jiazi-gala-poster-demo/scripts/export_pdf.py

"""将文字画页面导出为 PDF，画面四边各留恰好 1em。

页面为 28 × 44 格文字画，1em 即一个汉字格。导出页尺寸为 30em × 46em，
场景四周各留 1em；底色铺满整页。

默认把 page1、page2、page3 依次合成一份三页 PDF。

用法：
    uv run python scripts/export_pdf.py                                   # 三页 → dist/三页demo.pdf
    uv run python scripts/export_pdf.py --page page2.html -o dist/page2.pdf
    uv run python scripts/export_pdf.py --pages page1.html page3.html -o dist/选页.pdf
    uv run python scripts/export_pdf.py --em 16 -o dist/三页demo-16.pdf
"""

from __future__ import annotations

import argparse
import io
import pathlib

from playwright.sync_api import sync_playwright
from pypdf import PdfWriter

ROOT = pathlib.Path(__file__).resolve().parents[1]

COLS = 28
ROWS = 44
MARGIN_EM = 1
DEFAULT_PAGES = ("page1.html", "page2.html", "page3.html")


def export_css(em: float) -> tuple[str, float, float]:
    page_w = (COLS + MARGIN_EM * 2) * em
    page_h = (ROWS + MARGIN_EM * 2) * em
    css = f"""
@page {{ size: {page_w:g}px {page_h:g}px; margin: 0; }}

html, body {{
  width: {page_w:g}px !important;
  height: {page_h:g}px !important;
  min-height: 0 !important;
  margin: 0 !important;
  display: block !important;
  overflow: hidden !important;
  background: #000 !important;
}}

body {{ padding: {em:g}px !important; }}

.scene {{
  font-size: {em:g}px !important;
  width: {COLS}em !important;
  height: {ROWS}em !important;
  margin: 0 !important;
}}

* {{
  -webkit-print-color-adjust: exact;
  print-color-adjust: exact;
}}
"""
    return css, page_w, page_h


def launch_browser(pw):
    for name, kwargs in (
        ("Microsoft Edge", {"channel": "msedge"}),
        ("Google Chrome", {"channel": "chrome"}),
        ("Playwright Chromium", {}),
    ):
        try:
            return pw.chromium.launch(**kwargs), name
        except Exception:
            continue
    raise SystemExit(
        "未找到可用浏览器。请先运行：uv run playwright install chromium"
    )


def render_pdf(browser, html, css, em, page_w, page_h):
    """渲染单页并返回 PDF 字节；导出前校验画面四边恰好留 1em。"""
    page = browser.new_page(
        viewport={"width": round(page_w), "height": round(page_h)},
        device_scale_factor=1,
    )
    try:
        page.goto(html.as_uri(), wait_until="load")
        page.evaluate("() => document.fonts.ready")
        if not page.evaluate("document.fonts.check('16px Zpix')"):
            raise SystemExit(f"{html.name}：Zpix 字体未加载，无法导出。")

        page.add_style_tag(content=css)
        page.wait_for_timeout(200)

        metrics = page.evaluate(
            """() => {
                const scene = document.querySelector('.scene').getBoundingClientRect();
                const body = getComputedStyle(document.body);
                return {
                    x: scene.x, y: scene.y, w: scene.width, h: scene.height,
                    pad: [body.paddingTop, body.paddingRight, body.paddingBottom, body.paddingLeft],
                };
            }"""
        )
        expected_pad = f"{em:g}px"
        if (
            abs(metrics["x"] - em) > 0.5
            or abs(metrics["y"] - em) > 0.5
            or abs(metrics["w"] - COLS * em) > 0.5
            or abs(metrics["h"] - ROWS * em) > 0.5
            or any(v != expected_pad for v in metrics["pad"])
        ):
            raise SystemExit(f"{html.name}：导出布局校验失败：{metrics}")

        return page.pdf(
            width=f"{page_w:g}px",
            height=f"{page_h:g}px",
            margin={"top": "0", "right": "0", "bottom": "0", "left": "0"},
            print_background=True,
            prefer_css_page_size=True,
        )
    finally:
        page.close()


def export(htmls, out, em):
    css, page_w, page_h = export_css(em)

    with sync_playwright() as pw:
        browser, browser_name = launch_browser(pw)
        blobs = [render_pdf(browser, html, css, em, page_w, page_h) for html in htmls]
        browser.close()

    writer = PdfWriter()
    for blob in blobs:
        writer.append(io.BytesIO(blob))

    out.parent.mkdir(parents=True, exist_ok=True)
    with out.open("wb") as fh:
        writer.write(fh)
    writer.close()

    mm_w = page_w / 96 * 25.4
    mm_h = page_h / 96 * 25.4
    print(f"浏览器：{browser_name}")
    print(f"页面：{page_w:g} × {page_h:g}px（约 {mm_w:.0f} × {mm_h:.0f} mm），四边各 {em:g}px（1em）")
    for index, html in enumerate(htmls, start=1):
        print(f"  第 {index} 页：{html.name}")
    print(f"已导出 {len(htmls)} 页：{out.relative_to(ROOT)}")


def main() -> None:
    parser = argparse.ArgumentParser(
        description="将文字画页面导出为 PDF（四周各留 1em，可多页合一）"
    )
    parser.add_argument(
        "--pages",
        "--page",
        dest="pages",
        nargs="+",
        default=list(DEFAULT_PAGES),
        help="要导出的 HTML，按顺序合成多页；默认 page1 page2 page3",
    )
    parser.add_argument(
        "--em",
        type=float,
        default=24,
        help="导出时 1em（一个汉字格）的像素大小，默认 24",
    )
    parser.add_argument(
        "-o",
        "--out",
        type=pathlib.Path,
        default=ROOT / "dist" / "三页demo.pdf",
        help="输出文件路径，默认 dist/三页demo.pdf",
    )
    args = parser.parse_args()
    if args.em <= 0:
        parser.error("--em 必须大于 0")

    htmls = []
    for name in args.pages:
        html = (ROOT / name).resolve()
        if not html.is_file():
            parser.error(f"页面不存在：{name}")
        htmls.append(html)

    export(htmls, args.out.resolve(), args.em)


if __name__ == "__main__":
    main()