Coverage for src/qdrant_loader/core/file_conversion/conversion_config.py: 100%

1"""Configuration models for file conversion settings.""" 

2 

3from pydantic import BaseModel, Field 

4 

5 

6class MarkItDownConfig(BaseModel): 

7 """Configuration for MarkItDown-specific settings.""" 

8 

9 enable_llm_descriptions: bool = Field( 

10 default=False, description="Enable LLM integration for image descriptions" 

11 ) 

12 

13 llm_model: str = Field( 

14 default="gpt-4o", description="LLM model for image descriptions (when enabled)" 

15 ) 

16 

17 llm_endpoint: str = Field( 

18 default="https://api.openai.com/v1", description="LLM endpoint (when enabled)" 

19 ) 

20 

21 llm_api_key: str | None = Field( 

22 default=None, 

23 description="API key for LLM service (required when enable_llm_descriptions is True)", 

24 ) 

25 

26 

27class FileConversionConfig(BaseModel): 

28 """Configuration for file conversion operations.""" 

29 

30 max_file_size: int = Field( 

31 default=52428800, # 50MB 

32 description="Maximum file size for conversion (in bytes)", 

33 gt=0, 

34 le=104857600, # 100MB 

35 ) 

36 

37 conversion_timeout: int = Field( 

38 default=300, # 5 minutes 

39 description="Timeout for conversion operations (in seconds)", 

40 gt=0, 

41 le=3600, # 1 hour 

42 ) 

43 

44 markitdown: MarkItDownConfig = Field( 

45 default_factory=MarkItDownConfig, description="MarkItDown specific settings" 

46 ) 

47 

48 def get_max_file_size_mb(self) -> float: 

49 """Get maximum file size in megabytes. 

50 

51 Returns: 

52 Maximum file size in MB 

53 """ 

54 return self.max_file_size / (1024 * 1024) 

55 

56 def is_file_size_allowed(self, file_size: int) -> bool: 

57 """Check if file size is within allowed limits. 

58 

59 Args: 

60 file_size: File size in bytes 

61 

62 Returns: 

63 True if file size is allowed, False otherwise 

64 """ 

65 return file_size <= self.max_file_size 

66 

67 

68class ConnectorFileConversionConfig(BaseModel): 

69 """Configuration for file conversion at the connector level.""" 

70 

71 enable_file_conversion: bool = Field( 

72 default=False, description="Enable file conversion for this connector" 

73 ) 

74 

75 download_attachments: bool | None = Field( 

76 default=None, 

77 description="Download and process attachments (for Confluence/JIRA/PublicDocs)", 

78 ) 

79 

80 def should_download_attachments(self) -> bool: 

81 """Check if attachments should be downloaded. 

82 

83 Returns: 

84 True if attachments should be downloaded, False otherwise 

85 """ 

86 # Default to False if not specified 

87 return self.download_attachments is True