knowledge.py 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647
  1. from typing import List, Optional
  2. from pydantic import BaseModel
  3. from fastapi import APIRouter, Depends, HTTPException, status, Request
  4. import logging
  5. from open_webui.models.knowledge import (
  6. Knowledges,
  7. KnowledgeForm,
  8. KnowledgeResponse,
  9. KnowledgeUserResponse,
  10. )
  11. from open_webui.models.files import Files, FileModel
  12. from open_webui.retrieval.vector.connector import VECTOR_DB_CLIENT
  13. from open_webui.routers.retrieval import (
  14. process_file,
  15. ProcessFileForm,
  16. process_files_batch,
  17. BatchProcessFilesForm,
  18. )
  19. from open_webui.storage.provider import Storage
  20. from open_webui.constants import ERROR_MESSAGES
  21. from open_webui.utils.auth import get_verified_user
  22. from open_webui.utils.access_control import has_access, has_permission
  23. from open_webui.env import SRC_LOG_LEVELS
  24. from open_webui.models.models import Models, ModelForm
  25. log = logging.getLogger(__name__)
  26. log.setLevel(SRC_LOG_LEVELS["MODELS"])
  27. router = APIRouter()
  28. ############################
  29. # getKnowledgeBases
  30. ############################
  31. @router.get("/", response_model=list[KnowledgeUserResponse])
  32. async def get_knowledge(user=Depends(get_verified_user)):
  33. knowledge_bases = []
  34. if user.role == "admin":
  35. knowledge_bases = Knowledges.get_knowledge_bases()
  36. else:
  37. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "read")
  38. # Get files for each knowledge base
  39. knowledge_with_files = []
  40. for knowledge_base in knowledge_bases:
  41. files = []
  42. if knowledge_base.data:
  43. files = Files.get_file_metadatas_by_ids(
  44. knowledge_base.data.get("file_ids", [])
  45. )
  46. # Check if all files exist
  47. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  48. missing_files = list(
  49. set(knowledge_base.data.get("file_ids", []))
  50. - set([file.id for file in files])
  51. )
  52. if missing_files:
  53. data = knowledge_base.data or {}
  54. file_ids = data.get("file_ids", [])
  55. for missing_file in missing_files:
  56. file_ids.remove(missing_file)
  57. data["file_ids"] = file_ids
  58. Knowledges.update_knowledge_data_by_id(
  59. id=knowledge_base.id, data=data
  60. )
  61. files = Files.get_file_metadatas_by_ids(file_ids)
  62. knowledge_with_files.append(
  63. KnowledgeUserResponse(
  64. **knowledge_base.model_dump(),
  65. files=files,
  66. )
  67. )
  68. return knowledge_with_files
  69. @router.get("/list", response_model=list[KnowledgeUserResponse])
  70. async def get_knowledge_list(user=Depends(get_verified_user)):
  71. knowledge_bases = []
  72. if user.role == "admin":
  73. knowledge_bases = Knowledges.get_knowledge_bases()
  74. else:
  75. knowledge_bases = Knowledges.get_knowledge_bases_by_user_id(user.id, "write")
  76. # Get files for each knowledge base
  77. knowledge_with_files = []
  78. for knowledge_base in knowledge_bases:
  79. files = []
  80. if knowledge_base.data:
  81. files = Files.get_file_metadatas_by_ids(
  82. knowledge_base.data.get("file_ids", [])
  83. )
  84. # Check if all files exist
  85. if len(files) != len(knowledge_base.data.get("file_ids", [])):
  86. missing_files = list(
  87. set(knowledge_base.data.get("file_ids", []))
  88. - set([file.id for file in files])
  89. )
  90. if missing_files:
  91. data = knowledge_base.data or {}
  92. file_ids = data.get("file_ids", [])
  93. for missing_file in missing_files:
  94. file_ids.remove(missing_file)
  95. data["file_ids"] = file_ids
  96. Knowledges.update_knowledge_data_by_id(
  97. id=knowledge_base.id, data=data
  98. )
  99. files = Files.get_file_metadatas_by_ids(file_ids)
  100. knowledge_with_files.append(
  101. KnowledgeUserResponse(
  102. **knowledge_base.model_dump(),
  103. files=files,
  104. )
  105. )
  106. return knowledge_with_files
  107. ############################
  108. # CreateNewKnowledge
  109. ############################
  110. @router.post("/create", response_model=Optional[KnowledgeResponse])
  111. async def create_new_knowledge(
  112. request: Request, form_data: KnowledgeForm, user=Depends(get_verified_user)
  113. ):
  114. if user.role != "admin" and not has_permission(
  115. user.id, "workspace.knowledge", request.app.state.config.USER_PERMISSIONS
  116. ):
  117. raise HTTPException(
  118. status_code=status.HTTP_401_UNAUTHORIZED,
  119. detail=ERROR_MESSAGES.UNAUTHORIZED,
  120. )
  121. knowledge = Knowledges.insert_new_knowledge(user.id, form_data)
  122. if knowledge:
  123. return knowledge
  124. else:
  125. raise HTTPException(
  126. status_code=status.HTTP_400_BAD_REQUEST,
  127. detail=ERROR_MESSAGES.FILE_EXISTS,
  128. )
  129. ############################
  130. # GetKnowledgeById
  131. ############################
  132. class KnowledgeFilesResponse(KnowledgeResponse):
  133. files: list[FileModel]
  134. @router.get("/{id}", response_model=Optional[KnowledgeFilesResponse])
  135. async def get_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  136. knowledge = Knowledges.get_knowledge_by_id(id=id)
  137. if knowledge:
  138. if (
  139. user.role == "admin"
  140. or knowledge.user_id == user.id
  141. or has_access(user.id, "read", knowledge.access_control)
  142. ):
  143. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  144. files = Files.get_files_by_ids(file_ids)
  145. return KnowledgeFilesResponse(
  146. **knowledge.model_dump(),
  147. files=files,
  148. )
  149. else:
  150. raise HTTPException(
  151. status_code=status.HTTP_401_UNAUTHORIZED,
  152. detail=ERROR_MESSAGES.NOT_FOUND,
  153. )
  154. ############################
  155. # UpdateKnowledgeById
  156. ############################
  157. @router.post("/{id}/update", response_model=Optional[KnowledgeFilesResponse])
  158. async def update_knowledge_by_id(
  159. id: str,
  160. form_data: KnowledgeForm,
  161. user=Depends(get_verified_user),
  162. ):
  163. knowledge = Knowledges.get_knowledge_by_id(id=id)
  164. if not knowledge:
  165. raise HTTPException(
  166. status_code=status.HTTP_400_BAD_REQUEST,
  167. detail=ERROR_MESSAGES.NOT_FOUND,
  168. )
  169. # Is the user the original creator, in a group with write access, or an admin
  170. if (
  171. knowledge.user_id != user.id
  172. and not has_access(user.id, "write", knowledge.access_control)
  173. and user.role != "admin"
  174. ):
  175. raise HTTPException(
  176. status_code=status.HTTP_400_BAD_REQUEST,
  177. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  178. )
  179. knowledge = Knowledges.update_knowledge_by_id(id=id, form_data=form_data)
  180. if knowledge:
  181. file_ids = knowledge.data.get("file_ids", []) if knowledge.data else []
  182. files = Files.get_files_by_ids(file_ids)
  183. return KnowledgeFilesResponse(
  184. **knowledge.model_dump(),
  185. files=files,
  186. )
  187. else:
  188. raise HTTPException(
  189. status_code=status.HTTP_400_BAD_REQUEST,
  190. detail=ERROR_MESSAGES.ID_TAKEN,
  191. )
  192. ############################
  193. # AddFileToKnowledge
  194. ############################
  195. class KnowledgeFileIdForm(BaseModel):
  196. file_id: str
  197. @router.post("/{id}/file/add", response_model=Optional[KnowledgeFilesResponse])
  198. def add_file_to_knowledge_by_id(
  199. request: Request,
  200. id: str,
  201. form_data: KnowledgeFileIdForm,
  202. user=Depends(get_verified_user),
  203. ):
  204. knowledge = Knowledges.get_knowledge_by_id(id=id)
  205. if not knowledge:
  206. raise HTTPException(
  207. status_code=status.HTTP_400_BAD_REQUEST,
  208. detail=ERROR_MESSAGES.NOT_FOUND,
  209. )
  210. if knowledge.user_id != user.id and user.role != "admin":
  211. raise HTTPException(
  212. status_code=status.HTTP_400_BAD_REQUEST,
  213. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  214. )
  215. file = Files.get_file_by_id(form_data.file_id)
  216. if not file:
  217. raise HTTPException(
  218. status_code=status.HTTP_400_BAD_REQUEST,
  219. detail=ERROR_MESSAGES.NOT_FOUND,
  220. )
  221. if not file.data:
  222. raise HTTPException(
  223. status_code=status.HTTP_400_BAD_REQUEST,
  224. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  225. )
  226. # Add content to the vector database
  227. try:
  228. process_file(
  229. request,
  230. ProcessFileForm(file_id=form_data.file_id, collection_name=id),
  231. user=user
  232. )
  233. except Exception as e:
  234. log.debug(e)
  235. raise HTTPException(
  236. status_code=status.HTTP_400_BAD_REQUEST,
  237. detail=str(e),
  238. )
  239. if knowledge:
  240. data = knowledge.data or {}
  241. file_ids = data.get("file_ids", [])
  242. if form_data.file_id not in file_ids:
  243. file_ids.append(form_data.file_id)
  244. data["file_ids"] = file_ids
  245. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  246. if knowledge:
  247. files = Files.get_files_by_ids(file_ids)
  248. return KnowledgeFilesResponse(
  249. **knowledge.model_dump(),
  250. files=files,
  251. )
  252. else:
  253. raise HTTPException(
  254. status_code=status.HTTP_400_BAD_REQUEST,
  255. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  256. )
  257. else:
  258. raise HTTPException(
  259. status_code=status.HTTP_400_BAD_REQUEST,
  260. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  261. )
  262. else:
  263. raise HTTPException(
  264. status_code=status.HTTP_400_BAD_REQUEST,
  265. detail=ERROR_MESSAGES.NOT_FOUND,
  266. )
  267. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  268. def update_file_from_knowledge_by_id(
  269. request: Request,
  270. id: str,
  271. form_data: KnowledgeFileIdForm,
  272. user=Depends(get_verified_user),
  273. ):
  274. knowledge = Knowledges.get_knowledge_by_id(id=id)
  275. if not knowledge:
  276. raise HTTPException(
  277. status_code=status.HTTP_400_BAD_REQUEST,
  278. detail=ERROR_MESSAGES.NOT_FOUND,
  279. )
  280. if knowledge.user_id != user.id and user.role != "admin":
  281. raise HTTPException(
  282. status_code=status.HTTP_400_BAD_REQUEST,
  283. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  284. )
  285. file = Files.get_file_by_id(form_data.file_id)
  286. if not file:
  287. raise HTTPException(
  288. status_code=status.HTTP_400_BAD_REQUEST,
  289. detail=ERROR_MESSAGES.NOT_FOUND,
  290. )
  291. # Remove content from the vector database
  292. VECTOR_DB_CLIENT.delete(
  293. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  294. )
  295. # Add content to the vector database
  296. try:
  297. process_file(
  298. request,
  299. ProcessFileForm(file_id=form_data.file_id, collection_name=id),
  300. user=user
  301. )
  302. except Exception as e:
  303. raise HTTPException(
  304. status_code=status.HTTP_400_BAD_REQUEST,
  305. detail=str(e),
  306. )
  307. if knowledge:
  308. data = knowledge.data or {}
  309. file_ids = data.get("file_ids", [])
  310. files = Files.get_files_by_ids(file_ids)
  311. return KnowledgeFilesResponse(
  312. **knowledge.model_dump(),
  313. files=files,
  314. )
  315. else:
  316. raise HTTPException(
  317. status_code=status.HTTP_400_BAD_REQUEST,
  318. detail=ERROR_MESSAGES.NOT_FOUND,
  319. )
  320. ############################
  321. # RemoveFileFromKnowledge
  322. ############################
  323. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  324. def remove_file_from_knowledge_by_id(
  325. id: str,
  326. form_data: KnowledgeFileIdForm,
  327. user=Depends(get_verified_user),
  328. ):
  329. knowledge = Knowledges.get_knowledge_by_id(id=id)
  330. if not knowledge:
  331. raise HTTPException(
  332. status_code=status.HTTP_400_BAD_REQUEST,
  333. detail=ERROR_MESSAGES.NOT_FOUND,
  334. )
  335. if knowledge.user_id != user.id and user.role != "admin":
  336. raise HTTPException(
  337. status_code=status.HTTP_400_BAD_REQUEST,
  338. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  339. )
  340. file = Files.get_file_by_id(form_data.file_id)
  341. if not file:
  342. raise HTTPException(
  343. status_code=status.HTTP_400_BAD_REQUEST,
  344. detail=ERROR_MESSAGES.NOT_FOUND,
  345. )
  346. # Remove content from the vector database
  347. VECTOR_DB_CLIENT.delete(
  348. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  349. )
  350. # Remove the file's collection from vector database
  351. file_collection = f"file-{form_data.file_id}"
  352. if VECTOR_DB_CLIENT.has_collection(collection_name=file_collection):
  353. VECTOR_DB_CLIENT.delete_collection(collection_name=file_collection)
  354. # Delete physical file
  355. if file.path:
  356. Storage.delete_file(file.path)
  357. # Delete file from database
  358. Files.delete_file_by_id(form_data.file_id)
  359. if knowledge:
  360. data = knowledge.data or {}
  361. file_ids = data.get("file_ids", [])
  362. if form_data.file_id in file_ids:
  363. file_ids.remove(form_data.file_id)
  364. data["file_ids"] = file_ids
  365. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  366. if knowledge:
  367. files = Files.get_files_by_ids(file_ids)
  368. return KnowledgeFilesResponse(
  369. **knowledge.model_dump(),
  370. files=files,
  371. )
  372. else:
  373. raise HTTPException(
  374. status_code=status.HTTP_400_BAD_REQUEST,
  375. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  376. )
  377. else:
  378. raise HTTPException(
  379. status_code=status.HTTP_400_BAD_REQUEST,
  380. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  381. )
  382. else:
  383. raise HTTPException(
  384. status_code=status.HTTP_400_BAD_REQUEST,
  385. detail=ERROR_MESSAGES.NOT_FOUND,
  386. )
  387. ############################
  388. # DeleteKnowledgeById
  389. ############################
  390. @router.delete("/{id}/delete", response_model=bool)
  391. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  392. knowledge = Knowledges.get_knowledge_by_id(id=id)
  393. if not knowledge:
  394. raise HTTPException(
  395. status_code=status.HTTP_400_BAD_REQUEST,
  396. detail=ERROR_MESSAGES.NOT_FOUND,
  397. )
  398. if knowledge.user_id != user.id and user.role != "admin":
  399. raise HTTPException(
  400. status_code=status.HTTP_400_BAD_REQUEST,
  401. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  402. )
  403. log.info(f"Deleting knowledge base: {id} (name: {knowledge.name})")
  404. # Get all models
  405. models = Models.get_all_models()
  406. log.info(f"Found {len(models)} models to check for knowledge base {id}")
  407. # Update models that reference this knowledge base
  408. for model in models:
  409. if model.meta and hasattr(model.meta, "knowledge"):
  410. knowledge_list = model.meta.knowledge or []
  411. # Filter out the deleted knowledge base
  412. updated_knowledge = [k for k in knowledge_list if k.get("id") != id]
  413. # If the knowledge list changed, update the model
  414. if len(updated_knowledge) != len(knowledge_list):
  415. log.info(f"Updating model {model.id} to remove knowledge base {id}")
  416. model.meta.knowledge = updated_knowledge
  417. # Create a ModelForm for the update
  418. model_form = ModelForm(
  419. id=model.id,
  420. name=model.name,
  421. base_model_id=model.base_model_id,
  422. meta=model.meta,
  423. params=model.params,
  424. access_control=model.access_control,
  425. is_active=model.is_active,
  426. )
  427. Models.update_model_by_id(model.id, model_form)
  428. # Clean up vector DB
  429. try:
  430. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  431. except Exception as e:
  432. log.debug(e)
  433. pass
  434. result = Knowledges.delete_knowledge_by_id(id=id)
  435. return result
  436. ############################
  437. # ResetKnowledgeById
  438. ############################
  439. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  440. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  441. knowledge = Knowledges.get_knowledge_by_id(id=id)
  442. if not knowledge:
  443. raise HTTPException(
  444. status_code=status.HTTP_400_BAD_REQUEST,
  445. detail=ERROR_MESSAGES.NOT_FOUND,
  446. )
  447. if knowledge.user_id != user.id and user.role != "admin":
  448. raise HTTPException(
  449. status_code=status.HTTP_400_BAD_REQUEST,
  450. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  451. )
  452. try:
  453. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  454. except Exception as e:
  455. log.debug(e)
  456. pass
  457. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  458. return knowledge
  459. ############################
  460. # AddFilesToKnowledge
  461. ############################
  462. @router.post("/{id}/files/batch/add", response_model=Optional[KnowledgeFilesResponse])
  463. def add_files_to_knowledge_batch(
  464. request: Request,
  465. id: str,
  466. form_data: list[KnowledgeFileIdForm],
  467. user=Depends(get_verified_user),
  468. ):
  469. """
  470. Add multiple files to a knowledge base
  471. """
  472. knowledge = Knowledges.get_knowledge_by_id(id=id)
  473. if not knowledge:
  474. raise HTTPException(
  475. status_code=status.HTTP_400_BAD_REQUEST,
  476. detail=ERROR_MESSAGES.NOT_FOUND,
  477. )
  478. if knowledge.user_id != user.id and user.role != "admin":
  479. raise HTTPException(
  480. status_code=status.HTTP_400_BAD_REQUEST,
  481. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  482. )
  483. # Get files content
  484. print(f"files/batch/add - {len(form_data)} files")
  485. files: List[FileModel] = []
  486. for form in form_data:
  487. file = Files.get_file_by_id(form.file_id)
  488. if not file:
  489. raise HTTPException(
  490. status_code=status.HTTP_400_BAD_REQUEST,
  491. detail=f"File {form.file_id} not found",
  492. )
  493. files.append(file)
  494. # Process files
  495. try:
  496. result = process_files_batch(
  497. request=request,
  498. form_data=BatchProcessFilesForm(files=files, collection_name=id),
  499. user=user,
  500. )
  501. except Exception as e:
  502. log.error(
  503. f"add_files_to_knowledge_batch: Exception occurred: {e}", exc_info=True
  504. )
  505. raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e))
  506. # Add successful files to knowledge base
  507. data = knowledge.data or {}
  508. existing_file_ids = data.get("file_ids", [])
  509. # Only add files that were successfully processed
  510. successful_file_ids = [r.file_id for r in result.results if r.status == "completed"]
  511. for file_id in successful_file_ids:
  512. if file_id not in existing_file_ids:
  513. existing_file_ids.append(file_id)
  514. data["file_ids"] = existing_file_ids
  515. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  516. # If there were any errors, include them in the response
  517. if result.errors:
  518. error_details = [f"{err.file_id}: {err.error}" for err in result.errors]
  519. return KnowledgeFilesResponse(
  520. **knowledge.model_dump(),
  521. files=Files.get_files_by_ids(existing_file_ids),
  522. warnings={
  523. "message": "Some files failed to process",
  524. "errors": error_details,
  525. },
  526. )
  527. return KnowledgeFilesResponse(
  528. **knowledge.model_dump(), files=Files.get_files_by_ids(existing_file_ids)
  529. )