/**
 * Konvertiert WordPress HTML-Exports in saubere Markdown-Dateien
 * mit Frontmatter fuer das Next.js-Projekt.
 */
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");

// Duplikate/veraltete Slugs (die mit Datum-Suffix)
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",
];

// Kategorie-Zuordnung basierend auf Slug-Patterns
const categorize = (slug, title) => {
  const s = slug.toLowerCase();
  const t = title.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";
};

// Schwierigkeitsgrad
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";
};

const nhm = new NodeHtmlMarkdown({
  keepDataImages: false,
  useLinkReferenceDefinitions: false,
});

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

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

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

  // Felder parsen
  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)) {
    console.log(`  SKIP (Duplikat): ${slug}`);
    return null;
  }

  // HTML -> Markdown
  let markdown = nhm.translate(htmlContent);

  // Cleanup: Elementor-Artefakte, leere Divs, etc.
  markdown = markdown
    .replace(/\[vc_[^\]]*\]/g, "")
    .replace(/\[\/vc_[^\]]*\]/g, "")
    .replace(/\[elementor[^\]]*\]/g, "")
    .replace(/\n{3,}/g, "\n\n")
    .trim();

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

  const frontmatter = [
    "---",
    `title: "${title}"`,
    `slug: "${slug}"`,
    `category: "${cat}"`,
    `difficulty: "${diff}"`,
    `excerpt: "${excerpt}"`,
    "---",
  ].join("\n");

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

// Main
console.log("=== WordPress -> Markdown Konvertierung ===\n");

fs.mkdirSync(OUT_GUIDES, { recursive: true });
fs.mkdirSync(OUT_PAGES, { recursive: true });

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

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