routes.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474
  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_data_element_service():
  20. from app.core.data_research.data_elements import DataElementService
  21. from app.core.data_research.repository import SqlAlchemyDataElementRepository
  22. from app.core.events.outbox import enqueue_outbox
  23. return DataElementService(
  24. SqlAlchemyDataElementRepository(db.session),
  25. outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
  26. )
  27. def get_candidate_decision_service():
  28. from app.core.data_research.candidate_decisions import CandidateDecisionService
  29. from app.core.data_research.data_elements import DataElementService
  30. from app.core.data_research.repository import (
  31. SqlAlchemyCandidateDecisionRepository,
  32. SqlAlchemyDataElementRepository,
  33. )
  34. return CandidateDecisionService(
  35. SqlAlchemyCandidateDecisionRepository(db.session),
  36. data_elements=DataElementService(SqlAlchemyDataElementRepository(db.session)),
  37. )
  38. def _artifact_storage():
  39. from minio import Minio
  40. from app.core.data_research.artifacts import MinioArtifactStorage
  41. client = Minio(
  42. current_app.config["MINIO_HOST"],
  43. access_key=current_app.config["MINIO_USER"],
  44. secret_key=current_app.config["MINIO_PASSWORD"],
  45. secure=bool(current_app.config.get("MINIO_SECURE")),
  46. )
  47. return MinioArtifactStorage(client, current_app.config["MINIO_BUCKET"])
  48. def get_artifact_service():
  49. from app.core.common.identifiers import new_governance_uid
  50. from app.core.data_research.artifacts import (
  51. ArtifactService,
  52. SqlAlchemyArtifactRepository,
  53. )
  54. from app.core.data_research.file_policy import FilePolicy
  55. return ArtifactService(
  56. SqlAlchemyArtifactRepository(db.session),
  57. _artifact_storage(),
  58. FilePolicy(
  59. max_bytes=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_BYTES", 25 * 1024 * 1024)),
  60. max_pages=int(current_app.config.get("DATA_RESEARCH_FILE_MAX_PAGES", 500)),
  61. ),
  62. uid_factory=new_governance_uid,
  63. )
  64. def get_evidence_service():
  65. from app.core.data_research.artifacts import EvidenceService
  66. return EvidenceService(db.session, _artifact_storage())
  67. def get_ontology_service():
  68. from app.core.data_research.ontology.publication import (
  69. OntologyApplicationService,
  70. OntologyPublicationService,
  71. )
  72. from app.core.data_research.ontology.repository import SqlAlchemyOntologyRepository
  73. from app.core.events.outbox import enqueue_outbox
  74. repository = SqlAlchemyOntologyRepository(db.session)
  75. publication = OntologyPublicationService(
  76. repository,
  77. outbox_enqueue=lambda **event: enqueue_outbox(db.session, **event),
  78. commit=db.session.commit,
  79. rollback=db.session.rollback,
  80. )
  81. return OntologyApplicationService(
  82. repository,
  83. publication=publication,
  84. commit=db.session.commit,
  85. rollback=db.session.rollback,
  86. )
  87. def _identity():
  88. return getattr(g, "current_user", {}) or {}
  89. def _record(record):
  90. return {
  91. "uid": str(record.uid),
  92. "source_uid": str(record.source_uid),
  93. "artifact_uid": str(record.artifact_uid) if record.artifact_uid else None,
  94. "job_type": record.job_type,
  95. "parser_version": record.parser_version,
  96. "status": record.status,
  97. "parameters": dict(record.parameters or {}),
  98. "statistics": dict(record.statistics or {}),
  99. "last_error": record.last_error,
  100. "actor_uid": record.actor_uid,
  101. "created_at": record.created_at.isoformat() if record.created_at else None,
  102. "updated_at": record.updated_at.isoformat() if record.updated_at else None,
  103. "started_at": record.started_at.isoformat() if record.started_at else None,
  104. "finished_at": record.finished_at.isoformat() if record.finished_at else None,
  105. }
  106. def _element(record):
  107. return {
  108. "uid": str(record.uid),
  109. "code": record.code,
  110. "status": record.status,
  111. "current_version": int(record.current_version),
  112. "snapshot": dict(record.snapshot or {}),
  113. "created_by": record.created_by,
  114. "updated_by": record.updated_by,
  115. }
  116. def _decision(record):
  117. return {
  118. "uid": str(record.uid),
  119. "candidate_uid": str(record.candidate_uid),
  120. "action": record.action,
  121. "data_element_uid": (
  122. str(record.data_element_uid) if record.data_element_uid else None
  123. ),
  124. "evidence_uids": list(record.evidence_uids),
  125. "actor_uid": record.actor_uid,
  126. "reason": record.reason,
  127. }
  128. def _artifact(record):
  129. return {
  130. "uid": str(record.uid),
  131. "source_uid": str(record.source_uid),
  132. "filename": record.filename,
  133. "media_type": record.media_type,
  134. "size_bytes": int(record.size_bytes),
  135. "content_hash": record.content_hash,
  136. "parser_version": record.parser_version,
  137. }
  138. def _ontology(record):
  139. return {
  140. "uid": str(record.uid),
  141. "code": record.code,
  142. "name": record.name,
  143. "owner_uid": record.owner_uid,
  144. "status": record.status,
  145. "draft_revision": int(record.draft_revision),
  146. "active_version_uid": record.active_version_uid,
  147. "domain_links": [
  148. {"domain_uid": link.domain_uid, "role": link.role}
  149. for link in record.domain_links
  150. ],
  151. }
  152. def _ontology_version(record):
  153. return {
  154. "uid": str(record.uid),
  155. "ontology_uid": str(record.ontology_uid),
  156. "version": int(record.version),
  157. "parent_version_uid": record.parent_version_uid,
  158. "status": record.status,
  159. "content_hash": record.content_hash,
  160. "graph_document": record.graph_document.to_dict(),
  161. "created_by": record.created_by,
  162. }
  163. def _error(error):
  164. if isinstance(error, DataResearchError):
  165. return (
  166. jsonify(
  167. failed(
  168. str(error),
  169. code=error.http_status,
  170. error={"code": error.code},
  171. )
  172. ),
  173. error.http_status,
  174. )
  175. logger.exception("data-research ingestion request failed")
  176. return (
  177. jsonify(
  178. failed(
  179. "数据采集任务处理失败",
  180. code=500,
  181. error={"code": "DATA_RESEARCH_ERROR"},
  182. )
  183. ),
  184. 500,
  185. )
  186. @bp.route("/ingestion-jobs", methods=["POST"])
  187. def create_ingestion_job():
  188. payload = request.get_json(silent=True) or {}
  189. try:
  190. record, created = get_ingestion_service().create_job(
  191. payload,
  192. actor_uid=_identity().get("id") or _identity().get("sub"),
  193. )
  194. return jsonify(success(_record(record))), 201 if created else 200
  195. except Exception as error:
  196. return _error(error)
  197. @bp.route("/ingestion-jobs", methods=["GET"])
  198. def list_ingestion_jobs():
  199. filters = {
  200. name: request.args.get(name)
  201. for name in ("status", "source_uid")
  202. if request.args.get(name)
  203. }
  204. try:
  205. records = get_ingestion_service().list_jobs(filters)
  206. return jsonify(
  207. success({"records": [_record(item) for item in records], "total": len(records)})
  208. ), 200
  209. except Exception as error:
  210. return _error(error)
  211. @bp.route("/ingestion-jobs/<job_uid>", methods=["GET"])
  212. def get_ingestion_job(job_uid):
  213. try:
  214. return jsonify(success(_record(get_ingestion_service().get_job(job_uid)))), 200
  215. except Exception as error:
  216. return _error(error)
  217. @bp.route("/ingestion-jobs/<job_uid>/retry", methods=["POST"])
  218. def retry_ingestion_job(job_uid):
  219. try:
  220. return jsonify(success(_record(get_ingestion_service().retry(job_uid)))), 200
  221. except Exception as error:
  222. return _error(error)
  223. @bp.route("/ingestion-jobs/<job_uid>/cancel", methods=["POST"])
  224. def cancel_ingestion_job(job_uid):
  225. try:
  226. service = get_ingestion_service()
  227. record = service.get_job(job_uid)
  228. identity = _identity()
  229. permissions = set(identity.get("permissions") or [])
  230. actor_uid = identity.get("id") or identity.get("sub")
  231. if record.actor_uid != actor_uid and "ingestion:admin" not in permissions:
  232. return jsonify(failed("权限不足", code=403)), 403
  233. return jsonify(success(_record(service.cancel(job_uid)))), 200
  234. except Exception as error:
  235. return _error(error)
  236. @bp.route("/data-elements", methods=["POST"])
  237. def create_data_element():
  238. try:
  239. record = get_data_element_service().create_draft(
  240. request.get_json(silent=True) or {},
  241. actor_uid=_identity().get("id") or _identity().get("sub"),
  242. )
  243. db.session.commit()
  244. return jsonify(success(_element(record))), 201
  245. except Exception as error:
  246. db.session.rollback()
  247. return _error(error)
  248. @bp.route("/data-elements/<element_uid>/transitions", methods=["POST"])
  249. def transition_data_element(element_uid):
  250. payload = request.get_json(silent=True) or {}
  251. identity = _identity()
  252. target_status = str(payload.get("target_status") or "")
  253. if (
  254. target_status in {"published", "deprecated", "retired"}
  255. and "data-elements:publish" not in set(identity.get("permissions") or [])
  256. ):
  257. return jsonify(failed("权限不足", code=403)), 403
  258. try:
  259. record = get_data_element_service().transition(
  260. element_uid,
  261. target_status,
  262. expected_version=int(payload.get("expected_version")),
  263. actor_uid=identity.get("id") or identity.get("sub"),
  264. )
  265. db.session.commit()
  266. return jsonify(success(_element(record))), 200
  267. except Exception as error:
  268. db.session.rollback()
  269. return _error(error)
  270. @bp.route("/candidate-decisions", methods=["POST"])
  271. def decide_candidates():
  272. payload = request.get_json(silent=True) or {}
  273. try:
  274. records = get_candidate_decision_service().decide(
  275. payload.get("decisions") or [],
  276. actor_uid=_identity().get("id") or _identity().get("sub"),
  277. )
  278. return jsonify(success([_decision(record) for record in records])), 200
  279. except Exception as error:
  280. return _error(error)
  281. @bp.route("/sources/files", methods=["POST"])
  282. def upload_source_file():
  283. uploaded = request.files.get("file")
  284. if uploaded is None:
  285. return jsonify(failed("缺少上传文件", code=400)), 400
  286. try:
  287. record, created = get_artifact_service().store(
  288. request.form.get("source_uid"),
  289. uploaded.filename,
  290. uploaded.mimetype,
  291. uploaded.read(),
  292. request.form.get("parser_version") or "auto-v1",
  293. )
  294. db.session.commit()
  295. return jsonify(success(_artifact(record))), 201 if created else 200
  296. except Exception as error:
  297. db.session.rollback()
  298. return _error(error)
  299. @bp.route("/evidence/<evidence_uid>", methods=["GET"])
  300. def get_evidence(evidence_uid):
  301. identity = _identity()
  302. try:
  303. service = get_evidence_service()
  304. if request.args.get("download") in {"1", "true", "yes"}:
  305. if "evidence:download" not in set(identity.get("permissions") or []):
  306. return jsonify(failed("权限不足", code=403)), 403
  307. content, filename, media_type = service.download(evidence_uid)
  308. return send_file(
  309. io.BytesIO(content),
  310. mimetype=media_type,
  311. as_attachment=True,
  312. download_name=filename,
  313. )
  314. preview = service.preview(evidence_uid)
  315. from app.core.data_research.artifacts import redact_excerpt
  316. preview["excerpt"] = redact_excerpt(preview.get("excerpt"))
  317. return jsonify(success(preview)), 200
  318. except Exception as error:
  319. return _error(error)
  320. @bp.route("/ontologies", methods=["GET"])
  321. def list_ontologies():
  322. try:
  323. return jsonify(success([_ontology(item) for item in get_ontology_service().list()])), 200
  324. except Exception as error:
  325. return _error(error)
  326. @bp.route("/ontologies", methods=["POST"])
  327. def create_ontology():
  328. try:
  329. record = get_ontology_service().create(
  330. request.get_json(silent=True) or {},
  331. _identity().get("id") or _identity().get("sub"),
  332. )
  333. return jsonify(success(_ontology(record))), 201
  334. except Exception as error:
  335. return _error(error)
  336. @bp.route("/ontologies/<ontology_uid>/graph", methods=["PATCH"])
  337. def save_ontology_graph(ontology_uid):
  338. etag = str(request.headers.get("If-Match") or "").strip().strip('"')
  339. if not etag.isdigit():
  340. return jsonify(failed("缺少有效 If-Match 修订号", code=428)), 428
  341. try:
  342. version = get_ontology_service().save_draft(
  343. ontology_uid,
  344. request.get_json(silent=True) or {},
  345. int(etag),
  346. _identity().get("id") or _identity().get("sub"),
  347. )
  348. response = jsonify(success(_ontology_version(version)))
  349. response.headers["ETag"] = f'"{int(etag) + 1}"'
  350. return response, 200
  351. except Exception as error:
  352. return _error(error)
  353. @bp.route("/ontologies/<ontology_uid>/validate", methods=["POST"])
  354. def validate_ontology(ontology_uid):
  355. try:
  356. issues = get_ontology_service().validate(ontology_uid)
  357. data = [
  358. {"code": item.code, "message": item.message, "path": item.path}
  359. for item in issues
  360. ]
  361. return jsonify(success(data)), 422 if data else 200
  362. except Exception as error:
  363. return _error(error)
  364. @bp.route("/ontologies/<ontology_uid>/publish", methods=["POST"])
  365. def publish_ontology(ontology_uid):
  366. try:
  367. version = get_ontology_service().publish(
  368. ontology_uid,
  369. request.headers.get("Idempotency-Key") or f"{ontology_uid}:publish",
  370. _identity().get("id") or _identity().get("sub"),
  371. )
  372. return jsonify(success(_ontology_version(version))), 200
  373. except Exception as error:
  374. return _error(error)
  375. @bp.route("/ontologies/<ontology_uid>/diff", methods=["GET"])
  376. def diff_ontology(ontology_uid):
  377. try:
  378. return jsonify(success(get_ontology_service().diff(
  379. ontology_uid, request.args.get("left"), request.args.get("right")
  380. ))), 200
  381. except Exception as error:
  382. return _error(error)
  383. @bp.route("/ontologies/<ontology_uid>/rollback", methods=["POST"])
  384. def rollback_ontology(ontology_uid):
  385. payload = request.get_json(silent=True) or {}
  386. try:
  387. version = get_ontology_service().rollback(
  388. ontology_uid,
  389. payload.get("target_version_uid"),
  390. int(payload.get("expected_revision")),
  391. _identity().get("id") or _identity().get("sub"),
  392. )
  393. return jsonify(success(_ontology_version(version))), 201
  394. except Exception as error:
  395. return _error(error)