molzi3d-next/scripts/reconvert-content.mjs

/**
 * Re-Konvertierung: Direkt aus Original-HTML-Backup.
 * Entfernt WordPress-Nav/Footer, extrahiert nur den Content-Block,
 * konvertiert HTML-Tabellen sauber zu Markdown-Tabellen.
 */
import fs from "fs";
import path from "path";
import { NodeHtmlMarkdown } from "node-html-markdown";

const BACKUP_POSTS = path.resolve("backup/content/posts");
const BACKUP_PAGES = path.resolve("backup/content/pages");
const OUT_GUIDES = path.resolve("app/src/content/guides");
const OUT_PAGES = path.resolve("app/src/content/pages");

const SKIP_SLUGS = [
  "guide-1-pla-perfekt-einstellen-2026-03-25",
  "guide-2-stringing-reduzieren-2026-03-25",
  "guide-1-warping-vermeiden-2026-03-26",
  "guide-2-petg-ohne-frust-2026-03-26",
];

const categorize = (slug) => {
  const s = slug.toLowerCase();
  if (s.includes("guide-orcaslicer") || s.includes("guide-cura") || s.includes("guide-bambu") || s.includes("guide-prusaslicer") || s.includes("slicer")) return "Slicer";
  if (s.includes("pla") || s.includes("petg") || s.includes("tpu") || s.includes("asa") || s.includes("abs") || s.includes("nylon") || s.includes("carbon") || s.includes("resin") || s.includes("filament") || s.includes("bed-adhesion")) return "Materialien";
  if (s.includes("stringing") || s.includes("warping") || s.includes("unterextrusion") || s.includes("layer-separation") || s.includes("elefantenfuss") || s.includes("verstopfte")) return "Fehlerbehebung";
  if (s.includes("retraction") || s.includes("flow-rate") || s.includes("pressure-advance") || s.includes("input-shaping") || s.includes("temperaturturm") || s.includes("speed-tower") || s.includes("erste-schicht") || s.includes("druckbett-leveln")) return "Kalibrierung";
  if (s.includes("adaptive") || s.includes("modifier") || s.includes("ironing") || s.includes("fuzzy") || s.includes("multi-material") || s.includes("klipper")) return "Fortgeschritten";
  if (s.includes("erstes-modell") || s.includes("support") || s.includes("infill") || s.includes("duesenwechsel") || s.includes("druckzeit") || s.includes("masshaltigkeit") || s.includes("bruecken") || s.includes("nachbearbeiten") || s.includes("gridfinity") || s.includes("naht")) return "Grundlagen";
  return "Allgemein";
};

const difficulty = (slug, category) => {
  if (slug.includes("erstes-modell") || slug.includes("erste-schicht") || slug.includes("druckbett-leveln")) return "einsteiger";
  if (category === "Fortgeschritten" || slug.includes("klipper") || slug.includes("pressure-advance") || slug.includes("input-shaping") || slug.includes("carbon") || slug.includes("nylon-pa")) return "experte";
  return "fortgeschritten";
};

// Erstelle NHM mit besserer Konfiguration
const nhm = new NodeHtmlMarkdown({
  keepDataImages: false,
  useLinkReferenceDefinitions: false,
  ignore: ["style", "script", "nav", "footer", "button"],
  // Beibehalte Tabellen-Struktur
});

const extractContent = (rawHtml) => {
  // 1. Alles vor dem eigentlichen Content entfernen
  //    Content beginnt nach dem letzten Nav/Mobile-Menu-Block beim ersten <h1> oder <h2>
  let html = rawHtml;

  // Nav-Bloecke entfernen (alles in <nav> tags und v2-nav divs)
  html = html.replace(/<nav[^>]*>[\s\S]*?<\/nav>/gi, "");
  html = html.replace(/<div[^>]*class="v2-mobile-menu"[^>]*>[\s\S]*?<\/div>/gi, "");

  // Footer entfernen
  html = html.replace(/<footer[^>]*>[\s\S]*?<\/footer>/gi, "");

  // Inline styles aus Tabellen entfernen (die verursachen die Backslash-Probleme)
  html = html.replace(/\sstyle="[^"]*"/gi, "");

  // Wrapping divs mit Klassen entfernen, Content behalten
  html = html.replace(/<div[^>]*class="(hub-section|v2-guide|v2-footer)[^"]*"[^>]*>/gi, "");

  // Lead-Paragraphen: class entfernen
  html = html.replace(/<p\s+class="lead"[^>]*>/gi, "<p>");

  // Leere divs entfernen
  html = html.replace(/<div[^>]*>\s*<\/div>/gi, "");

  return html;
};

const processFile = (filePath, outDir) => {
  const raw = fs.readFileSync(filePath, "utf-8");

  const fmMatch = raw.match(/^---\n([\s\S]*?)\n---\n([\s\S]*)$/);
  if (!fmMatch) return null;

  const fmBlock = fmMatch[1];
  const htmlContent = fmMatch[2].trim();

  const titleMatch = fmBlock.match(/title:\s*"(.+?)"/);
  const slugMatch = fmBlock.match(/slug:\s*"(.+?)"/);
  const excerptMatch = fmBlock.match(/excerpt:\s*"(.*)"/);

  const title = titleMatch?.[1] ?? path.basename(filePath, ".html");
  const slug = slugMatch?.[1] ?? path.basename(filePath, ".html");
  const excerpt = excerptMatch?.[1] ?? "";

  if (SKIP_SLUGS.includes(slug)) return null;

  // Content extrahieren und bereinigen
  const cleanHtml = extractContent(htmlContent);

  // HTML -> Markdown
  let md = nhm.translate(cleanHtml);

  // Post-Cleanup
  md = md
    // WordPress-Kommentare entfernen
    .replace(/<!-- \/?wp:[^\s]+ -->/g, "")
    // Escaped newlines
    .replace(/\\n/g, "")
    // Einzelne Backslashes auf eigener Zeile
    .replace(/^\s*\\\s*$/gm, "")
    // Backslash am Zeilenanfang/-ende (Artefakte von <br>)
    .replace(/^\\ /gm, "")
    .replace(/ \\$/gm, "")
    // Inline-Backslashes vor/nach Text
    .replace(/\\\s{2,}/g, " ")
    // Doppelte Leerzeilen reduzieren
    .replace(/\n{3,}/g, "\n\n")
    // WordPress-Nav Reste
    .replace(/M0LZI[_]3D[\s\S]*?\[SYS_BOOT[^\]]*\]/g, "")
    .replace(/\[WISSEN\].*\[FAQ\]/gs, "")
    .replace(/JOIN_NETWORK.*?→/g, "")
    .replace(/M0LZI[_]3D © \d{4}/g, "")
    .replace(/\[WHATSAPP\].*?\[DATENSCHUTZ\].*?$/gm, "")
    .trim();

  // Kaputte Tabellen fixen: Zeilen die nur aus | \ \ \ | bestehen → entfernen
  md = md.replace(/^\|[\s\\|]+\|$/gm, "");
  // Leere Tabellenkoepfe entfernen
  md = md.replace(/^\|\s*\|\s*$/gm, "");

  // Nochmal doppelte Leerzeilen
  md = md.replace(/\n{3,}/g, "\n\n");

  const cat = categorize(slug);
  const diff = difficulty(slug, cat);

  const frontmatter = `---\ntitle: "${title}"\nslug: "${slug}"\ncategory: "${cat}"\ndifficulty: "${diff}"\nexcerpt: "${excerpt}"\n---`;

  const output = `${frontmatter}\n\n${md}\n`;
  const outPath = path.join(outDir, `${slug}.md`);
  fs.writeFileSync(outPath, output, "utf-8");
  return slug;
};

// Main
console.log("=== Re-Konvertierung aus Original-HTML ===\n");

console.log("Posts -> Guides:");
const postFiles = fs.readdirSync(BACKUP_POSTS).filter(f => f.endsWith(".html"));
let ok = 0;
for (const file of postFiles) {
  const result = processFile(path.join(BACKUP_POSTS, file), OUT_GUIDES);
  if (result) { console.log(`  OK: ${result}`); ok++; }
}
console.log(`\n${ok} Guides konvertiert.\n`);

console.log("Pages:");
const pageFiles = fs.readdirSync(BACKUP_PAGES).filter(f => f.endsWith(".html"));
let pOk = 0;
for (const file of pageFiles) {
  const result = processFile(path.join(BACKUP_PAGES, file), OUT_PAGES);
  if (result) { console.log(`  OK: ${result}`); pOk++; }
}
console.log(`\n${pOk} Pages konvertiert.`);

// Verifizierung
console.log("\n=== Verifizierung ===");
const guideFiles = fs.readdirSync(OUT_GUIDES).filter(f => f.endsWith(".md"));
let issues = 0;
for (const file of guideFiles) {
  const content = fs.readFileSync(path.join(OUT_GUIDES, file), "utf-8");
  const backslashLines = (content.match(/^\s*\\\s*$/gm) || []).length;
  const brokenTables = (content.match(/\| \\ \\ \\/g) || []).length;
  if (backslashLines > 0 || brokenTables > 0) {
    console.log(`  ISSUE: ${file} (${backslashLines} backslash-lines, ${brokenTables} broken tables)`);
    issues++;
  }
}
console.log(issues === 0 ? "  Alle Guides sauber!" : `  ${issues} Guides mit Problemen`);