knowledge.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516
  1. import json
  2. from typing import Optional, Union
  3. from pydantic import BaseModel
  4. from fastapi import APIRouter, Depends, HTTPException, status, Request
  5. import logging
  6. from open_webui.models.knowledge import (
  7. Knowledges,
  8. KnowledgeForm,
  9. KnowledgeResponse,
  10. KnowledgeUserResponse,
  11. )
  12. from open_webui.models.files import Files, FileModel
  13. from open_webui.retrieval.vector.connector import VECTOR_DB_CLIENT
  14. from open_webui.routers.retrieval import process_file, ProcessFileForm
  15. from open_webui.constants import ERROR_MESSAGES
  16. from open_webui.utils.auth import get_admin_user, get_verified_user
  17. from open_webui.utils.access_control import has_access, has_permission
  18. from open_webui.env import SRC_LOG_LEVELS
  19. log = logging.getLogger(__name__)
  20. log.setLevel(SRC_LOG_LEVELS["MODELS"])
  21. router = APIRouter()
  22. ############################
  23. # getKnowledgeBases
  24. ############################
  25. @router.get("/", response_model=list[KnowledgeUserResponse])
  26. async def get_knowledge(user=Depends(get_verified_user)):
  27. knowledge_bases = []
  28. if user.role == "admin":
  29. knowledge_bases = Knowledges.get_knowledge_bases()
  30. else:
  31. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "read")
  32. # Get files for each knowledge base
  33. knowledge_with_files = []
  34. for knowledge_base in knowledge_bases:
  35. files = []
  36. if knowledge_base.data:
  37. files = Files.get_file_metadatas_by_ids(
  38. knowledge_base.data.get("file_ids", [])
  39. )
  40. # Check if all files exist
  41. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  42. missing_files = list(
  43. set(knowledge_base.data.get("file_ids", []))
  44. - set([file.id for file in files])
  45. )
  46. if missing_files:
  47. data = knowledge_base.data or {}
  48. file_ids = data.get("file_ids", [])
  49. for missing_file in missing_files:
  50. file_ids.remove(missing_file)
  51. data["file_ids"] = file_ids
  52. Knowledges.update_knowledge_data_by_id(
  53. id=knowledge_base.id, data=data
  54. )
  55. files = Files.get_file_metadatas_by_ids(file_ids)
  56. knowledge_with_files.append(
  57. KnowledgeUserResponse(
  58. **knowledge_base.model_dump(),
  59. files=files,
  60. )
  61. )
  62. return knowledge_with_files
  63. @router.get("/list", response_model=list[KnowledgeUserResponse])
  64. async def get_knowledge_list(user=Depends(get_verified_user)):
  65. knowledge_bases = []
  66. if user.role == "admin":
  67. knowledge_bases = Knowledges.get_knowledge_bases()
  68. else:
  69. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  70. # Get files for each knowledge base
  71. knowledge_with_files = []
  72. for knowledge_base in knowledge_bases:
  73. files = []
  74. if knowledge_base.data:
  75. files = Files.get_file_metadatas_by_ids(
  76. knowledge_base.data.get("file_ids", [])
  77. )
  78. # Check if all files exist
  79. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  80. missing_files = list(
  81. set(knowledge_base.data.get("file_ids", []))
  82. - set([file.id for file in files])
  83. )
  84. if missing_files:
  85. data = knowledge_base.data or {}
  86. file_ids = data.get("file_ids", [])
  87. for missing_file in missing_files:
  88. file_ids.remove(missing_file)
  89. data["file_ids"] = file_ids
  90. Knowledges.update_knowledge_data_by_id(
  91. id=knowledge_base.id, data=data
  92. )
  93. files = Files.get_file_metadatas_by_ids(file_ids)
  94. knowledge_with_files.append(
  95. KnowledgeUserResponse(
  96. **knowledge_base.model_dump(),
  97. files=files,
  98. )
  99. )
  100. return knowledge_with_files
  101. ############################
  102. # CreateNewKnowledge
  103. ############################
  104. @router.post("/create", response_model=Optional[KnowledgeResponse])
  105. async def create_new_knowledge(
  106. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  107. ):
  108. if user.role != "admin" and not has_permission(
  109. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  110. ):
  111. raise HTTPException(
  112. status_code=status.HTTP_401_UNAUTHORIZED,
  113. detail=ERROR_MESSAGES.UNAUTHORIZED,
  114. )
  115. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  116. if knowledge:
  117. return knowledge
  118. else:
  119. raise HTTPException(
  120. status_code=status.HTTP_400_BAD_REQUEST,
  121. detail=ERROR_MESSAGES.FILE_EXISTS,
  122. )
  123. ############################
  124. # GetKnowledgeById
  125. ############################
  126. class KnowledgeFilesResponse(KnowledgeResponse):
  127. files: list[FileModel]
  128. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  129. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  130. knowledge = Knowledges.get_knowledge_by_id(id=id)
  131. if knowledge:
  132. if (
  133. user.role == "admin"
  134. or knowledge.user_id == user.id
  135. or has_access(user.id, "read", knowledge.access_control)
  136. ):
  137. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  138. files = Files.get_files_by_ids(file_ids)
  139. return KnowledgeFilesResponse(
  140. **knowledge.model_dump(),
  141. files=files,
  142. )
  143. else:
  144. raise HTTPException(
  145. status_code=status.HTTP_401_UNAUTHORIZED,
  146. detail=ERROR_MESSAGES.NOT_FOUND,
  147. )
  148. ############################
  149. # UpdateKnowledgeById
  150. ############################
  151. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  152. async def update_knowledge_by_id(
  153. id: str,
  154. form_data: KnowledgeForm,
  155. user=Depends(get_verified_user),
  156. ):
  157. knowledge = Knowledges.get_knowledge_by_id(id=id)
  158. if not knowledge:
  159. raise HTTPException(
  160. status_code=status.HTTP_400_BAD_REQUEST,
  161. detail=ERROR_MESSAGES.NOT_FOUND,
  162. )
  163. if knowledge.user_id != user.id and user.role != "admin":
  164. raise HTTPException(
  165. status_code=status.HTTP_400_BAD_REQUEST,
  166. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  167. )
  168. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  169. if knowledge:
  170. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  171. files = Files.get_files_by_ids(file_ids)
  172. return KnowledgeFilesResponse(
  173. **knowledge.model_dump(),
  174. files=files,
  175. )
  176. else:
  177. raise HTTPException(
  178. status_code=status.HTTP_400_BAD_REQUEST,
  179. detail=ERROR_MESSAGES.ID_TAKEN,
  180. )
  181. ############################
  182. # AddFileToKnowledge
  183. ############################
  184. class KnowledgeFileIdForm(BaseModel):
  185. file_id: str
  186. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  187. def add_file_to_knowledge_by_id(
  188. request: Request,
  189. id: str,
  190. form_data: KnowledgeFileIdForm,
  191. user=Depends(get_verified_user),
  192. ):
  193. knowledge = Knowledges.get_knowledge_by_id(id=id)
  194. if not knowledge:
  195. raise HTTPException(
  196. status_code=status.HTTP_400_BAD_REQUEST,
  197. detail=ERROR_MESSAGES.NOT_FOUND,
  198. )
  199. if knowledge.user_id != user.id and user.role != "admin":
  200. raise HTTPException(
  201. status_code=status.HTTP_400_BAD_REQUEST,
  202. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  203. )
  204. file = Files.get_file_by_id(form_data.file_id)
  205. if not file:
  206. raise HTTPException(
  207. status_code=status.HTTP_400_BAD_REQUEST,
  208. detail=ERROR_MESSAGES.NOT_FOUND,
  209. )
  210. if not file.data:
  211. raise HTTPException(
  212. status_code=status.HTTP_400_BAD_REQUEST,
  213. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  214. )
  215. # Add content to the vector database
  216. try:
  217. process_file(
  218. request, ProcessFileForm(file_id=form_data.file_id, collection_name=id)
  219. )
  220. except Exception as e:
  221. log.debug(e)
  222. raise HTTPException(
  223. status_code=status.HTTP_400_BAD_REQUEST,
  224. detail=str(e),
  225. )
  226. if knowledge:
  227. data = knowledge.data or {}
  228. file_ids = data.get("file_ids", [])
  229. if form_data.file_id not in file_ids:
  230. file_ids.append(form_data.file_id)
  231. data["file_ids"] = file_ids
  232. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  233. if knowledge:
  234. files = Files.get_files_by_ids(file_ids)
  235. return KnowledgeFilesResponse(
  236. **knowledge.model_dump(),
  237. files=files,
  238. )
  239. else:
  240. raise HTTPException(
  241. status_code=status.HTTP_400_BAD_REQUEST,
  242. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  243. )
  244. else:
  245. raise HTTPException(
  246. status_code=status.HTTP_400_BAD_REQUEST,
  247. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  248. )
  249. else:
  250. raise HTTPException(
  251. status_code=status.HTTP_400_BAD_REQUEST,
  252. detail=ERROR_MESSAGES.NOT_FOUND,
  253. )
  254. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  255. def update_file_from_knowledge_by_id(
  256. request: Request,
  257. id: str,
  258. form_data: KnowledgeFileIdForm,
  259. user=Depends(get_verified_user),
  260. ):
  261. knowledge = Knowledges.get_knowledge_by_id(id=id)
  262. if not knowledge:
  263. raise HTTPException(
  264. status_code=status.HTTP_400_BAD_REQUEST,
  265. detail=ERROR_MESSAGES.NOT_FOUND,
  266. )
  267. if knowledge.user_id != user.id and user.role != "admin":
  268. raise HTTPException(
  269. status_code=status.HTTP_400_BAD_REQUEST,
  270. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  271. )
  272. file = Files.get_file_by_id(form_data.file_id)
  273. if not file:
  274. raise HTTPException(
  275. status_code=status.HTTP_400_BAD_REQUEST,
  276. detail=ERROR_MESSAGES.NOT_FOUND,
  277. )
  278. # Remove content from the vector database
  279. VECTOR_DB_CLIENT.delete(
  280. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  281. )
  282. # Add content to the vector database
  283. try:
  284. process_file(
  285. request, ProcessFileForm(file_id=form_data.file_id, collection_name=id)
  286. )
  287. except Exception as e:
  288. raise HTTPException(
  289. status_code=status.HTTP_400_BAD_REQUEST,
  290. detail=str(e),
  291. )
  292. if knowledge:
  293. data = knowledge.data or {}
  294. file_ids = data.get("file_ids", [])
  295. files = Files.get_files_by_ids(file_ids)
  296. return KnowledgeFilesResponse(
  297. **knowledge.model_dump(),
  298. files=files,
  299. )
  300. else:
  301. raise HTTPException(
  302. status_code=status.HTTP_400_BAD_REQUEST,
  303. detail=ERROR_MESSAGES.NOT_FOUND,
  304. )
  305. ############################
  306. # RemoveFileFromKnowledge
  307. ############################
  308. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  309. def remove_file_from_knowledge_by_id(
  310. id: str,
  311. form_data: KnowledgeFileIdForm,
  312. user=Depends(get_verified_user),
  313. ):
  314. knowledge = Knowledges.get_knowledge_by_id(id=id)
  315. if not knowledge:
  316. raise HTTPException(
  317. status_code=status.HTTP_400_BAD_REQUEST,
  318. detail=ERROR_MESSAGES.NOT_FOUND,
  319. )
  320. if knowledge.user_id != user.id and user.role != "admin":
  321. raise HTTPException(
  322. status_code=status.HTTP_400_BAD_REQUEST,
  323. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  324. )
  325. file = Files.get_file_by_id(form_data.file_id)
  326. if not file:
  327. raise HTTPException(
  328. status_code=status.HTTP_400_BAD_REQUEST,
  329. detail=ERROR_MESSAGES.NOT_FOUND,
  330. )
  331. # Remove content from the vector database
  332. VECTOR_DB_CLIENT.delete(
  333. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  334. )
  335. result = VECTOR_DB_CLIENT.query(
  336. collection_name=knowledge.id,
  337. filter={"file_id": form_data.file_id},
  338. )
  339. Files.delete_file_by_id(form_data.file_id)
  340. if knowledge:
  341. data = knowledge.data or {}
  342. file_ids = data.get("file_ids", [])
  343. if form_data.file_id in file_ids:
  344. file_ids.remove(form_data.file_id)
  345. data["file_ids"] = file_ids
  346. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  347. if knowledge:
  348. files = Files.get_files_by_ids(file_ids)
  349. return KnowledgeFilesResponse(
  350. **knowledge.model_dump(),
  351. files=files,
  352. )
  353. else:
  354. raise HTTPException(
  355. status_code=status.HTTP_400_BAD_REQUEST,
  356. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  357. )
  358. else:
  359. raise HTTPException(
  360. status_code=status.HTTP_400_BAD_REQUEST,
  361. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  362. )
  363. else:
  364. raise HTTPException(
  365. status_code=status.HTTP_400_BAD_REQUEST,
  366. detail=ERROR_MESSAGES.NOT_FOUND,
  367. )
  368. ############################
  369. # DeleteKnowledgeById
  370. ############################
  371. @router.delete("/{id}/delete", response_model=bool)
  372. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  373. knowledge = Knowledges.get_knowledge_by_id(id=id)
  374. if not knowledge:
  375. raise HTTPException(
  376. status_code=status.HTTP_400_BAD_REQUEST,
  377. detail=ERROR_MESSAGES.NOT_FOUND,
  378. )
  379. if knowledge.user_id != user.id and user.role != "admin":
  380. raise HTTPException(
  381. status_code=status.HTTP_400_BAD_REQUEST,
  382. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  383. )
  384. try:
  385. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  386. except Exception as e:
  387. log.debug(e)
  388. pass
  389. result = Knowledges.delete_knowledge_by_id(id=id)
  390. return result
  391. ############################
  392. # ResetKnowledgeById
  393. ############################
  394. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  395. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  396. knowledge = Knowledges.get_knowledge_by_id(id=id)
  397. if not knowledge:
  398. raise HTTPException(
  399. status_code=status.HTTP_400_BAD_REQUEST,
  400. detail=ERROR_MESSAGES.NOT_FOUND,
  401. )
  402. if knowledge.user_id != user.id and user.role != "admin":
  403. raise HTTPException(
  404. status_code=status.HTTP_400_BAD_REQUEST,
  405. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  406. )
  407. try:
  408. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  409. except Exception as e:
  410. log.debug(e)
  411. pass
  412. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  413. return knowledge