MediaWiki:Gadget-LabelScan.js: Unterschied zwischen den Versionen

Version vom 5. November 2025, 23:20 Uhr

/* global mw, Tesseract */
(function () {
  'use strict';

  // =============================
  //   KONFIGURATION
  // =============================

  // ← Für Tests leer lassen: const ADOS_CATEGORIES = [];
  const ADOS_CATEGORIES = [
    'Alle A Dream of Scotland Abfüllungen',
    'Alle A Dream of Ireland Abfüllungen',
    'Alle A Dream of... – Der Rest der Welt Abfüllungen',
    'Friendly Mr. Z Whiskytainment Abfüllungen',
    'Die Whisky Elfen Abfüllungen',
    'The Fine Art of Whisky Abfüllungen',
    'Alle Rumbastic Abfüllungen'
  ];

  const KNOWN_TOKENS = [
    // Serien / Reihen
    'The Tasteful 8', 'Heroes of Childhood', 'A Dream of Scotland',
    'A Dream of Ireland', 'The Fine Art of Whisky', 'Friendly Mr. Z',
    'Die Whisky Elfen', 'Rumbastic',

    // Brennereien / Regionen
    'Ardbeg','Ardmore','Arran','Auchroisk','Ben Nevis','Blair Athol','Bowmore',
    'Caol Ila','Clynelish','Glenallachie','Glenrothes','Longmorn','Lagavulin',
    'Tullibardine','Dalmore','Benrinnes','Mortlach','Glenlivet','Inchgower',
    'Islay','Speyside','Highland','Lowland','Campbeltown','Ireland','Irland',

    // typische Label-Wörter
    'Cask Strength','First Fill','Bourbon Barrel','Sherry','PX','Oloroso'
  ];

  // =============================
  //   UI-Hilfen
  // =============================

  function hasUI () {
    return !!document.getElementById('ados-scan-run') &&
           !!document.getElementById('ados-scan-file');
  }

  function setStatus (t) {
    var el = document.getElementById('ados-scan-status');
    if (el) el.textContent = t || '';
  }

  function setProgress (p) {
    var bar = document.getElementById('ados-scan-progress');
    if (!bar) return;
    if (p == null) { bar.hidden = true; bar.value = 0; }
    else { bar.hidden = false; bar.value = Math.max(0, Math.min(1, p)); }
  }

  function showPreview (file) {
    var url = URL.createObjectURL(file);
    var prev = document.getElementById('ados-scan-preview');
    if (prev) {
      prev.innerHTML = '<img alt="Vorschau" src="' + url + '">';
      prev.setAttribute('aria-hidden', 'false');
    }
  }

  function showOCRText (t) {
    var el = document.getElementById('ados-scan-ocr');
    if (el) el.textContent = (t || '').trim();
  }

  // =============================
  //   Tesseract laden (nur 1x)
  // =============================

  var tesseractReady;
  function ensureTesseract () {
    if (tesseractReady) return tesseractReady;
    tesseractReady = new Promise(function (resolve, reject) {
      if (window.Tesseract) return resolve();
      var s = document.createElement('script');
      s.src = 'https://cdn.jsdelivr.net/npm/tesseract.js@5/dist/tesseract.min.js';
      s.async = true;
      s.onload = resolve;
      s.onerror = function () {
        var s2 = document.createElement('script');
        s2.src = 'https://unpkg.com/tesseract.js@5/dist/tesseract.min.js';
        s2.async = true;
        s2.onload = resolve;
        s2.onerror = function () { reject(new Error('Tesseract konnte nicht geladen werden')); };
        document.head.appendChild(s2);
      };
      document.head.appendChild(s);
    });
    return tesseractReady;
  }

  // =============================
  //   Vorverarbeitung (OCR)
  //   Graustufen + Unsharp + adaptive Schwelle
  // =============================

  async function preprocessImage (file) {
    const img = await new Promise((res, rej) => {
      const o = new Image();
      o.onload = () => res(o);
      o.onerror = rej;
      o.src = URL.createObjectURL(file);
    });

    const MAX = 1800;
    const s = Math.min(1, (img.width > img.height) ? MAX / img.width : MAX / img.height);
    const w = Math.round(img.width * s), h = Math.round(img.height * s);

    const c = document.createElement('canvas'); c.width = w; c.height = h;
    const g = c.getContext('2d', { willReadFrequently: true });
    g.imageSmoothingEnabled = true;
    g.drawImage(img, 0, 0, w, h);

    // → Graustufen
    let id = g.getImageData(0, 0, w, h), d = id.data;
    for (let i=0;i<d.length;i+=4){
      const y = 0.2126*d[i] + 0.7152*d[i+1] + 0.0722*d[i+2];
      d[i]=d[i+1]=d[i+2]=y;
    }
    g.putImageData(id, 0, 0);

    // → Unsharp (leichter Hochpass)
    id = g.getImageData(0,0,w,h); d = id.data;
    const copy = new Uint8ClampedArray(d);
    const idx = (x,y)=>4*(y*w+x);
    for (let y=1;y<h-1;y++){
      for (let x=1;x<w-1;x++){
        const i0=idx(x,y), a=copy[i0], b=copy[idx(x-1,y)], c0=copy[idx(x+1,y)],
              d0=copy[idx(x,y-1)], e=copy[idx(x,y+1)];
        const lap = 4*a - b - c0 - d0 - e;
        const v = Math.max(0, Math.min(255, a + 0.3*lap));
        d[i0]=d[i0+1]=d[i0+2]=v;
      }
    }
    g.putImageData(id,0,0);

    // → adaptive Schwelle (lokaler Mittelwert)
    const win = 25, half = (win|0);
    id = g.getImageData(0,0,w,h); d = id.data;
    for (let y=0;y<h;y++){
      for (let x=0;x<w;x++){
        let sum=0, cnt=0;
        for (let yy=Math.max(0,y-half); yy<=Math.min(h-1,y+half); yy+=5){
          for (let xx=Math.max(0,x-half); xx<=Math.min(w-1,x+half); xx+=5){
            sum += d[4*(yy*w+xx)];
            cnt++;
          }
        }
        const thr = (sum/cnt) - 6;
        const i = 4*(y*w+x);
        const v = d[i] < thr ? 0 : 255;
        d[i]=d[i+1]=d[i+2]=v;
      }
    }
    g.putImageData(id,0,0);

    return c;
  }

  // Hilfsfunktionen für Varianten
  function crop(canvas, x, y, w, h){
    const c = document.createElement('canvas'); c.width=w; c.height=h;
    c.getContext('2d').drawImage(canvas, x, y, w, h, 0, 0, w, h);
    return c;
  }
  function rotate(canvas, deg){
    const r = document.createElement('canvas');
    const ctx = r.getContext('2d');
    if (deg % 180 === 0){ r.width=canvas.width; r.height=canvas.height; }
    else { r.width=canvas.height; r.height=canvas.width; }
    ctx.translate(r.width/2, r.height/2);
    ctx.rotate(deg*Math.PI/180);
    ctx.drawImage(canvas, -canvas.width/2, -canvas.height/2);
    return r;
  }

  async function ocrOne(canvas, lang) {
    const res = await Tesseract.recognize(canvas, lang, {
      // Sparse text funktioniert bei Labels (verschieden orientierte Textblöcke)
      tessedit_pageseg_mode: 11,
      preserve_interword_spaces: 1
    });
    return { text: (res?.data?.text||'').trim(), conf: res?.data?.confidence||0 };
  }

  // =============================
  //   Mehrfach-OCR (Rotationen/Regionen) + Fallback-Sprache
  // =============================

  async function runOCR(file){
    await ensureTesseract();
    setProgress(0.01);
    const base = await preprocessImage(file);

    // Kandidatenflächen
    const variants = [];
    variants.push(base); // komplett
    variants.push(crop(base, 0, 0, Math.round(base.width*0.4), base.height)); // linke Spalte
    variants.push(crop(base, 0, Math.round(base.height*0.72), base.width, Math.round(base.height*0.28))); // unteres Banner

    // + Rotationen
    const more = [];
    for (const v of variants){
      more.push(v, rotate(v, 90), rotate(v, -90));
    }

    // zwei Sprachmodi testen
    const results = [];
    for (const canv of more){
      for (const lang of ['deu+eng','eng']){
        try {
          const r = await ocrOne(canv, lang);
          results.push(r);
        } catch(e){ /* einzelne Fehlschläge ignorieren */ }
      }
    }
    setProgress(null);

    results.sort((a,b)=> (b.conf||0)-(a.conf||0));
    return (results[0]?.text)||'';
  }

  // =============================
  //   Hinweise aus OCR
  // =============================

  function extractHints (text) {
    const raw = String(text || '').replace(/\s+/g, ' ').trim();

    const foundNames = [];
    KNOWN_TOKENS.forEach(t => {
      const re = new RegExp('\\b' + t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&') + '\\b', 'i');
      if (re.test(raw)) foundNames.push(t);
    });

    const ages = [];
    const ageRe = /\b([1-9]\d?)\s?(?:years?|yo|jahr(?:e)?)\b/gi;
    let m;
    while ((m = ageRe.exec(raw)) !== null) {
      const n = m[1]; if (!ages.includes(n)) ages.push(n);
    }

    const years = [];
    const yearRe = /\b(19|20)\d{2}\b/g;
    while ((m = yearRe.exec(raw)) !== null) {
      if (!years.includes(m[0])) years.push(m[0]);
    }

    const wordRe = /\b[A-ZÄÖÜ][A-Za-zÄÖÜäöüß\-]{3,}\b/g;
    const uniq = new Set(); let w; const words = [];
    while ((w = wordRe.exec(raw)) !== null) {
      const s = w[0];
      if (!uniq.has(s)) { uniq.add(s); words.push(s); if (words.length >= 8) break; }
    }

    return { names: foundNames, ages, years, words, raw };
  }

  // =============================
  //   Suche (3 Pässe) + Fallbacks
  // =============================

  function esc (s) { return mw.html.escape(String(s || '')); }

  function incatStr () {
    return (ADOS_CATEGORIES || []).map(c => 'incategory:"' + c + '"').join(' ');
  }

  async function searchWikiSmart (hints, limit) {
    await mw.loader.using(['mediawiki.api','mediawiki.util','mediawiki.html']);
    const api = new mw.Api();
    const ns0 = 0;
    const MAX = limit || 12;

    // PASS 1: intitle-Kombis (präzise)
    const pass1 = [];
    if (hints.names.length) {
      hints.names.forEach(n => {
        if (hints.ages.length) hints.ages.forEach(a => pass1.push(`intitle:"${n}" intitle:${a} ${incatStr()}`));
        if (hints.years.length) hints.years.forEach(y => pass1.push(`intitle:"${n}" "${y}" ${incatStr()}`));
        pass1.push(`intitle:"${n}" ${incatStr()}`);
      });
    }

    // PASS 2: gewichtete Volltextsuche
    const key = []
      .concat(hints.names.slice(0, 2), hints.ages.slice(0, 1), hints.years.slice(0, 1), hints.words.slice(0, 3))
      .map(x => `"${x}"`).join(' ');
    const pass2 = key ? [ `${key} ${incatStr()}` ] : [];

    // PASS 3: Prefix auf Titel
    const pass3 = [];
    if (hints.names.length) pass3.push(hints.names[0]);
    if (!pass3.length && hints.words.length) pass3.push(hints.words[0]);

    const seen = new Set(); const out = [];

    async function runSr (q) {
      const r = await api.get({ action: 'query', list: 'search', srsearch: q, srnamespace: ns0, srlimit: MAX, formatversion: 2 });
      (r.query?.search || []).forEach(it => {
        const k = it.title;
        if (seen.has(k)) return;
        seen.add(k);
        out.push(it);
      });
    }

    for (const q of pass1) { await runSr(q); if (out.length >= MAX) return out.slice(0, MAX); }
    for (const q of pass2) { await runSr(q); if (out.length >= MAX) return out.slice(0, MAX); }

    // Prefix (list=prefixsearch)
    for (const p of pass3) {
      const r = await api.get({ action: 'query', list: 'prefixsearch', pssearch: p, psnamespace: ns0, pslimit: MAX });
      (r.query?.prefixsearch || []).forEach(it => {
        const title = it.title || it['*'];
        const k = title;
        if (seen.has(k)) return;
        seen.add(k);
        out.push({ title, snippet: '' });
      });
      if (out.length >= MAX) break;
    }

    return out.slice(0, MAX);
  }

  // ganz einfacher Fuzzy-Fallback auf Suchergebnissen
  function scoreTitle(title, hints){
    const t = String(title||'').toLowerCase();
    let s = 0;
    hints.names.forEach(n => { if (t.includes(n.toLowerCase())) s += 1.0; });
    hints.words.forEach(n => { if (t.includes(n.toLowerCase())) s += 0.4; });
    hints.ages.forEach(a => { if (t.includes(String(a))) s += 0.4; });
    hints.years.forEach(y => { if (t.includes(String(y))) s += 0.4; });
    return s;
  }

  async function fallbackFuzzyTitles(hints, limit){
    await mw.loader.using('mediawiki.api');
    const api = new mw.Api();
    const MAX = limit || 12;

    // Breite Suche mit Tokens (mit/ohne Kategorie)
    const q1 = []
      .concat(hints.names.slice(0,2), hints.ages.slice(0,1), hints.years.slice(0,1), hints.words.slice(0,3))
      .map(x => `"${x}"`).join(' ');
    const q = `${q1} ${incatStr()}`.trim();

    const r = await api.get({ action:'query', list:'search', srsearch:q || hints.raw.split(/\s+/).slice(0,6).join(' '), srlimit:50, formatversion:2 });
    const items = (r.query?.search || []);
    const scored = items.map(it => ({ ...it, _score: scoreTitle(it.title, hints) }));
    scored.sort((a,b)=> b._score - a._score);
    const top = scored.slice(0, MAX).filter(x=> x._score >= 0.10); // großzügiger
    return top;
  }

  async function broadSearchNoCategory(hints, limit){
    await mw.loader.using('mediawiki.api');
    const api = new mw.Api();
    const MAX = limit || 12;

    const parts = []
      .concat(hints.names.slice(0,2), hints.ages.slice(0,1), hints.years.slice(0,1), hints.words.slice(0,3))
      .map(x => `"${x}"`);
    const q = parts.length ? parts.join(' ') : hints.raw.split(/\s+/).slice(0,6).join(' ');

    const r = await api.get({ action:'query', list:'search', srsearch:q, srlimit:MAX, formatversion:2 });
    return (r.query?.search || []);
  }

  // =============================
  //   Ergebnisse rendern
  // =============================

  function renderResults (items) {
    var box = document.getElementById('ados-scan-results');
    if (!box) return;
    box.innerHTML = '';
    if (!items || !items.length) {
      box.innerHTML = '<div class="ados-hit">Keine klaren Treffer. Bitte anderes Foto oder manuell suchen.</div>';
      return;
    }
    items.slice(0, 12).forEach(function (it) {
      var title = it.title || '';
      var link = mw.util.getUrl(title.replace(/ /g, '_'));
      var snip = String(it.snippet || '').replace(/<\/?span[^>]*>/g, '').replace(/&quot;/g, '"');
      var div = document.createElement('div');
      div.className = 'ados-hit';
      div.innerHTML =
        '<b><a href="' + link + '">' + esc(title) + '</a></b>' +
        (snip ? '<div class="meta">' + snip + '</div>' : '');
      box.appendChild(div);
    });
  }

  // =============================
  //   Binding
  // =============================

  var BOUND = false;
  function bind () {
    if (BOUND || !hasUI()) return;

    var runBtn = document.getElementById('ados-scan-run');
    var fileIn = document.getElementById('ados-scan-file');
    var bigBtn = document.getElementById('ados-scan-bigbtn');
    var form = document.getElementById('ados-scan-form');

    if (!runBtn || !fileIn) return;
    if (runBtn.dataset.bound === '1') return;
    runBtn.dataset.bound = '1'; BOUND = true;

    if (bigBtn) bigBtn.addEventListener('click', function () { fileIn.click(); });
    fileIn.addEventListener('change', function () {
      if (this.files && this.files[0]) showPreview(this.files[0]);
    });

    function onSubmit(ev){
      ev.preventDefault();
      if (!(fileIn.files && fileIn.files[0])) { alert('Bitte ein Foto auswählen oder aufnehmen.'); return; }
      var f = fileIn.files[0];
      (async function(){
        try {
          runBtn.disabled = true; runBtn.textContent = 'Erkenne …';
          setStatus('Erkenne Label …');
          const text = await runOCR(f);
          showOCRText(text);

          setStatus('Suche im Wiki …');
          const hints = extractHints(text);

          let hits = await searchWikiSmart(hints, 12);
          if (!hits || !hits.length) {
            setStatus('Kein direkter Treffer – Fuzzy über Kategorien …');
            hits = await fallbackFuzzyTitles(hints, 12);
          }
          if (!hits || !hits.length) {
            setStatus('Kein Treffer – breite Suche ohne Kategorien …');
            hits = await broadSearchNoCategory(hints, 12);
          }

          renderResults(hits);
          setStatus('Fertig.');
        } catch (e) {
          console.error('[LabelScan]', e);
          setStatus('Fehler bei Erkennung/Suche. Bitte erneut versuchen.');
        } finally {
          runBtn.disabled = false; runBtn.textContent = 'Erkennen & suchen';
        }
      })();
    }

    runBtn.addEventListener('click', onSubmit);
    if (form) form.addEventListener('submit', onSubmit);

    // Sicherheit
    var wrap = document.getElementById('ados-labelscan');
    if (wrap) wrap.style.position = 'relative';
    runBtn.style.position = 'relative';
    runBtn.style.zIndex = '9999';
    runBtn.style.pointerEvents = 'auto';
  }

  if (document.readyState === 'loading') {
    document.addEventListener('DOMContentLoaded', bind);
  } else {
    bind();
  }
  setTimeout(bind, 250);
  setTimeout(bind, 1000);
  var mo = new MutationObserver(function () { if (!BOUND) bind(); });
  mo.observe(document.documentElement || document.body, { childList: true, subtree: true });

})();