ADOS Wiki

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

← Zum vorherigen Versionsunterschied
Keine Bearbeitungszusammenfassung
Keine Bearbeitungszusammenfassung
 
(73 dazwischenliegende Versionen desselben Benutzers werden nicht angezeigt)
Zeile 1: Zeile 1:
/* global mw, Tesseract */
/* global mw */
(function () {
(() => {
   'use strict';
   'use strict';


   // ========== ADOS: Kategorien & Distillery-Tokens ==========
   const CFG = {
  const ADOS_CATEGORIES = []; // <- Filter AUS zum Testen
    // ---- Daten & Model ----
    indexTitle: (window.LabelScanConfig && window.LabelScanConfig.indexTitle) ||
                'MediaWiki:Gadget-LabelScan-index.json',
    transformersURL: 'https://cdn.jsdelivr.net/npm/@xenova/transformers@2.15.0',
    modelId: 'Xenova/clip-vit-base-patch32',
    localModelPath: '/models',          // <<— deine Modelle liegen hier


    topK: 3,                            // <<— MAX. 3 TREFFER
    maxSide: 1280,                      // Downscale vor Auto-Crop (Performance)


  const KNOWN_TOKENS = [
    // ---- Auto-Crop ----
     'Ardbeg','Ardmore','Arran','Auchroisk','Ben Nevis','Blair Athol','Bowmore',
     autoCrop: true,
     'Caol Ila','Clynelish','Glenallachie','Glenrothes','Longmorn','Lagavulin',
     edgeKeepRatio: 0.10,               // oberste 10% Kanten als Maske
     'Tullibardine','Dalmore','Benrinnes','Mortlach','Glenlivet','Inchgower',
     cropPadding: 0.08,                 // 8% Randzugabe um die Box
     'Islay','Speyside','Highland','Lowland','Campbeltown','Ireland'
     cropMinRel: 0.40,                   // min. 40% der kleineren Bildkante
  ];


  // ========== Utility: Normalisierung & Fuzzy ==========
    // ---- Score-Badges ----
  function norm(s){
    showNumericScore: false,            // true = Zahlen zeigen, false = Badges
     return String(s||'')
    confidenceBands: [0.90, 0.80],      // hoch ≥0.90, mittel ≥0.80, sonst niedrig
      .toLowerCase()
 
      .normalize('NFD').replace(/[\u0300-\u036f]/g,'')
    // ---- Sonstiges ----
      .replace(/[^a-z0-9\s\-]/g,' ')
    debug: true
      .replace(/\s+/g,' ')
  };
       .trim();
 
  // --------- Helpers ----------
  const log=(...a)=>{ if(CFG.debug) console.log('[LabelScan]',...a); };
  const warn=(...a)=>{ if(CFG.debug) console.warn('[LabelScan]',...a); };
  const err=(...a)=>{ console.error('[LabelScan]',...a); };
  const qs=id=>document.getElementById(id);
  const setStatus=t=>{ const el=qs('ados-scan-status'); if(el) el.textContent=t||''; };
  const setProgress=p=>{
     const bar=qs('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 resetResultsBox(msg){
    const r = qs('ados-scan-results');
    if (r) {
       r.innerHTML = `<div class="empty">${msg || 'Hier erscheinen Treffer.'}</div>`;
    }
   }
   }


   function levenshtein(a,b){
   function showPreview(file){
     a = norm(a); b = norm(b);
     const url=URL.createObjectURL(file);
     const m = a.length, n = b.length;
     const prev=qs('ados-scan-preview');
    if (!m) return n; if (!n) return m;
     if(prev){
     const dp = new Array(n+1);
      prev.innerHTML=`<img alt="Vorschau" style="max-width:260px;width:100%;height:auto;border-radius:8px;display:block;margin:0 auto;" src="${url}">`;
    for (let j=0;j<=n;j++) dp[j]=j;
      prev.setAttribute('aria-hidden','false');
    for (let i=1;i<=m;i++){
      let prev = dp[0], tmp; dp[0]=i;
      for (let j=1;j<=n;j++){
        tmp = dp[j];
        dp[j] = (a[i-1]===b[j-1]) ? prev : 1 + Math.min(prev, dp[j-1], dp[j]);
        prev = tmp;
      }
     }
     }
    return dp[n];
   }
   }


   function tokenSet(str){ return new Set(norm(str).split(' ').filter(Boolean)); }
   function base64ToFloat32(b64){
  function overlapScore(a,b){
    const bin=atob(b64), len=bin.length;
     const A = tokenSet(a), B = tokenSet(b);
     const buf=new ArrayBuffer(len), view=new Uint8Array(buf);
     if (!A.size || !B.size) return 0;
     for(let i=0;i<len;i++) view[i]=bin.charCodeAt(i);
    let inter=0; A.forEach(t=>{ if (B.has(t)) inter++; });
     return new Float32Array(buf);
     return inter / Math.max(A.size, B.size);
   }
   }


   function escHTML (s) {
   // --------- Index ----------
     return String(s ?? '')
  let INDEX=[], INDEX_EMB=[];
      .replace(/&/g,'&amp;')
  async function loadIndex({ ui=true }={}){
      .replace(/</g,'&lt;')
     if(INDEX.length) return INDEX;
      .replace(/>/g,'&gt;')
    if(ui){ setStatus('Index laden …'); setProgress(0.03); }
      .replace(/"/g,'&quot;')
    const rawURL = mw.util.getUrl(CFG.indexTitle,{ action:'raw', ctype:'application/json' });
      .replace(/'/g,'&#39;');
    const res = await fetch(rawURL,{ cache:'reload' });
    if(!res.ok) throw new Error('Index nicht ladbar: '+res.status);
    const json = await res.json();
    if(!Array.isArray(json)) throw new Error('Index ist keine Array-JSON');
    INDEX = json;
    INDEX_EMB = INDEX.map(it => (typeof it.embed==='string' && it.embed.length) ? base64ToFloat32(it.embed) : null);
    log('Index geladen:', INDEX.length, 'Einträge');
    log('Embeddings vorhanden:', INDEX_EMB.filter(v=>v&&v.length).length, '/', INDEX.length);
    if(ui) setProgress(0.06);
    return INDEX;
   }
   }


   // ========== UI Präsenz & Helfer ==========
   // --------- Transformers (lokal) ----------
  function hasUI () {
  let _visionLoadPromise=null;
    return !!document.getElementById('ados-scan-run') &&
   async function ensureClipVision(){
          !!document.getElementById('ados-scan-file');
     if(_visionLoadPromise) return _visionLoadPromise;
  }
 
  function setStatus (t) {
     setStatus('Modell laden …'); setProgress(0.08);
    var el = document.getElementById('ados-scan-status');
 
    if (el) el.textContent = t || '';
     _visionLoadPromise = (async()=>{
  }
       const mod = await import(/* webpackIgnore: true */ CFG.transformersURL);
  function showOCRText (t) {
  var el = document.getElementById('ados-scan-ocr');
  if (el) el.textContent = (t || '').trim();
   }
  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');
    }
  }


  // ========== Tesseract bei Bedarf laden ==========
      // Nur lokal laden
  var tesseractReady;
       mod.env.allowLocalModels = true;
  function ensureTesseract () {
       mod.env.allowRemoteModels = false;
    if (tesseractReady) return tesseractReady;
       mod.env.localModelPath  = CFG.localModelPath;
    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;
  }


  // ========== Bild-Vorverarbeitung (Canvas) ==========
      // WASM-Runtime (ort-wasm-simd.wasm) von transformers-CDN
  async function preprocessImage (file) {
      mod.env.backends = mod.env.backends || {};
    const img = await new Promise((res, rej) => {
       mod.env.backends.onnx = mod.env.backends.onnx || {};
       const o = new Image();
       mod.env.backends.onnx.wasm = mod.env.backends.onnx.wasm || {};
       o.onload = () => res(o);
       mod.env.backends.onnx.wasm.wasmPaths =
       o.onerror = rej;
        'https://cdn.jsdelivr.net/npm/@xenova/transformers@2.15.0/dist/';
      o.src = URL.createObjectURL(file);
    });


    // Längste Kante auf ~1800px skalieren (besser für OCR)
      const [processor, model] = await Promise.all([
    const MAX = 1800;
        mod.AutoProcessor.from_pretrained(CFG.modelId),
    const scale = (img.width > img.height) ? (MAX / img.width) : (MAX / img.height);
        mod.CLIPVisionModelWithProjection.from_pretrained(CFG.modelId, { quantized: true })
    const w = Math.round(img.width * scale);
      ]);
    const h = Math.round(img.height * scale);


    const c = document.createElement('canvas');
      let backend='unknown';
    c.width = w; c.height = h;
      try { backend = model?.session?.executionProvider || backend; } catch(_){}
    const ctx = c.getContext('2d');
      log('CLIP ready (vision, local):', model?.constructor?.name || 'unknown', '| Backend:', backend);
    ctx.imageSmoothingEnabled = true;
    ctx.drawImage(img, 0, 0, w, h);


    // Graustufen + leichter Kontrastboost
      return { mod, processor, model };
    const id = ctx.getImageData(0, 0, w, h);
     })();
    const d = id.data;
    for (let i = 0; i < d.length; i += 4) {
      const g = 0.2126 * d[i] + 0.7152 * d[i + 1] + 0.0722 * d[i + 2];
      const v = Math.max(0, Math.min(255, (g - 128) * 1.15 + 128));
      d[i] = d[i + 1] = d[i + 2] = v;
     }
    ctx.putImageData(id, 0, 0);


     return c; // Canvas an Tesseract übergeben
     return _visionLoadPromise;
   }
   }


   // ========== OCR ==========
   // --------- Auto-Crop Heuristik ----------
async function runOCR (file) {
  function toCanvasScaled(img, maxSide){
  await ensureTesseract();
    const c=document.createElement('canvas');
  setProgress(0);
    let { width:w, height:h } = img;
  const canvas = await preprocessImage(file);
     const s = Math.min(1, maxSide / Math.max(w,h));
  try {
     w = Math.round(w*s); h = Math.round(h*s);
     const res = await Tesseract.recognize(canvas, 'deu+eng', {
     c.width=w; c.height=h;
      tessedit_pageseg_mode: 6,
     const g=c.getContext('2d', { willReadFrequently:true });
      preserve_interword_spaces: 1,
     g.imageSmoothingEnabled = true;
      logger: m => { if (m?.status==='recognizing text' && typeof m.progress==='number') setProgress(m.progress); }
     g.drawImage(img,0,0,w,h);
     });
     return c;
    setProgress(null);
     return (res && res.data && res.data.text) || '';
  } catch (e) {
    // Fallback nur ENG
    console.warn('[LabelScan] deu+eng fehlgeschlagen, versuche eng:', e);
     const res = await Tesseract.recognize(canvas, 'eng', {
      tessedit_pageseg_mode: 6,
      preserve_interword_spaces: 1,
      logger: m => { if (m?.status==='recognizing text' && typeof m.progress==='number') setProgress(m.progress); }
     });
     setProgress(null);
     return (res && res.data && res.data.text) || '';
   }
   }
}


  function autoCropCanvas(inCanvas){
    const w=inCanvas.width, h=inCanvas.height;
    const ctx=inCanvas.getContext('2d', { willReadFrequently:true });
    const imgData=ctx.getImageData(0,0,w,h);
    const data=imgData.data;


  // ========== Hinweise aus OCR extrahieren ==========
    // Graustufen
  function extractHints (text) {
    const gray=new Uint8ClampedArray(w*h);
    const raw = String(text || '').replace(/\s+/g, ' ').trim();
    for(let y=0, p=0, i=0; y<h; y++){
      for(let x=0; x<w; x++, i++, p+=4){
        const r=data[p], g=data[p+1], b=data[p+2];
        gray[i] = (0.299*r + 0.587*g + 0.114*b)|0;
      }
    }


     // Distillery-/Marken-Token
     // Sobel-Kanten (Magnitude)
     const foundNames = [];
    const mag=new Float32Array(w*h);
     KNOWN_TOKENS.forEach(t => {
    const sobelX=[-1,0,1,-2,0,2,-1,0,1];
       const re = new RegExp('\\b' + t.replace(/[.*+?^${}()|[\]\\]/g, '\\$&') + '\\b', 'i');
     const sobelY=[-1,-2,-1,0,0,0,1,2,1];
       if (re.test(raw)) foundNames.push(t);
     for(let y=1; y<h-1; y++){
     });
       for(let x=1; x<w-1; x++){
        let gx=0, gy=0, k=0;
        for(let j=-1;j<=1;j++){
          for(let i=-1;i<=1;i++,k++){
            const v=gray[(y+j)*w + (x+i)];
            gx += sobelX[k]*v; gy += sobelY[k]*v;
          }
        }
        mag[y*w+x] = Math.hypot(gx,gy);
       }
     }


     // Alter
     // Schwellwert: oberes x%-Quantil
     const ageRe = /\b([1-9]\d?)\s?(?:years?|yo|jahr(?:e)?)\b/gi;
     const vals = Array.from(mag).sort((a,b)=>a-b);
     const ages = []; let m;
     const keep = CFG.edgeKeepRatio;
     while ((m = ageRe.exec(raw)) !== null){ const n = m[1]; if (!ages.includes(n)) ages.push(n); }
     const tIdx = Math.max(0, Math.min(vals.length-1, Math.floor(vals.length*(1-keep))));
    const thr = vals[tIdx];


     // Jahrgänge
     // Bounding-Box der Pixel > thr
     const yearRe = /\b(19|20)\d{2}\b/g;
     let minX=w, minY=h, maxX=0, maxY=0, count=0;
    const years=[]; while ((m = yearRe.exec(raw)) !== null){ if(!years.includes(m[0])) years.push(m[0]); }
    for(let y=0;y<h;y++){
      for(let x=0;x<w;x++){
        const m=mag[y*w+x];
        if(m>thr){ count++; if(x<minX)minX=x; if(y<minY)minY=y; if(x>maxX)maxX=x; if(y>maxY)maxY=y; }
      }
    }


     // ABV
     if(count<50) return inCanvas; // zu wenig Signal → return original
    const abvRe = /\b(\d{2}(?:[.,]\d)?)\s?%\b/g;
    const abvs=[]; while ((m = abvRe.exec(raw)) !== null){ abvs.push(m[1].replace(',','.')); }


     // Serien-Keywords (leichtgewichtig)
     // Padding
     const seriesMap = [
     const pad = Math.round(CFG.cropPadding * Math.max(w,h));
      ['A Dream of Scotland','dream of scotland'],
     minX = Math.max(0, minX - pad);
      ['A Dream of Ireland','dream of ireland'],
     minY = Math.max(0, minY - pad);
      ['A Dream of... – Der Rest der Welt','rest der welt','dream of','rest of the world'],
    maxX = Math.min(w-1, maxX + pad);
      ['Friendly Mr. Z Whiskytainment Abfüllungen','friendly mr z','mr z'],
     maxY = Math.min(h-1, maxY + pad);
      ['Die Whisky Elfen Abfüllungen','whisky elfen','elfen'],
      ['The Fine Art of Whisky Abfüllungen','fine art of whisky','fine art'],
      ['Alle Rumbastic Abfüllungen','rumbastic']
    ];
     const low = norm(raw); const series=[];
     seriesMap.forEach(([label, hint])=>{
      if (low.includes(norm(hint))) series.push(label);
     });


     // „Promi-Wörter“
     // Mindestgröße
     const wordRe = /\b[A-ZÄÖÜ][A-Za-zÄÖÜäöüß\-]{3,}\b/g;
     const boxW=maxX-minX+1, boxH=maxY-minY+1;
     const uniq = new Set(); let w; const words=[];
     const minLen = Math.round(CFG.cropMinRel * Math.min(w,h));
     while ((w = wordRe.exec(raw)) !== null){
    let cx=minX, cy=minY, cw=boxW, ch=boxH;
       const s = w[0]; if(!uniq.has(s)){ uniq.add(s); words.push(s); if(words.length>=8) break; }
     if(cw<minLen || ch<minLen){
      const needW = Math.max(minLen, cw);
      const needH = Math.max(minLen, ch);
       const centerX = Math.round((minX+maxX)/2);
      const centerY = Math.round((minY+maxY)/2);
      cx = Math.max(0, Math.min(w-needW, centerX - Math.floor(needW/2)));
      cy = Math.max(0, Math.min(h-needH, centerY - Math.floor(needH/2)));
      cw = needW; ch = needH;
     }
     }


     return { names: foundNames, ages, years, words, abvs, series, raw };
     const out=document.createElement('canvas');
    out.width=cw; out.height=ch;
    out.getContext('2d').drawImage(inCanvas, cx, cy, cw, ch, 0, 0, cw, ch);
    return out;
   }
   }


   // ========== Smarte Wiki-Suche (3 Pässe) ==========
   // --------- Embedding-Pipeline ---------
   async function searchWikiSmart (hints, limit) {
   async function embedFileImage(file){
     await mw.loader.using('mediawiki.api');
     const { mod, processor, model } = await ensureClipVision();
     const api = new mw.Api();
 
    const ns0 = 0;
     setStatus('Bild vorbereiten …'); setProgress(0.20);
    const MAX = limit || 12;


     function incatStr () {
     // 1) Bild laden
       return ADOS_CATEGORIES.map(c => 'incategory:"' + c + '"').join(' ');
    const img = await new Promise((res,rej)=>{
       const url=URL.createObjectURL(file);
      const image=new Image();
      image.crossOrigin='anonymous';
      image.onload=()=>{ URL.revokeObjectURL(url); res(image); };
      image.onerror=e=>{ URL.revokeObjectURL(url); rej(e); };
      image.src=url;
    });
 
    // 2) Scale → Auto-Crop
    let canvas = toCanvasScaled(img, CFG.maxSide);
    if(CFG.autoCrop){
      setStatus('Auto-Crop …'); setProgress(0.30);
      canvas = autoCropCanvas(canvas);
     }
     }


async function broadSearchNoCategory(hints, limit){
    // 3) Canvas → Blob → RawImage (robust für Processor)
  await mw.loader.using('mediawiki.api');
    const blob = await new Promise(r => canvas.toBlob(r, 'image/jpeg', 0.95));
  const api = new mw.Api();
    const imageRaw = await mod.RawImage.fromBlob(blob);
  const ns0 = 0;
  const MAX = limit || 12;


  // baue eine breite Suchphrase aus Distillery/Alter/Jahr/Wörtern
     setStatus('Bild analysieren …'); setProgress(0.45);
  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, srnamespace:ns0, srlimit:MAX, formatversion:2 });
    // 4) Processor & Model
  return (r.query?.search || []);
    const inputs = await processor(imageRaw, { return_tensors: 'pt' });
}
    const out = await model.forward({ pixel_values: inputs.pixel_values });


    // PASS 1: intitle-Kombis (präzise)
     const vec = out?.image_embeds?.data || out?.image_embeds;
     const pass1 = [];
    if(!(vec instanceof Float32Array)) throw new Error('Embedding-Format unerwartet');
    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
     // 5) Normieren
     const key = []
     let n=0; for(let i=0;i<vec.length;i++) n+=vec[i]*vec[i];
      .concat(hints.names.slice(0, 2), hints.ages.slice(0, 1), hints.years.slice(0, 1), hints.words.slice(0, 3))
    const norm = Math.sqrt(n)||1;
      .map(x => `"${x}"`).join(' ');
    const v = new Float32Array(vec.length);
     const pass2 = key ? [ `${key} ${incatStr()}` ] : [];
     for(let i=0;i<vec.length;i++) v[i]=vec[i]/norm;
    return v;
  }


    // PASS 3: Prefix auf Titel
  function cosine(a,b){ let s=0,L=Math.min(a.length,b.length); for(let i=0;i<L;i++) s+=a[i]*b[i]; return s; }
    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 = [];
  // vorher: slice(0, topK) hier
  // jetzt: ALLE sortiert zurückgeben, damit wir danach deduplizieren können
  function rankByCosine(q){
     const s=[];
    for(let i=0;i<INDEX.length;i++){
      const v=INDEX_EMB[i];
      if(!v) continue;
      s.push({ i, score: cosine(q,v) });
    }
    s.sort((a,b)=> b.score-a.score);
    return s;
  }


    async function runSr (q) {
  // NEU: pro Titel nur bester Treffer
      const r = await api.get({ action: 'query', list: 'search', srsearch: q, srnamespace: ns0, srlimit: MAX, formatversion: 2 });
  function dedupeByTitle(ranked){
      (r.query?.search || []).forEach(it => {
    const bestByTitle = Object.create(null);
        const k = it.title;
    for (let k = 0; k < ranked.length; k++) {
        if (seen.has(k)) return;
      const hit = ranked[k];
         seen.add(k);
      const it = INDEX[hit.i];
         out.push(it);
      const rawTitle = it && it.title ? String(it.title) : '';
       });
      const key = rawTitle.trim().toLowerCase();
      if (!key) continue;
      const prev = bestByTitle[key];
      if (!prev || hit.score > prev.score) {
         bestByTitle[key] = hit;
      }
    }
    const arr = [];
    for (const key in bestByTitle) {
      if (Object.prototype.hasOwnProperty.call(bestByTitle, key)) {
         arr.push(bestByTitle[key]);
       }
     }
     }
    arr.sort(function(a,b){ return b.score - a.score; });
    return arr;
  }


    for (const q of pass1) { await runSr(q); if (out.length >= MAX) return out.slice(0, MAX); }
  // --------- Score-Badges ---------
     for (const q of pass2) { await runSr(q); if (out.length >= MAX) return out.slice(0, MAX); }
  function scoreBadge(score){
 
     if (CFG.showNumericScore) {
    // Prefix (list=prefixsearch)
    return `<span style="font-variant-numeric:tabular-nums;color:#666">${score.toFixed(3)}</span>`;
    for (const p of pass3) {
      const r = await api.get({ action: 'query', list: 'prefixsearch', pssearch: p, psnamespace: ns0, pslimit: MAX });
      (r.query?.prefixsearch || []).forEach(it=>{
        const title = it.title || it['*'];
        const k = title; if (seen.has(k)) return; seen.add(k);
        out.push({ title, snippet: '' });
      });
      if (out.length >= MAX) break;
     }
     }
    const [hi, mid] = CFG.confidenceBands || [0.90, 0.80];
    let txt = 'niedrig', bg = '#f1f5f9', fg = '#334155';
    if (score >= hi) { txt = 'hoch';  bg = '#e6ffed'; fg = '#0a7d2c'; }
    else if (score >= mid) { txt = 'mittel'; bg = '#fff7e6'; fg = '#a45500'; }
    return `<span style="display:inline-block;padding:.12rem .45rem;border-radius:999px;background:${bg};color:${fg};font-weight:600;font-size:.85em;line-height:1">${txt}</span>`;
  }


    return out.slice(0, MAX);
  // --------- Rendering (max. 3 Treffer) ---------
  }
// Score-Badge ausgeblendet
// <div>${scoreBadge(score)}</div>


   // ========== Titel aus Kategorien + Fuzzy-Fallback ==========
   function renderResults(ranked){
  async function fetchTitlesFromCategories(){
     const box=qs('ados-scan-results');
     await mw.loader.using('mediawiki.api');
     if(!box) return;
     const api = new mw.Api();
     box.innerHTML='';
     const titles = new Set();


     for (const cat of ADOS_CATEGORIES){
     if(!ranked || !ranked.length){
       let cmcontinue = undefined;
       box.innerHTML='<div class="empty">Keine klaren Treffer. Bitte ein anderes Foto oder näher am Frontlabel.</div>';
      do {
       return;
        const r = await api.get({
          action:'query',
          list:'categorymembers',
          cmtitle:'Kategorie:' + cat,
          cmtype:'page',
          cmlimit:'max',
          cmcontinue
        });
        (r.query?.categorymembers||[]).forEach(it=> titles.add(it.title));
        cmcontinue = r.continue && r.continue.cmcontinue;
       } while (cmcontinue);
     }
     }
    return Array.from(titles);
  }


  function scoreTitleAgainstHints(title, hints){
    // NEU: Dedupe nach Titel, DANN auf topK begrenzen
    const normTitle = norm(title);
    const uniqueRanked = dedupeByTitle(ranked).slice(0, CFG.topK);
 
    const makeCard = (it, score) => `
      <div class="ados-card" style="display:grid;grid-template-columns:120px 1fr;gap:14px;align-items:center;padding:12px;border:1px solid #e6e6e6;border-radius:14px;box-shadow:0 1px 8px rgba(0,0,0,.04);">
        ${it.thumb?`<img src="${it.thumb}" alt="" style="width:120px;height:auto;border-radius:10px;">`
                  :`<div style="width:120px;height:90px;background:#f3f3f3;border-radius:10px;"></div>`}
        <div style="display:flex;flex-direction:column;gap:8px;">
          <div style="font-weight:700;font-size:1.05rem;line-height:1.2;">
            <a href="${mw.util.getUrl((it.title||'').replace(/ /g,'_'))}">${mw.html.escape(it.title||'')}</a>
          </div>
          <div>
            <a href="${mw.util.getUrl((it.title||'').replace(/ /g,'_'))}" class="mw-ui-button" style="display:inline-block;padding:.4rem .7rem;border-radius:8px;background:#2a4b8d;color:#fff;text-decoration:none;">Artikel öffnen</a>
          </div>
        </div>
      </div>`;


     // 1) Token-Overlap
     const grid = document.createElement('div');
     const base = overlapScore(title, hints.raw);
    grid.style.display='grid';
     grid.style.gridTemplateColumns='1fr';
    grid.style.gap='12px';


     // 2) Distillery-Boost
     // max. CFG.topK (=3) Karten nach Dedupe
     let nameBoost = 0;
     uniqueRanked.forEach(function(hit){
    hints.names.forEach(n=>{
       const it = INDEX[hit.i];
      const t = norm(n);
       grid.innerHTML += makeCard(it, hit.score);
       const d = levenshtein(normTitle, t);
       if (normTitle.includes(t)) nameBoost = Math.max(nameBoost, 0.35);
      else if (d <= 4) nameBoost = Math.max(nameBoost, 0.2);
     });
     });


     // 3) Alter/Jahr/ABV im Titel
     box.appendChild(grid);
    let numBoost = 0;
  }
    hints.ages.forEach(a=>{
 
      if (new RegExp('\\b'+a+'\\b').test(normTitle)) numBoost = Math.max(numBoost, 0.15);
  // --------- UI / Flow ----------
     });
  let BOUND=false;
     hints.years.forEach(y=>{
  function bindUI(){
      if (normTitle.includes(y)) numBoost = Math.max(numBoost, 0.15);
    if(BOUND) return;
     });
 
     hints.abvs.forEach(p=>{
    const btnRun = qs('ados-scan-run');
      if (normTitle.includes(p.replace('.',''))) numBoost = Math.max(numBoost, 0.1);
    const inCam  = qs('ados-scan-file-camera');
     });
     const inGal  = qs('ados-scan-file-gallery');
     const btnCam = qs('ados-scan-btn-camera');
     const btnGal = qs('ados-scan-btn-gallery');
     const drop  = qs('ados-scan-drop');
     const btnReset = qs('ados-scan-reset');


     // 4) Serien-Bonus
     if(!btnRun || !inCam || !inGal) return;
    let seriesBoost = 0;
    hints.series.forEach(s=>{
      const key = norm(s.split(' Abfüllungen')[0]); // Kern
      if (normTitle.includes(key)) seriesBoost = Math.max(seriesBoost, 0.15);
    });


     // 5) leichte Strafe bei sehr kleinem Overlap
     // Hilfsfunktion: wenn neues Bild gewählt → Vorschau & Ergebnisse zurücksetzen
     const penalty = base < 0.2 ? -0.05 : 0;
     const onNewImage = (file) => {
      if (!file) return;
      showPreview(file);
      resetResultsBox('Hier erscheinen Treffer.');
      setStatus('Bereit.');
      setProgress(null);
    };


     return Math.max(0, base + nameBoost + numBoost + seriesBoost + penalty);
     btnCam && btnCam.addEventListener('click', ()=> inCam.click());
  }
    btnGal && btnGal.addEventListener('click', ()=> inGal.click());


  async function fallbackFuzzyTitles(hints, limit){
     const pick = e => {
     const titles = await fetchTitlesFromCategories();
      const f=e.target.files?.[0];
    const scored = titles.map(t => ({ title: t, _score: scoreTitleAgainstHints(t, hints) }));
      if(f) onNewImage(f);
     scored.sort((a,b)=> b._score - a._score);
     };
     const top = scored.slice(0, limit||12).filter(x=> x._score >= 0.10);
     inCam.addEventListener('change', pick);
     return top.map(x=> ({ title: x.title, snippet: '' }));
     inGal.addEventListener('change', pick);
  }


  // ========== Treffer-Rendering ==========
    if(drop){
  function renderResults (items) {
      drop.addEventListener('dragover', function(ev){ ev.preventDefault(); drop.classList.add('is-over'); });
    var box = document.getElementById('ados-scan-results');
      drop.addEventListener('dragleave', function(){ drop.classList.remove('is-over'); });
    if (!box) return;
      drop.addEventListener('drop', function(ev){
    box.innerHTML = '';
        ev.preventDefault(); drop.classList.remove('is-over');
    if (!items || !items.length) {
        const f = ev.dataTransfer && ev.dataTransfer.files && ev.dataTransfer.files[0];
      box.innerHTML = '<div class="ados-hit">Keine klaren Treffer. Bitte anderes Foto oder manuell suchen.</div>';
        if(f){
       return;
          const dt=new DataTransfer(); dt.items.add(f);
          inGal.files=dt.files;
          onNewImage(f);
        }
       });
     }
     }
     items.slice(0, 12).forEach(function (it) {
 
       var title = it.title || '';
     btnReset && btnReset.addEventListener('click', function(){
      var link = (mw && mw.util) ? mw.util.getUrl(title.replace(/ /g, '_')) : ('/wiki/' + encodeURIComponent(title.replace(/ /g,'_')));
       setStatus('Bereit.'); setProgress(null);
      var snip = String(it.snippet || '').replace(/<\/?span[^>]*>/g, '').replace(/&quot;/g, '"');
       const p=qs('ados-scan-preview'); if(p) p.innerHTML='<div class="note">Noch keine Vorschau.</div>';
       var div = document.createElement('div');
      resetResultsBox('Hier erscheinen Treffer.');
      div.className = 'ados-hit';
       inCam.value=''; inGal.value='';
      div.innerHTML =
        '<b><a href="' + link + '">' + escHTML(title) + '</a></b>' +
        (snip ? '<div class="meta">' + snip + '</div>' : '');
       box.appendChild(div);
     });
     });
    btnRun.addEventListener('click', onRunClick);
    BOUND=true; log('UI gebunden.');
   }
   }


   // ========== Binding ==========
   async function onRunClick(){
  var BOUND = false;
    const btnRun = qs('ados-scan-run');
  function bind () {
    const inCam  = qs('ados-scan-file-camera');
     if (BOUND || !hasUI()) return;
     const inGal  = qs('ados-scan-file-gallery');


     var runBtn = document.getElementById('ados-scan-run');
     try{
    var fileIn = document.getElementById('ados-scan-file');
      const file = (inCam.files && inCam.files[0]) || (inGal.files && inGal.files[0]);
    var bigBtn = document.getElementById('ados-scan-bigbtn');
      if(!file){ alert('Bitte zuerst ein Foto auswählen.'); return; }
    var form  = document.getElementById('ados-scan-form');


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


    if (bigBtn) bigBtn.addEventListener('click', function () { fileIn.click(); });
       // Ergebnisse direkt leeren / „Suche läuft …“
    fileIn.addEventListener('change', function () {
       resetResultsBox('Suche läuft …');
       if (this.files && this.files[0]) showPreview(this.files[0]);
    });
    if (form) {
       form.addEventListener('submit', function(ev){
        ev.preventDefault();
        runBtn.click();
      });
    }


    runBtn.addEventListener('click', async function (ev) {
       await loadIndex({ ui:true });
       ev.preventDefault();
       await ensureClipVision(); // warmup
      if (!(fileIn.files && fileIn.files[0])) { alert('Bitte ein Foto auswählen oder aufnehmen.'); return; }
       var f = fileIn.files[0];
      try {
        runBtn.disabled = true; runBtn.textContent = 'Erkenne …';
        setStatus('Erkenne Label …');
        var text = await runOCR(f);
        showOCRText(text);


        setStatus('Suche im Wiki …');
      const q = await embedFileImage(file);
        var hints = extractHints(text);
var hits = await searchWikiSmart(hints, 12);


if (!hits || !hits.length) {
      setProgress(0.70);
  setStatus('Kein direkter Treffer – Fuzzy über Kategorien …');
      setStatus('Abgleich …');
  hits = await fallbackFuzzyTitles(hints, 12);
}


if (!hits || !hits.length) {
      const ranked = rankByCosine(q);
  setStatus('Kein Treffer – breite Suche ohne Kategorien …');
      renderResults(ranked);
  hits = await broadSearchNoCategory(hints, 12);
}


renderResults(hits);
      setStatus('Fertig.');
      setProgress(null);
    } catch(e){
      err('Fehler', e);
      setStatus('Fehler bei Erkennung.');
      resetResultsBox('Fehler bei der Erkennung – bitte ein anderes Foto versuchen.');
      setProgress(null);
    } finally {
      if(btnRun) btnRun.disabled = false;
    }
  }


        setStatus(hits && hits.length ? 'Fertig.' : 'Keine klaren Treffer.');
  function init(){
       } catch (e) {
    if(document.readyState==='loading'){
        console.error('[LabelScan]', e);
      document.addEventListener('DOMContentLoaded', bindUI, { once: true });
        setStatus('Fehler bei Erkennung/Suche. Bitte erneut versuchen.');
    } else {
       } finally {
       bindUI();
        runBtn.disabled = false; runBtn.textContent = 'Erkennen & suchen';
    }
       }
    loadIndex({ ui:false }).catch(err).finally(()=>{
      setStatus('Bereit.');
       resetResultsBox('Hier erscheinen Treffer.');
       setProgress(null);
     });
     });
  }


    // Sicherheit gegen Overlays
  log('gadget file loaded');
    var wrap = document.getElementById('ados-labelscan');
  init();
    if (wrap) wrap.style.position = 'relative';
    runBtn.style.position = 'relative';
    runBtn.style.zIndex = '9999';
    runBtn.style.pointerEvents = 'auto';
  }


  // Erstbindung + Fallbacks + Observer
  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 });
})();
})();
Abgerufen von „https://ados-wiki.de/wiki/MediaWiki:Gadget-LabelScan.js“