PDF to OFD High-Fidelity Converter

🎯 Purpose

A specialized skill for converting PDF documents into the Chinese National Standard OFD (GB/T 33190-2016) format. Optimized for Electronic Invoices (OFD版式发票) with advanced rendering capabilities that exceed standard conversion libraries.

✨ Key Features

• High-Fidelity Text Placement: Uses character-level positioning (DeltaX arrays) and baseline origin data extracted via rawdict to ensure text layout is 100% identical to the source PDF.
• Advanced Vector Graphics: Directly extracts original stroke colors, fill colors, and line widths. Supports complex path types and fill instructions.
• Transparency Preservation: Fully supports Alpha and FillOpacity for vector paths and SMask transparency for images (e.g., electronic seals and signatures).
• Cross-Platform Font Mapping: Intelligent mapping of macOS-specific (STSong, STKaiti) and Windows-specific font names to standardized OFD font names (宋体, 楷体, 黑体).
• In-Memory Packaging: Generates the final OFD zip structure entirely in memory to avoid temporary file clutter and ensure security.
• Color Snapping: Heuristic "Invoice Red" correction (128 0 0) for financial documents while preserving non-standard colors.

🛠️ Usage Instructions

When a user asks to convert a PDF or a "High-Fidelity" invoice to OFD:

1. Direct Execution:

```bash

python3 pdf2ofd.py [output_path.ofd]

```

1. Plugin Integration:

The script implements a PDF2OFDConverter class that can be easily imported and used in other Python workflows.

Example Output

Success: /path/to/invoice.ofd

📦 Requirements

Dependencies required in the environment:

• PyMuPDF (fitz): For advanced PDF parsing and raw character data extraction.
• Pillow: For image processing and transparency handling.
• easyofd: The base library for OFD structure (extended via internal monkey patches).
• xmltodict: For XML manipulation.

💡 Notes

• This skill uses deep monkey-patching on easyofd to fix known library limitations regarding character positioning and resource ID tracking.
• The conversion process assumes standard Chinese fonts (SimSun, KaiTi, SimHei) are available on the viewing system.
• Zero-copy resource handling: Images are extracted and re-compressed as PNG/JPG only when necessary to preserve quality.