message_file_parser.py 9.1 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221
  1. import re
  2. from collections.abc import Mapping, Sequence
  3. from typing import Any, Union
  4. from urllib.parse import parse_qs, urlparse
  5. import requests
  6. from core.app.app_config.entities import FileExtraConfig
  7. from core.file.file_obj import FileBelongsTo, FileTransferMethod, FileType, FileVar
  8. from extensions.ext_database import db
  9. from models.account import Account
  10. from models.model import EndUser, MessageFile, UploadFile
  11. from services.file_service import IMAGE_EXTENSIONS
  12. class MessageFileParser:
  13. def __init__(self, tenant_id: str, app_id: str) -> None:
  14. self.tenant_id = tenant_id
  15. self.app_id = app_id
  16. def validate_and_transform_files_arg(self, files: Sequence[Mapping[str, Any]], file_extra_config: FileExtraConfig,
  17. user: Union[Account, EndUser]) -> list[FileVar]:
  18. """
  19. validate and transform files arg
  20. :param files:
  21. :param file_extra_config:
  22. :param user:
  23. :return:
  24. """
  25. for file in files:
  26. if not isinstance(file, dict):
  27. raise ValueError('Invalid file format, must be dict')
  28. if not file.get('type'):
  29. raise ValueError('Missing file type')
  30. FileType.value_of(file.get('type'))
  31. if not file.get('transfer_method'):
  32. raise ValueError('Missing file transfer method')
  33. FileTransferMethod.value_of(file.get('transfer_method'))
  34. if file.get('transfer_method') == FileTransferMethod.REMOTE_URL.value:
  35. if not file.get('url'):
  36. raise ValueError('Missing file url')
  37. if not file.get('url').startswith('http'):
  38. raise ValueError('Invalid file url')
  39. if file.get('transfer_method') == FileTransferMethod.LOCAL_FILE.value and not file.get('upload_file_id'):
  40. raise ValueError('Missing file upload_file_id')
  41. if file.get('transform_method') == FileTransferMethod.TOOL_FILE.value and not file.get('tool_file_id'):
  42. raise ValueError('Missing file tool_file_id')
  43. # transform files to file objs
  44. type_file_objs = self._to_file_objs(files, file_extra_config)
  45. # validate files
  46. new_files = []
  47. for file_type, file_objs in type_file_objs.items():
  48. if file_type == FileType.IMAGE:
  49. # parse and validate files
  50. image_config = file_extra_config.image_config
  51. # check if image file feature is enabled
  52. if not image_config:
  53. continue
  54. # Validate number of files
  55. if len(files) > image_config['number_limits']:
  56. raise ValueError(f"Number of image files exceeds the maximum limit {image_config['number_limits']}")
  57. for file_obj in file_objs:
  58. # Validate transfer method
  59. if file_obj.transfer_method.value not in image_config['transfer_methods']:
  60. raise ValueError(f'Invalid transfer method: {file_obj.transfer_method.value}')
  61. # Validate file type
  62. if file_obj.type != FileType.IMAGE:
  63. raise ValueError(f'Invalid file type: {file_obj.type}')
  64. if file_obj.transfer_method == FileTransferMethod.REMOTE_URL:
  65. # check remote url valid and is image
  66. result, error = self._check_image_remote_url(file_obj.url)
  67. if result is False:
  68. raise ValueError(error)
  69. elif file_obj.transfer_method == FileTransferMethod.LOCAL_FILE:
  70. # get upload file from upload_file_id
  71. upload_file = (db.session.query(UploadFile)
  72. .filter(
  73. UploadFile.id == file_obj.related_id,
  74. UploadFile.tenant_id == self.tenant_id,
  75. UploadFile.created_by == user.id,
  76. UploadFile.created_by_role == ('account' if isinstance(user, Account) else 'end_user'),
  77. UploadFile.extension.in_(IMAGE_EXTENSIONS)
  78. ).first())
  79. # check upload file is belong to tenant and user
  80. if not upload_file:
  81. raise ValueError('Invalid upload file')
  82. new_files.append(file_obj)
  83. # return all file objs
  84. return new_files
  85. def transform_message_files(self, files: list[MessageFile], file_extra_config: FileExtraConfig) -> list[FileVar]:
  86. """
  87. transform message files
  88. :param files:
  89. :param file_extra_config:
  90. :return:
  91. """
  92. # transform files to file objs
  93. type_file_objs = self._to_file_objs(files, file_extra_config)
  94. # return all file objs
  95. return [file_obj for file_objs in type_file_objs.values() for file_obj in file_objs]
  96. def _to_file_objs(self, files: list[Union[dict, MessageFile]],
  97. file_extra_config: FileExtraConfig) -> dict[FileType, list[FileVar]]:
  98. """
  99. transform files to file objs
  100. :param files:
  101. :param file_extra_config:
  102. :return:
  103. """
  104. type_file_objs: dict[FileType, list[FileVar]] = {
  105. # Currently only support image
  106. FileType.IMAGE: []
  107. }
  108. if not files:
  109. return type_file_objs
  110. # group by file type and convert file args or message files to FileObj
  111. for file in files:
  112. if isinstance(file, MessageFile):
  113. if file.belongs_to == FileBelongsTo.ASSISTANT.value:
  114. continue
  115. file_obj = self._to_file_obj(file, file_extra_config)
  116. if file_obj.type not in type_file_objs:
  117. continue
  118. type_file_objs[file_obj.type].append(file_obj)
  119. return type_file_objs
  120. def _to_file_obj(self, file: Union[dict, MessageFile], file_extra_config: FileExtraConfig) -> FileVar:
  121. """
  122. transform file to file obj
  123. :param file:
  124. :return:
  125. """
  126. if isinstance(file, dict):
  127. transfer_method = FileTransferMethod.value_of(file.get('transfer_method'))
  128. if transfer_method != FileTransferMethod.TOOL_FILE:
  129. return FileVar(
  130. tenant_id=self.tenant_id,
  131. type=FileType.value_of(file.get('type')),
  132. transfer_method=transfer_method,
  133. url=file.get('url') if transfer_method == FileTransferMethod.REMOTE_URL else None,
  134. related_id=file.get('upload_file_id') if transfer_method == FileTransferMethod.LOCAL_FILE else None,
  135. extra_config=file_extra_config
  136. )
  137. return FileVar(
  138. tenant_id=self.tenant_id,
  139. type=FileType.value_of(file.get('type')),
  140. transfer_method=transfer_method,
  141. url=None,
  142. related_id=file.get('tool_file_id'),
  143. extra_config=file_extra_config
  144. )
  145. else:
  146. return FileVar(
  147. id=file.id,
  148. tenant_id=self.tenant_id,
  149. type=FileType.value_of(file.type),
  150. transfer_method=FileTransferMethod.value_of(file.transfer_method),
  151. url=file.url,
  152. related_id=file.upload_file_id or None,
  153. extra_config=file_extra_config
  154. )
  155. def _check_image_remote_url(self, url):
  156. try:
  157. headers = {
  158. "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
  159. }
  160. def is_s3_presigned_url(url):
  161. try:
  162. parsed_url = urlparse(url)
  163. if 'amazonaws.com' not in parsed_url.netloc:
  164. return False
  165. query_params = parse_qs(parsed_url.query)
  166. required_params = ['Signature', 'Expires']
  167. for param in required_params:
  168. if param not in query_params:
  169. return False
  170. if not query_params['Expires'][0].isdigit():
  171. return False
  172. signature = query_params['Signature'][0]
  173. if not re.match(r'^[A-Za-z0-9+/]+={0,2}$', signature):
  174. return False
  175. return True
  176. except Exception:
  177. return False
  178. if is_s3_presigned_url(url):
  179. response = requests.get(url, headers=headers, allow_redirects=True)
  180. if response.status_code in {200, 304}:
  181. return True, ""
  182. response = requests.head(url, headers=headers, allow_redirects=True)
  183. if response.status_code in {200, 304}:
  184. return True, ""
  185. else:
  186. return False, "URL does not exist."
  187. except requests.RequestException as e:
  188. return False, f"Error checking URL: {e}"