Note

Adds a new DeepSeekOllamaOCR provider using Ollama’s deepseek-ocr:3b model and wires it into infer_provider via deepseek_ollama.

• OCR Providers:
  • New DeepSeekOllamaOCR in src/upsonic/ocr/deepseek_ollama.py:
    • Initializes Ollama Client, validates dependency, and warns on unsupported languages.
    • Processes images via temp file, calls client.chat with <image> prompt, returns OCRResult with metadata.
    • Provides supported_languages, customizable host, model, and prompt.
• Factory Wiring:
  • Adds deepseek_ollama -> upsonic.ocr.deepseek_ollama.DeepSeekOllamaOCR to infer_provider map in src/upsonic/ocr/ocr.py.

^{Written by Cursor Bugbot for commit 782a452. This will update automatically on new commits. Configure here.}