r2r.py 5.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170
  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, "/v3")
  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, "/v3")
  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. id=None,
  51. )
  52. async def ingest_fileinfo(self, file: UploadFile, metadata: Optional[dict]):
  53. await self._check_login()
  54. return await self.client.documents.create(
  55. file=file,
  56. metadata=metadata if metadata else None,
  57. id=None,
  58. )
  59. def search(self, query: str, filters: dict[str, Any]):
  60. self._check_login_sync()
  61. print(
  62. "aaaaaaaaaaaaaaaaaaaaaaaaaaaasssssssssssssssssssssssssssssssssssssssssgggggggggggggggggggg"
  63. )
  64. search_response = self.client_sync.retrieval.search(
  65. query=query,
  66. search_settings={
  67. "filters": filters,
  68. "limit": tool_settings.R2R_SEARCH_LIMIT,
  69. },
  70. )
  71. print("vvvvvvvvvvvvvvvvvvmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm")
  72. print(search_response.get("results"))
  73. return search_response.get("results").get("chunk_search_results")
  74. def list(
  75. self,
  76. ids: Optional[list[str]] = None,
  77. offset: Optional[int] = 0,
  78. limit: Optional[int] = 100,
  79. ):
  80. self._check_login_sync()
  81. print("aaaaaaaaaaaaaaaaaaaaaaaaaaaaassssssssssssssssssssssssssssssssssss")
  82. print(ids)
  83. """
  84. listed = mutable_client.documents.list(limit=2, offset=0)
  85. results = listed.results
  86. assert len(results) == 2, "Expected 2 results for paginated listing"
  87. """
  88. print("listlistlistlistlistlistlistlistlistlistlistlistlistlistlistlistlist")
  89. if len(ids) > 0:
  90. listed = self.client_sync.documents.list(
  91. ids=ids, limit=limit, offset=offset
  92. )
  93. print(listed.get("results"))
  94. return listed.get("results")
  95. else:
  96. return []
  97. def list_documents(
  98. self,
  99. id: Optional[str] = "",
  100. offset: Optional[int] = 0,
  101. limit: Optional[int] = 100,
  102. ):
  103. self._check_login_sync()
  104. """
  105. docs = client.collections.list_documents(empty_coll_id).results
  106. assert len(docs) == 0, "Expected no documents in a new empty collection"
  107. """
  108. print(
  109. "collectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollections"
  110. )
  111. if id != "":
  112. try:
  113. listed = self.client_sync.collections.list_documents(
  114. id=id, limit=limit, offset=offset
  115. )
  116. print(listed.get("results"))
  117. return listed.get("results")
  118. except Exception as e:
  119. print(e)
  120. listed = []
  121. return listed
  122. else:
  123. return []
  124. async def _check_login(self):
  125. if not self.auth_enabled:
  126. return
  127. if not self.client.access_token and verify_jwt_expiration(
  128. self.client.access_token
  129. ):
  130. return
  131. else:
  132. await self.init()
  133. def _check_login_sync(self):
  134. if not self.auth_enabled:
  135. return
  136. try:
  137. if not self.client_sync.access_token and verify_jwt_expiration(
  138. self.client_sync.access_token
  139. ):
  140. return
  141. except Exception as e:
  142. print(e)
  143. self.client_sync.users.login(
  144. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  145. )
  146. # 创建 R2R 实例
  147. r2r = R2R()
  148. # 在您的应用程序启动时调用 initialize_r2r()
  149. async def initialize_r2r():
  150. await r2r.init()