knowledge.py 9.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329
  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. return [
  41. KnowledgeResponse(**knowledge.model_dump())
  42. for knowledge in Knowledges.get_knowledge_items()
  43. ]
  44. ############################
  45. # CreateNewKnowledge
  46. ############################
  47. @router.post("/create", response_model=Optional[KnowledgeResponse])
  48. async def create_new_knowledge(form_data: KnowledgeForm, user=Depends(get_admin_user)):
  49. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  50. if knowledge:
  51. return knowledge
  52. else:
  53. raise HTTPException(
  54. status_code=status.HTTP_400_BAD_REQUEST,
  55. detail=ERROR_MESSAGES.FILE_EXISTS,
  56. )
  57. ############################
  58. # GetKnowledgeById
  59. ############################
  60. class KnowledgeFilesResponse(KnowledgeResponse):
  61. files: list[FileModel]
  62. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  63. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  64. knowledge = Knowledges.get_knowledge_by_id(id=id)
  65. if knowledge:
  66. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  67. files = Files.get_files_by_ids(file_ids)
  68. return KnowledgeFilesResponse(
  69. **knowledge.model_dump(),
  70. files=files,
  71. )
  72. else:
  73. raise HTTPException(
  74. status_code=status.HTTP_401_UNAUTHORIZED,
  75. detail=ERROR_MESSAGES.NOT_FOUND,
  76. )
  77. ############################
  78. # UpdateKnowledgeById
  79. ############################
  80. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  81. async def update_knowledge_by_id(
  82. id: str,
  83. form_data: KnowledgeUpdateForm,
  84. user=Depends(get_admin_user),
  85. ):
  86. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  87. if knowledge:
  88. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  89. files = Files.get_files_by_ids(file_ids)
  90. return KnowledgeFilesResponse(
  91. **knowledge.model_dump(),
  92. files=files,
  93. )
  94. else:
  95. raise HTTPException(
  96. status_code=status.HTTP_400_BAD_REQUEST,
  97. detail=ERROR_MESSAGES.ID_TAKEN,
  98. )
  99. ############################
  100. # AddFileToKnowledge
  101. ############################
  102. class KnowledgeFileIdForm(BaseModel):
  103. file_id: str
  104. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  105. def add_file_to_knowledge_by_id(
  106. id: str,
  107. form_data: KnowledgeFileIdForm,
  108. user=Depends(get_admin_user),
  109. ):
  110. knowledge = Knowledges.get_knowledge_by_id(id=id)
  111. file = Files.get_file_by_id(form_data.file_id)
  112. if not file:
  113. raise HTTPException(
  114. status_code=status.HTTP_400_BAD_REQUEST,
  115. detail=ERROR_MESSAGES.NOT_FOUND,
  116. )
  117. if not file.data:
  118. raise HTTPException(
  119. status_code=status.HTTP_400_BAD_REQUEST,
  120. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  121. )
  122. # Add content to the vector database
  123. try:
  124. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  125. except Exception as e:
  126. log.debug(e)
  127. raise HTTPException(
  128. status_code=status.HTTP_400_BAD_REQUEST,
  129. detail=str(e),
  130. )
  131. if knowledge:
  132. data = knowledge.data or {}
  133. file_ids = data.get("file_ids", [])
  134. if form_data.file_id not in file_ids:
  135. file_ids.append(form_data.file_id)
  136. data["file_ids"] = file_ids
  137. knowledge = Knowledges.update_knowledge_by_id(
  138. id=id, form_data=KnowledgeUpdateForm(data=data)
  139. )
  140. if knowledge:
  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.DEFAULT("knowledge"),
  150. )
  151. else:
  152. raise HTTPException(
  153. status_code=status.HTTP_400_BAD_REQUEST,
  154. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  155. )
  156. else:
  157. raise HTTPException(
  158. status_code=status.HTTP_400_BAD_REQUEST,
  159. detail=ERROR_MESSAGES.NOT_FOUND,
  160. )
  161. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  162. def update_file_from_knowledge_by_id(
  163. id: str,
  164. form_data: KnowledgeFileIdForm,
  165. user=Depends(get_admin_user),
  166. ):
  167. knowledge = Knowledges.get_knowledge_by_id(id=id)
  168. file = Files.get_file_by_id(form_data.file_id)
  169. if not file:
  170. raise HTTPException(
  171. status_code=status.HTTP_400_BAD_REQUEST,
  172. detail=ERROR_MESSAGES.NOT_FOUND,
  173. )
  174. # Remove content from the vector database
  175. VECTOR_DB_CLIENT.delete(
  176. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  177. )
  178. # Add content to the vector database
  179. try:
  180. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  181. except Exception as e:
  182. raise HTTPException(
  183. status_code=status.HTTP_400_BAD_REQUEST,
  184. detail=str(e),
  185. )
  186. if knowledge:
  187. data = knowledge.data or {}
  188. file_ids = data.get("file_ids", [])
  189. files = Files.get_files_by_ids(file_ids)
  190. return KnowledgeFilesResponse(
  191. **knowledge.model_dump(),
  192. files=files,
  193. )
  194. else:
  195. raise HTTPException(
  196. status_code=status.HTTP_400_BAD_REQUEST,
  197. detail=ERROR_MESSAGES.NOT_FOUND,
  198. )
  199. ############################
  200. # RemoveFileFromKnowledge
  201. ############################
  202. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  203. def remove_file_from_knowledge_by_id(
  204. id: str,
  205. form_data: KnowledgeFileIdForm,
  206. user=Depends(get_admin_user),
  207. ):
  208. knowledge = Knowledges.get_knowledge_by_id(id=id)
  209. file = Files.get_file_by_id(form_data.file_id)
  210. if not file:
  211. raise HTTPException(
  212. status_code=status.HTTP_400_BAD_REQUEST,
  213. detail=ERROR_MESSAGES.NOT_FOUND,
  214. )
  215. # Remove content from the vector database
  216. VECTOR_DB_CLIENT.delete(
  217. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  218. )
  219. result = VECTOR_DB_CLIENT.query(
  220. collection_name=knowledge.id,
  221. filter={"file_id": form_data.file_id},
  222. )
  223. Files.delete_file_by_id(form_data.file_id)
  224. if knowledge:
  225. data = knowledge.data or {}
  226. file_ids = data.get("file_ids", [])
  227. if form_data.file_id in file_ids:
  228. file_ids.remove(form_data.file_id)
  229. data["file_ids"] = file_ids
  230. knowledge = Knowledges.update_knowledge_by_id(
  231. id=id, form_data=KnowledgeUpdateForm(data=data)
  232. )
  233. if knowledge:
  234. files = Files.get_files_by_ids(file_ids)
  235. return KnowledgeFilesResponse(
  236. **knowledge.model_dump(),
  237. files=files,
  238. )
  239. else:
  240. raise HTTPException(
  241. status_code=status.HTTP_400_BAD_REQUEST,
  242. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  243. )
  244. else:
  245. raise HTTPException(
  246. status_code=status.HTTP_400_BAD_REQUEST,
  247. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  248. )
  249. else:
  250. raise HTTPException(
  251. status_code=status.HTTP_400_BAD_REQUEST,
  252. detail=ERROR_MESSAGES.NOT_FOUND,
  253. )
  254. ############################
  255. # DeleteKnowledgeById
  256. ############################
  257. @router.delete("/{id}/delete", response_model=bool)
  258. async def delete_knowledge_by_id(id: str, user=Depends(get_admin_user)):
  259. try:
  260. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  261. except Exception as e:
  262. log.debug(e)
  263. pass
  264. result = Knowledges.delete_knowledge_by_id(id=id)
  265. return result