#!/usr/bin/env python3 from __future__ import annotations import argparse import html import json import os import shutil import subprocess import sys import tempfile from pathlib import Path DEFAULT_TITLE = "Audit export" DEFAULT_FORMAT = "A4" PREMIUM_PRINT_CSS = r''' :root { color-scheme: light; --pdf-bg: #f5f7fb; --pdf-surface: #ffffff; --pdf-surface-soft: #f8fafc; --pdf-ink: #0f172a; --pdf-muted: #475569; --pdf-border: rgba(15, 23, 42, 0.10); --pdf-shadow: 0 12px 28px rgba(15, 23, 42, 0.08); --pdf-accent: #2563eb; --pdf-success: #16a34a; --pdf-warning: #d97706; --pdf-danger: #dc2626; } @page { size: A4 landscape; margin: 12mm; } html, body { margin: 0 !important; padding: 0 !important; background: var(--pdf-bg) !important; color: var(--pdf-ink) !important; font-family: Inter, ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", Arial, sans-serif !important; font-size: 12px !important; line-height: 1.45 !important; -webkit-print-color-adjust: exact !important; print-color-adjust: exact !important; } * { box-sizing: border-box !important; } a { color: inherit !important; text-decoration: none !important; } body { min-height: 100vh; } .simple-chat-export-root { width: 100% !important; } .simple-chat-message__artifacts, .simple-chat-export-root > * { width: 100% !important; max-width: none !important; } .simple-chat-export-root button, .simple-chat-export-root .js-artifact-export-button, .simple-chat-export-root .simple-chat-artifact__actions, .simple-chat-export-root .simple-chat-message__artifacts-toolbar, .simple-chat-export-root .audit-feed-block__actions, .simple-chat-export-root [data-export-ignore="1"] { display: none !important; } .simple-chat-export-root [data-export-text-value="1"] { white-space: pre-wrap !important; min-height: 36px !important; } .simple-chat-export-root input, .simple-chat-export-root textarea, .simple-chat-export-root select, .simple-chat-export-root .audit-export-field-value, .simple-chat-export-root [data-export-text-value="1"] { display: block !important; width: 100% !important; border: 1px solid var(--pdf-border) !important; border-radius: 12px !important; background: #fff !important; color: var(--pdf-ink) !important; padding: 10px 12px !important; box-shadow: none !important; } .simple-chat-export-root .simple-chat-message__section, .simple-chat-export-root .simple-audit-dashboard__section, .simple-chat-export-root .simple-chat-chart-card, .simple-chat-export-root .audit-feed-block, .simple-chat-export-root .simple-chat-linked-post, .simple-chat-export-root .simple-chat-linked-comment, .simple-chat-export-root .simple-chat-artifact, .simple-chat-export-root .simple-chat-artifact__content, .simple-chat-export-root .simple-audit-card, .simple-chat-export-root .simple-audit-pill { break-inside: avoid !important; page-break-inside: avoid !important; } .simple-chat-export-root .simple-chat-message__section, .simple-chat-export-root .simple-audit-dashboard__section, .simple-chat-export-root .audit-feed-block, .simple-chat-export-root .simple-chat-linked-post, .simple-chat-export-root .simple-chat-linked-comment, .simple-chat-export-root .simple-chat-artifact, .simple-chat-export-root .simple-chat-artifact__content { margin: 0 0 14px !important; padding: 14px 16px !important; border: 1px solid var(--pdf-border) !important; border-radius: 18px !important; background: var(--pdf-surface) !important; box-shadow: var(--pdf-shadow) !important; } .simple-chat-export-root .simple-audit-dashboard__head { display: flex !important; justify-content: space-between !important; align-items: flex-start !important; gap: 16px !important; margin: 0 0 14px !important; padding: 0 0 12px !important; border-bottom: 1px solid var(--pdf-border) !important; } .simple-chat-export-root h1, .simple-chat-export-root h2, .simple-chat-export-root h3, .simple-chat-export-root h4, .simple-chat-export-root h5, .simple-chat-export-root h6 { margin: 0 0 8px !important; color: var(--pdf-ink) !important; line-height: 1.2 !important; } .simple-chat-export-root p, .simple-chat-export-root li, .simple-chat-export-root span, .simple-chat-export-root div, .simple-chat-export-root td, .simple-chat-export-root th, .simple-chat-export-root small, .simple-chat-export-root label { color: inherit !important; } .simple-chat-export-root .simple-audit-overview-grid, .simple-chat-export-root .simple-audit-comparison-grid, .simple-chat-export-root .simple-audit-comments-grid, .simple-chat-export-root .simple-audit-dashboard__split, .simple-chat-export-root .simple-audit-embedded-charts, .simple-chat-export-root .simple-chat-chart__grid, .simple-chat-export-root .simple-chat-audit-blocks, .simple-chat-export-root .simple-audit-dashboard__cards, .simple-chat-export-root .simple-audit-dashboard__groups, .simple-chat-export-root .simple-audit-pills { display: flex !important; flex-wrap: wrap !important; gap: 12px !important; align-items: stretch !important; } .simple-chat-export-root .simple-audit-overview-grid__cards, .simple-chat-export-root .simple-audit-overview-grid__charts, .simple-chat-export-root .simple-audit-comparison-grid__table, .simple-chat-export-root .simple-audit-comparison-grid__charts, .simple-chat-export-root .simple-audit-comments-grid__charts, .simple-chat-export-root .simple-audit-comments-grid__groups, .simple-chat-export-root .simple-audit-dashboard__split-col { flex: 1 1 320px !important; min-width: 280px !important; } .simple-chat-export-root .simple-audit-card, .simple-chat-export-root .simple-audit-pill, .simple-chat-export-root .simple-chat-chart-card, .simple-chat-export-root [class*="metric"], .simple-chat-export-root [class*="kpi"], .simple-chat-export-root [class*="stat"] { flex: 1 1 220px !important; min-width: 180px !important; padding: 14px !important; border-radius: 16px !important; border: 1px solid var(--pdf-border) !important; background: var(--pdf-surface-soft) !important; box-shadow: none !important; } .simple-chat-export-root .simple-audit-card__label, .simple-chat-export-root .simple-audit-pill__label, .simple-chat-export-root .simple-chat-chart-card__head-sub, .simple-chat-export-root .simple-chat-chart-card__range, .simple-chat-export-root .audit-feed-block__description, .simple-chat-export-root .audit-feed-block__connector-summary, .simple-chat-export-root .simple-chat-chart__note, .simple-chat-export-root .simple-audit-dashboard__meta { color: var(--pdf-muted) !important; font-size: 10px !important; } .simple-chat-export-root .simple-audit-card__value, .simple-chat-export-root .simple-audit-pill__value { display: block !important; margin-top: 6px !important; font-size: 24px !important; font-weight: 700 !important; line-height: 1.05 !important; color: var(--pdf-ink) !important; } .simple-chat-export-root .simple-audit-pill--positive { background: #ecfdf5 !important; border-color: rgba(22,163,74,.20) !important; } .simple-chat-export-root .simple-audit-pill--negative { background: #fef2f2 !important; border-color: rgba(220,38,38,.18) !important; } .simple-chat-export-root .simple-audit-pill--warning { background: #fffbeb !important; border-color: rgba(217,119,6,.20) !important; } .simple-chat-export-root .simple-audit-pill--neutral { background: #eff6ff !important; border-color: rgba(37,99,235,.18) !important; } .simple-chat-export-root table { width: 100% !important; border-collapse: collapse !important; background: #fff !important; border-radius: 12px !important; overflow: hidden !important; } .simple-chat-export-root thead { display: table-header-group !important; } .simple-chat-export-root th, .simple-chat-export-root td { padding: 10px 12px !important; border: 1px solid rgba(15, 23, 42, 0.08) !important; text-align: left !important; vertical-align: top !important; } .simple-chat-export-root th { background: #eff6ff !important; font-weight: 700 !important; } .simple-chat-export-root tr, .simple-chat-export-root img, .simple-chat-export-root svg, .simple-chat-export-root canvas, .simple-chat-export-root .chart, .simple-chat-export-root .graph { break-inside: avoid !important; page-break-inside: avoid !important; } .simple-chat-export-root img, .simple-chat-export-root svg, .simple-chat-export-root canvas { max-width: 100% !important; height: auto !important; } ''' HEADER_TEMPLATE = """
""".strip() FOOTER_TEMPLATE = """
Ubisoft Audit Export /
""".strip() class PdfExportError(RuntimeError): pass def parse_args() -> argparse.Namespace: parser = argparse.ArgumentParser( description="Render simple-chat-message__artifacts HTML to PDF via PrinceXML or Playwright/Chromium." ) parser.add_argument("output_positional", nargs="?", help="Output PDF path (legacy positional form)") parser.add_argument("--input", help="Input HTML fragment file path. Reads stdin if omitted.") parser.add_argument("--output", help="Output PDF path.") parser.add_argument("--title", default=DEFAULT_TITLE, help="Document title") parser.add_argument("--format", default=DEFAULT_FORMAT, help="Paper format for Playwright fallback") parser.add_argument("--landscape", action="store_true", help="Landscape orientation") parser.add_argument("--base-url", default="", help="Base URL used to resolve relative assets") parser.add_argument("--prefer-prince", action="store_true", help="Try PrinceXML first") parser.add_argument("--engine", choices=["auto", "prince", "playwright"], default="auto") parser.add_argument("--timeout-ms", type=int, default=45000, help="Render timeout in milliseconds") args = parser.parse_args() if not args.output: args.output = args.output_positional if not args.output: parser.error("an output path is required via --output or positional output") return args def read_input_payload(path: str | None) -> tuple[str, dict]: if path: return Path(path).read_text(encoding="utf-8"), {} raw = sys.stdin.read() if not raw.strip(): raise PdfExportError("No input HTML provided.") try: payload = json.loads(raw) except json.JSONDecodeError: return raw, {} if isinstance(payload, dict): html_value = str(payload.get("html") or "") if html_value.strip(): return html_value, payload return raw, {} def normalize_fragment(fragment_html: str) -> str: body = fragment_html.strip() if not body: raise PdfExportError("Empty artifacts HTML.") return body def build_document(fragment_html: str, title: str, base_url: str, landscape: bool, paper_format: str) -> str: safe_title = html.escape(title or DEFAULT_TITLE, quote=True) safe_base = html.escape(base_url, quote=True) if base_url else "" page_css = f"@page {{ size: {paper_format}{' landscape' if landscape else ''}; margin: 12mm; }}" return f""" {safe_title} {f'' if safe_base else ''}
{fragment_html}
""" def ensure_pdf(path: Path) -> None: if not path.exists(): raise PdfExportError("No PDF file was generated.") if path.stat().st_size < 800: raise PdfExportError("Generated PDF is too small.") with path.open("rb") as handle: if not handle.read(5).startswith(b"%PDF-"): raise PdfExportError("Generated file is not a valid PDF.") def discover_playwright_browser() -> str | None: for name in ("chromium", "google-chrome", "google-chrome-stable", "chrome", "chromium-browser"): path = shutil.which(name) if path: return path return None def render_with_prince(document_html: str, output_path: Path, title: str) -> None: prince_bin = shutil.which("prince") if not prince_bin: raise PdfExportError("PrinceXML not found in PATH.") with tempfile.TemporaryDirectory(prefix="audit_pdf_prince_") as tmp_dir: html_path = Path(tmp_dir) / "document.html" css_path = Path(tmp_dir) / "print.css" html_path.write_text(document_html, encoding="utf-8") css_path.write_text(PREMIUM_PRINT_CSS, encoding="utf-8") cmd = [ prince_bin, str(html_path), "--output", str(output_path), "--style", str(css_path), "--pdf-title", title or DEFAULT_TITLE, "--pdf-profile", "pdf/ua-1", "--tagged-pdf", ] proc = subprocess.run(cmd, capture_output=True, text=True) if proc.returncode != 0: stderr = (proc.stderr or proc.stdout or "").strip() raise PdfExportError(f"PrinceXML render failed: {stderr}") ensure_pdf(output_path) def render_with_playwright(document_html: str, output_path: Path, title: str, base_url: str, paper_format: str, landscape: bool, timeout_ms: int) -> None: try: from playwright.sync_api import sync_playwright except Exception as exc: # noqa: BLE001 raise PdfExportError(f"Playwright unavailable: {exc}") from exc executable_path = discover_playwright_browser() launch_options = {"headless": True} if executable_path: launch_options["executable_path"] = executable_path with sync_playwright() as p: browser = p.chromium.launch(**launch_options) try: page = browser.new_page(viewport={"width": 1600, "height": 1100}, device_scale_factor=1.25) page.emulate_media(media="screen") page.set_content(document_html, wait_until="load") page.wait_for_timeout(350) page.evaluate( """ async () => { if (document.fonts && document.fonts.ready) { try { await document.fonts.ready; } catch (e) {} } const waitForImages = Array.from(document.images || []).map((img) => { if (img.complete) return Promise.resolve(); return new Promise((resolve) => { img.addEventListener('load', resolve, { once: true }); img.addEventListener('error', resolve, { once: true }); }); }); await Promise.all(waitForImages); } """ ) page.pdf( path=str(output_path), format=paper_format, landscape=landscape, print_background=True, prefer_css_page_size=True, display_header_footer=True, header_template=HEADER_TEMPLATE, footer_template=FOOTER_TEMPLATE, margin={"top": "18mm", "right": "10mm", "bottom": "16mm", "left": "10mm"}, outline=True, tagged=True, timeout=timeout_ms, ) finally: browser.close() ensure_pdf(output_path) def main() -> int: args = parse_args() output_path = Path(args.output).expanduser().resolve() output_path.parent.mkdir(parents=True, exist_ok=True) try: input_html, payload = read_input_payload(args.input) title = str(payload.get("title") or args.title or DEFAULT_TITLE) base_url = str(payload.get("base_url") or args.base_url or "") paper_format = str(payload.get("format") or args.format or DEFAULT_FORMAT) landscape = bool(payload.get("landscape")) or bool(args.landscape) fragment_html = normalize_fragment(input_html) document_html = build_document(fragment_html, title=title, base_url=base_url, landscape=landscape, paper_format=paper_format) errors: list[str] = [] engine = args.engine order = [] if engine == "prince": order = ["prince"] elif engine == "playwright": order = ["playwright"] elif args.prefer_prince: order = ["prince", "playwright"] else: order = ["playwright", "prince"] for selected in order: try: if selected == "playwright": render_with_playwright( document_html=document_html, output_path=output_path, title=title, base_url=base_url, paper_format=paper_format, landscape=landscape, timeout_ms=args.timeout_ms, ) else: render_with_prince(document_html=document_html, output_path=output_path, title=title) return 0 except Exception as exc: # noqa: BLE001 errors.append(f"{selected}: {exc}") raise PdfExportError(" | ".join(errors) if errors else "No PDF engine succeeded.") except Exception as exc: # noqa: BLE001 sys.stderr.write(f"PDF generation failed: {exc}") return 1 if __name__ == "__main__": raise SystemExit(main())