data_research.py 31 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962
  1. from __future__ import annotations
  2. from typing import Any
  3. from sqlalchemy.dialects.postgresql import JSONB, UUID
  4. from app import db
  5. from app.core.common.identifiers import new_governance_uid
  6. from app.core.common.timezone_utils import now_china, now_china_naive
  7. JOB_STATUSES = (
  8. "created",
  9. "queued",
  10. "extracting",
  11. "normalizing",
  12. "matching",
  13. "awaiting_review",
  14. "published",
  15. "partial",
  16. "failed",
  17. "cancelled",
  18. )
  19. def _iso(value: Any) -> str | None:
  20. return value.isoformat() if value is not None else None
  21. class IngestionSource(db.Model):
  22. __tablename__ = "ingestion_sources"
  23. __table_args__ = (
  24. db.CheckConstraint(
  25. "source_type IN ('database','file','ddl')",
  26. name="ck_ingestion_source_type",
  27. ),
  28. db.CheckConstraint(
  29. "status IN ('active','disabled')",
  30. name="ck_ingestion_source_status",
  31. ),
  32. {"schema": "public"},
  33. )
  34. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  35. source_type = db.Column(db.String(20), nullable=False)
  36. name = db.Column(db.String(300), nullable=False)
  37. config = db.Column(JSONB, nullable=False, default=dict)
  38. permission_scope = db.Column(JSONB, nullable=False, default=dict)
  39. status = db.Column(db.String(20), nullable=False, default="active")
  40. created_by = db.Column(db.String(100))
  41. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  42. updated_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  43. def to_dict(self) -> dict[str, Any]:
  44. return {
  45. "uid": str(self.uid),
  46. "source_type": self.source_type,
  47. "name": self.name,
  48. "permission_scope": dict(self.permission_scope or {}),
  49. "status": self.status,
  50. "created_by": self.created_by,
  51. "created_at": _iso(self.created_at),
  52. "updated_at": _iso(self.updated_at),
  53. }
  54. class SourceArtifact(db.Model):
  55. __tablename__ = "source_artifacts"
  56. __table_args__ = (
  57. db.CheckConstraint("size_bytes >= 0", name="ck_source_artifact_size"),
  58. db.UniqueConstraint(
  59. "source_uid",
  60. "content_hash",
  61. "parser_version",
  62. name="uq_source_artifact_parser_content",
  63. ),
  64. {"schema": "public"},
  65. )
  66. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  67. source_uid = db.Column(
  68. UUID(as_uuid=False),
  69. db.ForeignKey("public.ingestion_sources.uid"),
  70. nullable=False,
  71. )
  72. filename = db.Column(db.String(500), nullable=False)
  73. media_type = db.Column(db.String(200), nullable=False)
  74. size_bytes = db.Column(db.BigInteger, nullable=False)
  75. content_hash = db.Column(db.String(64), nullable=False)
  76. storage_ref = db.Column(db.Text, nullable=False)
  77. parser_version = db.Column(db.String(100), nullable=False)
  78. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  79. def to_dict(self) -> dict[str, Any]:
  80. return {
  81. "uid": str(self.uid),
  82. "source_uid": str(self.source_uid),
  83. "filename": self.filename,
  84. "media_type": self.media_type,
  85. "size_bytes": int(self.size_bytes),
  86. "content_hash": self.content_hash,
  87. "parser_version": self.parser_version,
  88. "created_at": _iso(self.created_at),
  89. }
  90. class IngestionJob(db.Model):
  91. __tablename__ = "ingestion_jobs"
  92. __table_args__ = (
  93. db.CheckConstraint(
  94. "status IN (" + ",".join(f"'{value}'" for value in JOB_STATUSES) + ")",
  95. name="ck_ingestion_job_status",
  96. ),
  97. db.UniqueConstraint(
  98. "idempotency_key",
  99. name="uq_ingestion_job_idempotency_key",
  100. ),
  101. db.CheckConstraint(
  102. "attempt_count >= 0",
  103. name="ck_ingestion_job_attempt_count",
  104. ),
  105. {"schema": "public"},
  106. )
  107. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  108. source_uid = db.Column(
  109. UUID(as_uuid=False),
  110. db.ForeignKey("public.ingestion_sources.uid"),
  111. nullable=False,
  112. )
  113. artifact_uid = db.Column(
  114. UUID(as_uuid=False),
  115. db.ForeignKey("public.source_artifacts.uid"),
  116. )
  117. job_type = db.Column(db.String(30), nullable=False)
  118. status = db.Column(db.String(30), nullable=False, default="created")
  119. idempotency_key = db.Column(db.String(64), nullable=False, unique=True)
  120. parser_version = db.Column(db.String(100), nullable=False)
  121. parameters = db.Column(JSONB, nullable=False, default=dict)
  122. statistics = db.Column(JSONB, nullable=False, default=dict)
  123. last_error = db.Column(db.String(1000))
  124. attempt_count = db.Column(db.Integer, nullable=False, default=0)
  125. failure_stage = db.Column(db.String(30))
  126. actor_uid = db.Column(db.String(100))
  127. force_rerun = db.Column(db.Boolean, nullable=False, default=False)
  128. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  129. updated_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  130. started_at = db.Column(db.DateTime)
  131. finished_at = db.Column(db.DateTime)
  132. def to_dict(self) -> dict[str, Any]:
  133. return {
  134. "uid": str(self.uid),
  135. "source_uid": str(self.source_uid),
  136. "artifact_uid": str(self.artifact_uid) if self.artifact_uid else None,
  137. "job_type": self.job_type,
  138. "status": self.status,
  139. "parser_version": self.parser_version,
  140. "parameters": dict(self.parameters or {}),
  141. "statistics": dict(self.statistics or {}),
  142. "last_error": self.last_error,
  143. "attempt_count": int(self.attempt_count or 0),
  144. "failure_stage": self.failure_stage,
  145. "actor_uid": self.actor_uid,
  146. "created_at": _iso(self.created_at),
  147. "updated_at": _iso(self.updated_at),
  148. "started_at": _iso(self.started_at),
  149. "finished_at": _iso(self.finished_at),
  150. }
  151. class CatalogSnapshot(db.Model):
  152. __tablename__ = "catalog_snapshots"
  153. __table_args__ = (
  154. db.CheckConstraint(
  155. "attempt > 0",
  156. name="ck_catalog_snapshot_attempt",
  157. ),
  158. db.UniqueConstraint(
  159. "job_uid",
  160. "attempt",
  161. name="uq_catalog_snapshot_job_attempt",
  162. ),
  163. {"schema": "public"},
  164. )
  165. uid = db.Column(
  166. UUID(as_uuid=False),
  167. primary_key=True,
  168. default=new_governance_uid,
  169. )
  170. job_uid = db.Column(
  171. UUID(as_uuid=False),
  172. db.ForeignKey("public.ingestion_jobs.uid", ondelete="CASCADE"),
  173. nullable=False,
  174. )
  175. source_uid = db.Column(
  176. UUID(as_uuid=False),
  177. db.ForeignKey("public.ingestion_sources.uid"),
  178. nullable=False,
  179. )
  180. attempt = db.Column(db.Integer, nullable=False)
  181. database_type = db.Column(db.String(20), nullable=False)
  182. content_hash = db.Column(db.String(64), nullable=False)
  183. snapshot = db.Column(JSONB, nullable=False)
  184. created_at = db.Column(
  185. db.DateTime,
  186. nullable=False,
  187. default=now_china_naive,
  188. )
  189. def to_dict(self) -> dict[str, Any]:
  190. return {
  191. "uid": str(self.uid),
  192. "job_uid": str(self.job_uid),
  193. "source_uid": str(self.source_uid),
  194. "attempt": int(self.attempt),
  195. "database_type": self.database_type,
  196. "content_hash": self.content_hash,
  197. "snapshot": dict(self.snapshot or {}),
  198. "created_at": _iso(self.created_at),
  199. }
  200. class DeviceAsset(db.Model):
  201. __tablename__ = "device_assets"
  202. __table_args__ = (
  203. db.CheckConstraint(
  204. "asset_type IN ("
  205. "'device','component','measurement_point',"
  206. "'alarm','maintenance_record'"
  207. ")",
  208. name="ck_device_asset_type",
  209. ),
  210. db.CheckConstraint(
  211. "status IN ('active','retired')",
  212. name="ck_device_asset_status",
  213. ),
  214. db.CheckConstraint(
  215. "current_version > 0",
  216. name="ck_device_asset_current_version",
  217. ),
  218. {"schema": "public"},
  219. )
  220. uid = db.Column(
  221. UUID(as_uuid=False),
  222. primary_key=True,
  223. default=new_governance_uid,
  224. )
  225. asset_type = db.Column(db.String(40), nullable=False)
  226. name = db.Column(db.String(300), nullable=False)
  227. status = db.Column(db.String(20), nullable=False, default="active")
  228. current_version = db.Column(db.Integer, nullable=False, default=1)
  229. content_hash = db.Column(db.String(64), nullable=False)
  230. location = db.Column(db.String(300))
  231. organization = db.Column(db.String(300))
  232. responsible_person = db.Column(db.String(300))
  233. attributes = db.Column(JSONB, nullable=False, default=dict)
  234. created_by = db.Column(db.String(100))
  235. updated_by = db.Column(db.String(100))
  236. created_at = db.Column(
  237. db.DateTime(timezone=True),
  238. nullable=False,
  239. default=now_china,
  240. )
  241. updated_at = db.Column(
  242. db.DateTime(timezone=True),
  243. nullable=False,
  244. default=now_china,
  245. )
  246. class DeviceAssetSourceMapping(db.Model):
  247. __tablename__ = "device_asset_source_mappings"
  248. __table_args__ = (
  249. db.CheckConstraint(
  250. "asset_type IN ("
  251. "'device','component','measurement_point',"
  252. "'alarm','maintenance_record'"
  253. ")",
  254. name="ck_device_asset_mapping_type",
  255. ),
  256. db.UniqueConstraint(
  257. "source_uid",
  258. "source_entity",
  259. "asset_type",
  260. "source_code",
  261. name="uq_device_asset_source_identity",
  262. ),
  263. {"schema": "public"},
  264. )
  265. uid = db.Column(
  266. UUID(as_uuid=False),
  267. primary_key=True,
  268. default=new_governance_uid,
  269. )
  270. asset_uid = db.Column(
  271. UUID(as_uuid=False),
  272. db.ForeignKey("public.device_assets.uid", ondelete="CASCADE"),
  273. nullable=False,
  274. )
  275. source_uid = db.Column(
  276. UUID(as_uuid=False),
  277. db.ForeignKey("public.ingestion_sources.uid"),
  278. nullable=False,
  279. )
  280. source_entity = db.Column(db.String(300), nullable=False)
  281. asset_type = db.Column(db.String(40), nullable=False)
  282. source_code = db.Column(db.String(300), nullable=False)
  283. source_updated_at = db.Column(db.DateTime(timezone=True))
  284. first_seen_at = db.Column(
  285. db.DateTime(timezone=True),
  286. nullable=False,
  287. default=now_china,
  288. )
  289. last_seen_at = db.Column(
  290. db.DateTime(timezone=True),
  291. nullable=False,
  292. default=now_china,
  293. )
  294. class DeviceAssetVersion(db.Model):
  295. __tablename__ = "device_asset_versions"
  296. __table_args__ = (
  297. db.CheckConstraint(
  298. "version > 0",
  299. name="ck_device_asset_version_number",
  300. ),
  301. db.UniqueConstraint(
  302. "asset_uid",
  303. "version",
  304. name="uq_device_asset_version",
  305. ),
  306. {"schema": "public"},
  307. )
  308. uid = db.Column(
  309. UUID(as_uuid=False),
  310. primary_key=True,
  311. default=new_governance_uid,
  312. )
  313. asset_uid = db.Column(
  314. UUID(as_uuid=False),
  315. db.ForeignKey("public.device_assets.uid", ondelete="CASCADE"),
  316. nullable=False,
  317. )
  318. version = db.Column(db.Integer, nullable=False)
  319. content_hash = db.Column(db.String(64), nullable=False)
  320. snapshot = db.Column(JSONB, nullable=False)
  321. source_mapping_uid = db.Column(
  322. UUID(as_uuid=False),
  323. db.ForeignKey(
  324. "public.device_asset_source_mappings.uid",
  325. ondelete="RESTRICT",
  326. ),
  327. nullable=False,
  328. )
  329. actor_uid = db.Column(db.String(100))
  330. created_at = db.Column(
  331. db.DateTime(timezone=True),
  332. nullable=False,
  333. default=now_china,
  334. )
  335. class DeviceSemanticCode(db.Model):
  336. __tablename__ = "device_semantic_codes"
  337. __table_args__ = (
  338. db.CheckConstraint(
  339. "code_type IN ('fault','cause','action')",
  340. name="ck_device_semantic_code_type",
  341. ),
  342. db.CheckConstraint(
  343. "status IN ("
  344. "'draft','in_review','published','rejected','retired'"
  345. ")",
  346. name="ck_device_semantic_code_status",
  347. ),
  348. db.CheckConstraint(
  349. "current_version > 0",
  350. name="ck_device_semantic_code_current_version",
  351. ),
  352. db.CheckConstraint(
  353. "suggestion_source IN ('manual','rule','ai')",
  354. name="ck_device_semantic_suggestion_source",
  355. ),
  356. db.CheckConstraint(
  357. "confidence IS NULL OR (confidence >= 0 AND confidence <= 1)",
  358. name="ck_device_semantic_confidence",
  359. ),
  360. db.UniqueConstraint(
  361. "ontology_uid",
  362. "code_type",
  363. "canonical_code",
  364. name="uq_device_semantic_identity",
  365. ),
  366. {"schema": "public"},
  367. )
  368. uid = db.Column(
  369. UUID(as_uuid=False),
  370. primary_key=True,
  371. default=new_governance_uid,
  372. )
  373. ontology_uid = db.Column(
  374. UUID(as_uuid=False),
  375. db.ForeignKey("public.ontologies.uid", ondelete="RESTRICT"),
  376. nullable=False,
  377. )
  378. code_type = db.Column(db.String(20), nullable=False)
  379. canonical_code = db.Column(db.String(120), nullable=False)
  380. canonical_name = db.Column(db.String(300), nullable=False)
  381. definition = db.Column(db.Text)
  382. status = db.Column(db.String(20), nullable=False, default="draft")
  383. current_version = db.Column(db.Integer, nullable=False, default=1)
  384. source_mappings = db.Column(JSONB, nullable=False, default=list)
  385. evidence_uids = db.Column(JSONB, nullable=False, default=list)
  386. suggestion_source = db.Column(
  387. db.String(20),
  388. nullable=False,
  389. default="manual",
  390. )
  391. confidence = db.Column(db.Float)
  392. created_by = db.Column(db.String(100))
  393. updated_by = db.Column(db.String(100))
  394. created_at = db.Column(
  395. db.DateTime(timezone=True),
  396. nullable=False,
  397. default=now_china,
  398. )
  399. updated_at = db.Column(
  400. db.DateTime(timezone=True),
  401. nullable=False,
  402. default=now_china,
  403. )
  404. class DeviceSemanticCodeVersion(db.Model):
  405. __tablename__ = "device_semantic_code_versions"
  406. __table_args__ = (
  407. db.CheckConstraint(
  408. "version > 0",
  409. name="ck_device_semantic_code_version_number",
  410. ),
  411. db.UniqueConstraint(
  412. "code_uid",
  413. "version",
  414. name="uq_device_semantic_code_version",
  415. ),
  416. {"schema": "public"},
  417. )
  418. uid = db.Column(
  419. UUID(as_uuid=False),
  420. primary_key=True,
  421. default=new_governance_uid,
  422. )
  423. code_uid = db.Column(
  424. UUID(as_uuid=False),
  425. db.ForeignKey(
  426. "public.device_semantic_codes.uid",
  427. ondelete="CASCADE",
  428. ),
  429. nullable=False,
  430. )
  431. version = db.Column(db.Integer, nullable=False)
  432. snapshot = db.Column(JSONB, nullable=False)
  433. created_by = db.Column(db.String(100))
  434. created_at = db.Column(
  435. db.DateTime(timezone=True),
  436. nullable=False,
  437. default=now_china,
  438. )
  439. class DeviceSemanticCodeReview(db.Model):
  440. __tablename__ = "device_semantic_code_reviews"
  441. __table_args__ = (
  442. db.CheckConstraint(
  443. "version > 0",
  444. name="ck_device_semantic_review_version",
  445. ),
  446. db.CheckConstraint(
  447. "decision IN ('approve','reject')",
  448. name="ck_device_semantic_review_decision",
  449. ),
  450. {"schema": "public"},
  451. )
  452. uid = db.Column(
  453. UUID(as_uuid=False),
  454. primary_key=True,
  455. default=new_governance_uid,
  456. )
  457. code_uid = db.Column(
  458. UUID(as_uuid=False),
  459. db.ForeignKey(
  460. "public.device_semantic_codes.uid",
  461. ondelete="CASCADE",
  462. ),
  463. nullable=False,
  464. )
  465. version = db.Column(db.Integer, nullable=False)
  466. decision = db.Column(db.String(20), nullable=False)
  467. reason = db.Column(db.String(1000), nullable=False, default="")
  468. actor_uid = db.Column(db.String(100), nullable=False)
  469. created_at = db.Column(
  470. db.DateTime(timezone=True),
  471. nullable=False,
  472. default=now_china,
  473. )
  474. class DeviceEntityMatchCandidate(db.Model):
  475. __tablename__ = "device_entity_match_candidates"
  476. __table_args__ = (
  477. db.CheckConstraint(
  478. "left_asset_uid <> right_asset_uid",
  479. name="ck_device_entity_candidate_distinct_assets",
  480. ),
  481. db.CheckConstraint(
  482. "status IN ('pending','merged','rejected','rolled_back')",
  483. name="ck_device_entity_candidate_status",
  484. ),
  485. db.CheckConstraint(
  486. "suggestion_source IN ('rule','ai','manual')",
  487. name="ck_device_entity_candidate_source",
  488. ),
  489. db.CheckConstraint(
  490. "confidence >= 0 AND confidence <= 1",
  491. name="ck_device_entity_candidate_confidence",
  492. ),
  493. db.CheckConstraint(
  494. "current_version > 0",
  495. name="ck_device_entity_candidate_version",
  496. ),
  497. db.Index(
  498. "uq_device_entity_open_pair",
  499. "left_asset_uid",
  500. "right_asset_uid",
  501. unique=True,
  502. postgresql_where=db.text(
  503. "status IN ('pending','merged')"
  504. ),
  505. ),
  506. {"schema": "public"},
  507. )
  508. uid = db.Column(
  509. UUID(as_uuid=False),
  510. primary_key=True,
  511. default=new_governance_uid,
  512. )
  513. left_asset_uid = db.Column(
  514. UUID(as_uuid=False),
  515. db.ForeignKey("public.device_assets.uid", ondelete="RESTRICT"),
  516. nullable=False,
  517. )
  518. right_asset_uid = db.Column(
  519. UUID(as_uuid=False),
  520. db.ForeignKey("public.device_assets.uid", ondelete="RESTRICT"),
  521. nullable=False,
  522. )
  523. canonical_asset_uid = db.Column(
  524. UUID(as_uuid=False),
  525. db.ForeignKey("public.device_assets.uid", ondelete="RESTRICT"),
  526. )
  527. status = db.Column(db.String(20), nullable=False, default="pending")
  528. suggestion_source = db.Column(db.String(20), nullable=False)
  529. confidence = db.Column(db.Float, nullable=False)
  530. explanation = db.Column(JSONB, nullable=False, default=list)
  531. evidence_uids = db.Column(JSONB, nullable=False, default=list)
  532. model_provider = db.Column(db.String(200))
  533. model_name = db.Column(db.String(200))
  534. current_version = db.Column(db.Integer, nullable=False, default=1)
  535. created_by = db.Column(db.String(100))
  536. reviewed_by = db.Column(db.String(100))
  537. created_at = db.Column(
  538. db.DateTime(timezone=True),
  539. nullable=False,
  540. default=now_china,
  541. )
  542. updated_at = db.Column(
  543. db.DateTime(timezone=True),
  544. nullable=False,
  545. default=now_china,
  546. )
  547. class DeviceEntityMatchReview(db.Model):
  548. __tablename__ = "device_entity_match_reviews"
  549. __table_args__ = (
  550. db.CheckConstraint(
  551. "version > 0",
  552. name="ck_device_entity_review_version",
  553. ),
  554. db.CheckConstraint(
  555. "decision IN ('approve','reject','auto_approve')",
  556. name="ck_device_entity_review_decision",
  557. ),
  558. {"schema": "public"},
  559. )
  560. uid = db.Column(
  561. UUID(as_uuid=False),
  562. primary_key=True,
  563. default=new_governance_uid,
  564. )
  565. candidate_uid = db.Column(
  566. UUID(as_uuid=False),
  567. db.ForeignKey(
  568. "public.device_entity_match_candidates.uid",
  569. ondelete="CASCADE",
  570. ),
  571. nullable=False,
  572. )
  573. version = db.Column(db.Integer, nullable=False)
  574. decision = db.Column(db.String(20), nullable=False)
  575. reason = db.Column(db.String(1000), nullable=False)
  576. actor_uid = db.Column(db.String(100), nullable=False)
  577. created_at = db.Column(
  578. db.DateTime(timezone=True),
  579. nullable=False,
  580. default=now_china,
  581. )
  582. class DeviceEntityMergeEvent(db.Model):
  583. __tablename__ = "device_entity_merge_events"
  584. __table_args__ = (
  585. db.UniqueConstraint(
  586. "candidate_uid",
  587. name="uq_device_entity_merge_candidate",
  588. ),
  589. {"schema": "public"},
  590. )
  591. uid = db.Column(
  592. UUID(as_uuid=False),
  593. primary_key=True,
  594. default=new_governance_uid,
  595. )
  596. candidate_uid = db.Column(
  597. UUID(as_uuid=False),
  598. db.ForeignKey(
  599. "public.device_entity_match_candidates.uid",
  600. ondelete="CASCADE",
  601. ),
  602. nullable=False,
  603. )
  604. canonical_asset_uid = db.Column(
  605. UUID(as_uuid=False),
  606. db.ForeignKey("public.device_assets.uid", ondelete="RESTRICT"),
  607. nullable=False,
  608. )
  609. member_asset_uid = db.Column(
  610. UUID(as_uuid=False),
  611. db.ForeignKey("public.device_assets.uid", ondelete="RESTRICT"),
  612. nullable=False,
  613. )
  614. review_uid = db.Column(
  615. UUID(as_uuid=False),
  616. db.ForeignKey(
  617. "public.device_entity_match_reviews.uid",
  618. ondelete="RESTRICT",
  619. ),
  620. nullable=False,
  621. )
  622. snapshot = db.Column(JSONB, nullable=False)
  623. actor_uid = db.Column(db.String(100), nullable=False)
  624. created_at = db.Column(
  625. db.DateTime(timezone=True),
  626. nullable=False,
  627. default=now_china,
  628. )
  629. class DeviceEntityMergeRollback(db.Model):
  630. __tablename__ = "device_entity_merge_rollbacks"
  631. __table_args__ = (
  632. db.UniqueConstraint(
  633. "merge_uid",
  634. name="uq_device_entity_merge_rollback",
  635. ),
  636. {"schema": "public"},
  637. )
  638. uid = db.Column(
  639. UUID(as_uuid=False),
  640. primary_key=True,
  641. default=new_governance_uid,
  642. )
  643. merge_uid = db.Column(
  644. UUID(as_uuid=False),
  645. db.ForeignKey(
  646. "public.device_entity_merge_events.uid",
  647. ondelete="CASCADE",
  648. ),
  649. nullable=False,
  650. )
  651. candidate_uid = db.Column(
  652. UUID(as_uuid=False),
  653. db.ForeignKey(
  654. "public.device_entity_match_candidates.uid",
  655. ondelete="CASCADE",
  656. ),
  657. nullable=False,
  658. )
  659. reason = db.Column(db.String(1000), nullable=False)
  660. snapshot = db.Column(JSONB, nullable=False)
  661. actor_uid = db.Column(db.String(100), nullable=False)
  662. created_at = db.Column(
  663. db.DateTime(timezone=True),
  664. nullable=False,
  665. default=now_china,
  666. )
  667. class EvidenceFragment(db.Model):
  668. __tablename__ = "evidence_fragments"
  669. __table_args__ = (
  670. db.CheckConstraint(
  671. "confidence IS NULL OR (confidence >= 0 AND confidence <= 1)",
  672. name="ck_evidence_confidence",
  673. ),
  674. {"schema": "public"},
  675. )
  676. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  677. job_uid = db.Column(
  678. UUID(as_uuid=False),
  679. db.ForeignKey("public.ingestion_jobs.uid", ondelete="CASCADE"),
  680. nullable=False,
  681. )
  682. artifact_uid = db.Column(
  683. UUID(as_uuid=False),
  684. db.ForeignKey("public.source_artifacts.uid"),
  685. )
  686. locator = db.Column(JSONB, nullable=False)
  687. excerpt = db.Column(db.Text)
  688. confidence = db.Column(db.Float)
  689. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  690. def to_dict(self) -> dict[str, Any]:
  691. return {
  692. "uid": str(self.uid),
  693. "job_uid": str(self.job_uid),
  694. "artifact_uid": str(self.artifact_uid) if self.artifact_uid else None,
  695. "locator": dict(self.locator or {}),
  696. "excerpt": self.excerpt,
  697. "confidence": self.confidence,
  698. "created_at": _iso(self.created_at),
  699. }
  700. class ExtractionCandidate(db.Model):
  701. __tablename__ = "extraction_candidates"
  702. __table_args__ = (
  703. db.CheckConstraint(
  704. "confidence >= 0 AND confidence <= 1",
  705. name="ck_candidate_confidence",
  706. ),
  707. db.CheckConstraint(
  708. "status IN ('candidate','accepted','rejected','ignored')",
  709. name="ck_candidate_status",
  710. ),
  711. {"schema": "public"},
  712. )
  713. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  714. job_uid = db.Column(
  715. UUID(as_uuid=False),
  716. db.ForeignKey("public.ingestion_jobs.uid", ondelete="CASCADE"),
  717. nullable=False,
  718. )
  719. candidate_type = db.Column(db.String(40), nullable=False)
  720. normalized_data = db.Column(JSONB, nullable=False)
  721. evidence_uids = db.Column(JSONB, nullable=False, default=list)
  722. confidence = db.Column(db.Float, nullable=False)
  723. status = db.Column(db.String(20), nullable=False, default="candidate")
  724. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  725. updated_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  726. def to_dict(self) -> dict[str, Any]:
  727. return {
  728. "uid": str(self.uid),
  729. "job_uid": str(self.job_uid),
  730. "candidate_type": self.candidate_type,
  731. "normalized_data": dict(self.normalized_data or {}),
  732. "evidence_uids": list(self.evidence_uids or []),
  733. "confidence": float(self.confidence),
  734. "status": self.status,
  735. "created_at": _iso(self.created_at),
  736. "updated_at": _iso(self.updated_at),
  737. }
  738. class DataElement(db.Model):
  739. __tablename__ = "data_elements"
  740. __table_args__ = (
  741. db.CheckConstraint(
  742. "status IN ('draft','in_review','published','deprecated','retired')",
  743. name="ck_data_element_status",
  744. ),
  745. {"schema": "public"},
  746. )
  747. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  748. code = db.Column(db.String(120), nullable=False, unique=True)
  749. current_version = db.Column(db.Integer, nullable=False, default=1)
  750. status = db.Column(db.String(20), nullable=False, default="draft")
  751. business_domain_uids = db.Column(JSONB, nullable=False, default=list)
  752. created_by = db.Column(db.String(100))
  753. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  754. updated_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  755. versions = db.relationship(
  756. "DataElementVersion",
  757. lazy="selectin",
  758. order_by="DataElementVersion.version",
  759. cascade="all, delete-orphan",
  760. )
  761. class DataElementVersion(db.Model):
  762. __tablename__ = "data_element_versions"
  763. __table_args__ = (
  764. db.UniqueConstraint(
  765. "data_element_uid",
  766. "version",
  767. name="uq_data_element_version",
  768. ),
  769. {"schema": "public"},
  770. )
  771. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  772. data_element_uid = db.Column(
  773. UUID(as_uuid=False),
  774. db.ForeignKey("public.data_elements.uid"),
  775. nullable=False,
  776. )
  777. version = db.Column(db.Integer, nullable=False)
  778. status = db.Column(db.String(20), nullable=False)
  779. snapshot = db.Column(JSONB, nullable=False)
  780. evidence_uids = db.Column(JSONB, nullable=False, default=list)
  781. created_by = db.Column(db.String(100))
  782. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  783. class CandidateDecisionRecord(db.Model):
  784. __tablename__ = "candidate_decisions"
  785. __table_args__ = (
  786. db.CheckConstraint(
  787. "action IN ('reuse','create','map','ignore')",
  788. name="ck_candidate_decision_action",
  789. ),
  790. db.UniqueConstraint("candidate_uid", name="uq_candidate_decision"),
  791. {"schema": "public"},
  792. )
  793. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  794. candidate_uid = db.Column(
  795. UUID(as_uuid=False),
  796. db.ForeignKey("public.extraction_candidates.uid"),
  797. nullable=False,
  798. )
  799. action = db.Column(db.String(20), nullable=False)
  800. data_element_uid = db.Column(
  801. UUID(as_uuid=False),
  802. db.ForeignKey("public.data_elements.uid"),
  803. )
  804. evidence_uids = db.Column(JSONB, nullable=False, default=list)
  805. payload = db.Column(JSONB, nullable=False, default=dict)
  806. actor_uid = db.Column(db.String(100))
  807. reason = db.Column(db.String(1000))
  808. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  809. class OntologyModel(db.Model):
  810. __tablename__ = "ontologies"
  811. __table_args__ = ({"schema": "public"},)
  812. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  813. code = db.Column(db.String(120), nullable=False, unique=True)
  814. name = db.Column(db.String(300), nullable=False)
  815. owner_uid = db.Column(db.String(100), nullable=False)
  816. status = db.Column(db.String(20), nullable=False, default="draft")
  817. draft_revision = db.Column(db.Integer, nullable=False, default=0)
  818. active_version_uid = db.Column(UUID(as_uuid=False))
  819. created_by = db.Column(db.String(100))
  820. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  821. updated_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  822. class OntologyVersionModel(db.Model):
  823. __tablename__ = "ontology_versions"
  824. __table_args__ = (
  825. db.UniqueConstraint("ontology_uid", "version", name="uq_ontology_version"),
  826. {"schema": "public"},
  827. )
  828. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  829. ontology_uid = db.Column(
  830. UUID(as_uuid=False), db.ForeignKey("public.ontologies.uid"), nullable=False
  831. )
  832. version = db.Column(db.Integer, nullable=False)
  833. parent_version_uid = db.Column(
  834. UUID(as_uuid=False), db.ForeignKey("public.ontology_versions.uid")
  835. )
  836. status = db.Column(db.String(20), nullable=False, default="draft")
  837. graph_document = db.Column(JSONB, nullable=False)
  838. content_hash = db.Column(db.String(64), nullable=False)
  839. created_by = db.Column(db.String(100))
  840. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  841. published_at = db.Column(db.DateTime)
  842. class OntologyDomainLinkModel(db.Model):
  843. __tablename__ = "ontology_domain_links"
  844. __table_args__ = ({"schema": "public"},)
  845. ontology_uid = db.Column(
  846. UUID(as_uuid=False), db.ForeignKey("public.ontologies.uid"), primary_key=True
  847. )
  848. domain_uid = db.Column(UUID(as_uuid=False), primary_key=True)
  849. role = db.Column(db.String(20), nullable=False)
  850. class OntologyChangeSetModel(db.Model):
  851. __tablename__ = "ontology_change_sets"
  852. __table_args__ = ({"schema": "public"},)
  853. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  854. ontology_uid = db.Column(
  855. UUID(as_uuid=False), db.ForeignKey("public.ontologies.uid"), nullable=False
  856. )
  857. base_version_uid = db.Column(UUID(as_uuid=False))
  858. status = db.Column(db.String(20), nullable=False, default="draft")
  859. changes = db.Column(JSONB, nullable=False, default=list)
  860. decisions = db.Column(JSONB, nullable=False, default=list)
  861. created_by = db.Column(db.String(100))
  862. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  863. class OntologyPublishRunModel(db.Model):
  864. __tablename__ = "ontology_publish_runs"
  865. __table_args__ = ({"schema": "public"},)
  866. uid = db.Column(UUID(as_uuid=False), primary_key=True, default=new_governance_uid)
  867. ontology_uid = db.Column(
  868. UUID(as_uuid=False), db.ForeignKey("public.ontologies.uid"), nullable=False
  869. )
  870. version_uid = db.Column(
  871. UUID(as_uuid=False), db.ForeignKey("public.ontology_versions.uid"), nullable=False
  872. )
  873. idempotency_key = db.Column(db.String(128), nullable=False, unique=True)
  874. status = db.Column(db.String(20), nullable=False)
  875. validation_result = db.Column(JSONB, nullable=False, default=dict)
  876. actor_uid = db.Column(db.String(100))
  877. created_at = db.Column(db.DateTime, nullable=False, default=now_china_naive)
  878. finished_at = db.Column(db.DateTime)