r2r.py 5.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158
  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. 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_settings={
  68. "filters": filters,
  69. "limit": tool_settings.R2R_SEARCH_LIMIT,
  70. },
  71. )
  72. print("vvvvvvvvvvvvvvvvvvmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm")
  73. print(search_response.get("results"))
  74. return search_response.get("results").get("chunk_search_results")
  75. def list_chunks(self, ids: list[str] = []):
  76. self._check_login_sync()
  77. print(
  78. "retrieve_documentsretrieve_documentsretrieve_documentsretrieve_documentsretrieve_documents"
  79. )
  80. print(ids)
  81. allfile = []
  82. for id in ids:
  83. listed = self.client_sync.documents.list_chunks(id=id)
  84. allfile += listed.get("results")
  85. return allfile
  86. def list_documents(
  87. self,
  88. id: Optional[str] = "",
  89. offset: Optional[int] = 0,
  90. limit: Optional[int] = 100,
  91. ):
  92. self._check_login_sync()
  93. """
  94. docs = client.collections.list_documents(empty_coll_id).results
  95. assert len(docs) == 0, "Expected no documents in a new empty collection"
  96. """
  97. print(
  98. "collectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollections"
  99. )
  100. if id != "":
  101. try:
  102. listed = self.client_sync.collections.list_documents(
  103. id=id, limit=limit, offset=offset
  104. )
  105. print(listed.get("results"))
  106. return listed.get("results")
  107. except Exception as e:
  108. print(e)
  109. listed = []
  110. return listed
  111. else:
  112. return []
  113. async def _check_login(self):
  114. if not self.auth_enabled:
  115. return
  116. if not self.client.access_token and verify_jwt_expiration(
  117. self.client.access_token
  118. ):
  119. return
  120. else:
  121. await self.init()
  122. def _check_login_sync(self):
  123. if not self.auth_enabled:
  124. return
  125. try:
  126. if not self.client_sync.access_token and verify_jwt_expiration(
  127. self.client_sync.access_token
  128. ):
  129. return
  130. except Exception as e:
  131. print(e)
  132. self.client_sync.users.login(
  133. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  134. )
  135. # 创建 R2R 实例
  136. r2r = R2R()
  137. # 在您的应用程序启动时调用 initialize_r2r()
  138. async def initialize_r2r():
  139. await r2r.init()