PYTHON pdf_generator.py

# ─── PDF Generation ────────────────────────────────────────────────────────────
#
# WeasyPrint PDF generation and PDF storage operations.
#
# ─── Imports ───────────────────────────────────────────────────────────────────

from pathlib import Path
import uuid

from weasyprint import HTML


# ─── Constants ─────────────────────────────────────────────────────────────────

# Where generated PDFs live on disk
PDF_STORAGE_DIR = Path("/app/data/documents")


# ─── PDF Generation ────────────────────────────────────────────────────────────

def generate_pdf(html_content: str, style: dict = None) -> bytes:
    """Generate PDF from HTML content using WeasyPrint.

    Args:
        html_content: Rendered HTML string (templates own their @page rules)
        style: Style config (passed to templates via render_document_html)

    Returns:
        PDF bytes

    Note: Templates define their own @page rules inline. We do NOT inject
    an external stylesheet here — that was overriding template @page/body
    rules and breaking full-bleed layouts (Stripe, HubSpot, Notion, etc.)
    which use @page { margin: 0 } with internal padding.
    """
    html_doc = HTML(string=html_content, base_url=str(Path(__file__).parent.parent))
    return html_doc.write_pdf()


# ─── PDF Storage ───────────────────────────────────────────────────────────────

def save_pdf(pdf_bytes: bytes, document_id: str) -> str:
    """Save PDF bytes to disk. Returns relative path."""
    PDF_STORAGE_DIR.mkdir(parents=True, exist_ok=True)
    filename = f"{document_id}.pdf"
    filepath = PDF_STORAGE_DIR / filename
    filepath.write_bytes(pdf_bytes)
    return str(filepath)


def get_pdf_path(document_id: str) -> Path:
    """Get absolute path for a stored PDF."""
    return PDF_STORAGE_DIR / f"{document_id}.pdf"