knowledge.py 20 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668
  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 (
  211. knowledge.user_id != user.id
  212. and not has_access(user.id, "write", knowledge.access_control)
  213. and user.role != "admin"
  214. ):
  215. raise HTTPException(
  216. status_code=status.HTTP_400_BAD_REQUEST,
  217. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  218. )
  219. file = Files.get_file_by_id(form_data.file_id)
  220. if not file:
  221. raise HTTPException(
  222. status_code=status.HTTP_400_BAD_REQUEST,
  223. detail=ERROR_MESSAGES.NOT_FOUND,
  224. )
  225. if not file.data:
  226. raise HTTPException(
  227. status_code=status.HTTP_400_BAD_REQUEST,
  228. detail=ERROR_MESSAGES.FILE_NOT_PROCESSED,
  229. )
  230. # Add content to the vector database
  231. try:
  232. process_file(
  233. request,
  234. ProcessFileForm(file_id=form_data.file_id, collection_name=id),
  235. user=user,
  236. )
  237. except Exception as e:
  238. log.debug(e)
  239. raise HTTPException(
  240. status_code=status.HTTP_400_BAD_REQUEST,
  241. detail=str(e),
  242. )
  243. if knowledge:
  244. data = knowledge.data or {}
  245. file_ids = data.get("file_ids", [])
  246. if form_data.file_id not in file_ids:
  247. file_ids.append(form_data.file_id)
  248. data["file_ids"] = file_ids
  249. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  250. if knowledge:
  251. files = Files.get_files_by_ids(file_ids)
  252. return KnowledgeFilesResponse(
  253. **knowledge.model_dump(),
  254. files=files,
  255. )
  256. else:
  257. raise HTTPException(
  258. status_code=status.HTTP_400_BAD_REQUEST,
  259. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  260. )
  261. else:
  262. raise HTTPException(
  263. status_code=status.HTTP_400_BAD_REQUEST,
  264. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  265. )
  266. else:
  267. raise HTTPException(
  268. status_code=status.HTTP_400_BAD_REQUEST,
  269. detail=ERROR_MESSAGES.NOT_FOUND,
  270. )
  271. @router.post("/{id}/file/update", response_model=Optional[KnowledgeFilesResponse])
  272. def update_file_from_knowledge_by_id(
  273. request: Request,
  274. id: str,
  275. form_data: KnowledgeFileIdForm,
  276. user=Depends(get_verified_user),
  277. ):
  278. knowledge = Knowledges.get_knowledge_by_id(id=id)
  279. if not knowledge:
  280. raise HTTPException(
  281. status_code=status.HTTP_400_BAD_REQUEST,
  282. detail=ERROR_MESSAGES.NOT_FOUND,
  283. )
  284. if (
  285. knowledge.user_id != user.id
  286. and not has_access(user.id, "write", knowledge.access_control)
  287. and user.role != "admin"
  288. ):
  289. raise HTTPException(
  290. status_code=status.HTTP_400_BAD_REQUEST,
  291. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  292. )
  293. file = Files.get_file_by_id(form_data.file_id)
  294. if not file:
  295. raise HTTPException(
  296. status_code=status.HTTP_400_BAD_REQUEST,
  297. detail=ERROR_MESSAGES.NOT_FOUND,
  298. )
  299. # Remove content from the vector database
  300. VECTOR_DB_CLIENT.delete(
  301. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  302. )
  303. # Add content to the vector database
  304. try:
  305. process_file(
  306. request,
  307. ProcessFileForm(file_id=form_data.file_id, collection_name=id),
  308. user=user,
  309. )
  310. except Exception as e:
  311. raise HTTPException(
  312. status_code=status.HTTP_400_BAD_REQUEST,
  313. detail=str(e),
  314. )
  315. if knowledge:
  316. data = knowledge.data or {}
  317. file_ids = data.get("file_ids", [])
  318. files = Files.get_files_by_ids(file_ids)
  319. return KnowledgeFilesResponse(
  320. **knowledge.model_dump(),
  321. files=files,
  322. )
  323. else:
  324. raise HTTPException(
  325. status_code=status.HTTP_400_BAD_REQUEST,
  326. detail=ERROR_MESSAGES.NOT_FOUND,
  327. )
  328. ############################
  329. # RemoveFileFromKnowledge
  330. ############################
  331. @router.post("/{id}/file/remove", response_model=Optional[KnowledgeFilesResponse])
  332. def remove_file_from_knowledge_by_id(
  333. id: str,
  334. form_data: KnowledgeFileIdForm,
  335. user=Depends(get_verified_user),
  336. ):
  337. knowledge = Knowledges.get_knowledge_by_id(id=id)
  338. if not knowledge:
  339. raise HTTPException(
  340. status_code=status.HTTP_400_BAD_REQUEST,
  341. detail=ERROR_MESSAGES.NOT_FOUND,
  342. )
  343. if (
  344. knowledge.user_id != user.id
  345. and not has_access(user.id, "write", knowledge.access_control)
  346. and user.role != "admin"
  347. ):
  348. raise HTTPException(
  349. status_code=status.HTTP_400_BAD_REQUEST,
  350. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  351. )
  352. file = Files.get_file_by_id(form_data.file_id)
  353. if not file:
  354. raise HTTPException(
  355. status_code=status.HTTP_400_BAD_REQUEST,
  356. detail=ERROR_MESSAGES.NOT_FOUND,
  357. )
  358. # Remove content from the vector database
  359. VECTOR_DB_CLIENT.delete(
  360. collection_name=knowledge.id, filter={"file_id": form_data.file_id}
  361. )
  362. # Remove the file's collection from vector database
  363. file_collection = f"file-{form_data.file_id}"
  364. if VECTOR_DB_CLIENT.has_collection(collection_name=file_collection):
  365. VECTOR_DB_CLIENT.delete_collection(collection_name=file_collection)
  366. # Delete file from database
  367. Files.delete_file_by_id(form_data.file_id)
  368. if knowledge:
  369. data = knowledge.data or {}
  370. file_ids = data.get("file_ids", [])
  371. if form_data.file_id in file_ids:
  372. file_ids.remove(form_data.file_id)
  373. data["file_ids"] = file_ids
  374. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  375. if knowledge:
  376. files = Files.get_files_by_ids(file_ids)
  377. return KnowledgeFilesResponse(
  378. **knowledge.model_dump(),
  379. files=files,
  380. )
  381. else:
  382. raise HTTPException(
  383. status_code=status.HTTP_400_BAD_REQUEST,
  384. detail=ERROR_MESSAGES.DEFAULT("knowledge"),
  385. )
  386. else:
  387. raise HTTPException(
  388. status_code=status.HTTP_400_BAD_REQUEST,
  389. detail=ERROR_MESSAGES.DEFAULT("file_id"),
  390. )
  391. else:
  392. raise HTTPException(
  393. status_code=status.HTTP_400_BAD_REQUEST,
  394. detail=ERROR_MESSAGES.NOT_FOUND,
  395. )
  396. ############################
  397. # DeleteKnowledgeById
  398. ############################
  399. @router.delete("/{id}/delete", response_model=bool)
  400. async def delete_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  401. knowledge = Knowledges.get_knowledge_by_id(id=id)
  402. if not knowledge:
  403. raise HTTPException(
  404. status_code=status.HTTP_400_BAD_REQUEST,
  405. detail=ERROR_MESSAGES.NOT_FOUND,
  406. )
  407. if (
  408. knowledge.user_id != user.id
  409. and not has_access(user.id, "write", knowledge.access_control)
  410. and user.role != "admin"
  411. ):
  412. raise HTTPException(
  413. status_code=status.HTTP_400_BAD_REQUEST,
  414. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  415. )
  416. log.info(f"Deleting knowledge base: {id} (name: {knowledge.name})")
  417. # Get all models
  418. models = Models.get_all_models()
  419. log.info(f"Found {len(models)} models to check for knowledge base {id}")
  420. # Update models that reference this knowledge base
  421. for model in models:
  422. if model.meta and hasattr(model.meta, "knowledge"):
  423. knowledge_list = model.meta.knowledge or []
  424. # Filter out the deleted knowledge base
  425. updated_knowledge = [k for k in knowledge_list if k.get("id") != id]
  426. # If the knowledge list changed, update the model
  427. if len(updated_knowledge) != len(knowledge_list):
  428. log.info(f"Updating model {model.id} to remove knowledge base {id}")
  429. model.meta.knowledge = updated_knowledge
  430. # Create a ModelForm for the update
  431. model_form = ModelForm(
  432. id=model.id,
  433. name=model.name,
  434. base_model_id=model.base_model_id,
  435. meta=model.meta,
  436. params=model.params,
  437. access_control=model.access_control,
  438. is_active=model.is_active,
  439. )
  440. Models.update_model_by_id(model.id, model_form)
  441. # Clean up vector DB
  442. try:
  443. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  444. except Exception as e:
  445. log.debug(e)
  446. pass
  447. result = Knowledges.delete_knowledge_by_id(id=id)
  448. return result
  449. ############################
  450. # ResetKnowledgeById
  451. ############################
  452. @router.post("/{id}/reset", response_model=Optional[KnowledgeResponse])
  453. async def reset_knowledge_by_id(id: str, user=Depends(get_verified_user)):
  454. knowledge = Knowledges.get_knowledge_by_id(id=id)
  455. if not knowledge:
  456. raise HTTPException(
  457. status_code=status.HTTP_400_BAD_REQUEST,
  458. detail=ERROR_MESSAGES.NOT_FOUND,
  459. )
  460. if (
  461. knowledge.user_id != user.id
  462. and not has_access(user.id, "write", knowledge.access_control)
  463. and user.role != "admin"
  464. ):
  465. raise HTTPException(
  466. status_code=status.HTTP_400_BAD_REQUEST,
  467. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  468. )
  469. try:
  470. VECTOR_DB_CLIENT.delete_collection(collection_name=id)
  471. except Exception as e:
  472. log.debug(e)
  473. pass
  474. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data={"file_ids": []})
  475. return knowledge
  476. ############################
  477. # AddFilesToKnowledge
  478. ############################
  479. @router.post("/{id}/files/batch/add", response_model=Optional[KnowledgeFilesResponse])
  480. def add_files_to_knowledge_batch(
  481. request: Request,
  482. id: str,
  483. form_data: list[KnowledgeFileIdForm],
  484. user=Depends(get_verified_user),
  485. ):
  486. """
  487. Add multiple files to a knowledge base
  488. """
  489. knowledge = Knowledges.get_knowledge_by_id(id=id)
  490. if not knowledge:
  491. raise HTTPException(
  492. status_code=status.HTTP_400_BAD_REQUEST,
  493. detail=ERROR_MESSAGES.NOT_FOUND,
  494. )
  495. if (
  496. knowledge.user_id != user.id
  497. and not has_access(user.id, "write", knowledge.access_control)
  498. and user.role != "admin"
  499. ):
  500. raise HTTPException(
  501. status_code=status.HTTP_400_BAD_REQUEST,
  502. detail=ERROR_MESSAGES.ACCESS_PROHIBITED,
  503. )
  504. # Get files content
  505. log.info(f"files/batch/add - {len(form_data)} files")
  506. files: List[FileModel] = []
  507. for form in form_data:
  508. file = Files.get_file_by_id(form.file_id)
  509. if not file:
  510. raise HTTPException(
  511. status_code=status.HTTP_400_BAD_REQUEST,
  512. detail=f"File {form.file_id} not found",
  513. )
  514. files.append(file)
  515. # Process files
  516. try:
  517. result = process_files_batch(
  518. request=request,
  519. form_data=BatchProcessFilesForm(files=files, collection_name=id),
  520. user=user,
  521. )
  522. except Exception as e:
  523. log.error(
  524. f"add_files_to_knowledge_batch: Exception occurred: {e}", exc_info=True
  525. )
  526. raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e))
  527. # Add successful files to knowledge base
  528. data = knowledge.data or {}
  529. existing_file_ids = data.get("file_ids", [])
  530. # Only add files that were successfully processed
  531. successful_file_ids = [r.file_id for r in result.results if r.status == "completed"]
  532. for file_id in successful_file_ids:
  533. if file_id not in existing_file_ids:
  534. existing_file_ids.append(file_id)
  535. data["file_ids"] = existing_file_ids
  536. knowledge = Knowledges.update_knowledge_data_by_id(id=id, data=data)
  537. # If there were any errors, include them in the response
  538. if result.errors:
  539. error_details = [f"{err.file_id}: {err.error}" for err in result.errors]
  540. return KnowledgeFilesResponse(
  541. **knowledge.model_dump(),
  542. files=Files.get_files_by_ids(existing_file_ids),
  543. warnings={
  544. "message": "Some files failed to process",
  545. "errors": error_details,
  546. },
  547. )
  548. return KnowledgeFilesResponse(
  549. **knowledge.model_dump(), files=Files.get_files_by_ids(existing_file_ids)
  550. )