knowledge.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591
  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.apps.webui.models.knowledge import (
  6. Knowledges,
  7. KnowledgeForm,
  8. KnowledgeResponse,
  9. KnowledgeUserResponse,
  10. )
  11. from open_webui.apps.webui.models.files import Files, FileModel
  12. from open_webui.apps.retrieval.vector.connector import VECTOR_DB_CLIENT
  13. from open_webui.apps.retrieval.main import BatchProcessFilesForm, process_file, ProcessFileForm, process_files_batch
  14. from open_webui.constants import ERROR_MESSAGES
  15. from open_webui.utils.auth import get_verified_user
  16. from open_webui.utils.access_control import has_access, has_permission
  17. from open_webui.env import SRC_LOG_LEVELS
  18. log = logging.getLogger(__name__)
  19. log.setLevel(SRC_LOG_LEVELS["MODELS"])
  20. router = APIRouter()
  21. ############################
  22. # getKnowledgeBases
  23. ############################
  24. @router.get("/", response_model=list[KnowledgeUserResponse])
  25. async def get_knowledge(user=Depends(get_verified_user)):
  26. knowledge_bases = []
  27. if user.role == "admin":
  28. knowledge_bases = Knowledges.get_knowledge_bases()
  29. else:
  30. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "read")
  31. # Get files for each knowledge base
  32. knowledge_with_files = []
  33. for knowledge_base in knowledge_bases:
  34. files = []
  35. if knowledge_base.data:
  36. files = Files.get_file_metadatas_by_ids(
  37. knowledge_base.data.get("file_ids", [])
  38. )
  39. # Check if all files exist
  40. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  41. missing_files = list(
  42. set(knowledge_base.data.get("file_ids", []))
  43. - set([file.id for file in files])
  44. )
  45. if missing_files:
  46. data = knowledge_base.data or {}
  47. file_ids = data.get("file_ids", [])
  48. for missing_file in missing_files:
  49. file_ids.remove(missing_file)
  50. data["file_ids"] = file_ids
  51. Knowledges.update_knowledge_data_by_id(
  52. id=knowledge_base.id, data=data
  53. )
  54. files = Files.get_file_metadatas_by_ids(file_ids)
  55. knowledge_with_files.append(
  56. KnowledgeUserResponse(
  57. **knowledge_base.model_dump(),
  58. files=files,
  59. )
  60. )
  61. return knowledge_with_files
  62. @router.get("/list", response_model=list[KnowledgeUserResponse])
  63. async def get_knowledge_list(user=Depends(get_verified_user)):
  64. knowledge_bases = []
  65. if user.role == "admin":
  66. knowledge_bases = Knowledges.get_knowledge_bases()
  67. else:
  68. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  69. # Get files for each knowledge base
  70. knowledge_with_files = []
  71. for knowledge_base in knowledge_bases:
  72. files = []
  73. if knowledge_base.data:
  74. files = Files.get_file_metadatas_by_ids(
  75. knowledge_base.data.get("file_ids", [])
  76. )
  77. # Check if all files exist
  78. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  79. missing_files = list(
  80. set(knowledge_base.data.get("file_ids", []))
  81. - set([file.id for file in files])
  82. )
  83. if missing_files:
  84. data = knowledge_base.data or {}
  85. file_ids = data.get("file_ids", [])
  86. for missing_file in missing_files:
  87. file_ids.remove(missing_file)
  88. data["file_ids"] = file_ids
  89. Knowledges.update_knowledge_data_by_id(
  90. id=knowledge_base.id, data=data
  91. )
  92. files = Files.get_file_metadatas_by_ids(file_ids)
  93. knowledge_with_files.append(
  94. KnowledgeUserResponse(
  95. **knowledge_base.model_dump(),
  96. files=files,
  97. )
  98. )
  99. return knowledge_with_files
  100. ############################
  101. # CreateNewKnowledge
  102. ############################
  103. @router.post("/create", response_model=Optional[KnowledgeResponse])
  104. async def create_new_knowledge(
  105. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  106. ):
  107. if user.role != "admin" and not has_permission(
  108. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  109. ):
  110. raise HTTPException(
  111. status_code=status.HTTP_401_UNAUTHORIZED,
  112. detail=ERROR_MESSAGES.UNAUTHORIZED,
  113. )
  114. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  115. if knowledge:
  116. return knowledge
  117. else:
  118. raise HTTPException(
  119. status_code=status.HTTP_400_BAD_REQUEST,
  120. detail=ERROR_MESSAGES.FILE_EXISTS,
  121. )
  122. ############################
  123. # GetKnowledgeById
  124. ############################
  125. class KnowledgeFilesResponse(KnowledgeResponse):
  126. files: list[FileModel]
  127. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  128. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  129. knowledge = Knowledges.get_knowledge_by_id(id=id)
  130. if knowledge:
  131. if (
  132. user.role == "admin"
  133. or knowledge.user_id == user.id
  134. or has_access(user.id, "read", knowledge.access_control)
  135. ):
  136. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  137. files = Files.get_files_by_ids(file_ids)
  138. return KnowledgeFilesResponse(
  139. **knowledge.model_dump(),
  140. files=files,
  141. )
  142. else:
  143. raise HTTPException(
  144. status_code=status.HTTP_401_UNAUTHORIZED,
  145. detail=ERROR_MESSAGES.NOT_FOUND,
  146. )
  147. ############################
  148. # UpdateKnowledgeById
  149. ############################
  150. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  151. async def update_knowledge_by_id(
  152. id: str,
  153. form_data: KnowledgeForm,
  154. user=Depends(get_verified_user),
  155. ):
  156. knowledge = Knowledges.get_knowledge_by_id(id=id)
  157. if not knowledge:
  158. raise HTTPException(
  159. status_code=status.HTTP_400_BAD_REQUEST,
  160. detail=ERROR_MESSAGES.NOT_FOUND,
  161. )
  162. if knowledge.user_id != user.id and user.role != "admin":
  163. raise HTTPException(
  164. status_code=status.HTTP_400_BAD_REQUEST,
  165. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  166. )
  167. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  168. if knowledge:
  169. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  170. files = Files.get_files_by_ids(file_ids)
  171. return KnowledgeFilesResponse(
  172. **knowledge.model_dump(),
  173. files=files,
  174. )
  175. else:
  176. raise HTTPException(
  177. status_code=status.HTTP_400_BAD_REQUEST,
  178. detail=ERROR_MESSAGES.ID_TAKEN,
  179. )
  180. ############################
  181. # AddFileToKnowledge
  182. ############################
  183. class KnowledgeFileIdForm(BaseModel):
  184. file_id: str
  185. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  186. def add_file_to_knowledge_by_id(
  187. id: str,
  188. form_data: KnowledgeFileIdForm,
  189. user=Depends(get_verified_user),
  190. ):
  191. knowledge = Knowledges.get_knowledge_by_id(id=id)
  192. if not knowledge:
  193. raise HTTPException(
  194. status_code=status.HTTP_400_BAD_REQUEST,
  195. detail=ERROR_MESSAGES.NOT_FOUND,
  196. )
  197. if knowledge.user_id != user.id and user.role != "admin":
  198. raise HTTPException(
  199. status_code=status.HTTP_400_BAD_REQUEST,
  200. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  201. )
  202. file = Files.get_file_by_id(form_data.file_id)
  203. if not file:
  204. raise HTTPException(
  205. status_code=status.HTTP_400_BAD_REQUEST,
  206. detail=ERROR_MESSAGES.NOT_FOUND,
  207. )
  208. if not file.data:
  209. raise HTTPException(
  210. status_code=status.HTTP_400_BAD_REQUEST,
  211. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  212. )
  213. # Add content to the vector database
  214. try:
  215. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  216. except Exception as e:
  217. log.debug(e)
  218. raise HTTPException(
  219. status_code=status.HTTP_400_BAD_REQUEST,
  220. detail=str(e),
  221. )
  222. if knowledge:
  223. data = knowledge.data or {}
  224. file_ids = data.get("file_ids", [])
  225. if form_data.file_id not in file_ids:
  226. file_ids.append(form_data.file_id)
  227. data["file_ids"] = file_ids
  228. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  229. if knowledge:
  230. files = Files.get_files_by_ids(file_ids)
  231. return KnowledgeFilesResponse(
  232. **knowledge.model_dump(),
  233. files=files,
  234. )
  235. else:
  236. raise HTTPException(
  237. status_code=status.HTTP_400_BAD_REQUEST,
  238. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  239. )
  240. else:
  241. raise HTTPException(
  242. status_code=status.HTTP_400_BAD_REQUEST,
  243. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  244. )
  245. else:
  246. raise HTTPException(
  247. status_code=status.HTTP_400_BAD_REQUEST,
  248. detail=ERROR_MESSAGES.NOT_FOUND,
  249. )
  250. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  251. def update_file_from_knowledge_by_id(
  252. id: str,
  253. form_data: KnowledgeFileIdForm,
  254. user=Depends(get_verified_user),
  255. ):
  256. knowledge = Knowledges.get_knowledge_by_id(id=id)
  257. if not knowledge:
  258. raise HTTPException(
  259. status_code=status.HTTP_400_BAD_REQUEST,
  260. detail=ERROR_MESSAGES.NOT_FOUND,
  261. )
  262. if knowledge.user_id != user.id and user.role != "admin":
  263. raise HTTPException(
  264. status_code=status.HTTP_400_BAD_REQUEST,
  265. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  266. )
  267. file = Files.get_file_by_id(form_data.file_id)
  268. if not file:
  269. raise HTTPException(
  270. status_code=status.HTTP_400_BAD_REQUEST,
  271. detail=ERROR_MESSAGES.NOT_FOUND,
  272. )
  273. # Remove content from the vector database
  274. VECTOR_DB_CLIENT.delete(
  275. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  276. )
  277. # Add content to the vector database
  278. try:
  279. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  280. except Exception as e:
  281. raise HTTPException(
  282. status_code=status.HTTP_400_BAD_REQUEST,
  283. detail=str(e),
  284. )
  285. if knowledge:
  286. data = knowledge.data or {}
  287. file_ids = data.get("file_ids", [])
  288. files = Files.get_files_by_ids(file_ids)
  289. return KnowledgeFilesResponse(
  290. **knowledge.model_dump(),
  291. files=files,
  292. )
  293. else:
  294. raise HTTPException(
  295. status_code=status.HTTP_400_BAD_REQUEST,
  296. detail=ERROR_MESSAGES.NOT_FOUND,
  297. )
  298. ############################
  299. # RemoveFileFromKnowledge
  300. ############################
  301. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  302. def remove_file_from_knowledge_by_id(
  303. id: str,
  304. form_data: KnowledgeFileIdForm,
  305. user=Depends(get_verified_user),
  306. ):
  307. knowledge = Knowledges.get_knowledge_by_id(id=id)
  308. if not knowledge:
  309. raise HTTPException(
  310. status_code=status.HTTP_400_BAD_REQUEST,
  311. detail=ERROR_MESSAGES.NOT_FOUND,
  312. )
  313. if knowledge.user_id != user.id and user.role != "admin":
  314. raise HTTPException(
  315. status_code=status.HTTP_400_BAD_REQUEST,
  316. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  317. )
  318. file = Files.get_file_by_id(form_data.file_id)
  319. if not file:
  320. raise HTTPException(
  321. status_code=status.HTTP_400_BAD_REQUEST,
  322. detail=ERROR_MESSAGES.NOT_FOUND,
  323. )
  324. # Remove content from the vector database
  325. VECTOR_DB_CLIENT.delete(
  326. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  327. )
  328. result = VECTOR_DB_CLIENT.query(
  329. collection_name=knowledge.id,
  330. filter={"file_id": form_data.file_id},
  331. )
  332. Files.delete_file_by_id(form_data.file_id)
  333. if knowledge:
  334. data = knowledge.data or {}
  335. file_ids = data.get("file_ids", [])
  336. if form_data.file_id in file_ids:
  337. file_ids.remove(form_data.file_id)
  338. data["file_ids"] = file_ids
  339. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  340. if knowledge:
  341. files = Files.get_files_by_ids(file_ids)
  342. return KnowledgeFilesResponse(
  343. **knowledge.model_dump(),
  344. files=files,
  345. )
  346. else:
  347. raise HTTPException(
  348. status_code=status.HTTP_400_BAD_REQUEST,
  349. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  350. )
  351. else:
  352. raise HTTPException(
  353. status_code=status.HTTP_400_BAD_REQUEST,
  354. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  355. )
  356. else:
  357. raise HTTPException(
  358. status_code=status.HTTP_400_BAD_REQUEST,
  359. detail=ERROR_MESSAGES.NOT_FOUND,
  360. )
  361. ############################
  362. # DeleteKnowledgeById
  363. ############################
  364. @router.delete("/{id}/delete", response_model=bool)
  365. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  366. knowledge = Knowledges.get_knowledge_by_id(id=id)
  367. if not knowledge:
  368. raise HTTPException(
  369. status_code=status.HTTP_400_BAD_REQUEST,
  370. detail=ERROR_MESSAGES.NOT_FOUND,
  371. )
  372. if knowledge.user_id != user.id and user.role != "admin":
  373. raise HTTPException(
  374. status_code=status.HTTP_400_BAD_REQUEST,
  375. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  376. )
  377. try:
  378. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  379. except Exception as e:
  380. log.debug(e)
  381. pass
  382. result = Knowledges.delete_knowledge_by_id(id=id)
  383. return result
  384. ############################
  385. # ResetKnowledgeById
  386. ############################
  387. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  388. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  389. knowledge = Knowledges.get_knowledge_by_id(id=id)
  390. if not knowledge:
  391. raise HTTPException(
  392. status_code=status.HTTP_400_BAD_REQUEST,
  393. detail=ERROR_MESSAGES.NOT_FOUND,
  394. )
  395. if knowledge.user_id != user.id and user.role != "admin":
  396. raise HTTPException(
  397. status_code=status.HTTP_400_BAD_REQUEST,
  398. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  399. )
  400. try:
  401. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  402. except Exception as e:
  403. log.debug(e)
  404. pass
  405. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  406. return knowledge
  407. ############################
  408. # AddFilesToKnowledge
  409. ############################
  410. @router.post("/{id}/files/batch/add", response_model=Optional[KnowledgeFilesResponse])
  411. def add_files_to_knowledge_batch(
  412. id: str,
  413. form_data: list[KnowledgeFileIdForm],
  414. user=Depends(get_verified_user),
  415. ):
  416. """
  417. Add multiple files to a knowledge base
  418. """
  419. knowledge = Knowledges.get_knowledge_by_id(id=id)
  420. if not knowledge:
  421. raise HTTPException(
  422. status_code=status.HTTP_400_BAD_REQUEST,
  423. detail=ERROR_MESSAGES.NOT_FOUND,
  424. )
  425. if knowledge.user_id != user.id and user.role != "admin":
  426. raise HTTPException(
  427. status_code=status.HTTP_400_BAD_REQUEST,
  428. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  429. )
  430. # Get files content
  431. print(f"files/batch/add - {len(form_data)} files")
  432. files: List[FileModel] = []
  433. for form in form_data:
  434. file = Files.get_file_by_id(form.file_id)
  435. if not file:
  436. raise HTTPException(
  437. status_code=status.HTTP_400_BAD_REQUEST,
  438. detail=f"File {form.file_id} not found",
  439. )
  440. files.append(file)
  441. # Process files
  442. try:
  443. result = process_files_batch(BatchProcessFilesForm(
  444. files=files,
  445. collection_name=id
  446. ))
  447. except Exception as e:
  448. log.error(f"add_files_to_knowledge_batch: Exception occurred: {e}", exc_info=True)
  449. raise HTTPException(
  450. status_code=status.HTTP_400_BAD_REQUEST,
  451. detail=str(e)
  452. )
  453. # Add successful files to knowledge base
  454. data = knowledge.data or {}
  455. existing_file_ids = data.get("file_ids", [])
  456. # Only add files that were successfully processed
  457. successful_file_ids = [r.file_id for r in result.results if r.status == "completed"]
  458. for file_id in successful_file_ids:
  459. if file_id not in existing_file_ids:
  460. existing_file_ids.append(file_id)
  461. data["file_ids"] = existing_file_ids
  462. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  463. # If there were any errors, include them in the response
  464. if result.errors:
  465. error_details = [f"{err.file_id}: {err.error}" for err in result.errors]
  466. return KnowledgeFilesResponse(
  467. **knowledge.model_dump(),
  468. files=Files.get_files_by_ids(existing_file_ids),
  469. warnings={
  470. "message": "Some files failed to process",
  471. "errors": error_details
  472. }
  473. )
  474. return KnowledgeFilesResponse(
  475. **knowledge.model_dump(),
  476. files=Files.get_files_by_ids(existing_file_ids)
  477. )