From 2abcda73e82b6db3b87694bff85abba8a01d7f07 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Sebastian=20Kr=C3=BCger?= <valknar@pivoine.art>
Date: Sun, 3 May 2026 19:08:35 +0200
Subject: [PATCH] Add Ghostscript post-processing to pdf.js for KDP compliance

- Puppeteer writes to a temp kaleidoskop-raw.pdf
- Ghostscript re-processes with -dPDFSETTINGS=/prepress to fully embed
  all fonts and prevent subsetting (KDP checklist item 6)
- PDF info dict is written via PostScript pdfmarks, with title/author/
  subject encoded as UTF-16BE hex to support German umlauts (item 13)
- File size drops from ~108 MB to ~9 MB (placeholder art; will grow with
  real 300 DPI illustrations but stays well under KDP's 650 MB limit)
- GS failure falls back gracefully to the raw Puppeteer PDF

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
 scripts/pdf.js | 97 ++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 87 insertions(+), 10 deletions(-)

diff --git a/scripts/pdf.js b/scripts/pdf.js
index b5b810a..ce03c65 100644
--- a/scripts/pdf.js
+++ b/scripts/pdf.js
@@ -1,16 +1,76 @@
 import puppeteer from 'puppeteer';
 import { resolve, dirname } from 'path';
 import { fileURLToPath } from 'url';
-import { access } from 'fs/promises';
+import { access, stat, readFile, writeFile, rename, unlink } from 'fs/promises';
+import { execFile } from 'child_process';
+import { promisify } from 'util';
+import matter from 'gray-matter';
 
+const execFileAsync = promisify(execFile);
 const __dir = dirname(fileURLToPath(import.meta.url));
+const root = resolve(__dir, '..');
+const inputPath  = resolve(root, 'output', 'book.html');
+const rawPath    = resolve(root, 'output', 'kaleidoskop-raw.pdf');
+const marksPath  = resolve(root, 'output', 'pdfmarks.ps');
+const outputPath = resolve(root, 'output', 'kaleidoskop.pdf');
 
 async function fileExists(path) {
   try { await access(path); return true; } catch { return false; }
 }
-const root = resolve(__dir, '..');
-const inputPath = resolve(root, 'output', 'book.html');
-const outputPath = resolve(root, 'output', 'kaleidoskop.pdf');
+
+// Encode a JS string as UTF-16BE hex for PostScript pdfmarks (handles umlauts etc.)
+function toUtf16BeHex(str) {
+  const hex = [];
+  for (let i = 0; i < str.length; i++) {
+    const cp = str.charCodeAt(i);
+    hex.push(((cp >> 8) & 0xff).toString(16).padStart(2, '0'));
+    hex.push((cp & 0xff).toString(16).padStart(2, '0'));
+  }
+  return 'FEFF' + hex.join('').toUpperCase();
+}
+
+function buildPdfmarks({ title, author, subject }) {
+  return [
+    '[ /Title <' + toUtf16BeHex(title)   + '>',
+    '  /Author <' + toUtf16BeHex(author)  + '>',
+    '  /Subject <' + toUtf16BeHex(subject) + '>',
+    '  /Creator (Puppeteer + Ghostscript)',
+    '  /Producer (kaleidoskop build pipeline)',
+    '  /DOCINFO pdfmark',
+  ].join('\n');
+}
+
+async function runGhostscript({ title, author, subject }) {
+  const marks = buildPdfmarks({ title, author, subject });
+  await writeFile(marksPath, marks, 'utf-8');
+
+  const gs = 'gs';
+  const args = [
+    '-dBATCH', '-dNOPAUSE', '-dQUIET',
+    '-sDEVICE=pdfwrite',
+    '-dCompatibilityLevel=1.4',
+    '-dPDFSETTINGS=/prepress',
+    '-dEmbedAllFonts=true',
+    '-dSubsetFonts=false',
+    '-dDownsampleColorImages=false',
+    '-dDownsampleGrayImages=false',
+    '-dDownsampleMonoImages=false',
+    `-sOutputFile=${outputPath}`,
+    marksPath,
+    rawPath,
+  ];
+
+  try {
+    await execFileAsync(gs, args);
+  } catch (err) {
+    console.error('Ghostscript failed:', err.message);
+    // Fall back to raw PDF so the build still produces output
+    await rename(rawPath, outputPath);
+    console.warn('Falling back to raw Puppeteer PDF (no GS post-processing).');
+    return false;
+  }
+  return true;
+}
 
 async function generate() {
   try {
@@ -20,9 +80,13 @@ async function generate() {
     process.exit(1);
   }
 
-  // On ARM64 (e.g. WSL2 on Apple Silicon / Raspberry Pi), Puppeteer's bundled
-  // Chrome is x86-64 and won't run. Use the system Chromium instead:
-  //   sudo apt-get install -y chromium
+  // Read metadata from front-matter for PDF info dict
+  const fmRaw = await readFile(resolve(root, 'content', '00-front-matter.md'), 'utf-8');
+  const { data: fm } = matter(fmRaw);
+  const title   = fm.title   || 'Das Kaleidoskop der Schlummerwelten';
+  const author  = fm.author  || '';
+  const subject = fm.subtitle || 'Kinderbuch';
+
   const systemChromium = '/usr/bin/chromium';
   const useSystem = await fileExists(systemChromium);
 
@@ -35,9 +99,9 @@ async function generate() {
 
   await page.goto(`file://${inputPath}`, { waitUntil: 'networkidle0' });
 
-  // 8.75 × 8.75 inches = trim (8.5×8.5) + 0.125in bleed on each side
+  // Write to temp file first so Ghostscript can read it
   await page.pdf({
-    path: outputPath,
+    path: rawPath,
     width: '8.75in',
     height: '8.75in',
     printBackground: true,
@@ -45,7 +109,20 @@ async function generate() {
   });
 
   await browser.close();
-  console.log(`PDF written to output/kaleidoskop.pdf`);
+
+  const rawSize = (await stat(rawPath)).size;
+  console.log(`Raw PDF: ${(rawSize / 1_048_576).toFixed(1)} MB`);
+
+  console.log('Running Ghostscript (font embedding + metadata)…');
+  const gsOk = await runGhostscript({ title, author, subject });
+
+  if (gsOk) {
+    const finalSize = (await stat(outputPath)).size;
+    console.log(`Final PDF: ${(finalSize / 1_048_576).toFixed(1)} MB`);
+    await unlink(rawPath).catch(() => {});
+    await unlink(marksPath).catch(() => {});
+    console.log('PDF written to output/kaleidoskop.pdf (fonts embedded, metadata set)');
+  }
 }
 
 generate().catch(err => { console.error(err); process.exit(1); });