knowledge.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417
  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.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_by_id(
  51. id=knowledge_base.id, form_data=KnowledgeUpdateForm(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_by_id(
  86. id=knowledge_base.id, form_data=KnowledgeUpdateForm(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. form_data: KnowledgeForm, user=Depends(get_verified_user)
  100. ):
  101. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  102. if knowledge:
  103. return knowledge
  104. else:
  105. raise HTTPException(
  106. status_code=status.HTTP_400_BAD_REQUEST,
  107. detail=ERROR_MESSAGES.FILE_EXISTS,
  108. )
  109. ############################
  110. # GetKnowledgeById
  111. ############################
  112. class KnowledgeFilesResponse(KnowledgeResponse):
  113. files: list[FileModel]
  114. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  115. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  116. knowledge = Knowledges.get_knowledge_by_id(id=id)
  117. if knowledge:
  118. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  119. files = Files.get_files_by_ids(file_ids)
  120. return KnowledgeFilesResponse(
  121. **knowledge.model_dump(),
  122. files=files,
  123. )
  124. else:
  125. raise HTTPException(
  126. status_code=status.HTTP_401_UNAUTHORIZED,
  127. detail=ERROR_MESSAGES.NOT_FOUND,
  128. )
  129. ############################
  130. # UpdateKnowledgeById
  131. ############################
  132. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  133. async def update_knowledge_by_id(
  134. id: str,
  135. form_data: KnowledgeUpdateForm,
  136. user=Depends(get_verified_user),
  137. ):
  138. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  139. if knowledge:
  140. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  141. files = Files.get_files_by_ids(file_ids)
  142. return KnowledgeFilesResponse(
  143. **knowledge.model_dump(),
  144. files=files,
  145. )
  146. else:
  147. raise HTTPException(
  148. status_code=status.HTTP_400_BAD_REQUEST,
  149. detail=ERROR_MESSAGES.ID_TAKEN,
  150. )
  151. ############################
  152. # AddFileToKnowledge
  153. ############################
  154. class KnowledgeFileIdForm(BaseModel):
  155. file_id: str
  156. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  157. def add_file_to_knowledge_by_id(
  158. id: str,
  159. form_data: KnowledgeFileIdForm,
  160. user=Depends(get_verified_user),
  161. ):
  162. knowledge = Knowledges.get_knowledge_by_id(id=id)
  163. file = Files.get_file_by_id(form_data.file_id)
  164. if not file:
  165. raise HTTPException(
  166. status_code=status.HTTP_400_BAD_REQUEST,
  167. detail=ERROR_MESSAGES.NOT_FOUND,
  168. )
  169. if not file.data:
  170. raise HTTPException(
  171. status_code=status.HTTP_400_BAD_REQUEST,
  172. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  173. )
  174. # Add content to the vector database
  175. try:
  176. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  177. except Exception as e:
  178. log.debug(e)
  179. raise HTTPException(
  180. status_code=status.HTTP_400_BAD_REQUEST,
  181. detail=str(e),
  182. )
  183. if knowledge:
  184. data = knowledge.data or {}
  185. file_ids = data.get("file_ids", [])
  186. if form_data.file_id not in file_ids:
  187. file_ids.append(form_data.file_id)
  188. data["file_ids"] = file_ids
  189. knowledge = Knowledges.update_knowledge_by_id(
  190. id=id, form_data=KnowledgeUpdateForm(data=data)
  191. )
  192. if knowledge:
  193. files = Files.get_files_by_ids(file_ids)
  194. return KnowledgeFilesResponse(
  195. **knowledge.model_dump(),
  196. files=files,
  197. )
  198. else:
  199. raise HTTPException(
  200. status_code=status.HTTP_400_BAD_REQUEST,
  201. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  202. )
  203. else:
  204. raise HTTPException(
  205. status_code=status.HTTP_400_BAD_REQUEST,
  206. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  207. )
  208. else:
  209. raise HTTPException(
  210. status_code=status.HTTP_400_BAD_REQUEST,
  211. detail=ERROR_MESSAGES.NOT_FOUND,
  212. )
  213. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  214. def update_file_from_knowledge_by_id(
  215. id: str,
  216. form_data: KnowledgeFileIdForm,
  217. user=Depends(get_verified_user),
  218. ):
  219. knowledge = Knowledges.get_knowledge_by_id(id=id)
  220. file = Files.get_file_by_id(form_data.file_id)
  221. if not file:
  222. raise HTTPException(
  223. status_code=status.HTTP_400_BAD_REQUEST,
  224. detail=ERROR_MESSAGES.NOT_FOUND,
  225. )
  226. # Remove content from the vector database
  227. VECTOR_DB_CLIENT.delete(
  228. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  229. )
  230. # Add content to the vector database
  231. try:
  232. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  233. except Exception as e:
  234. raise HTTPException(
  235. status_code=status.HTTP_400_BAD_REQUEST,
  236. detail=str(e),
  237. )
  238. if knowledge:
  239. data = knowledge.data or {}
  240. file_ids = data.get("file_ids", [])
  241. files = Files.get_files_by_ids(file_ids)
  242. return KnowledgeFilesResponse(
  243. **knowledge.model_dump(),
  244. files=files,
  245. )
  246. else:
  247. raise HTTPException(
  248. status_code=status.HTTP_400_BAD_REQUEST,
  249. detail=ERROR_MESSAGES.NOT_FOUND,
  250. )
  251. ############################
  252. # RemoveFileFromKnowledge
  253. ############################
  254. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  255. def remove_file_from_knowledge_by_id(
  256. id: str,
  257. form_data: KnowledgeFileIdForm,
  258. user=Depends(get_verified_user),
  259. ):
  260. knowledge = Knowledges.get_knowledge_by_id(id=id)
  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. result = VECTOR_DB_CLIENT.query(
  272. collection_name=knowledge.id,
  273. filter={"file_id": form_data.file_id},
  274. )
  275. Files.delete_file_by_id(form_data.file_id)
  276. if knowledge:
  277. data = knowledge.data or {}
  278. file_ids = data.get("file_ids", [])
  279. if form_data.file_id in file_ids:
  280. file_ids.remove(form_data.file_id)
  281. data["file_ids"] = file_ids
  282. knowledge = Knowledges.update_knowledge_by_id(
  283. id=id, form_data=KnowledgeUpdateForm(data=data)
  284. )
  285. if knowledge:
  286. files = Files.get_files_by_ids(file_ids)
  287. return KnowledgeFilesResponse(
  288. **knowledge.model_dump(),
  289. files=files,
  290. )
  291. else:
  292. raise HTTPException(
  293. status_code=status.HTTP_400_BAD_REQUEST,
  294. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  295. )
  296. else:
  297. raise HTTPException(
  298. status_code=status.HTTP_400_BAD_REQUEST,
  299. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  300. )
  301. else:
  302. raise HTTPException(
  303. status_code=status.HTTP_400_BAD_REQUEST,
  304. detail=ERROR_MESSAGES.NOT_FOUND,
  305. )
  306. ############################
  307. # ResetKnowledgeById
  308. ############################
  309. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  310. async def reset_knowledge_by_id(id: str, user=Depends(get_admin_user)):
  311. try:
  312. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  313. except Exception as e:
  314. log.debug(e)
  315. pass
  316. knowledge = Knowledges.update_knowledge_by_id(
  317. id=id, form_data=KnowledgeUpdateForm(data={"file_ids": []})
  318. )
  319. return knowledge
  320. ############################
  321. # DeleteKnowledgeById
  322. ############################
  323. @router.delete("/{id}/delete", response_model=bool)
  324. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  325. try:
  326. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  327. except Exception as e:
  328. log.debug(e)
  329. pass
  330. result = Knowledges.delete_knowledge_by_id(id=id)
  331. return result