Feat: Support for PDF

[tushar-prox]

Convert PDF pages to JPEG instead of PNG

Unable to test locally so this is in draft state

Changes

• Modified PDF conversion options to use JPEG format
• Updated MIME types in data URLs to consistently use image/jpeg
• Simplified image format handling in API calls

flowchart TD
    A[Start OCR] --> B{Check File Type}
    B -->|PDF| C{Is Remote?}
    B -->|Image| D{Is Remote?}
    
    C -->|Yes| E[Fetch PDF Buffer]
    C -->|No| F[Read Local PDF]
    
    E --> G[Convert PDF to JPEG]
    F --> G
    
    D -->|Yes| H[Use Direct URL]
    D -->|No| I[Encode to Base64]
    
    G --> J[Create Base64 Image URL]
    H --> K[Process with Together AI]
    I --> K
    J --> K
    
    K --> L[Return Markdown]

    style A fill:#f9f,stroke:#333
    style L fill:#9f9,stroke:#333

Loading

Why

• JPEG format provides better compression for document images while maintaining readable quality
• Smaller file sizes lead to faster processing and reduced memory usage
• More consistent handling of image formats throughout the pipeline

Testing

• Test PDF conversion with local files
• Test PDF conversion with remote files
• Verify image quality is sufficient for OCR processing

[Nutlope]

Love this @tushar-prox! Lemme know when this is ready to review, was planning to work on this feature anyway