_service().get(ontology_uid)))), 200
  923. except Exception as error:
  924. return _error(error)
  925. @bp.route("/ontologies/<ontology_uid>/versions", methods=["GET"])
  926. def list_ontology_versions(ontology_uid):
  927. try:
  928. return jsonify(
  929. success(
  930. [
  931. _ontology_version(item)
  932. for item in get_ontology_service().list_versions(ontology_uid)
  933. ]
  934. )
  935. ), 200
  936. except Exception as error:
  937. return _error(error)
  938. @bp.route("/ontologies/<ontology_uid>/graph", methods=["GET"])
  939. def get_ontology_graph(ontology_uid):
  940. try:
  941. service = get_ontology_service()
  942. ontology = service.get(ontology_uid)
  943. version = service.latest_version(ontology_uid)
  944. data = (
  945. _ontology_version(version)
  946. if version is not None
  947. else {
  948. "uid": None,
  949. "ontology_uid": str(ontology_uid),
  950. "version": 0,
  951. "parent_version_uid": None,
  952. "status": "draft",
  953. "content_hash": None,
  954. "graph_document": {
  955. name: []
  956. for name in (
  957. "classes",
  958. "properties",
  959. "relations",
  960. "constraints",
  961. "domain_links",
  962. "element_mappings",
  963. )
  964. },
  965. "created_by": None,
  966. }
  967. )
  968. response = jsonify(success(data))
  969. response.headers["ETag"] = f'"{int(ontology.draft_revision)}"'
  970. return response, 200
  971. except Exception as error:
  972. return _error(error)
  973. @bp.route("/ontologies/<ontology_uid>/graph", methods=["PATCH"])
  974. def save_ontology_graph(ontology_uid):
  975. etag = str(request.headers.get("If-Match") or "").strip().strip('"')
  976. if not etag.isdigit():
  977. return jsonify(failed("缺少有效 If-Match 修订号", code=428)), 428
  978. try:
  979. version = get_ontology_service().save_draft(
  980. ontology_uid,
  981. request.get_json(silent=True) or {},
  982. int(etag),
  983. _identity().get("id") or _identity().get("sub"),
  984. )
  985. response = jsonify(success(_ontology_version(version)))
  986. response.headers["ETag"] = f'"{int(etag) + 1}"'
  987. return response, 200
  988. except Exception as error:
  989. return _error(error)
  990. @bp.route("/ontologies/<ontology_uid>/validate", methods=["POST"])
  991. def validate_ontology(ontology_uid):
  992. try:
  993. issues = get_ontology_service().validate(ontology_uid)
  994. data = [
  995. {"code": item.code, "message": item.message, "path": item.path}
  996. for item in issues
  997. ]
  998. return jsonify(success(data)), 200
  999. except Exception as error:
  1000. return _error(error)
  1001. @bp.route("/ontologies/<ontology_uid>/publish", methods=["POST"])
  1002. def publish_ontology(ontology_uid):
  1003. try:
  1004. version = get_ontology_service().publish(
  1005. ontology_uid,
  1006. request.headers.get("Idempotency-Key") or f"{ontology_uid}:publish",
  1007. _identity().get("id") or _identity().get("sub"),
  1008. )
  1009. return jsonify(success(_ontology_version(version))), 200
  1010. except Exception as error:
  1011. return _error(error)
  1012. @bp.route("/ontologies/<ontology_uid>/diff", methods=["GET"])
  1013. def diff_ontology(ontology_uid):
  1014. try:
  1015. return jsonify(success(get_ontology_service().diff(
  1016. ontology_uid, request.args.get("left"), request.args.get("right")
  1017. ))), 200
  1018. except Exception as error:
  1019. return _error(error)
  1020. @bp.route("/ontologies/<ontology_uid>/rollback", methods=["POST"])
  1021. def rollback_ontology(ontology_uid):
  1022. payload = request.get_json(silent=True) or {}
  1023. try:
  1024. version = get_ontology_service().rollback(
  1025. ontology_uid,
  1026. payload.get("target_version_uid"),
  1027. int(payload.get("expected_revision")),
  1028. _identity().get("id") or _identity().get("sub"),
  1029. )
  1030. return jsonify(success(_ontology_version(version))), 201
  1031. except Exception as error:
  1032. return _error(error)
  1033. @bp.route("/ontologies/<ontology_uid>/suggestions", methods=["POST"])
  1034. def generate_ontology_suggestions(ontology_uid):
  1035. try:
  1036. result = get_ontology_dynamic_service().generate(
  1037. ontology_uid,
  1038. request.get_json(silent=True) or {},
  1039. _identity().get("id") or _identity().get("sub"),
  1040. )
  1041. records = (
  1042. result.get("suggestions") or []
  1043. if isinstance(result, dict)
  1044. else result
  1045. )
  1046. suggestions = [
  1047. item
  1048. if isinstance(item, dict)
  1049. else {
  1050. "uid": item.uid,
  1051. "kind": item.kind,
  1052. "payload": item.payload,
  1053. "evidence_uids": list(item.evidence_uids),
  1054. "confidence": item.confidence,
  1055. "source": item.source,
  1056. "model_version": item.model_version,
  1057. "prompt_version": item.prompt_version,
  1058. }
  1059. for item in records
  1060. ]
  1061. data = {
  1062. "change_set_uid": (
  1063. result.get("change_set_uid")
  1064. if isinstance(result, dict)
  1065. else None
  1066. ),
  1067. "suggestions": suggestions,
  1068. }
  1069. return jsonify(success(data)), 201
  1070. except Exception as error:
  1071. return _error(error)
  1072. @bp.route(
  1073. "/ontologies/<ontology_uid>/change-sets/<change_set_uid>/decisions",
  1074. methods=["POST"],
  1075. )
  1076. def decide_ontology_change_set(ontology_uid, change_set_uid):
  1077. payload = request.get_json(silent=True) or {}
  1078. try:
  1079. result = get_ontology_dynamic_service().decide(
  1080. ontology_uid,
  1081. change_set_uid,
  1082. payload.get("decisions") or [],
  1083. _identity().get("id") or _identity().get("sub"),
  1084. )
  1085. return jsonify(success(result)), 200
  1086. except Exception as error:
  1087. return _error(error)
  1088. @bp.route("/ontologies/<ontology_uid>/export", methods=["GET"])
  1089. def export_ontology(ontology_uid):
  1090. try:
  1091. content, media_type, filename = get_ontology_exchange_service().export(
  1092. ontology_uid, request.args.get("format") or "json"
  1093. )
  1094. return send_file(
  1095. io.BytesIO(content),
  1096. mimetype=media_type,
  1097. as_attachment=True,
  1098. download_name=filename,
  1099. )
  1100. except Exception as error:
  1101. return _error(error)
  1102. @bp.route("/ontologies/import", methods=["POST"])
  1103. def import_ontology():
  1104. try:
  1105. uploaded = request.files.get("file")
  1106. content = uploaded.read() if uploaded is not None else request.get_data(cache=False)
  1107. result = get_ontology_exchange_service().import_document(
  1108. content,
  1109. request.args.get("format") or "json",
  1110. _identity().get("id") or _identity().get("sub"),
  1111. )
  1112. return jsonify(success(result)), 201
  1113. except Exception as error:
  1114. return _error(error)
  1115. @bp.route("/semantic/properties/<property_uid>", methods=["GET"])
  1116. def query_semantic_property(property_uid):
  1117. domain = str(request.args.get("business_domain_uid") or "").strip()
  1118. identity = _identity()
  1119. scoped = set(identity.get("business_domains") or [domain])
  1120. if "*" not in scoped and domain not in scoped:
  1121. return jsonify(failed("权限不足", code=403)), 403
  1122. try:
  1123. result = get_semantic_query_service().trace_property(
  1124. property_uid,
  1125. allowed_domains={domain},
  1126. limit=int(request.args.get("limit") or 20),
  1127. after_uid=request.args.get("after_uid"),
  1128. )
  1129. return jsonify(success(result)), 200
  1130. except Exception as error:
  1131. return _error(error)