update

sync/src/scraper.ts

@@ -242,7 +242,7 @@ export async function scrapeAll(username: string, password: string, knownStNrs:
         const idInstanzen      = urlObj.searchParams.get('id_instanzen') ?? ID_INSTANZEN;
 
         // Ausbildungen
-        const quals = await scrapeAusbildungenFromDetailPage(mainFrame, member);
+        const quals = await scrapeAusbildungenFromDetailPage(mainFrame, member, idMitgliedschaft, idPersonen);
         ausbildungen.push(...quals);
 
         // Beförderungen
@@ -688,76 +688,186 @@ async function scrapeDetailProfileFields(frame: Frame): Promise<{
 }
 
 /**
- * Scrape Ausbildungen from the detail page (already loaded).
+ * Scrape Ausbildungen by navigating to the AusbildungenListEdit.aspx page.
- * Navigates to the Ausbildung sub-page if needed.
+ * This is a ListEdit page (like Fahrgenehmigungen) with <input>/<select> elements.
  */
-async function scrapeAusbildungenFromDetailPage(frame: Frame, member: FdiskMember): Promise<FdiskAusbildung[]> {
+async function scrapeAusbildungenFromDetailPage(
-  // Look for Ausbildungsliste section — it's likely a table or list
+  frame: Frame,
-  const ausbildungSection = frame.locator('text=Ausbildung, text=Ausbildungsliste').first();
+  member: FdiskMember,
-  const hasSec = await ausbildungSection.isVisible().catch(() => false);
+  idMitgliedschaft?: string | null,
-
+  idPersonen?: string | null,
-  if (!hasSec) {
+): Promise<FdiskAusbildung[]> {
-    // Try navigating to an Ausbildung tab/link if present
+  // If we don't have the IDs, we cannot navigate to the Ausbildungen page
-    const ausbildungLink = frame.locator('a:has-text("Ausbildung")').first();
+  if (!idMitgliedschaft || !idPersonen) {
-    const hasLink = await ausbildungLink.isVisible().catch(() => false);
+    log(`  Ausbildungen for StNr ${member.standesbuchNr}: missing mitgliedschaft/personen IDs, skipping`);
-    if (hasLink) {
+    return [];
-      await ausbildungLink.click();
-      await frame.waitForLoadState('networkidle').catch(() => {});
-    }
   }
 
-  // Parse the qualification table
+  const url = `${BASE_URL}/fdisk/module/mgvw/ausbildungen/AusbildungenListEdit.aspx`
-  const tables = await frame.$$('table');
+    + `?search=1&searchid_mitgliedschaften=${idMitgliedschaft}&id_personen=${idPersonen}`
-  const ausbildungen: FdiskAusbildung[] = [];
+    + `&id_mitgliedschaften=${idMitgliedschaft}&searchid_personen=${idPersonen}&searchid_maskmode=`;
+
+  await frame_goto(frame, url);
+
+  const landed = frame.url();
+  const title  = await frame.title().catch(() => '');
+  if (landed.includes('BLError') || landed.includes('support.aspx') || title.toLowerCase().includes('fehler')) {
+    log(`    → Ausbildungen ERROR page: ${landed}`);
+    return [];
+  }
+
+  // Dump HTML for debugging
+  await dumpHtml(frame, `ausbildungen_StNr${member.standesbuchNr}`);
+
+  // This is a ListEdit page — read form fields by ID pattern or parse table with extractCellValue
+  const ausbildungen = await frame.evaluate((stNr: string) => {
+    const extractCellValue = (cell: Element): string => {
+      const input = cell.querySelector('input[type="text"], input:not([type])') as HTMLInputElement | null;
+      if (input && input.value?.trim()) return input.value.trim();
+      const sel = cell.querySelector('select') as HTMLSelectElement | null;
+      if (sel) {
+        // Try selectedIndex first
+        const idx = sel.selectedIndex;
+        if (idx >= 0 && sel.options[idx]) {
+          const t = (sel.options[idx].text || sel.options[idx].value || '').trim();
+          if (t) return t;
+        }
+        // Fallback: read the selected attribute directly from HTML
+        const selectedOpt = sel.querySelector('option[selected]') as HTMLOptionElement | null;
+        if (selectedOpt) {
+          const t = (selectedOpt.text || selectedOpt.value || '').trim();
+          if (t) return t;
+        }
+      }
+      const anchor = cell.querySelector('a');
+      const atitle = anchor?.getAttribute('title')?.trim();
+      if (atitle) return atitle;
+      return cell.textContent?.replace(/\u00A0/g, ' ').trim() ?? '';
+    };
+
+    const results: Array<{
+      standesbuchNr: string;
+      kursname: string | null;
+      kursDatum: string | null;
+      ablaufdatum: string | null;
+      ort: string | null;
+      bemerkung: string | null;
+      syncKey: string;
+    }> = [];
+
+    // Collect rows from all tables, find the data table
+    const tables = Array.from(document.querySelectorAll('table'));
+    let bestRows: Array<{ cells: string[] }> = [];
+    let bestHeaders: string[] = [];
 
     for (const table of tables) {
-    const rows = await table.$$eval('tr', (rows) => {
+      const rows: Array<{ cells: string[] }> = [];
-      return rows.map(row => ({
+      const headerCells: string[] = [];
-        cells: Array.from(row.querySelectorAll('td, th')).map(c => (c as Element).textContent?.trim() ?? ''),
-      }));
-    });
 
-    if (rows.length < 2) continue;
+      // Get headers
+      for (const th of Array.from(table.querySelectorAll('thead th, tr:first-child th'))) {
+        headerCells.push(extractCellValue(th));
+      }
 
-    const header = rows[0].cells.map(c => c.toLowerCase());
+      // Get data rows
-    const isAusbildungTable =
+      for (const tr of Array.from(table.querySelectorAll('tr'))) {
-      header.some(h => h.includes('kurs') || h.includes('ausbildung') || h.includes('bezeichnung'));
+        if (tr.closest('table') !== table) continue;
+        const tds = Array.from(tr.querySelectorAll('td'));
+        if (tds.length < 2) continue;
+        if (tr.querySelectorAll('th').length > 0) continue;
+        rows.push({ cells: tds.map(td => extractCellValue(td)) });
+      }
 
-    if (!isAusbildungTable) continue;
+      if (rows.length > bestRows.length) {
+        bestRows = rows;
+        bestHeaders = headerCells;
+      }
+    }
 
-    const kursnameIdx = header.findIndex(h => h.includes('kurs') || h.includes('ausbildung') || h.includes('bezeichnung'));
+    if (bestRows.length === 0) return results;
-    const datumIdx = header.findIndex(h => h.includes('datum') || h.includes('abschluss'));
-    const ablaufIdx = header.findIndex(h => h.includes('ablauf') || h.includes('gültig'));
-    const ortIdx = header.findIndex(h => h.includes('ort'));
-    const bemIdx = header.findIndex(h => h.includes('bem') || h.includes('info'));
 
-    for (const row of rows.slice(1)) {
+    // Try to find column indices from headers
-      const kursname = cellText(row.cells[kursnameIdx >= 0 ? kursnameIdx : 0]);
+    const hdr = bestHeaders.map(h => h.toLowerCase());
+    let kursnameIdx = hdr.findIndex(h => h.includes('kurs') || h.includes('ausbildung') || h.includes('bezeichnung'));
+    let datumIdx = hdr.findIndex(h => h.includes('datum') || h.includes('abschluss'));
+    let ablaufIdx = hdr.findIndex(h => h.includes('ablauf') || h.includes('gültig'));
+    let ortIdx = hdr.findIndex(h => h.includes('ort'));
+    let bemIdx = hdr.findIndex(h => h.includes('bem') || h.includes('info'));
+
+    // If headers didn't help, scan data for date-like columns and text columns
+    if (kursnameIdx === -1 && bestRows.length > 0) {
+      const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+      // Find date columns
+      const dateCols = new Set<number>();
+      const textCols: number[] = [];
+      for (const row of bestRows.slice(0, 3)) {
+        for (let ci = 0; ci < row.cells.length; ci++) {
+          const v = row.cells[ci]?.trim();
+          if (!v) continue;
+          if (datePattern.test(v)) dateCols.add(ci);
+          else if (v.length > 2 && !/^[\d.,]+$/.test(v)) textCols.push(ci);
+        }
+      }
+      // The longest text column is likely the Kursname
+      if (textCols.length > 0) {
+        let maxLen = 0;
+        for (const ci of textCols) {
+          const len = (bestRows[0]?.cells[ci] ?? '').length;
+          if (len > maxLen) { maxLen = len; kursnameIdx = ci; }
+        }
+      }
+      // First date column is Datum, second is Ablaufdatum
+      const sortedDates = Array.from(dateCols).sort((a, b) => a - b);
+      if (sortedDates.length > 0 && datumIdx === -1) datumIdx = sortedDates[0];
+      if (sortedDates.length > 1 && ablaufIdx === -1) ablaufIdx = sortedDates[1];
+    }
+
+    for (const row of bestRows) {
+      const kursname = (kursnameIdx >= 0 ? row.cells[kursnameIdx] : row.cells[0])?.trim() || null;
       if (!kursname) continue;
+      // Skip header-like rows
+      if (/kurs|ausbildung|bezeichnung|datensätze|tiefennavigation/i.test(kursname)) continue;
 
-      const kursDatum = parseDate(datumIdx >= 0 ? row.cells[datumIdx] : null);
+      const rawDatum = datumIdx >= 0 ? row.cells[datumIdx]?.trim() : null;
-      const syncKey = `${member.standesbuchNr}::${kursname}::${kursDatum ?? ''}`;
+      const rawAblauf = ablaufIdx >= 0 ? row.cells[ablaufIdx]?.trim() : null;
+      const rawOrt = ortIdx >= 0 ? row.cells[ortIdx]?.trim() || null : null;
+      const rawBem = bemIdx >= 0 ? row.cells[bemIdx]?.trim() || null : null;
 
-      ausbildungen.push({
+      // parseDate is not available inside evaluate; return raw values
-        standesbuchNr: member.standesbuchNr,
+      results.push({
+        standesbuchNr: stNr,
         kursname,
-        kursDatum,
+        kursDatum: rawDatum || null,
-        ablaufdatum: parseDate(ablaufIdx >= 0 ? row.cells[ablaufIdx] : null),
+        ablaufdatum: rawAblauf || null,
-        ort: ortIdx >= 0 ? cellText(row.cells[ortIdx]) : null,
+        ort: rawOrt,
-        bemerkung: bemIdx >= 0 ? cellText(row.cells[bemIdx]) : null,
+        bemerkung: rawBem,
-        syncKey,
+        syncKey: `${stNr}::${kursname}::${rawDatum ?? ''}`,
       });
     }
 
-    break; // only process the first Ausbildung table found
+    return results;
-  }
+  }, member.standesbuchNr).catch(() => [] as FdiskAusbildung[]);
+
+  // Post-process: parse dates and rebuild syncKeys
+  const results: FdiskAusbildung[] = ausbildungen.map(a => {
+    const kursDatum = parseDate(a.kursDatum);
+    return {
+      standesbuchNr: a.standesbuchNr,
+      kursname: a.kursname,
+      kursDatum,
+      ablaufdatum: parseDate(a.ablaufdatum),
+      ort: a.ort,
+      bemerkung: a.bemerkung,
+      syncKey: `${a.standesbuchNr}::${a.kursname}::${kursDatum ?? ''}`,
+    };
+  });
 
   // Debug: dump HTML when no Ausbildungen found
-  if (ausbildungen.length === 0) {
+  if (results.length === 0) {
-    await dumpHtml(frame, `ausbildungen_StNr${member.standesbuchNr}`);
+    await dumpHtml(frame, `ausbildungen_empty_StNr${member.standesbuchNr}`);
   }
 
-  return ausbildungen;
+  return results;
 }
 
 /**
@@ -768,7 +878,7 @@ async function scrapeAusbildungenFromDetailPage(frame: Frame, member: FdiskMembe
 async function navigateAndGetTableRows(
   frame: Frame,
   url: string,
-): Promise<Array<{ cells: string[] }> | null> {
+): Promise<{ rows: Array<{ cells: string[] }>; dateColIdx: number } | null> {
   await frame_goto(frame, url);
 
   const landed = frame.url();
@@ -827,13 +937,29 @@ async function navigateAndGetTableRows(
   const fdcRows = allRows.filter(r => r.tableClass.includes('FdcLayList'));
   const resultRows = fdcRows.length > 0 ? fdcRows : allRows;
 
-  const mapped = resultRows.map(r => ({ cells: r.cells }));
+  // Strip \u00A0 (non-breaking space) from all cell values and trim
+  const mapped = resultRows.map(r => ({
+    cells: r.cells.map(c => c.replace(/\u00A0/g, ' ').trim()),
+  }));
 
-  // Filter: only keep rows where cells[0] looks like a DD.MM.YYYY date
+  // Find date column dynamically: look for a DD.MM.YYYY pattern in any column
   const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
-  const dataRows = mapped.filter(r => datePattern.test(r.cells[0]?.trim() ?? ''));
+  let dateColIdx = -1;
+  for (const r of mapped) {
+    for (let ci = 0; ci < r.cells.length; ci++) {
+      if (datePattern.test(r.cells[ci] ?? '')) {
+        dateColIdx = ci;
+        break;
+      }
+    }
+    if (dateColIdx >= 0) break;
+  }
 
-  log(`    → ${allRows.length} total rows, ${fdcRows.length} FdcLayList rows, ${dataRows.length} data rows (with date in cells[0])`);
+  const dataRows = dateColIdx >= 0
+    ? mapped.filter(r => datePattern.test(r.cells[dateColIdx] ?? ''))
+    : [];
+
+  log(`    → ${allRows.length} total rows, ${fdcRows.length} FdcLayList rows, ${dataRows.length} data rows (date in col ${dateColIdx})`);
 
   // Debug: dump HTML when no data rows found
   if (dataRows.length === 0) {
@@ -841,7 +967,7 @@ async function navigateAndGetTableRows(
     await dumpHtml(frame, `navigateAndGetTableRows_${urlSlug}`);
   }
 
-  return dataRows;
+  return { rows: dataRows, dateColIdx };
 }
 
 /**
@@ -857,13 +983,19 @@ async function scrapeMemberBefoerderungen(
     + `?search=1&searchid_mitgliedschaften=${idMitgliedschaft}&id_personen=${idPersonen}`
     + `&id_mitgliedschaften=${idMitgliedschaft}&searchid_personen=${idPersonen}&searchid_maskmode=`;
 
-  const rows = await navigateAndGetTableRows(frame, url);
+  const result = await navigateAndGetTableRows(frame, url);
-  if (!rows) return [];
+  if (!result) return [];
 
+  const { rows, dateColIdx } = result;
   const results: FdiskBefoerderung[] = [];
   for (const row of rows) {
-    const datum = parseDate(row.cells[0]);
+    const datum = parseDate(row.cells[dateColIdx]);
-    const dienstgrad = cellText(row.cells[1]) ?? '';
+    // The next non-empty column after the date holds the Dienstgrad
+    let dienstgrad = '';
+    for (let ci = dateColIdx + 1; ci < row.cells.length; ci++) {
+      const v = cellText(row.cells[ci]);
+      if (v) { dienstgrad = v; break; }
+    }
     const syncKey = `${standesbuchNr}::${dienstgrad}::${datum ?? ''}`;
     results.push({ standesbuchNr, datum, dienstgrad, syncKey });
   }
@@ -885,22 +1017,32 @@ async function scrapeMemberUntersuchungen(
     + `?search=1&searchid_mitgliedschaften=${idMitgliedschaft}&id_personen=${idPersonen}`
     + `&id_mitgliedschaften=${idMitgliedschaft}&searchid_personen=${idPersonen}&searchid_maskmode=`;
 
-  const rows = await navigateAndGetTableRows(frame, url);
+  const result = await navigateAndGetTableRows(frame, url);
-  if (!rows) return [];
+  if (!result) return [];
 
+  const { rows, dateColIdx } = result;
   const results: FdiskUntersuchung[] = [];
   for (const row of rows) {
-    // Columns: 0=Datum, 1=Anmerkungen, 2=Untersuchungsart, 3=Tauglichkeitsstufe
+    // Collect non-empty values from columns after the date column
-    const art = cellText(row.cells[2]);
+    const valueCols: string[] = [];
+    for (let ci = dateColIdx + 1; ci < row.cells.length; ci++) {
+      const v = cellText(row.cells[ci]);
+      if (v !== null) valueCols.push(v);
+    }
+    // Original layout: 0=Datum, 1=Anmerkungen, 2=Untersuchungsart, 3=Tauglichkeitsstufe
+    // With spacer columns the date may not be at 0; use relative offsets from collected values
+    const anmerkungen = valueCols[0] ?? null;
+    const art = valueCols[1] ?? null;
+    const ergebnis = valueCols[2] ?? null;
     if (!art) continue;
-    const datum = parseDate(row.cells[0]);
+    const datum = parseDate(row.cells[dateColIdx]);
     const syncKey = `${standesbuchNr}::${art}::${datum ?? ''}`;
     results.push({
       standesbuchNr,
       datum,
-      anmerkungen: cellText(row.cells[1]),
+      anmerkungen,
       art,
-      ergebnis: cellText(row.cells[3]),
+      ergebnis,
       syncKey,
     });
   }
@@ -911,8 +1053,9 @@ async function scrapeMemberUntersuchungen(
 
 /**
  * Navigate to the Gesetzliche Fahrgenehmigungen sub-page and scrape all entries.
- * This page is a ListEdit page with a different structure than normal list pages.
+ * This page is a ListEdit page with form fields named by row index pattern:
- * Uses its own page evaluation to read <th> headers + <td>/<input>/<select> data.
+ *   ausstellungsdatum_{i}, gueltig_bis_{i}, behoerde_{i}, nummer_{i}, id_fahrgenehmigungsklassen_{i}
+ * Falls back to table-based parsing if field IDs are not found.
  */
 async function scrapeMemberFahrgenehmigungen(
   frame: Frame,
@@ -935,7 +1078,88 @@ async function scrapeMemberFahrgenehmigungen(
     return [];
   }
 
-  // Custom page evaluation: extract headers (<th>) and data rows (<td> with input/select)
+  // Dump HTML for diagnostics
+  await dumpHtml(frame, `fahrgenehmigungen_StNr${standesbuchNr}`);
+
+  // Read form fields by ID pattern: {fieldname}_{rowIndex}
+  const rawRows = await frame.evaluate(() => {
+    const rows: Array<{
+      ausstellungsdatum: string;
+      gueltigBis: string;
+      behoerde: string;
+      nummer: string;
+      klasse: string;
+    }> = [];
+
+    for (let i = 0; i < 100; i++) {
+      // Try to find any field for this row index — if none exist, we've passed all rows
+      const ausstellungEl = document.querySelector(`input[name="ausstellungsdatum_${i}"], input[id="ausstellungsdatum_${i}"]`) as HTMLInputElement | null;
+      const gueltigEl     = document.querySelector(`input[name="gueltig_bis_${i}"], input[id="gueltig_bis_${i}"]`) as HTMLInputElement | null;
+      const behoerdeEl    = document.querySelector(`input[name="behoerde_${i}"], input[id="behoerde_${i}"]`) as HTMLInputElement | null;
+      const nummerEl      = document.querySelector(`input[name="nummer_${i}"], input[id="nummer_${i}"]`) as HTMLInputElement | null;
+      const klasseEl      = document.querySelector(`select[name="id_fahrgenehmigungsklassen_${i}"], select[id="id_fahrgenehmigungsklassen_${i}"]`) as HTMLSelectElement | null;
+
+      // If no field found at all, stop
+      if (!ausstellungEl && !gueltigEl && !behoerdeEl && !nummerEl && !klasseEl) break;
+
+      // Read klasse from select: try selectedIndex, then fallback to [selected] attribute
+      let klasse = '';
+      if (klasseEl) {
+        const idx = klasseEl.selectedIndex;
+        if (idx >= 0 && klasseEl.options[idx]) {
+          klasse = (klasseEl.options[idx].text || klasseEl.options[idx].value || '').trim();
+        }
+        if (!klasse) {
+          const selectedOpt = klasseEl.querySelector('option[selected]') as HTMLOptionElement | null;
+          if (selectedOpt) {
+            klasse = (selectedOpt.text || selectedOpt.value || '').trim();
+          }
+        }
+        if (!klasse && klasseEl.value?.trim()) {
+          klasse = klasseEl.value.trim();
+        }
+      }
+
+      rows.push({
+        ausstellungsdatum: ausstellungEl?.value?.trim() ?? '',
+        gueltigBis:        gueltigEl?.value?.trim() ?? '',
+        behoerde:          behoerdeEl?.value?.trim() ?? '',
+        nummer:            nummerEl?.value?.trim() ?? '',
+        klasse,
+      });
+    }
+
+    return rows;
+  }).catch(() => [] as Array<{ ausstellungsdatum: string; gueltigBis: string; behoerde: string; nummer: string; klasse: string }>);
+
+  log(`    → Fahrgenehmigungen form-field extraction: ${rawRows.length} rows found`);
+
+  // If form-field approach found rows, use them
+  if (rawRows.length > 0) {
+    const results: FdiskFahrgenehmigung[] = [];
+    for (const row of rawRows) {
+      const klasse = cellText(row.klasse);
+      if (!klasse) continue;
+      const ausstellungsdatum = parseDate(row.ausstellungsdatum);
+      const syncKey = `${standesbuchNr}::${klasse}::${ausstellungsdatum ?? ''}`;
+      results.push({
+        standesbuchNr,
+        ausstellungsdatum,
+        gueltigBis: parseDate(row.gueltigBis),
+        behoerde:   cellText(row.behoerde),
+        nummer:     cellText(row.nummer),
+        klasse,
+        syncKey,
+      });
+    }
+    log(`  Fahrgenehmigungen for StNr ${standesbuchNr}: ${results.length} rows`);
+    for (const f of results) log(`    ${f.ausstellungsdatum ?? '—'} [${f.klasse}] ${f.behoerde ?? ''} ${f.nummer ?? ''}`);
+    return results;
+  }
+
+  // Fallback: table-based parsing (original approach with extractCellValue)
+  log(`    → Fahrgenehmigungen: no form fields found, falling back to table parsing`);
+
   const pageData = await frame.evaluate(() => {
     const extractCellValue = (cell: Element): string => {
       const input = cell.querySelector('input[type="text"], input:not([type])') as HTMLInputElement | null;
@@ -947,13 +1171,13 @@ async function scrapeMemberFahrgenehmigungen(
           const t = (sel.options[idx].text || sel.options[idx].value || '').trim();
           if (t) return t;
         }
-        if (sel.value?.trim()) return sel.value.trim();
+        // Fallback: read the selected attribute directly from HTML
         const selectedOpt = sel.querySelector('option[selected]') as HTMLOptionElement | null;
         if (selectedOpt) {
           const t = (selectedOpt.text || selectedOpt.value || '').trim();
           if (t) return t;
         }
-        // fall through to textContent if select is empty
+        if (sel.value?.trim()) return sel.value.trim();
       }
       const anchor = cell.querySelector('a');
       const atitle = anchor?.getAttribute('title')?.trim();
@@ -969,16 +1193,13 @@ async function scrapeMemberFahrgenehmigungen(
 
     for (const table of Array.from(document.querySelectorAll('table'))) {
       const cls = table.className || '';
-      // Extract headers from <th> elements
       const thElements = Array.from(table.querySelectorAll('thead th, tr th'));
       const headers = thElements.map(th => extractCellValue(th));
-      // Extract data from <td> elements
       const dataRows: Array<{ cells: string[] }> = [];
       for (const tr of Array.from(table.querySelectorAll('tr'))) {
         if (tr.closest('table') !== table) continue;
         const tds = Array.from(tr.querySelectorAll('td'));
         if (tds.length < 2) continue;
-        // Skip rows that contain <th> (header rows)
         if (tr.querySelectorAll('th').length > 0) continue;
         dataRows.push({ cells: tds.map(td => extractCellValue(td)) });
       }
@@ -997,7 +1218,6 @@ async function scrapeMemberFahrgenehmigungen(
     }
   }
 
-  // Pick the best table: prefer FdcLayList tables, then largest table with data
   const bestTable = pageData.find(t => t.tableClass.includes('FdcLayList') && t.rows.length > 0)
     || pageData.filter(t => t.rows.length > 0).sort((a, b) => b.rows.length - a.rows.length)[0];
 
@@ -1009,24 +1229,23 @@ async function scrapeMemberFahrgenehmigungen(
   const headers = bestTable.headers.map(h => h.toLowerCase());
   log(`  Fahrgenehmigungen headers: [${headers.join(', ')}]`);
 
-  // Map column indices from headers
   let klasseIdx = headers.findIndex(h => h.includes('klasse') || h.includes('fahrgenehmigung'));
   let ausstellungIdx = headers.findIndex(h => h.includes('ausstellung'));
   let gueltigIdx = headers.findIndex(h => h.includes('gültig') || h.includes('gultig') || h.includes('ablauf'));
   let behoerdeIdx = headers.findIndex(h => h.includes('behörde') || h.includes('behorde'));
   let nummerIdx = headers.findIndex(h => h.includes('nummer') || h.includes('nr'));
 
-  // If headers didn't help, try scanning data rows for known Führerscheinklassen
   const KNOWN_KLASSEN = new Set([
     'AM', 'A1', 'A2', 'A', 'B', 'BE', 'C1', 'C1E', 'C', 'CE',
     'D1', 'D1E', 'D', 'DE', 'F', 'L', 'L17', 'B+E', 'C+E', 'D+E',
   ]);
 
   if (klasseIdx === -1) {
-    // Scan first 3 data rows to find which column contains a known Klasse
     for (const row of bestTable.rows.slice(0, 3)) {
       for (let ci = 0; ci < row.cells.length; ci++) {
-        if (KNOWN_KLASSEN.has(row.cells[ci]?.trim().toUpperCase())) {
+        const val = row.cells[ci]?.trim();
+        // Match known klassen or values containing "Führerschein" etc.
+        if (KNOWN_KLASSEN.has(val.toUpperCase()) || /führerschein|lenkberechtigung/i.test(val)) {
           klasseIdx = ci;
           log(`  Fahrgenehmigungen: found Klasse in column ${ci} by data inspection`);
           break;
@@ -1036,7 +1255,6 @@ async function scrapeMemberFahrgenehmigungen(
     }
   }
 
-  // If still no klasse column found, also try matching date columns for Ausstellung
   if (ausstellungIdx === -1) {
     const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
     for (const row of bestTable.rows.slice(0, 3)) {
@@ -1055,7 +1273,7 @@ async function scrapeMemberFahrgenehmigungen(
 
   if (klasseIdx === -1) {
     log(`  Fahrgenehmigungen for StNr ${standesbuchNr}: could not determine Klasse column. Returning empty.`);
-    await dumpHtml(frame, `fahrgenehmigungen_StNr${standesbuchNr}`);
+    await dumpHtml(frame, `fahrgenehmigungen_fallback_StNr${standesbuchNr}`);
     return [];
   }
 
@@ -1063,9 +1281,7 @@ async function scrapeMemberFahrgenehmigungen(
   for (const row of bestTable.rows) {
     const klasse = cellText(row.cells[klasseIdx]);
     if (!klasse) continue;
-    // Skip non-data rows (pagination, info text, header-like rows)
     if (/klasse|fahrgenehmigung|ausstellung|datensätze|information|tiefennavigation/i.test(klasse)) continue;
-    // Skip rows where klasse looks like a date (clearly wrong column)
     if (/^\d{2}\.\d{2}\.\d{4}$/.test(klasse)) continue;
 
     const ausstellungsdatum = parseDate(ausstellungIdx >= 0 ? row.cells[ausstellungIdx] : undefined);
@@ -1089,5 +1305,9 @@ async function scrapeMemberFahrgenehmigungen(
 export async function scrapeMemberAusbildung(frame: Frame, member: FdiskMember): Promise<FdiskAusbildung[]> {
   if (!member.detailUrl) return [];
   await frame_goto(frame, member.detailUrl);
-  return scrapeAusbildungenFromDetailPage(frame, member);
+  // Try to extract IDs from the detail URL
+  const urlObj = new URL(member.detailUrl, frame.url());
+  const idMitgliedschaft = urlObj.searchParams.get('id_mitgliedschaften');
+  const idPersonen = urlObj.searchParams.get('id_personen');
+  return scrapeAusbildungenFromDetailPage(frame, member, idMitgliedschaft, idPersonen);
 }