knowledge.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505
  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.apps.webui.models.knowledge import (
  7. Knowledges,
  8. KnowledgeForm,
  9. KnowledgeResponse,
  10. KnowledgeUserResponse,
  11. )
  12. from open_webui.apps.webui.models.files import Files, FileModel
  13. from open_webui.apps.retrieval.vector.connector import VECTOR_DB_CLIENT
  14. from open_webui.apps.retrieval.main import process_file, ProcessFileForm
  15. from open_webui.constants import ERROR_MESSAGES
  16. from open_webui.utils.utils 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. 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_base = KnowledgeResponse(
  56. **knowledge_base.model_dump(),
  57. files=files,
  58. )
  59. return knowledge_bases
  60. @router.get("/list", response_model=list[KnowledgeUserResponse])
  61. async def get_knowledge_list(user=Depends(get_verified_user)):
  62. knowledge_bases = []
  63. if user.role == "admin":
  64. knowledge_bases = Knowledges.get_knowledge_bases()
  65. else:
  66. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  67. # Get files for each knowledge base
  68. for knowledge_base in knowledge_bases:
  69. files = []
  70. if knowledge_base.data:
  71. files = Files.get_file_metadatas_by_ids(
  72. knowledge_base.data.get("file_ids", [])
  73. )
  74. # Check if all files exist
  75. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  76. missing_files = list(
  77. set(knowledge_base.data.get("file_ids", []))
  78. - set([file.id for file in files])
  79. )
  80. if missing_files:
  81. data = knowledge_base.data or {}
  82. file_ids = data.get("file_ids", [])
  83. for missing_file in missing_files:
  84. file_ids.remove(missing_file)
  85. data["file_ids"] = file_ids
  86. Knowledges.update_knowledge_data_by_id(
  87. id=knowledge_base.id, data=data
  88. )
  89. files = Files.get_file_metadatas_by_ids(file_ids)
  90. knowledge_base = KnowledgeResponse(
  91. **knowledge_base.model_dump(),
  92. files=files,
  93. )
  94. return knowledge_bases
  95. ############################
  96. # CreateNewKnowledge
  97. ############################
  98. @router.post("/create", response_model=Optional[KnowledgeResponse])
  99. async def create_new_knowledge(
  100. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  101. ):
  102. if user.role != "admin" and not has_permission(
  103. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  104. ):
  105. raise HTTPException(
  106. status_code=status.HTTP_401_UNAUTHORIZED,
  107. detail=ERROR_MESSAGES.UNAUTHORIZED,
  108. )
  109. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  110. if knowledge:
  111. return knowledge
  112. else:
  113. raise HTTPException(
  114. status_code=status.HTTP_400_BAD_REQUEST,
  115. detail=ERROR_MESSAGES.FILE_EXISTS,
  116. )
  117. ############################
  118. # GetKnowledgeById
  119. ############################
  120. class KnowledgeFilesResponse(KnowledgeResponse):
  121. files: list[FileModel]
  122. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  123. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  124. knowledge = Knowledges.get_knowledge_by_id(id=id)
  125. if knowledge:
  126. if (
  127. user.role == "admin"
  128. or knowledge.user_id == user.id
  129. or has_access(user.id, "read", knowledge.access_control)
  130. ):
  131. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  132. files = Files.get_files_by_ids(file_ids)
  133. return KnowledgeFilesResponse(
  134. **knowledge.model_dump(),
  135. files=files,
  136. )
  137. else:
  138. raise HTTPException(
  139. status_code=status.HTTP_401_UNAUTHORIZED,
  140. detail=ERROR_MESSAGES.NOT_FOUND,
  141. )
  142. ############################
  143. # UpdateKnowledgeById
  144. ############################
  145. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  146. async def update_knowledge_by_id(
  147. id: str,
  148. form_data: KnowledgeForm,
  149. user=Depends(get_verified_user),
  150. ):
  151. knowledge = Knowledges.get_knowledge_by_id(id=id)
  152. if not knowledge:
  153. raise HTTPException(
  154. status_code=status.HTTP_400_BAD_REQUEST,
  155. detail=ERROR_MESSAGES.NOT_FOUND,
  156. )
  157. if knowledge.user_id != user.id and user.role != "admin":
  158. raise HTTPException(
  159. status_code=status.HTTP_400_BAD_REQUEST,
  160. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  161. )
  162. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  163. if knowledge:
  164. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  165. files = Files.get_files_by_ids(file_ids)
  166. return KnowledgeFilesResponse(
  167. **knowledge.model_dump(),
  168. files=files,
  169. )
  170. else:
  171. raise HTTPException(
  172. status_code=status.HTTP_400_BAD_REQUEST,
  173. detail=ERROR_MESSAGES.ID_TAKEN,
  174. )
  175. ############################
  176. # AddFileToKnowledge
  177. ############################
  178. class KnowledgeFileIdForm(BaseModel):
  179. file_id: str
  180. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  181. def add_file_to_knowledge_by_id(
  182. id: str,
  183. form_data: KnowledgeFileIdForm,
  184. user=Depends(get_verified_user),
  185. ):
  186. knowledge = Knowledges.get_knowledge_by_id(id=id)
  187. if not knowledge:
  188. raise HTTPException(
  189. status_code=status.HTTP_400_BAD_REQUEST,
  190. detail=ERROR_MESSAGES.NOT_FOUND,
  191. )
  192. if knowledge.user_id != user.id and user.role != "admin":
  193. raise HTTPException(
  194. status_code=status.HTTP_400_BAD_REQUEST,
  195. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  196. )
  197. file = Files.get_file_by_id(form_data.file_id)
  198. if not file:
  199. raise HTTPException(
  200. status_code=status.HTTP_400_BAD_REQUEST,
  201. detail=ERROR_MESSAGES.NOT_FOUND,
  202. )
  203. if not file.data:
  204. raise HTTPException(
  205. status_code=status.HTTP_400_BAD_REQUEST,
  206. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  207. )
  208. # Add content to the vector database
  209. try:
  210. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  211. except Exception as e:
  212. log.debug(e)
  213. raise HTTPException(
  214. status_code=status.HTTP_400_BAD_REQUEST,
  215. detail=str(e),
  216. )
  217. if knowledge:
  218. data = knowledge.data or {}
  219. file_ids = data.get("file_ids", [])
  220. if form_data.file_id not in file_ids:
  221. file_ids.append(form_data.file_id)
  222. data["file_ids"] = file_ids
  223. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  224. if knowledge:
  225. files = Files.get_files_by_ids(file_ids)
  226. return KnowledgeFilesResponse(
  227. **knowledge.model_dump(),
  228. files=files,
  229. )
  230. else:
  231. raise HTTPException(
  232. status_code=status.HTTP_400_BAD_REQUEST,
  233. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  234. )
  235. else:
  236. raise HTTPException(
  237. status_code=status.HTTP_400_BAD_REQUEST,
  238. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  239. )
  240. else:
  241. raise HTTPException(
  242. status_code=status.HTTP_400_BAD_REQUEST,
  243. detail=ERROR_MESSAGES.NOT_FOUND,
  244. )
  245. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  246. def update_file_from_knowledge_by_id(
  247. id: str,
  248. form_data: KnowledgeFileIdForm,
  249. user=Depends(get_verified_user),
  250. ):
  251. knowledge = Knowledges.get_knowledge_by_id(id=id)
  252. if not knowledge:
  253. raise HTTPException(
  254. status_code=status.HTTP_400_BAD_REQUEST,
  255. detail=ERROR_MESSAGES.NOT_FOUND,
  256. )
  257. if knowledge.user_id != user.id and user.role != "admin":
  258. raise HTTPException(
  259. status_code=status.HTTP_400_BAD_REQUEST,
  260. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  261. )
  262. file = Files.get_file_by_id(form_data.file_id)
  263. if not file:
  264. raise HTTPException(
  265. status_code=status.HTTP_400_BAD_REQUEST,
  266. detail=ERROR_MESSAGES.NOT_FOUND,
  267. )
  268. # Remove content from the vector database
  269. VECTOR_DB_CLIENT.delete(
  270. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  271. )
  272. # Add content to the vector database
  273. try:
  274. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  275. except Exception as e:
  276. raise HTTPException(
  277. status_code=status.HTTP_400_BAD_REQUEST,
  278. detail=str(e),
  279. )
  280. if knowledge:
  281. data = knowledge.data or {}
  282. file_ids = data.get("file_ids", [])
  283. files = Files.get_files_by_ids(file_ids)
  284. return KnowledgeFilesResponse(
  285. **knowledge.model_dump(),
  286. files=files,
  287. )
  288. else:
  289. raise HTTPException(
  290. status_code=status.HTTP_400_BAD_REQUEST,
  291. detail=ERROR_MESSAGES.NOT_FOUND,
  292. )
  293. ############################
  294. # RemoveFileFromKnowledge
  295. ############################
  296. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  297. def remove_file_from_knowledge_by_id(
  298. id: str,
  299. form_data: KnowledgeFileIdForm,
  300. user=Depends(get_verified_user),
  301. ):
  302. knowledge = Knowledges.get_knowledge_by_id(id=id)
  303. if not knowledge:
  304. raise HTTPException(
  305. status_code=status.HTTP_400_BAD_REQUEST,
  306. detail=ERROR_MESSAGES.NOT_FOUND,
  307. )
  308. if knowledge.user_id != user.id and user.role != "admin":
  309. raise HTTPException(
  310. status_code=status.HTTP_400_BAD_REQUEST,
  311. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  312. )
  313. file = Files.get_file_by_id(form_data.file_id)
  314. if not file:
  315. raise HTTPException(
  316. status_code=status.HTTP_400_BAD_REQUEST,
  317. detail=ERROR_MESSAGES.NOT_FOUND,
  318. )
  319. # Remove content from the vector database
  320. VECTOR_DB_CLIENT.delete(
  321. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  322. )
  323. result = VECTOR_DB_CLIENT.query(
  324. collection_name=knowledge.id,
  325. filter={"file_id": form_data.file_id},
  326. )
  327. Files.delete_file_by_id(form_data.file_id)
  328. if knowledge:
  329. data = knowledge.data or {}
  330. file_ids = data.get("file_ids", [])
  331. if form_data.file_id in file_ids:
  332. file_ids.remove(form_data.file_id)
  333. data["file_ids"] = file_ids
  334. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  335. if knowledge:
  336. files = Files.get_files_by_ids(file_ids)
  337. return KnowledgeFilesResponse(
  338. **knowledge.model_dump(),
  339. files=files,
  340. )
  341. else:
  342. raise HTTPException(
  343. status_code=status.HTTP_400_BAD_REQUEST,
  344. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  345. )
  346. else:
  347. raise HTTPException(
  348. status_code=status.HTTP_400_BAD_REQUEST,
  349. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  350. )
  351. else:
  352. raise HTTPException(
  353. status_code=status.HTTP_400_BAD_REQUEST,
  354. detail=ERROR_MESSAGES.NOT_FOUND,
  355. )
  356. ############################
  357. # DeleteKnowledgeById
  358. ############################
  359. @router.delete("/{id}/delete", response_model=bool)
  360. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  361. knowledge = Knowledges.get_knowledge_by_id(id=id)
  362. if not knowledge:
  363. raise HTTPException(
  364. status_code=status.HTTP_400_BAD_REQUEST,
  365. detail=ERROR_MESSAGES.NOT_FOUND,
  366. )
  367. if knowledge.user_id != user.id and user.role != "admin":
  368. raise HTTPException(
  369. status_code=status.HTTP_400_BAD_REQUEST,
  370. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  371. )
  372. try:
  373. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  374. except Exception as e:
  375. log.debug(e)
  376. pass
  377. result = Knowledges.delete_knowledge_by_id(id=id)
  378. return result
  379. ############################
  380. # ResetKnowledgeById
  381. ############################
  382. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  383. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  384. knowledge = Knowledges.get_knowledge_by_id(id=id)
  385. if not knowledge:
  386. raise HTTPException(
  387. status_code=status.HTTP_400_BAD_REQUEST,
  388. detail=ERROR_MESSAGES.NOT_FOUND,
  389. )
  390. if knowledge.user_id != user.id and user.role != "admin":
  391. raise HTTPException(
  392. status_code=status.HTTP_400_BAD_REQUEST,
  393. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  394. )
  395. try:
  396. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  397. except Exception as e:
  398. log.debug(e)
  399. pass
  400. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  401. return knowledge