knowledge.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510
  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.apps.retrieval.vector.connector import VECTOR_DB_CLIENT
  14. from backend.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. id: str,
  189. form_data: KnowledgeFileIdForm,
  190. user=Depends(get_verified_user),
  191. ):
  192. knowledge = Knowledges.get_knowledge_by_id(id=id)
  193. if not knowledge:
  194. raise HTTPException(
  195. status_code=status.HTTP_400_BAD_REQUEST,
  196. detail=ERROR_MESSAGES.NOT_FOUND,
  197. )
  198. if knowledge.user_id != user.id and user.role != "admin":
  199. raise HTTPException(
  200. status_code=status.HTTP_400_BAD_REQUEST,
  201. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  202. )
  203. file = Files.get_file_by_id(form_data.file_id)
  204. if not file:
  205. raise HTTPException(
  206. status_code=status.HTTP_400_BAD_REQUEST,
  207. detail=ERROR_MESSAGES.NOT_FOUND,
  208. )
  209. if not file.data:
  210. raise HTTPException(
  211. status_code=status.HTTP_400_BAD_REQUEST,
  212. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  213. )
  214. # Add content to the vector database
  215. try:
  216. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  217. except Exception as e:
  218. log.debug(e)
  219. raise HTTPException(
  220. status_code=status.HTTP_400_BAD_REQUEST,
  221. detail=str(e),
  222. )
  223. if knowledge:
  224. data = knowledge.data or {}
  225. file_ids = data.get("file_ids", [])
  226. if form_data.file_id not in file_ids:
  227. file_ids.append(form_data.file_id)
  228. data["file_ids"] = file_ids
  229. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  230. if knowledge:
  231. files = Files.get_files_by_ids(file_ids)
  232. return KnowledgeFilesResponse(
  233. **knowledge.model_dump(),
  234. files=files,
  235. )
  236. else:
  237. raise HTTPException(
  238. status_code=status.HTTP_400_BAD_REQUEST,
  239. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  240. )
  241. else:
  242. raise HTTPException(
  243. status_code=status.HTTP_400_BAD_REQUEST,
  244. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  245. )
  246. else:
  247. raise HTTPException(
  248. status_code=status.HTTP_400_BAD_REQUEST,
  249. detail=ERROR_MESSAGES.NOT_FOUND,
  250. )
  251. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  252. def update_file_from_knowledge_by_id(
  253. id: str,
  254. form_data: KnowledgeFileIdForm,
  255. user=Depends(get_verified_user),
  256. ):
  257. knowledge = Knowledges.get_knowledge_by_id(id=id)
  258. if not knowledge:
  259. raise HTTPException(
  260. status_code=status.HTTP_400_BAD_REQUEST,
  261. detail=ERROR_MESSAGES.NOT_FOUND,
  262. )
  263. if knowledge.user_id != user.id and user.role != "admin":
  264. raise HTTPException(
  265. status_code=status.HTTP_400_BAD_REQUEST,
  266. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  267. )
  268. file = Files.get_file_by_id(form_data.file_id)
  269. if not file:
  270. raise HTTPException(
  271. status_code=status.HTTP_400_BAD_REQUEST,
  272. detail=ERROR_MESSAGES.NOT_FOUND,
  273. )
  274. # Remove content from the vector database
  275. VECTOR_DB_CLIENT.delete(
  276. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  277. )
  278. # Add content to the vector database
  279. try:
  280. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  281. except Exception as e:
  282. raise HTTPException(
  283. status_code=status.HTTP_400_BAD_REQUEST,
  284. detail=str(e),
  285. )
  286. if knowledge:
  287. data = knowledge.data or {}
  288. file_ids = data.get("file_ids", [])
  289. files = Files.get_files_by_ids(file_ids)
  290. return KnowledgeFilesResponse(
  291. **knowledge.model_dump(),
  292. files=files,
  293. )
  294. else:
  295. raise HTTPException(
  296. status_code=status.HTTP_400_BAD_REQUEST,
  297. detail=ERROR_MESSAGES.NOT_FOUND,
  298. )
  299. ############################
  300. # RemoveFileFromKnowledge
  301. ############################
  302. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  303. def remove_file_from_knowledge_by_id(
  304. id: str,
  305. form_data: KnowledgeFileIdForm,
  306. user=Depends(get_verified_user),
  307. ):
  308. knowledge = Knowledges.get_knowledge_by_id(id=id)
  309. if not knowledge:
  310. raise HTTPException(
  311. status_code=status.HTTP_400_BAD_REQUEST,
  312. detail=ERROR_MESSAGES.NOT_FOUND,
  313. )
  314. if knowledge.user_id != user.id and user.role != "admin":
  315. raise HTTPException(
  316. status_code=status.HTTP_400_BAD_REQUEST,
  317. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  318. )
  319. file = Files.get_file_by_id(form_data.file_id)
  320. if not file:
  321. raise HTTPException(
  322. status_code=status.HTTP_400_BAD_REQUEST,
  323. detail=ERROR_MESSAGES.NOT_FOUND,
  324. )
  325. # Remove content from the vector database
  326. VECTOR_DB_CLIENT.delete(
  327. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  328. )
  329. result = VECTOR_DB_CLIENT.query(
  330. collection_name=knowledge.id,
  331. filter={"file_id": form_data.file_id},
  332. )
  333. Files.delete_file_by_id(form_data.file_id)
  334. if knowledge:
  335. data = knowledge.data or {}
  336. file_ids = data.get("file_ids", [])
  337. if form_data.file_id in file_ids:
  338. file_ids.remove(form_data.file_id)
  339. data["file_ids"] = file_ids
  340. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  341. if knowledge:
  342. files = Files.get_files_by_ids(file_ids)
  343. return KnowledgeFilesResponse(
  344. **knowledge.model_dump(),
  345. files=files,
  346. )
  347. else:
  348. raise HTTPException(
  349. status_code=status.HTTP_400_BAD_REQUEST,
  350. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  351. )
  352. else:
  353. raise HTTPException(
  354. status_code=status.HTTP_400_BAD_REQUEST,
  355. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  356. )
  357. else:
  358. raise HTTPException(
  359. status_code=status.HTTP_400_BAD_REQUEST,
  360. detail=ERROR_MESSAGES.NOT_FOUND,
  361. )
  362. ############################
  363. # DeleteKnowledgeById
  364. ############################
  365. @router.delete("/{id}/delete", response_model=bool)
  366. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  367. knowledge = Knowledges.get_knowledge_by_id(id=id)
  368. if not knowledge:
  369. raise HTTPException(
  370. status_code=status.HTTP_400_BAD_REQUEST,
  371. detail=ERROR_MESSAGES.NOT_FOUND,
  372. )
  373. if knowledge.user_id != user.id and user.role != "admin":
  374. raise HTTPException(
  375. status_code=status.HTTP_400_BAD_REQUEST,
  376. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  377. )
  378. try:
  379. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  380. except Exception as e:
  381. log.debug(e)
  382. pass
  383. result = Knowledges.delete_knowledge_by_id(id=id)
  384. return result
  385. ############################
  386. # ResetKnowledgeById
  387. ############################
  388. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  389. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  390. knowledge = Knowledges.get_knowledge_by_id(id=id)
  391. if not knowledge:
  392. raise HTTPException(
  393. status_code=status.HTTP_400_BAD_REQUEST,
  394. detail=ERROR_MESSAGES.NOT_FOUND,
  395. )
  396. if knowledge.user_id != user.id and user.role != "admin":
  397. raise HTTPException(
  398. status_code=status.HTTP_400_BAD_REQUEST,
  399. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  400. )
  401. try:
  402. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  403. except Exception as e:
  404. log.debug(e)
  405. pass
  406. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  407. return knowledge