Metadata-Version: 2.4
Name: pdf2word-hebrew
Version: 0.1.0
Summary: Convert any PDF to RTL Hebrew .docx using Claude vision
License: MIT
Project-URL: Homepage, https://github.com/itamar-shamir/pdf2word-hebrew
Project-URL: Bug Tracker, https://github.com/itamar-shamir/pdf2word-hebrew/issues
Keywords: pdf,word,hebrew,rtl,docx,claude,ai
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.11
Description-Content-Type: text/markdown
Requires-Dist: anthropic>=0.100.0
Requires-Dist: python-docx>=1.1.0
Requires-Dist: markitdown>=0.1.0
Requires-Dist: pypdfium2>=4.0.0
Requires-Dist: Pillow>=10.0.0

# pdf2word-hebrew

Convert any PDF to a properly formatted RTL Hebrew `.docx` Word document.

Uses Claude AI vision to accurately read Hebrew text, tables, and diagrams — even from scanned or image-based PDFs.

## Install

```bash
pip install pdf2word-hebrew
```

## Usage

```bash
# Convert PDF to Word (Hebrew RTL, Claude vision)
pdf2word document.pdf

# Custom output path
pdf2word document.pdf output.docx

# Fast mode — text extraction only, no AI (may garble Hebrew)
pdf2word document.pdf --fast
```

## Setup

Requires an Anthropic API key:

```bash
export ANTHROPIC_API_KEY=sk-ant-...
```

Or a corporate LLM proxy:

```bash
export ANTHROPIC_BASE_URL=http://your-proxy:port/anthropic/
export ANTHROPIC_AUTH_TOKEN=your-token
```

## How it works

1. **Default mode** — rasterizes each PDF page at 144 DPI, sends to Claude vision, gets clean Markdown back.
2. **`--fast` mode** — extracts embedded text with markitdown (free, no AI, but may not handle Hebrew RTL correctly).
3. Converts Markdown → `.docx` with full RTL support: right-aligned paragraphs, Hebrew-aware tables, Arial font.

## License

MIT
