knowledge.py 9.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320
  1. import json
  2. from typing import Optional, Union
  3. from pydantic import BaseModel
  4. from fastapi import APIRouter, Depends, HTTPException, status
  5. from open_webui.apps.webui.models.knowledge import (
  6. Knowledges,
  7. KnowledgeUpdateForm,
  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. router = APIRouter()
  17. ############################
  18. # GetKnowledgeItems
  19. ############################
  20. @router.get(
  21. "/", response_model=Optional[Union[list[KnowledgeResponse], KnowledgeResponse]]
  22. )
  23. async def get_knowledge_items(
  24. id: Optional[str] = None, user=Depends(get_verified_user)
  25. ):
  26. if id:
  27. knowledge = Knowledges.get_knowledge_by_id(id=id)
  28. if knowledge:
  29. return knowledge
  30. else:
  31. raise HTTPException(
  32. status_code=status.HTTP_401_UNAUTHORIZED,
  33. detail=ERROR_MESSAGES.NOT_FOUND,
  34. )
  35. else:
  36. return [
  37. KnowledgeResponse(**knowledge.model_dump())
  38. for knowledge in Knowledges.get_knowledge_items()
  39. ]
  40. ############################
  41. # CreateNewKnowledge
  42. ############################
  43. @router.post("/create", response_model=Optional[KnowledgeResponse])
  44. async def create_new_knowledge(form_data: KnowledgeForm, user=Depends(get_admin_user)):
  45. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  46. if knowledge:
  47. return knowledge
  48. else:
  49. raise HTTPException(
  50. status_code=status.HTTP_400_BAD_REQUEST,
  51. detail=ERROR_MESSAGES.FILE_EXISTS,
  52. )
  53. ############################
  54. # GetKnowledgeById
  55. ############################
  56. class KnowledgeFilesResponse(KnowledgeResponse):
  57. files: list[FileModel]
  58. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  59. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  60. knowledge = Knowledges.get_knowledge_by_id(id=id)
  61. if knowledge:
  62. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  63. files = Files.get_files_by_ids(file_ids)
  64. return KnowledgeFilesResponse(
  65. **knowledge.model_dump(),
  66. files=files,
  67. )
  68. else:
  69. raise HTTPException(
  70. status_code=status.HTTP_401_UNAUTHORIZED,
  71. detail=ERROR_MESSAGES.NOT_FOUND,
  72. )
  73. ############################
  74. # UpdateKnowledgeById
  75. ############################
  76. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  77. async def update_knowledge_by_id(
  78. id: str,
  79. form_data: KnowledgeUpdateForm,
  80. user=Depends(get_admin_user),
  81. ):
  82. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  83. if knowledge:
  84. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  85. files = Files.get_files_by_ids(file_ids)
  86. return KnowledgeFilesResponse(
  87. **knowledge.model_dump(),
  88. files=files,
  89. )
  90. else:
  91. raise HTTPException(
  92. status_code=status.HTTP_400_BAD_REQUEST,
  93. detail=ERROR_MESSAGES.ID_TAKEN,
  94. )
  95. ############################
  96. # AddFileToKnowledge
  97. ############################
  98. class KnowledgeFileIdForm(BaseModel):
  99. file_id: str
  100. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  101. def add_file_to_knowledge_by_id(
  102. id: str,
  103. form_data: KnowledgeFileIdForm,
  104. user=Depends(get_admin_user),
  105. ):
  106. knowledge = Knowledges.get_knowledge_by_id(id=id)
  107. file = Files.get_file_by_id(form_data.file_id)
  108. if not file:
  109. raise HTTPException(
  110. status_code=status.HTTP_400_BAD_REQUEST,
  111. detail=ERROR_MESSAGES.NOT_FOUND,
  112. )
  113. if not file.data:
  114. raise HTTPException(
  115. status_code=status.HTTP_400_BAD_REQUEST,
  116. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  117. )
  118. # Add content to the vector database
  119. try:
  120. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  121. except Exception as e:
  122. raise HTTPException(
  123. status_code=status.HTTP_400_BAD_REQUEST,
  124. detail=str(e),
  125. )
  126. if knowledge:
  127. data = knowledge.data or {}
  128. file_ids = data.get("file_ids", [])
  129. if form_data.file_id not in file_ids:
  130. file_ids.append(form_data.file_id)
  131. data["file_ids"] = file_ids
  132. knowledge = Knowledges.update_knowledge_by_id(
  133. id=id, form_data=KnowledgeUpdateForm(data=data)
  134. )
  135. if knowledge:
  136. files = Files.get_files_by_ids(file_ids)
  137. return KnowledgeFilesResponse(
  138. **knowledge.model_dump(),
  139. files=files,
  140. )
  141. else:
  142. raise HTTPException(
  143. status_code=status.HTTP_400_BAD_REQUEST,
  144. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  145. )
  146. else:
  147. raise HTTPException(
  148. status_code=status.HTTP_400_BAD_REQUEST,
  149. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  150. )
  151. else:
  152. raise HTTPException(
  153. status_code=status.HTTP_400_BAD_REQUEST,
  154. detail=ERROR_MESSAGES.NOT_FOUND,
  155. )
  156. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  157. def update_file_from_knowledge_by_id(
  158. id: str,
  159. form_data: KnowledgeFileIdForm,
  160. user=Depends(get_admin_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. # Remove content from the vector database
  170. VECTOR_DB_CLIENT.delete(
  171. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  172. )
  173. # Add content to the vector database
  174. try:
  175. process_file(ProcessFileForm(file_id=form_data.file_id, collection_name=id))
  176. except Exception as e:
  177. raise HTTPException(
  178. status_code=status.HTTP_400_BAD_REQUEST,
  179. detail=str(e),
  180. )
  181. if knowledge:
  182. data = knowledge.data or {}
  183. file_ids = data.get("file_ids", [])
  184. files = Files.get_files_by_ids(file_ids)
  185. return KnowledgeFilesResponse(
  186. **knowledge.model_dump(),
  187. files=files,
  188. )
  189. else:
  190. raise HTTPException(
  191. status_code=status.HTTP_400_BAD_REQUEST,
  192. detail=ERROR_MESSAGES.NOT_FOUND,
  193. )
  194. ############################
  195. # RemoveFileFromKnowledge
  196. ############################
  197. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  198. def remove_file_from_knowledge_by_id(
  199. id: str,
  200. form_data: KnowledgeFileIdForm,
  201. user=Depends(get_admin_user),
  202. ):
  203. knowledge = Knowledges.get_knowledge_by_id(id=id)
  204. file = Files.get_file_by_id(form_data.file_id)
  205. if not file:
  206. raise HTTPException(
  207. status_code=status.HTTP_400_BAD_REQUEST,
  208. detail=ERROR_MESSAGES.NOT_FOUND,
  209. )
  210. # Remove content from the vector database
  211. VECTOR_DB_CLIENT.delete(
  212. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  213. )
  214. result = VECTOR_DB_CLIENT.query(
  215. collection_name=knowledge.id,
  216. filter={"file_id": form_data.file_id},
  217. )
  218. Files.delete_file_by_id(form_data.file_id)
  219. if knowledge:
  220. data = knowledge.data or {}
  221. file_ids = data.get("file_ids", [])
  222. if form_data.file_id in file_ids:
  223. file_ids.remove(form_data.file_id)
  224. data["file_ids"] = file_ids
  225. knowledge = Knowledges.update_knowledge_by_id(
  226. id=id, form_data=KnowledgeUpdateForm(data=data)
  227. )
  228. if knowledge:
  229. files = Files.get_files_by_ids(file_ids)
  230. return KnowledgeFilesResponse(
  231. **knowledge.model_dump(),
  232. files=files,
  233. )
  234. else:
  235. raise HTTPException(
  236. status_code=status.HTTP_400_BAD_REQUEST,
  237. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  238. )
  239. else:
  240. raise HTTPException(
  241. status_code=status.HTTP_400_BAD_REQUEST,
  242. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  243. )
  244. else:
  245. raise HTTPException(
  246. status_code=status.HTTP_400_BAD_REQUEST,
  247. detail=ERROR_MESSAGES.NOT_FOUND,
  248. )
  249. ############################
  250. # DeleteKnowledgeById
  251. ############################
  252. @router.delete("/{id}/delete", response_model=bool)
  253. async def delete_knowledge_by_id(id: str, user=Depends(get_admin_user)):
  254. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  255. result = Knowledges.delete_knowledge_by_id(id=id)
  256. return result