clawd/tools/generate_pdf.py

#!/usr/bin/env python3
"""
Generate PDF from markdown content.
Outputs PDF to stdout as binary.
Simple, robust approach focusing on text content.
"""

import sys
import json
from pathlib import Path

# Read JSON from stdin
input_data = json.load(sys.stdin)
markdown_content = input_data.get('markdown', '')
filename = input_data.get('filename', 'document.pdf')

try:
    from fpdf import FPDF
    import re
    
    # Create PDF
    pdf = FPDF(format='A4')
    pdf.add_page()
    pdf.set_margins(12, 12, 12)
    
    # Try to use DejaVu font for Romanian support
    try:
        dejavu_path = Path("/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf")
        if dejavu_path.exists():
            pdf.add_font("DejaVu", "", str(dejavu_path))
            pdf.add_font("DejaVu", "B", "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf")
            oblique_path = Path("/usr/share/fonts/truetype/dejavu/DejaVuSans-Oblique.ttf")
            if oblique_path.exists():
                pdf.add_font("DejaVu", "I", str(oblique_path))
            bold_oblique = Path("/usr/share/fonts/truetype/dejavu/DejaVuSans-BoldOblique.ttf")
            if bold_oblique.exists():
                pdf.add_font("DejaVu", "BI", str(bold_oblique))
            pdf.set_font("DejaVu", "", 10)
            use_dejavu = True
        else:
            raise Exception("DejaVu font not found")
    except:
        pdf.set_font("Helvetica", "", 10)
        use_dejavu = False
    
    def render_rich_text(pdf, text, use_dejavu, size):
        """Render text with inline **bold** and *italic* formatting."""
        font_name = "DejaVu" if use_dejavu else "Helvetica"
        # Remove links but keep text
        text = re.sub(r'\[(.*?)\]\(.*?\)', r'\1', text)
        # Split on bold/italic markers, process segments
        # Pattern: **bold**, __bold__, *italic*, _italic_
        parts = re.split(r'(\*\*.*?\*\*|__.*?__|\*.*?\*|_.*?_)', text)
        for part in parts:
            if not part:
                continue
            if part.startswith('**') and part.endswith('**'):
                pdf.set_font(font_name, "B", size)
                pdf.write(5, part[2:-2])
                pdf.set_font(font_name, "", size)
            elif part.startswith('__') and part.endswith('__'):
                pdf.set_font(font_name, "B", size)
                pdf.write(5, part[2:-2])
                pdf.set_font(font_name, "", size)
            elif part.startswith('*') and part.endswith('*') and len(part) > 2:
                try:
                    pdf.set_font(font_name, "I", size)
                except:
                    pass  # italic not available, keep current font
                pdf.write(5, part[1:-1])
                pdf.set_font(font_name, "", size)
            elif part.startswith('_') and part.endswith('_') and len(part) > 2:
                try:
                    pdf.set_font(font_name, "I", size)
                except:
                    pass
                pdf.write(5, part[1:-1])
                pdf.set_font(font_name, "", size)
            else:
                pdf.write(5, part)
    
    # Parse markdown line by line
    lines = markdown_content.split('\n')
    i = 0
    
    while i < len(lines):
        line = lines[i]
        
        # Skip empty lines but add spacing
        if not line.strip():
            pdf.ln(2)
            i += 1
            continue
        
        # H1 - Main heading
        if line.startswith('# '):
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "B", 16)
            text = line.replace('# ', '', 1).strip()
            pdf.multi_cell(0, 7, text, ln=True)
            pdf.ln(1)
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "", 10)
        
        # H2 - Section heading
        elif line.startswith('## '):
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "B", 12)
            text = line.replace('## ', '', 1).strip()
            pdf.multi_cell(0, 6, text, ln=True)
            pdf.ln(0.5)
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "", 10)
        
        # H3 - Subsection
        elif line.startswith('### '):
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "B", 11)
            text = line.replace('### ', '', 1).strip()
            pdf.multi_cell(0, 5, text, ln=True)
            pdf.ln(0.3)
            pdf.set_font("DejaVu" if use_dejavu else "Helvetica", "", 10)
        
        # Bullet point
        elif line.strip().startswith('- ') or line.strip().startswith('* '):
            text = line.strip().lstrip('-*').strip()
            pdf.write(5, '- ')
            render_rich_text(pdf, text, use_dejavu, 10)
            pdf.ln(5)
        
        # Numbered list
        elif re.match(r'^\s*(\d+\.)\s', line):
            m = re.match(r'^\s*(\d+\.)\s(.*)', line)
            prefix = m.group(1) + ' '
            text = m.group(2)
            pdf.write(5, prefix)
            render_rich_text(pdf, text, use_dejavu, 10)
            pdf.ln(5)
        
        # Regular text with formatting
        else:
            text = line.strip()
            if text:
                render_rich_text(pdf, text, use_dejavu, 10)
                pdf.ln(5)
        
        i += 1
    
    # Output PDF
    pdf_bytes = pdf.output()
    sys.stdout.buffer.write(pdf_bytes)
    sys.exit(0)

except Exception as e:
    error_json = json.dumps({'error': str(e)})
    sys.stderr.write(error_json)
    sys.exit(1)