r2r.py 5.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161
  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)
  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)
  31. return self.client_sync
  32. async def init(self):
  33. if not self.auth_enabled:
  34. return
  35. # if not self.client:
  36. self.client = R2RAsyncClient(tool_settings.R2R_BASE_URL)
  37. print(
  38. "1111111111111111111111111111111122222vvdgdfdf" + tool_settings.R2R_USERNAME
  39. )
  40. print(tool_settings.R2R_USERNAME)
  41. print(tool_settings.R2R_PASSWORD)
  42. await self.client.users.login(
  43. tool_settings.R2R_USERNAME, tool_settings.R2R_PASSWORD
  44. )
  45. print(self.client.access_token)
  46. return self.client
  47. async def ingest_file(self, file_path: str, metadata: Optional[dict]):
  48. client = await self._check_login()
  49. return await client.documents.create(
  50. file_path=file_path,
  51. metadata=metadata if metadata else None,
  52. ingestion_mode="fast",
  53. id=None,
  54. )
  55. async def ingest_fileinfo(self, file: UploadFile, metadata: Optional[dict]):
  56. client = await self._check_login()
  57. return await client.documents.create(
  58. file=file,
  59. metadata=metadata if metadata else None,
  60. id=None,
  61. )
  62. def search(self, query: str, filters: dict[str, Any]):
  63. client = self._check_login_sync()
  64. print(
  65. "aaaaaaaaaaaaaaaaaaaaaaaaaaaasssssssssssssssssssssssssssssssssssssssssgggggggggggggggggggg"
  66. )
  67. search_response = client.retrieval.search(
  68. query=query,
  69. search_mode="basic",
  70. search_settings={
  71. "filters": filters,
  72. "limit": tool_settings.R2R_SEARCH_LIMIT,
  73. },
  74. )
  75. print("vvvvvvvvvvvvvvvvvvmmmmmmmmmmmmmmmmmmmmmmmmmmmmmm")
  76. # print(search_response)
  77. print(search_response.results)
  78. return search_response.results.chunk_search_results
  79. def list_chunks(self, ids: list[str] = []):
  80. client = self._check_login_sync()
  81. print(
  82. "retrieve_documentsretrieve_documentsretrieve_documentsretrieve_documentsretrieve_documents"
  83. )
  84. print(ids)
  85. allfile = []
  86. for id in ids:
  87. listed = client.documents.list_chunks(id=id)
  88. allfile += listed.results
  89. return allfile
  90. def list_documents(
  91. self,
  92. id: Optional[str] = "",
  93. offset: Optional[int] = 0,
  94. limit: Optional[int] = 100,
  95. ):
  96. client = self._check_login_sync()
  97. """
  98. docs = client.collections.list_documents(empty_coll_id).results
  99. assert len(docs) == 0, "Expected no documents in a new empty collection"
  100. """
  101. print(
  102. "collectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollectionscollections"
  103. )
  104. if id != "":
  105. try:
  106. listed = client.collections.list_documents(
  107. id=id, limit=limit, offset=offset
  108. )
  109. print(listed.results)
  110. return listed.results
  111. except Exception as e:
  112. print(e)
  113. listed = []
  114. return listed
  115. else:
  116. return []
  117. async def _check_login(self):
  118. if not self.auth_enabled:
  119. return
  120. # if self.client.access_token and verify_jwt_expiration(self.client.access_token):
  121. # return
  122. # else:
  123. return await self.init()
  124. def _check_login_sync(self):
  125. print("access_tokenaccess_tokenaccess_tokenaccess_token")
  126. print(self.client_sync)
  127. if not self.auth_enabled:
  128. return
  129. # try:
  130. # if self.client_sync.access_token and verify_jwt_expiration(
  131. # self.client_sync.access_token
  132. # ):
  133. # print(self.client_sync.access_token)
  134. # return
  135. # except Exception as e:
  136. # print(e)
  137. return self.init_sync()
  138. # 创建 R2R 实例
  139. r2r = R2R()
  140. # 在您的应用程序启动时调用 initialize_r2r()
  141. # async def initialize_r2r():
  142. # await r2r.init()