knowledge.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381
  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. # GetKnowledgeItems
  23. ############################
  24. @router.get(
  25. "/", response_model=Optional[Union[list[KnowledgeResponse], KnowledgeResponse]]
  26. )
  27. async def get_knowledge_items(
  28. id: Optional[str] = None, user=Depends(get_verified_user)
  29. ):
  30. if id:
  31. knowledge = Knowledges.get_knowledge_by_id(id=id)
  32. if knowledge:
  33. return knowledge
  34. else:
  35. raise HTTPException(
  36. status_code=status.HTTP_401_UNAUTHORIZED,
  37. detail=ERROR_MESSAGES.NOT_FOUND,
  38. )
  39. else:
  40. knowledge_bases = []
  41. for knowledge in Knowledges.get_knowledge_items():
  42. files = []
  43. if knowledge.data:
  44. files = Files.get_file_metadatas_by_ids(
  45. knowledge.data.get("file_ids", [])
  46. )
  47. # Check if all files exist
  48. if len(files) != len(knowledge.data.get("file_ids", [])):
  49. missing_files = list(
  50. set(knowledge.data.get("file_ids", []))
  51. - set([file.id for file in files])
  52. )
  53. if missing_files:
  54. data = knowledge.data or {}
  55. file_ids = data.get("file_ids", [])
  56. for missing_file in missing_files:
  57. file_ids.remove(missing_file)
  58. data["file_ids"] = file_ids
  59. Knowledges.update_knowledge_by_id(
  60. id=knowledge.id, form_data=KnowledgeUpdateForm(data=data)
  61. )
  62. files = Files.get_file_metadatas_by_ids(file_ids)
  63. knowledge_bases.append(
  64. KnowledgeResponse(
  65. **knowledge.model_dump(),
  66. files=files,
  67. )
  68. )
  69. return knowledge_bases
  70. ############################
  71. # CreateNewKnowledge
  72. ############################
  73. @router.post("/create", response_model=Optional[KnowledgeResponse])
  74. async def create_new_knowledge(form_data: KnowledgeForm, user=Depends(get_admin_user)):
  75. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  76. if knowledge:
  77. return knowledge
  78. else:
  79. raise HTTPException(
  80. status_code=status.HTTP_400_BAD_REQUEST,
  81. detail=ERROR_MESSAGES.FILE_EXISTS,
  82. )
  83. ############################
  84. # GetKnowledgeById
  85. ############################
  86. class KnowledgeFilesResponse(KnowledgeResponse):
  87. files: list[FileModel]
  88. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  89. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  90. knowledge = Knowledges.get_knowledge_by_id(id=id)
  91. if knowledge:
  92. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  93. files = Files.get_files_by_ids(file_ids)
  94. return KnowledgeFilesResponse(
  95. **knowledge.model_dump(),
  96. files=files,
  97. )
  98. else:
  99. raise HTTPException(
  100. status_code=status.HTTP_401_UNAUTHORIZED,
  101. detail=ERROR_MESSAGES.NOT_FOUND,
  102. )
  103. ############################
  104. # UpdateKnowledgeById
  105. ############################
  106. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  107. async def update_knowledge_by_id(
  108. id: str,
  109. form_data: KnowledgeUpdateForm,
  110. user=Depends(get_admin_user),
  111. ):
  112. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  113. if knowledge:
  114. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  115. files = Files.get_files_by_ids(file_ids)
  116. return KnowledgeFilesResponse(
  117. **knowledge.model_dump(),
  118. files=files,
  119. )
  120. else:
  121. raise HTTPException(
  122. status_code=status.HTTP_400_BAD_REQUEST,
  123. detail=ERROR_MESSAGES.ID_TAKEN,
  124. )
  125. ############################
  126. # AddFileToKnowledge
  127. ############################
  128. class KnowledgeFileIdForm(BaseModel):
  129. file_id: str
  130. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  131. def add_file_to_knowledge_by_id(
  132. id: str,
  133. form_data: KnowledgeFileIdForm,
  134. user=Depends(get_admin_user),
  135. ):
  136. knowledge = Knowledges.get_knowledge_by_id(id=id)
  137. file = Files.get_file_by_id(form_data.file_id)
  138. if not file:
  139. raise HTTPException(
  140. status_code=status.HTTP_400_BAD_REQUEST,
  141. detail=ERROR_MESSAGES.NOT_FOUND,
  142. )
  143. if not file.data:
  144. raise HTTPException(
  145. status_code=status.HTTP_400_BAD_REQUEST,
  146. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  147. )
  148. # Add content to the vector database
  149. try:
  150. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  151. except Exception as e:
  152. log.debug(e)
  153. raise HTTPException(
  154. status_code=status.HTTP_400_BAD_REQUEST,
  155. detail=str(e),
  156. )
  157. if knowledge:
  158. data = knowledge.data or {}
  159. file_ids = data.get("file_ids", [])
  160. if form_data.file_id not in file_ids:
  161. file_ids.append(form_data.file_id)
  162. data["file_ids"] = file_ids
  163. knowledge = Knowledges.update_knowledge_by_id(
  164. id=id, form_data=KnowledgeUpdateForm(data=data)
  165. )
  166. if knowledge:
  167. files = Files.get_files_by_ids(file_ids)
  168. return KnowledgeFilesResponse(
  169. **knowledge.model_dump(),
  170. files=files,
  171. )
  172. else:
  173. raise HTTPException(
  174. status_code=status.HTTP_400_BAD_REQUEST,
  175. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  176. )
  177. else:
  178. raise HTTPException(
  179. status_code=status.HTTP_400_BAD_REQUEST,
  180. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  181. )
  182. else:
  183. raise HTTPException(
  184. status_code=status.HTTP_400_BAD_REQUEST,
  185. detail=ERROR_MESSAGES.NOT_FOUND,
  186. )
  187. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  188. def update_file_from_knowledge_by_id(
  189. id: str,
  190. form_data: KnowledgeFileIdForm,
  191. user=Depends(get_admin_user),
  192. ):
  193. knowledge = Knowledges.get_knowledge_by_id(id=id)
  194. file = Files.get_file_by_id(form_data.file_id)
  195. if not file:
  196. raise HTTPException(
  197. status_code=status.HTTP_400_BAD_REQUEST,
  198. detail=ERROR_MESSAGES.NOT_FOUND,
  199. )
  200. # Remove content from the vector database
  201. VECTOR_DB_CLIENT.delete(
  202. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  203. )
  204. # Add content to the vector database
  205. try:
  206. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  207. except Exception as e:
  208. raise HTTPException(
  209. status_code=status.HTTP_400_BAD_REQUEST,
  210. detail=str(e),
  211. )
  212. if knowledge:
  213. data = knowledge.data or {}
  214. file_ids = data.get("file_ids", [])
  215. files = Files.get_files_by_ids(file_ids)
  216. return KnowledgeFilesResponse(
  217. **knowledge.model_dump(),
  218. files=files,
  219. )
  220. else:
  221. raise HTTPException(
  222. status_code=status.HTTP_400_BAD_REQUEST,
  223. detail=ERROR_MESSAGES.NOT_FOUND,
  224. )
  225. ############################
  226. # RemoveFileFromKnowledge
  227. ############################
  228. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  229. def remove_file_from_knowledge_by_id(
  230. id: str,
  231. form_data: KnowledgeFileIdForm,
  232. user=Depends(get_admin_user),
  233. ):
  234. knowledge = Knowledges.get_knowledge_by_id(id=id)
  235. file = Files.get_file_by_id(form_data.file_id)
  236. if not file:
  237. raise HTTPException(
  238. status_code=status.HTTP_400_BAD_REQUEST,
  239. detail=ERROR_MESSAGES.NOT_FOUND,
  240. )
  241. # Remove content from the vector database
  242. VECTOR_DB_CLIENT.delete(
  243. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  244. )
  245. result = VECTOR_DB_CLIENT.query(
  246. collection_name=knowledge.id,
  247. filter={"file_id": form_data.file_id},
  248. )
  249. Files.delete_file_by_id(form_data.file_id)
  250. if knowledge:
  251. data = knowledge.data or {}
  252. file_ids = data.get("file_ids", [])
  253. if form_data.file_id in file_ids:
  254. file_ids.remove(form_data.file_id)
  255. data["file_ids"] = file_ids
  256. knowledge = Knowledges.update_knowledge_by_id(
  257. id=id, form_data=KnowledgeUpdateForm(data=data)
  258. )
  259. if knowledge:
  260. files = Files.get_files_by_ids(file_ids)
  261. return KnowledgeFilesResponse(
  262. **knowledge.model_dump(),
  263. files=files,
  264. )
  265. else:
  266. raise HTTPException(
  267. status_code=status.HTTP_400_BAD_REQUEST,
  268. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  269. )
  270. else:
  271. raise HTTPException(
  272. status_code=status.HTTP_400_BAD_REQUEST,
  273. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  274. )
  275. else:
  276. raise HTTPException(
  277. status_code=status.HTTP_400_BAD_REQUEST,
  278. detail=ERROR_MESSAGES.NOT_FOUND,
  279. )
  280. ############################
  281. # ResetKnowledgeById
  282. ############################
  283. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  284. async def reset_knowledge_by_id(id: str, user=Depends(get_admin_user)):
  285. try:
  286. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  287. except Exception as e:
  288. log.debug(e)
  289. pass
  290. knowledge = Knowledges.update_knowledge_by_id(
  291. id=id, form_data=KnowledgeUpdateForm(data={"file_ids": []})
  292. )
  293. return knowledge
  294. ############################
  295. # DeleteKnowledgeById
  296. ############################
  297. @router.delete("/{id}/delete", response_model=bool)
  298. async def delete_knowledge_by_id(id: str, user=Depends(get_admin_user)):
  299. try:
  300. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  301. except Exception as e:
  302. log.debug(e)
  303. pass
  304. result = Knowledges.delete_knowledge_by_id(id=id)
  305. return result