knowledge.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600
  1. from typing import List, Optional
  2. from pydantic import BaseModel
  3. from fastapi import APIRouter, Depends, HTTPException, status, Request
  4. import logging
  5. from open_webui.models.knowledge import (
  6. Knowledges,
  7. KnowledgeForm,
  8. KnowledgeResponse,
  9. KnowledgeUserResponse,
  10. )
  11. from open_webui.models.files import Files, FileModel
  12. from open_webui.retrieval.vector.connector import VECTOR_DB_CLIENT
  13. from open_webui.routers.retrieval import (
  14. process_file,
  15. ProcessFileForm,
  16. process_files_batch,
  17. BatchProcessFilesForm,
  18. )
  19. from open_webui.constants import ERROR_MESSAGES
  20. from open_webui.utils.auth import get_verified_user
  21. from open_webui.utils.access_control import has_access, has_permission
  22. from open_webui.env import SRC_LOG_LEVELS
  23. log = logging.getLogger(__name__)
  24. log.setLevel(SRC_LOG_LEVELS["MODELS"])
  25. router = APIRouter()
  26. ############################
  27. # getKnowledgeBases
  28. ############################
  29. @router.get("/", response_model=list[KnowledgeUserResponse])
  30. async def get_knowledge(user=Depends(get_verified_user)):
  31. knowledge_bases = []
  32. if user.role == "admin":
  33. knowledge_bases = Knowledges.get_knowledge_bases()
  34. else:
  35. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "read")
  36. # Get files for each knowledge base
  37. knowledge_with_files = []
  38. for knowledge_base in knowledge_bases:
  39. files = []
  40. if knowledge_base.data:
  41. files = Files.get_file_metadatas_by_ids(
  42. knowledge_base.data.get("file_ids", [])
  43. )
  44. # Check if all files exist
  45. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  46. missing_files = list(
  47. set(knowledge_base.data.get("file_ids", []))
  48. - set([file.id for file in files])
  49. )
  50. if missing_files:
  51. data = knowledge_base.data or {}
  52. file_ids = data.get("file_ids", [])
  53. for missing_file in missing_files:
  54. file_ids.remove(missing_file)
  55. data["file_ids"] = file_ids
  56. Knowledges.update_knowledge_data_by_id(
  57. id=knowledge_base.id, data=data
  58. )
  59. files = Files.get_file_metadatas_by_ids(file_ids)
  60. knowledge_with_files.append(
  61. KnowledgeUserResponse(
  62. **knowledge_base.model_dump(),
  63. files=files,
  64. )
  65. )
  66. return knowledge_with_files
  67. @router.get("/list", response_model=list[KnowledgeUserResponse])
  68. async def get_knowledge_list(user=Depends(get_verified_user)):
  69. knowledge_bases = []
  70. if user.role == "admin":
  71. knowledge_bases = Knowledges.get_knowledge_bases()
  72. else:
  73. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  74. # Get files for each knowledge base
  75. knowledge_with_files = []
  76. for knowledge_base in knowledge_bases:
  77. files = []
  78. if knowledge_base.data:
  79. files = Files.get_file_metadatas_by_ids(
  80. knowledge_base.data.get("file_ids", [])
  81. )
  82. # Check if all files exist
  83. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  84. missing_files = list(
  85. set(knowledge_base.data.get("file_ids", []))
  86. - set([file.id for file in files])
  87. )
  88. if missing_files:
  89. data = knowledge_base.data or {}
  90. file_ids = data.get("file_ids", [])
  91. for missing_file in missing_files:
  92. file_ids.remove(missing_file)
  93. data["file_ids"] = file_ids
  94. Knowledges.update_knowledge_data_by_id(
  95. id=knowledge_base.id, data=data
  96. )
  97. files = Files.get_file_metadatas_by_ids(file_ids)
  98. knowledge_with_files.append(
  99. KnowledgeUserResponse(
  100. **knowledge_base.model_dump(),
  101. files=files,
  102. )
  103. )
  104. return knowledge_with_files
  105. ############################
  106. # CreateNewKnowledge
  107. ############################
  108. @router.post("/create", response_model=Optional[KnowledgeResponse])
  109. async def create_new_knowledge(
  110. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  111. ):
  112. if user.role != "admin" and not has_permission(
  113. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  114. ):
  115. raise HTTPException(
  116. status_code=status.HTTP_401_UNAUTHORIZED,
  117. detail=ERROR_MESSAGES.UNAUTHORIZED,
  118. )
  119. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  120. if knowledge:
  121. return knowledge
  122. else:
  123. raise HTTPException(
  124. status_code=status.HTTP_400_BAD_REQUEST,
  125. detail=ERROR_MESSAGES.FILE_EXISTS,
  126. )
  127. ############################
  128. # GetKnowledgeById
  129. ############################
  130. class KnowledgeFilesResponse(KnowledgeResponse):
  131. files: list[FileModel]
  132. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  133. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  134. knowledge = Knowledges.get_knowledge_by_id(id=id)
  135. if knowledge:
  136. if (
  137. user.role == "admin"
  138. or knowledge.user_id == user.id
  139. or has_access(user.id, "read", knowledge.access_control)
  140. ):
  141. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  142. files = Files.get_files_by_ids(file_ids)
  143. return KnowledgeFilesResponse(
  144. **knowledge.model_dump(),
  145. files=files,
  146. )
  147. else:
  148. raise HTTPException(
  149. status_code=status.HTTP_401_UNAUTHORIZED,
  150. detail=ERROR_MESSAGES.NOT_FOUND,
  151. )
  152. ############################
  153. # UpdateKnowledgeById
  154. ############################
  155. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  156. async def update_knowledge_by_id(
  157. id: str,
  158. form_data: KnowledgeForm,
  159. user=Depends(get_verified_user),
  160. ):
  161. knowledge = Knowledges.get_knowledge_by_id(id=id)
  162. if not knowledge:
  163. raise HTTPException(
  164. status_code=status.HTTP_400_BAD_REQUEST,
  165. detail=ERROR_MESSAGES.NOT_FOUND,
  166. )
  167. if knowledge.user_id != user.id and user.role != "admin":
  168. raise HTTPException(
  169. status_code=status.HTTP_400_BAD_REQUEST,
  170. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  171. )
  172. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  173. if knowledge:
  174. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  175. files = Files.get_files_by_ids(file_ids)
  176. return KnowledgeFilesResponse(
  177. **knowledge.model_dump(),
  178. files=files,
  179. )
  180. else:
  181. raise HTTPException(
  182. status_code=status.HTTP_400_BAD_REQUEST,
  183. detail=ERROR_MESSAGES.ID_TAKEN,
  184. )
  185. ############################
  186. # AddFileToKnowledge
  187. ############################
  188. class KnowledgeFileIdForm(BaseModel):
  189. file_id: str
  190. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  191. def add_file_to_knowledge_by_id(
  192. request: Request,
  193. id: str,
  194. form_data: KnowledgeFileIdForm,
  195. user=Depends(get_verified_user),
  196. ):
  197. knowledge = Knowledges.get_knowledge_by_id(id=id)
  198. if not knowledge:
  199. raise HTTPException(
  200. status_code=status.HTTP_400_BAD_REQUEST,
  201. detail=ERROR_MESSAGES.NOT_FOUND,
  202. )
  203. if knowledge.user_id != user.id and user.role != "admin":
  204. raise HTTPException(
  205. status_code=status.HTTP_400_BAD_REQUEST,
  206. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  207. )
  208. file = Files.get_file_by_id(form_data.file_id)
  209. if not file:
  210. raise HTTPException(
  211. status_code=status.HTTP_400_BAD_REQUEST,
  212. detail=ERROR_MESSAGES.NOT_FOUND,
  213. )
  214. if not file.data:
  215. raise HTTPException(
  216. status_code=status.HTTP_400_BAD_REQUEST,
  217. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  218. )
  219. # Add content to the vector database
  220. try:
  221. process_file(
  222. request, ProcessFileForm(file_id=form_data.file_id, collection_name=id)
  223. )
  224. except Exception as e:
  225. log.debug(e)
  226. raise HTTPException(
  227. status_code=status.HTTP_400_BAD_REQUEST,
  228. detail=str(e),
  229. )
  230. if knowledge:
  231. data = knowledge.data or {}
  232. file_ids = data.get("file_ids", [])
  233. if form_data.file_id not in file_ids:
  234. file_ids.append(form_data.file_id)
  235. data["file_ids"] = file_ids
  236. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  237. if knowledge:
  238. files = Files.get_files_by_ids(file_ids)
  239. return KnowledgeFilesResponse(
  240. **knowledge.model_dump(),
  241. files=files,
  242. )
  243. else:
  244. raise HTTPException(
  245. status_code=status.HTTP_400_BAD_REQUEST,
  246. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  247. )
  248. else:
  249. raise HTTPException(
  250. status_code=status.HTTP_400_BAD_REQUEST,
  251. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  252. )
  253. else:
  254. raise HTTPException(
  255. status_code=status.HTTP_400_BAD_REQUEST,
  256. detail=ERROR_MESSAGES.NOT_FOUND,
  257. )
  258. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  259. def update_file_from_knowledge_by_id(
  260. request: Request,
  261. id: str,
  262. form_data: KnowledgeFileIdForm,
  263. user=Depends(get_verified_user),
  264. ):
  265. knowledge = Knowledges.get_knowledge_by_id(id=id)
  266. if not knowledge:
  267. raise HTTPException(
  268. status_code=status.HTTP_400_BAD_REQUEST,
  269. detail=ERROR_MESSAGES.NOT_FOUND,
  270. )
  271. if knowledge.user_id != user.id and user.role != "admin":
  272. raise HTTPException(
  273. status_code=status.HTTP_400_BAD_REQUEST,
  274. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  275. )
  276. file = Files.get_file_by_id(form_data.file_id)
  277. if not file:
  278. raise HTTPException(
  279. status_code=status.HTTP_400_BAD_REQUEST,
  280. detail=ERROR_MESSAGES.NOT_FOUND,
  281. )
  282. # Remove content from the vector database
  283. VECTOR_DB_CLIENT.delete(
  284. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  285. )
  286. # Add content to the vector database
  287. try:
  288. process_file(
  289. request, ProcessFileForm(file_id=form_data.file_id, collection_name=id)
  290. )
  291. except Exception as e:
  292. raise HTTPException(
  293. status_code=status.HTTP_400_BAD_REQUEST,
  294. detail=str(e),
  295. )
  296. if knowledge:
  297. data = knowledge.data or {}
  298. file_ids = data.get("file_ids", [])
  299. files = Files.get_files_by_ids(file_ids)
  300. return KnowledgeFilesResponse(
  301. **knowledge.model_dump(),
  302. files=files,
  303. )
  304. else:
  305. raise HTTPException(
  306. status_code=status.HTTP_400_BAD_REQUEST,
  307. detail=ERROR_MESSAGES.NOT_FOUND,
  308. )
  309. ############################
  310. # RemoveFileFromKnowledge
  311. ############################
  312. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  313. def remove_file_from_knowledge_by_id(
  314. id: str,
  315. form_data: KnowledgeFileIdForm,
  316. user=Depends(get_verified_user),
  317. ):
  318. knowledge = Knowledges.get_knowledge_by_id(id=id)
  319. if not knowledge:
  320. raise HTTPException(
  321. status_code=status.HTTP_400_BAD_REQUEST,
  322. detail=ERROR_MESSAGES.NOT_FOUND,
  323. )
  324. if knowledge.user_id != user.id and user.role != "admin":
  325. raise HTTPException(
  326. status_code=status.HTTP_400_BAD_REQUEST,
  327. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  328. )
  329. file = Files.get_file_by_id(form_data.file_id)
  330. if not file:
  331. raise HTTPException(
  332. status_code=status.HTTP_400_BAD_REQUEST,
  333. detail=ERROR_MESSAGES.NOT_FOUND,
  334. )
  335. # Remove content from the vector database
  336. VECTOR_DB_CLIENT.delete(
  337. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  338. )
  339. result = VECTOR_DB_CLIENT.query(
  340. collection_name=knowledge.id,
  341. filter={"file_id": form_data.file_id},
  342. )
  343. Files.delete_file_by_id(form_data.file_id)
  344. if knowledge:
  345. data = knowledge.data or {}
  346. file_ids = data.get("file_ids", [])
  347. if form_data.file_id in file_ids:
  348. file_ids.remove(form_data.file_id)
  349. data["file_ids"] = file_ids
  350. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  351. if knowledge:
  352. files = Files.get_files_by_ids(file_ids)
  353. return KnowledgeFilesResponse(
  354. **knowledge.model_dump(),
  355. files=files,
  356. )
  357. else:
  358. raise HTTPException(
  359. status_code=status.HTTP_400_BAD_REQUEST,
  360. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  361. )
  362. else:
  363. raise HTTPException(
  364. status_code=status.HTTP_400_BAD_REQUEST,
  365. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  366. )
  367. else:
  368. raise HTTPException(
  369. status_code=status.HTTP_400_BAD_REQUEST,
  370. detail=ERROR_MESSAGES.NOT_FOUND,
  371. )
  372. ############################
  373. # DeleteKnowledgeById
  374. ############################
  375. @router.delete("/{id}/delete", response_model=bool)
  376. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  377. knowledge = Knowledges.get_knowledge_by_id(id=id)
  378. if not knowledge:
  379. raise HTTPException(
  380. status_code=status.HTTP_400_BAD_REQUEST,
  381. detail=ERROR_MESSAGES.NOT_FOUND,
  382. )
  383. if knowledge.user_id != user.id and user.role != "admin":
  384. raise HTTPException(
  385. status_code=status.HTTP_400_BAD_REQUEST,
  386. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  387. )
  388. try:
  389. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  390. except Exception as e:
  391. log.debug(e)
  392. pass
  393. result = Knowledges.delete_knowledge_by_id(id=id)
  394. return result
  395. ############################
  396. # ResetKnowledgeById
  397. ############################
  398. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  399. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  400. knowledge = Knowledges.get_knowledge_by_id(id=id)
  401. if not knowledge:
  402. raise HTTPException(
  403. status_code=status.HTTP_400_BAD_REQUEST,
  404. detail=ERROR_MESSAGES.NOT_FOUND,
  405. )
  406. if knowledge.user_id != user.id and user.role != "admin":
  407. raise HTTPException(
  408. status_code=status.HTTP_400_BAD_REQUEST,
  409. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  410. )
  411. try:
  412. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  413. except Exception as e:
  414. log.debug(e)
  415. pass
  416. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  417. return knowledge
  418. ############################
  419. # AddFilesToKnowledge
  420. ############################
  421. @router.post("/{id}/files/batch/add", response_model=Optional[KnowledgeFilesResponse])
  422. def add_files_to_knowledge_batch(
  423. id: str,
  424. form_data: list[KnowledgeFileIdForm],
  425. user=Depends(get_verified_user),
  426. ):
  427. """
  428. Add multiple files to a knowledge base
  429. """
  430. knowledge = Knowledges.get_knowledge_by_id(id=id)
  431. if not knowledge:
  432. raise HTTPException(
  433. status_code=status.HTTP_400_BAD_REQUEST,
  434. detail=ERROR_MESSAGES.NOT_FOUND,
  435. )
  436. if knowledge.user_id != user.id and user.role != "admin":
  437. raise HTTPException(
  438. status_code=status.HTTP_400_BAD_REQUEST,
  439. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  440. )
  441. # Get files content
  442. print(f"files/batch/add - {len(form_data)} files")
  443. files: List[FileModel] = []
  444. for form in form_data:
  445. file = Files.get_file_by_id(form.file_id)
  446. if not file:
  447. raise HTTPException(
  448. status_code=status.HTTP_400_BAD_REQUEST,
  449. detail=f"File {form.file_id} not found",
  450. )
  451. files.append(file)
  452. # Process files
  453. try:
  454. result = process_files_batch(
  455. BatchProcessFilesForm(files=files, collection_name=id)
  456. )
  457. except Exception as e:
  458. log.error(
  459. f"add_files_to_knowledge_batch: Exception occurred: {e}", exc_info=True
  460. )
  461. raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e))
  462. # Add successful files to knowledge base
  463. data = knowledge.data or {}
  464. existing_file_ids = data.get("file_ids", [])
  465. # Only add files that were successfully processed
  466. successful_file_ids = [r.file_id for r in result.results if r.status == "completed"]
  467. for file_id in successful_file_ids:
  468. if file_id not in existing_file_ids:
  469. existing_file_ids.append(file_id)
  470. data["file_ids"] = existing_file_ids
  471. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  472. # If there were any errors, include them in the response
  473. if result.errors:
  474. error_details = [f"{err.file_id}: {err.error}" for err in result.errors]
  475. return KnowledgeFilesResponse(
  476. **knowledge.model_dump(),
  477. files=Files.get_files_by_ids(existing_file_ids),
  478. warnings={
  479. "message": "Some files failed to process",
  480. "errors": error_details,
  481. },
  482. )
  483. return KnowledgeFilesResponse(
  484. **knowledge.model_dump(), files=Files.get_files_by_ids(existing_file_ids)
  485. )