knowledge.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504
  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. )
  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 process_file, ProcessFileForm
  14. from open_webui.constants import ERROR_MESSAGES
  15. from open_webui.utils.utils import get_admin_user, 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[KnowledgeResponse])
  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. for knowledge_base in knowledge_bases:
  33. files = []
  34. if knowledge_base.data:
  35. files = Files.get_file_metadatas_by_ids(
  36. knowledge_base.data.get("file_ids", [])
  37. )
  38. # Check if all files exist
  39. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  40. missing_files = list(
  41. set(knowledge_base.data.get("file_ids", []))
  42. - set([file.id for file in files])
  43. )
  44. if missing_files:
  45. data = knowledge_base.data or {}
  46. file_ids = data.get("file_ids", [])
  47. for missing_file in missing_files:
  48. file_ids.remove(missing_file)
  49. data["file_ids"] = file_ids
  50. Knowledges.update_knowledge_data_by_id(
  51. id=knowledge_base.id, data=data
  52. )
  53. files = Files.get_file_metadatas_by_ids(file_ids)
  54. knowledge_base = KnowledgeResponse(
  55. **knowledge_base.model_dump(),
  56. files=files,
  57. )
  58. return knowledge_bases
  59. @router.get("/list", response_model=list[KnowledgeResponse])
  60. async def get_knowledge_list(user=Depends(get_verified_user)):
  61. knowledge_bases = []
  62. if user.role == "admin":
  63. knowledge_bases = Knowledges.get_knowledge_bases()
  64. else:
  65. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  66. # Get files for each knowledge base
  67. for knowledge_base in knowledge_bases:
  68. files = []
  69. if knowledge_base.data:
  70. files = Files.get_file_metadatas_by_ids(
  71. knowledge_base.data.get("file_ids", [])
  72. )
  73. # Check if all files exist
  74. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  75. missing_files = list(
  76. set(knowledge_base.data.get("file_ids", []))
  77. - set([file.id for file in files])
  78. )
  79. if missing_files:
  80. data = knowledge_base.data or {}
  81. file_ids = data.get("file_ids", [])
  82. for missing_file in missing_files:
  83. file_ids.remove(missing_file)
  84. data["file_ids"] = file_ids
  85. Knowledges.update_knowledge_data_by_id(
  86. id=knowledge_base.id, data=data
  87. )
  88. files = Files.get_file_metadatas_by_ids(file_ids)
  89. knowledge_base = KnowledgeResponse(
  90. **knowledge_base.model_dump(),
  91. files=files,
  92. )
  93. return knowledge_bases
  94. ############################
  95. # CreateNewKnowledge
  96. ############################
  97. @router.post("/create", response_model=Optional[KnowledgeResponse])
  98. async def create_new_knowledge(
  99. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  100. ):
  101. if user.role != "admin" and not has_permission(
  102. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  103. ):
  104. raise HTTPException(
  105. status_code=status.HTTP_401_UNAUTHORIZED,
  106. detail=ERROR_MESSAGES.UNAUTHORIZED,
  107. )
  108. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  109. if knowledge:
  110. return knowledge
  111. else:
  112. raise HTTPException(
  113. status_code=status.HTTP_400_BAD_REQUEST,
  114. detail=ERROR_MESSAGES.FILE_EXISTS,
  115. )
  116. ############################
  117. # GetKnowledgeById
  118. ############################
  119. class KnowledgeFilesResponse(KnowledgeResponse):
  120. files: list[FileModel]
  121. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  122. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  123. knowledge = Knowledges.get_knowledge_by_id(id=id)
  124. if knowledge:
  125. if (
  126. user.role == "admin"
  127. or knowledge.user_id == user.id
  128. or has_access(user.id, "read", knowledge.access_control)
  129. ):
  130. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  131. files = Files.get_files_by_ids(file_ids)
  132. return KnowledgeFilesResponse(
  133. **knowledge.model_dump(),
  134. files=files,
  135. )
  136. else:
  137. raise HTTPException(
  138. status_code=status.HTTP_401_UNAUTHORIZED,
  139. detail=ERROR_MESSAGES.NOT_FOUND,
  140. )
  141. ############################
  142. # UpdateKnowledgeById
  143. ############################
  144. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  145. async def update_knowledge_by_id(
  146. id: str,
  147. form_data: KnowledgeForm,
  148. user=Depends(get_verified_user),
  149. ):
  150. knowledge = Knowledges.get_knowledge_by_id(id=id)
  151. if not knowledge:
  152. raise HTTPException(
  153. status_code=status.HTTP_400_BAD_REQUEST,
  154. detail=ERROR_MESSAGES.NOT_FOUND,
  155. )
  156. if knowledge.user_id != user.id and user.role != "admin":
  157. raise HTTPException(
  158. status_code=status.HTTP_400_BAD_REQUEST,
  159. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  160. )
  161. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  162. if knowledge:
  163. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  164. files = Files.get_files_by_ids(file_ids)
  165. return KnowledgeFilesResponse(
  166. **knowledge.model_dump(),
  167. files=files,
  168. )
  169. else:
  170. raise HTTPException(
  171. status_code=status.HTTP_400_BAD_REQUEST,
  172. detail=ERROR_MESSAGES.ID_TAKEN,
  173. )
  174. ############################
  175. # AddFileToKnowledge
  176. ############################
  177. class KnowledgeFileIdForm(BaseModel):
  178. file_id: str
  179. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  180. def add_file_to_knowledge_by_id(
  181. id: str,
  182. form_data: KnowledgeFileIdForm,
  183. user=Depends(get_verified_user),
  184. ):
  185. knowledge = Knowledges.get_knowledge_by_id(id=id)
  186. if not knowledge:
  187. raise HTTPException(
  188. status_code=status.HTTP_400_BAD_REQUEST,
  189. detail=ERROR_MESSAGES.NOT_FOUND,
  190. )
  191. if knowledge.user_id != user.id and user.role != "admin":
  192. raise HTTPException(
  193. status_code=status.HTTP_400_BAD_REQUEST,
  194. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  195. )
  196. file = Files.get_file_by_id(form_data.file_id)
  197. if not file:
  198. raise HTTPException(
  199. status_code=status.HTTP_400_BAD_REQUEST,
  200. detail=ERROR_MESSAGES.NOT_FOUND,
  201. )
  202. if not file.data:
  203. raise HTTPException(
  204. status_code=status.HTTP_400_BAD_REQUEST,
  205. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  206. )
  207. # Add content to the vector database
  208. try:
  209. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  210. except Exception as e:
  211. log.debug(e)
  212. raise HTTPException(
  213. status_code=status.HTTP_400_BAD_REQUEST,
  214. detail=str(e),
  215. )
  216. if knowledge:
  217. data = knowledge.data or {}
  218. file_ids = data.get("file_ids", [])
  219. if form_data.file_id not in file_ids:
  220. file_ids.append(form_data.file_id)
  221. data["file_ids"] = file_ids
  222. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  223. if knowledge:
  224. files = Files.get_files_by_ids(file_ids)
  225. return KnowledgeFilesResponse(
  226. **knowledge.model_dump(),
  227. files=files,
  228. )
  229. else:
  230. raise HTTPException(
  231. status_code=status.HTTP_400_BAD_REQUEST,
  232. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  233. )
  234. else:
  235. raise HTTPException(
  236. status_code=status.HTTP_400_BAD_REQUEST,
  237. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  238. )
  239. else:
  240. raise HTTPException(
  241. status_code=status.HTTP_400_BAD_REQUEST,
  242. detail=ERROR_MESSAGES.NOT_FOUND,
  243. )
  244. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  245. def update_file_from_knowledge_by_id(
  246. id: str,
  247. form_data: KnowledgeFileIdForm,
  248. user=Depends(get_verified_user),
  249. ):
  250. knowledge = Knowledges.get_knowledge_by_id(id=id)
  251. if not knowledge:
  252. raise HTTPException(
  253. status_code=status.HTTP_400_BAD_REQUEST,
  254. detail=ERROR_MESSAGES.NOT_FOUND,
  255. )
  256. if knowledge.user_id != user.id and user.role != "admin":
  257. raise HTTPException(
  258. status_code=status.HTTP_400_BAD_REQUEST,
  259. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  260. )
  261. file = Files.get_file_by_id(form_data.file_id)
  262. if not file:
  263. raise HTTPException(
  264. status_code=status.HTTP_400_BAD_REQUEST,
  265. detail=ERROR_MESSAGES.NOT_FOUND,
  266. )
  267. # Remove content from the vector database
  268. VECTOR_DB_CLIENT.delete(
  269. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  270. )
  271. # Add content to the vector database
  272. try:
  273. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  274. except Exception as e:
  275. raise HTTPException(
  276. status_code=status.HTTP_400_BAD_REQUEST,
  277. detail=str(e),
  278. )
  279. if knowledge:
  280. data = knowledge.data or {}
  281. file_ids = data.get("file_ids", [])
  282. files = Files.get_files_by_ids(file_ids)
  283. return KnowledgeFilesResponse(
  284. **knowledge.model_dump(),
  285. files=files,
  286. )
  287. else:
  288. raise HTTPException(
  289. status_code=status.HTTP_400_BAD_REQUEST,
  290. detail=ERROR_MESSAGES.NOT_FOUND,
  291. )
  292. ############################
  293. # RemoveFileFromKnowledge
  294. ############################
  295. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  296. def remove_file_from_knowledge_by_id(
  297. id: str,
  298. form_data: KnowledgeFileIdForm,
  299. user=Depends(get_verified_user),
  300. ):
  301. knowledge = Knowledges.get_knowledge_by_id(id=id)
  302. if not knowledge:
  303. raise HTTPException(
  304. status_code=status.HTTP_400_BAD_REQUEST,
  305. detail=ERROR_MESSAGES.NOT_FOUND,
  306. )
  307. if knowledge.user_id != user.id and user.role != "admin":
  308. raise HTTPException(
  309. status_code=status.HTTP_400_BAD_REQUEST,
  310. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  311. )
  312. file = Files.get_file_by_id(form_data.file_id)
  313. if not file:
  314. raise HTTPException(
  315. status_code=status.HTTP_400_BAD_REQUEST,
  316. detail=ERROR_MESSAGES.NOT_FOUND,
  317. )
  318. # Remove content from the vector database
  319. VECTOR_DB_CLIENT.delete(
  320. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  321. )
  322. result = VECTOR_DB_CLIENT.query(
  323. collection_name=knowledge.id,
  324. filter={"file_id": form_data.file_id},
  325. )
  326. Files.delete_file_by_id(form_data.file_id)
  327. if knowledge:
  328. data = knowledge.data or {}
  329. file_ids = data.get("file_ids", [])
  330. if form_data.file_id in file_ids:
  331. file_ids.remove(form_data.file_id)
  332. data["file_ids"] = file_ids
  333. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  334. if knowledge:
  335. files = Files.get_files_by_ids(file_ids)
  336. return KnowledgeFilesResponse(
  337. **knowledge.model_dump(),
  338. files=files,
  339. )
  340. else:
  341. raise HTTPException(
  342. status_code=status.HTTP_400_BAD_REQUEST,
  343. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  344. )
  345. else:
  346. raise HTTPException(
  347. status_code=status.HTTP_400_BAD_REQUEST,
  348. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  349. )
  350. else:
  351. raise HTTPException(
  352. status_code=status.HTTP_400_BAD_REQUEST,
  353. detail=ERROR_MESSAGES.NOT_FOUND,
  354. )
  355. ############################
  356. # DeleteKnowledgeById
  357. ############################
  358. @router.delete("/{id}/delete", response_model=bool)
  359. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  360. knowledge = Knowledges.get_knowledge_by_id(id=id)
  361. if not knowledge:
  362. raise HTTPException(
  363. status_code=status.HTTP_400_BAD_REQUEST,
  364. detail=ERROR_MESSAGES.NOT_FOUND,
  365. )
  366. if knowledge.user_id != user.id and user.role != "admin":
  367. raise HTTPException(
  368. status_code=status.HTTP_400_BAD_REQUEST,
  369. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  370. )
  371. try:
  372. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  373. except Exception as e:
  374. log.debug(e)
  375. pass
  376. result = Knowledges.delete_knowledge_by_id(id=id)
  377. return result
  378. ############################
  379. # ResetKnowledgeById
  380. ############################
  381. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  382. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  383. knowledge = Knowledges.get_knowledge_by_id(id=id)
  384. if not knowledge:
  385. raise HTTPException(
  386. status_code=status.HTTP_400_BAD_REQUEST,
  387. detail=ERROR_MESSAGES.NOT_FOUND,
  388. )
  389. if knowledge.user_id != user.id and user.role != "admin":
  390. raise HTTPException(
  391. status_code=status.HTTP_400_BAD_REQUEST,
  392. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  393. )
  394. try:
  395. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  396. except Exception as e:
  397. log.debug(e)
  398. pass
  399. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  400. return knowledge