file_manager.py 4.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138
  1. import base64
  2. from configs import dify_config
  3. from core.file import file_repository
  4. from core.helper import ssrf_proxy
  5. from core.model_runtime.entities import (
  6. AudioPromptMessageContent,
  7. DocumentPromptMessageContent,
  8. ImagePromptMessageContent,
  9. VideoPromptMessageContent,
  10. )
  11. from extensions.ext_database import db
  12. from extensions.ext_storage import storage
  13. from . import helpers
  14. from .enums import FileAttribute
  15. from .models import File, FileTransferMethod, FileType
  16. from .tool_file_parser import ToolFileParser
  17. def get_attr(*, file: File, attr: FileAttribute):
  18. match attr:
  19. case FileAttribute.TYPE:
  20. return file.type.value
  21. case FileAttribute.SIZE:
  22. return file.size
  23. case FileAttribute.NAME:
  24. return file.filename
  25. case FileAttribute.MIME_TYPE:
  26. return file.mime_type
  27. case FileAttribute.TRANSFER_METHOD:
  28. return file.transfer_method.value
  29. case FileAttribute.URL:
  30. return file.remote_url
  31. case FileAttribute.EXTENSION:
  32. return file.extension
  33. def to_prompt_message_content(
  34. f: File,
  35. /,
  36. *,
  37. image_detail_config: ImagePromptMessageContent.DETAIL | None = None,
  38. ):
  39. if f.extension is None:
  40. raise ValueError("Missing file extension")
  41. if f.mime_type is None:
  42. raise ValueError("Missing file mime_type")
  43. params = {
  44. "base64_data": _get_encoded_string(f) if dify_config.MULTIMODAL_SEND_FORMAT == "base64" else "",
  45. "url": _to_url(f) if dify_config.MULTIMODAL_SEND_FORMAT == "url" else "",
  46. "format": f.extension.removeprefix("."),
  47. "mime_type": f.mime_type,
  48. }
  49. if f.type == FileType.IMAGE:
  50. params["detail"] = image_detail_config or ImagePromptMessageContent.DETAIL.LOW
  51. prompt_class_map = {
  52. FileType.IMAGE: ImagePromptMessageContent,
  53. FileType.AUDIO: AudioPromptMessageContent,
  54. FileType.VIDEO: VideoPromptMessageContent,
  55. FileType.DOCUMENT: DocumentPromptMessageContent,
  56. }
  57. try:
  58. return prompt_class_map[f.type](**params)
  59. except KeyError:
  60. raise ValueError(f"file type {f.type} is not supported")
  61. def download(f: File, /):
  62. if f.transfer_method == FileTransferMethod.TOOL_FILE:
  63. tool_file = file_repository.get_tool_file(session=db.session(), file=f)
  64. return _download_file_content(tool_file.file_key)
  65. elif f.transfer_method == FileTransferMethod.LOCAL_FILE:
  66. upload_file = file_repository.get_upload_file(session=db.session(), file=f)
  67. return _download_file_content(upload_file.key)
  68. # remote file
  69. response = ssrf_proxy.get(f.remote_url, follow_redirects=True)
  70. response.raise_for_status()
  71. return response.content
  72. def _download_file_content(path: str, /):
  73. """
  74. Download and return the contents of a file as bytes.
  75. This function loads the file from storage and ensures it's in bytes format.
  76. Args:
  77. path (str): The path to the file in storage.
  78. Returns:
  79. bytes: The contents of the file as a bytes object.
  80. Raises:
  81. ValueError: If the loaded file is not a bytes object.
  82. """
  83. data = storage.load(path, stream=False)
  84. if not isinstance(data, bytes):
  85. raise ValueError(f"file {path} is not a bytes object")
  86. return data
  87. def _get_encoded_string(f: File, /):
  88. match f.transfer_method:
  89. case FileTransferMethod.REMOTE_URL:
  90. response = ssrf_proxy.get(f.remote_url, follow_redirects=True)
  91. response.raise_for_status()
  92. data = response.content
  93. case FileTransferMethod.LOCAL_FILE:
  94. upload_file = file_repository.get_upload_file(session=db.session(), file=f)
  95. data = _download_file_content(upload_file.key)
  96. case FileTransferMethod.TOOL_FILE:
  97. tool_file = file_repository.get_tool_file(session=db.session(), file=f)
  98. data = _download_file_content(tool_file.file_key)
  99. encoded_string = base64.b64encode(data).decode("utf-8")
  100. return encoded_string
  101. def _to_url(f: File, /):
  102. if f.transfer_method == FileTransferMethod.REMOTE_URL:
  103. if f.remote_url is None:
  104. raise ValueError("Missing file remote_url")
  105. return f.remote_url
  106. elif f.transfer_method == FileTransferMethod.LOCAL_FILE:
  107. if f.related_id is None:
  108. raise ValueError("Missing file related_id")
  109. return f.remote_url or helpers.get_signed_file_url(upload_file_id=f.related_id)
  110. elif f.transfer_method == FileTransferMethod.TOOL_FILE:
  111. # add sign url
  112. if f.related_id is None or f.extension is None:
  113. raise ValueError("Missing file related_id or extension")
  114. return ToolFileParser.get_tool_file_manager().sign_file(tool_file_id=f.related_id, extension=f.extension)
  115. else:
  116. raise ValueError(f"Unsupported transfer method: {f.transfer_method}")