knowledge.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502
  1. import json
  2. from typing import Optional, Union
  3. from pydantic import BaseModel
  4. from fastapi import APIRouter, Depends, HTTPException, status
  5. import logging
  6. from open_webui.apps.webui.models.knowledge import (
  7. Knowledges,
  8. KnowledgeUpdateForm,
  9. KnowledgeForm,
  10. KnowledgeResponse,
  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
  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[KnowledgeResponse])
  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_by_id(
  52. id=knowledge_base.id, form_data=KnowledgeUpdateForm(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[KnowledgeResponse])
  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_by_id(
  87. id=knowledge_base.id, form_data=KnowledgeUpdateForm(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. form_data: KnowledgeForm, user=Depends(get_verified_user)
  101. ):
  102. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  103. if knowledge:
  104. return knowledge
  105. else:
  106. raise HTTPException(
  107. status_code=status.HTTP_400_BAD_REQUEST,
  108. detail=ERROR_MESSAGES.FILE_EXISTS,
  109. )
  110. ############################
  111. # GetKnowledgeById
  112. ############################
  113. class KnowledgeFilesResponse(KnowledgeResponse):
  114. files: list[FileModel]
  115. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  116. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  117. knowledge = Knowledges.get_knowledge_by_id(id=id)
  118. if knowledge:
  119. if (
  120. user.role == "admin"
  121. or knowledge.user_id == user.id
  122. or has_access(user.id, "read", knowledge.access_control)
  123. ):
  124. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  125. files = Files.get_files_by_ids(file_ids)
  126. return KnowledgeFilesResponse(
  127. **knowledge.model_dump(),
  128. files=files,
  129. )
  130. else:
  131. raise HTTPException(
  132. status_code=status.HTTP_401_UNAUTHORIZED,
  133. detail=ERROR_MESSAGES.NOT_FOUND,
  134. )
  135. ############################
  136. # UpdateKnowledgeById
  137. ############################
  138. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  139. async def update_knowledge_by_id(
  140. id: str,
  141. form_data: KnowledgeUpdateForm,
  142. user=Depends(get_verified_user),
  143. ):
  144. knowledge = Knowledges.get_knowledge_by_id(id=id)
  145. if not knowledge:
  146. raise HTTPException(
  147. status_code=status.HTTP_400_BAD_REQUEST,
  148. detail=ERROR_MESSAGES.NOT_FOUND,
  149. )
  150. if knowledge.user_id != user.id and user.role != "admin":
  151. raise HTTPException(
  152. status_code=status.HTTP_400_BAD_REQUEST,
  153. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  154. )
  155. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  156. if knowledge:
  157. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  158. files = Files.get_files_by_ids(file_ids)
  159. return KnowledgeFilesResponse(
  160. **knowledge.model_dump(),
  161. files=files,
  162. )
  163. else:
  164. raise HTTPException(
  165. status_code=status.HTTP_400_BAD_REQUEST,
  166. detail=ERROR_MESSAGES.ID_TAKEN,
  167. )
  168. ############################
  169. # AddFileToKnowledge
  170. ############################
  171. class KnowledgeFileIdForm(BaseModel):
  172. file_id: str
  173. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  174. def add_file_to_knowledge_by_id(
  175. id: str,
  176. form_data: KnowledgeFileIdForm,
  177. user=Depends(get_verified_user),
  178. ):
  179. knowledge = Knowledges.get_knowledge_by_id(id=id)
  180. if not knowledge:
  181. raise HTTPException(
  182. status_code=status.HTTP_400_BAD_REQUEST,
  183. detail=ERROR_MESSAGES.NOT_FOUND,
  184. )
  185. if knowledge.user_id != user.id and user.role != "admin":
  186. raise HTTPException(
  187. status_code=status.HTTP_400_BAD_REQUEST,
  188. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  189. )
  190. file = Files.get_file_by_id(form_data.file_id)
  191. if not file:
  192. raise HTTPException(
  193. status_code=status.HTTP_400_BAD_REQUEST,
  194. detail=ERROR_MESSAGES.NOT_FOUND,
  195. )
  196. if not file.data:
  197. raise HTTPException(
  198. status_code=status.HTTP_400_BAD_REQUEST,
  199. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  200. )
  201. # Add content to the vector database
  202. try:
  203. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  204. except Exception as e:
  205. log.debug(e)
  206. raise HTTPException(
  207. status_code=status.HTTP_400_BAD_REQUEST,
  208. detail=str(e),
  209. )
  210. if knowledge:
  211. data = knowledge.data or {}
  212. file_ids = data.get("file_ids", [])
  213. if form_data.file_id not in file_ids:
  214. file_ids.append(form_data.file_id)
  215. data["file_ids"] = file_ids
  216. knowledge = Knowledges.update_knowledge_by_id(
  217. id=id, form_data=KnowledgeUpdateForm(data=data)
  218. )
  219. if knowledge:
  220. files = Files.get_files_by_ids(file_ids)
  221. return KnowledgeFilesResponse(
  222. **knowledge.model_dump(),
  223. files=files,
  224. )
  225. else:
  226. raise HTTPException(
  227. status_code=status.HTTP_400_BAD_REQUEST,
  228. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  229. )
  230. else:
  231. raise HTTPException(
  232. status_code=status.HTTP_400_BAD_REQUEST,
  233. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  234. )
  235. else:
  236. raise HTTPException(
  237. status_code=status.HTTP_400_BAD_REQUEST,
  238. detail=ERROR_MESSAGES.NOT_FOUND,
  239. )
  240. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  241. def update_file_from_knowledge_by_id(
  242. id: str,
  243. form_data: KnowledgeFileIdForm,
  244. user=Depends(get_verified_user),
  245. ):
  246. knowledge = Knowledges.get_knowledge_by_id(id=id)
  247. if not knowledge:
  248. raise HTTPException(
  249. status_code=status.HTTP_400_BAD_REQUEST,
  250. detail=ERROR_MESSAGES.NOT_FOUND,
  251. )
  252. if knowledge.user_id != user.id and user.role != "admin":
  253. raise HTTPException(
  254. status_code=status.HTTP_400_BAD_REQUEST,
  255. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  256. )
  257. file = Files.get_file_by_id(form_data.file_id)
  258. if not file:
  259. raise HTTPException(
  260. status_code=status.HTTP_400_BAD_REQUEST,
  261. detail=ERROR_MESSAGES.NOT_FOUND,
  262. )
  263. # Remove content from the vector database
  264. VECTOR_DB_CLIENT.delete(
  265. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  266. )
  267. # Add content to the vector database
  268. try:
  269. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  270. except Exception as e:
  271. raise HTTPException(
  272. status_code=status.HTTP_400_BAD_REQUEST,
  273. detail=str(e),
  274. )
  275. if knowledge:
  276. data = knowledge.data or {}
  277. file_ids = data.get("file_ids", [])
  278. files = Files.get_files_by_ids(file_ids)
  279. return KnowledgeFilesResponse(
  280. **knowledge.model_dump(),
  281. files=files,
  282. )
  283. else:
  284. raise HTTPException(
  285. status_code=status.HTTP_400_BAD_REQUEST,
  286. detail=ERROR_MESSAGES.NOT_FOUND,
  287. )
  288. ############################
  289. # RemoveFileFromKnowledge
  290. ############################
  291. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  292. def remove_file_from_knowledge_by_id(
  293. id: str,
  294. form_data: KnowledgeFileIdForm,
  295. user=Depends(get_verified_user),
  296. ):
  297. knowledge = Knowledges.get_knowledge_by_id(id=id)
  298. if not knowledge:
  299. raise HTTPException(
  300. status_code=status.HTTP_400_BAD_REQUEST,
  301. detail=ERROR_MESSAGES.NOT_FOUND,
  302. )
  303. if knowledge.user_id != user.id and user.role != "admin":
  304. raise HTTPException(
  305. status_code=status.HTTP_400_BAD_REQUEST,
  306. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  307. )
  308. file = Files.get_file_by_id(form_data.file_id)
  309. if not file:
  310. raise HTTPException(
  311. status_code=status.HTTP_400_BAD_REQUEST,
  312. detail=ERROR_MESSAGES.NOT_FOUND,
  313. )
  314. # Remove content from the vector database
  315. VECTOR_DB_CLIENT.delete(
  316. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  317. )
  318. result = VECTOR_DB_CLIENT.query(
  319. collection_name=knowledge.id,
  320. filter={"file_id": form_data.file_id},
  321. )
  322. Files.delete_file_by_id(form_data.file_id)
  323. if knowledge:
  324. data = knowledge.data or {}
  325. file_ids = data.get("file_ids", [])
  326. if form_data.file_id in file_ids:
  327. file_ids.remove(form_data.file_id)
  328. data["file_ids"] = file_ids
  329. knowledge = Knowledges.update_knowledge_by_id(
  330. id=id, form_data=KnowledgeUpdateForm(data=data)
  331. )
  332. if knowledge:
  333. files = Files.get_files_by_ids(file_ids)
  334. return KnowledgeFilesResponse(
  335. **knowledge.model_dump(),
  336. files=files,
  337. )
  338. else:
  339. raise HTTPException(
  340. status_code=status.HTTP_400_BAD_REQUEST,
  341. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  342. )
  343. else:
  344. raise HTTPException(
  345. status_code=status.HTTP_400_BAD_REQUEST,
  346. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  347. )
  348. else:
  349. raise HTTPException(
  350. status_code=status.HTTP_400_BAD_REQUEST,
  351. detail=ERROR_MESSAGES.NOT_FOUND,
  352. )
  353. ############################
  354. # DeleteKnowledgeById
  355. ############################
  356. @router.delete("/{id}/delete", response_model=bool)
  357. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  358. knowledge = Knowledges.get_knowledge_by_id(id=id)
  359. if not knowledge:
  360. raise HTTPException(
  361. status_code=status.HTTP_400_BAD_REQUEST,
  362. detail=ERROR_MESSAGES.NOT_FOUND,
  363. )
  364. if knowledge.user_id != user.id and user.role != "admin":
  365. raise HTTPException(
  366. status_code=status.HTTP_400_BAD_REQUEST,
  367. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  368. )
  369. try:
  370. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  371. except Exception as e:
  372. log.debug(e)
  373. pass
  374. result = Knowledges.delete_knowledge_by_id(id=id)
  375. return result
  376. ############################
  377. # ResetKnowledgeById
  378. ############################
  379. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  380. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  381. knowledge = Knowledges.get_knowledge_by_id(id=id)
  382. if not knowledge:
  383. raise HTTPException(
  384. status_code=status.HTTP_400_BAD_REQUEST,
  385. detail=ERROR_MESSAGES.NOT_FOUND,
  386. )
  387. if knowledge.user_id != user.id and user.role != "admin":
  388. raise HTTPException(
  389. status_code=status.HTTP_400_BAD_REQUEST,
  390. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  391. )
  392. try:
  393. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  394. except Exception as e:
  395. log.debug(e)
  396. pass
  397. knowledge = Knowledges.update_knowledge_by_id(
  398. id=id, form_data=KnowledgeUpdateForm(data={"file_ids": []})
  399. )
  400. return knowledge