| name | Dynamic Image to PDF Converter |
| description | Automatically transcribes, structures, and compiles images or text drafts into professional, print-ready vector A4 PDFs using WeasyPrint. Universal - handles any content type: badges, code blocks, tables, blockquotes, RTL text, images, and more. |
Dependency Management
Important: Use uv for all package and environment management. Do NOT use pip directly. Refer to global standards in ~/.claude/docs/tooling/package-management.md.
Supported Markdown + HTML Features
The compiler supports a rich set of Markdown extensions and raw HTML:
| Feature | Syntax | Example |
|---|
| Badges/Chips | Raw HTML <span> | <span class="badge orange">REACT</span> |
| Code blocks | Fenced code | ```python ... ``` |
| Project trees | Fenced code (text) | ```text ... ``` |
| Tables | Markdown pipes | | Col1 | Col2 | |
| Blockquotes | > prefix | > quoted text |
| RTL text | HTML wrapper | <div class="urdu-text">...</div> |
| Images | Markdown/HTML |  or <img> |
| Horizontal rules | --- | --- |
| Lists | - or 1. | Unordered / ordered |
| Table of Contents | HTML <a> + .toc | <div class="toc"><a href="#sec1">Title</a></div> |
| Repeating Headers | <thead> in HTML | <table><thead>...</thead></table> |
| Section Headers | .section-header | <div class="section-header">My Section</div> |
Available badge classes: orange, grey, purple, pink, blue-purple, blue, light-blue, green, red, yellow, teal, indigo, cyan, rose, slate, amber, lime, emerald, sky, violet, fuchsia
Utility classes: text-center, text-left, text-right, text-sm, text-lg, mt-0, mb-0, page-break
Process Workflow
When a user provides images or text to convert into a professional PDF, execute this sequence:
-
Analysis & Structure Extraction
- Read and extract text from the images.
- Visual Audit: Analyze the source image for non-text elements:
- Badges/Chips: Identify labels and assign the closest color class.
- Code Blocks/Trees: Identify project structures or code snippets.
- Tables: Identify tabular data with headers and rows.
- Layout: Note specific alignments (centered headers, justified text).
- Separate content logically into sections using Markdown headers (
# Title), metadata, and bullet points.
- For multilingual content (e.g., Urdu), wrap in
<div class="urdu-text"> for RTL support.
-
Payload Generation
- Write the extracted content into a
.md file using Markdown + inline HTML where needed.
- Styling rules:
- Badges:
<div class="badge-container"><span class="badge color">TEXT</span></div>
- Code/trees: Use fenced code blocks (
text ... ) to preserve formatting.
- Tables: Use standard Markdown pipe syntax.
- Blockquotes: Use
> prefix.
- Call:
python scripts/compile_pdf.py <input_file.md> <output_file.pdf>
-
Verification & Iterative Refinement
- Confirm the output is a professional vector PDF.
- Check for: tree alignment, header spacing, badge colors, table formatting.
- If layout issues exist, refine the CSS in
scripts/compile_pdf.py and re-compile.
Customization
To add new badge colors or utility classes, edit scripts/compile_pdf.py and add CSS rules under the /* ---- Badges ---- */ or /* ---- Utility ---- */ sections. The compiler auto-bootstraps its venv on first run.
