r2r.py 5.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160
  1. from typing import Optional, Any
  2. from r2r import R2RAsyncClient
  3. from r2r import R2RClient
  4. from fastapi import UploadFile
  5. from app.libs.util import verify_jwt_expiration
  6. from config.llm import tool_settings
  7. import nest_asyncio
  8. # 使得异步代码可以在已运行的事件循环中嵌套
  9. nest_asyncio.apply()
  10. class R2R:
  11. client: R2RAsyncClient
  12. client_sync: R2RClient
  13. def __init__(self):
  14. self.auth_enabled = tool_settings.R2R_USERNAME and tool_settings.R2R_PASSWORD
  15. self.client = None
  16. self.client_sync = None
  17. def init_sync(self):
  18. if not self.auth_enabled:
  19. return
  20. if not self.client_sync:
  21. self.client_sync = R2RClient(tool_settings.R2R_BASE_URL, 300)
  22. self.client_sync.users.login(
  23. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  24. )
  25. print(
  26. "1111111111111111111111111111111122222vvdgdfdf" + tool_settings.R2R_USERNAME
  27. )
  28. print(tool_settings.R2R_USERNAME)
  29. print(tool_settings.R2R_PASSWORD)
  30. # print(self.client_sync.access_token)
  31. async def init(self):
  32. if not self.auth_enabled:
  33. return
  34. if not self.client:
  35. self.client = R2RAsyncClient(tool_settings.R2R_BASE_URL, 300)
  36. print(
  37. "1111111111111111111111111111111122222vvdgdfdf" + tool_settings.R2R_USERNAME
  38. )
  39. print(tool_settings.R2R_USERNAME)
  40. print(tool_settings.R2R_PASSWORD)
  41. await self.client.users.login(
  42. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  43. )
  44. print(self.client.access_token)
  45. async def ingest_file(self, file_path: str, metadata: Optional[dict]):
  46. await self._check_login()
  47. return await self.client.documents.create(
  48. file_path=file_path,
  49. metadata=metadata if metadata else None,
  50. ingestion_mode="fast",
  51. id=None,
  52. )
  53. async def ingest_fileinfo(self, file: UploadFile, metadata: Optional[dict]):
  54. await self._check_login()
  55. return await self.client.documents.create(
  56. file=file,
  57. metadata=metadata if metadata else None,
  58. id=None,
  59. )
  60. def search(self, query: str, filters: dict[str, Any]):
  61. self._check_login_sync()
  62. print(
  63. "aaaaaaaaaaaaaaaaaaaaaaaaaaaasssssssssssssssssssssssssssssssssssssssssgggggggggggggggggggg"
  64. )
  65. search_response = self.client_sync.retrieval.search(
  66. query=query,
  67. search_mode="basic",
  68. search_settings={
  69. "filters": filters,
  70. "limit": tool_settings.R2R_SEARCH_LIMIT,
  71. },
  72. )
  73. print("vvvvvvvvvvvvvvvvvvmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm")
  74. # print(search_response)
  75. print(search_response.get("results"))
  76. return search_response.get("results")["chunk_search_results"]
  77. def list_chunks(self, ids: list[str] = []):
  78. self._check_login_sync()
  79. print(
  80. "retrieve_documentsretrieve_documentsretrieve_documentsretrieve_documentsretrieve_documents"
  81. )
  82. print(ids)
  83. allfile = []
  84. for id in ids:
  85. listed = self.client_sync.documents.list_chunks(id=id)
  86. allfile += listed.get("results")
  87. return allfile
  88. def list_documents(
  89. self,
  90. id: Optional[str] = "",
  91. offset: Optional[int] = 0,
  92. limit: Optional[int] = 100,
  93. ):
  94. self._check_login_sync()
  95. """
  96. docs = client.collections.list_documents(empty_coll_id).results
  97. assert len(docs) == 0, "Expected no documents in a new empty collection"
  98. """
  99. print(
  100. "collectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollections"
  101. )
  102. if id != "":
  103. try:
  104. listed = self.client_sync.collections.list_documents(
  105. id=id, limit=limit, offset=offset
  106. )
  107. print(listed.get("results"))
  108. return listed.get("results")
  109. except Exception as e:
  110. print(e)
  111. listed = []
  112. return listed
  113. else:
  114. return []
  115. async def _check_login(self):
  116. if not self.auth_enabled:
  117. return
  118. if not self.client.access_token and verify_jwt_expiration(
  119. self.client.access_token
  120. ):
  121. return
  122. else:
  123. await self.init()
  124. def _check_login_sync(self):
  125. if not self.auth_enabled:
  126. return
  127. try:
  128. if not self.client_sync.access_token and verify_jwt_expiration(
  129. self.client_sync.access_token
  130. ):
  131. return
  132. except Exception as e:
  133. print(e)
  134. self.client_sync.users.login(
  135. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  136. )
  137. # 创建 R2R 实例
  138. r2r = R2R()
  139. # 在您的应用程序启动时调用 initialize_r2r()
  140. async def initialize_r2r():
  141. await r2r.init()