new features

sync/src/scraper.ts

@@ -241,9 +241,25 @@ export async function scrapeAll(username: string, password: string, knownStNrs:
         const idPersonen       = urlObj.searchParams.get('id_personen');
         const idInstanzen      = urlObj.searchParams.get('id_instanzen') ?? ID_INSTANZEN;
 
-        // Ausbildungen — disabled: requires different page/approach (TODO)
-        // const quals = await scrapeAusbildungenFromDetailPage(mainFrame, member, idMitgliedschaft, idPersonen);
-        // ausbildungen.push(...quals);
+        // Ausbildungen
+        if (idMitgliedschaft && idPersonen) {
+          try {
+            const quals = await scrapeAusbildungenFromDetailPage(mainFrame, member, idMitgliedschaft, idPersonen);
+            ausbildungen.push(...quals);
+            log(`  ${member.vorname} ${member.zuname}: ${quals.length} Ausbildungen`);
+          } catch (err: any) {
+            log(`  WARN: Ausbildungen scrape failed for ${member.vorname} ${member.zuname} (StNr ${member.standesbuchNr}): ${err.message}`);
+            // Always dump HTML on failure for diagnosis
+            try {
+              const debugDir = path.resolve(process.cwd(), 'debug');
+              fs.mkdirSync(debugDir, { recursive: true });
+              const html = await mainFrame.content();
+              const filePath = path.join(debugDir, `ausbildungen_error_StNr${member.standesbuchNr}.html`);
+              fs.writeFileSync(filePath, html, 'utf-8');
+              log(`  [debug] saved error HTML → ${filePath}`);
+            } catch { /* ignore dump errors */ }
+          }
+        }
 
         // Beförderungen
         const befos = (idMitgliedschaft && idPersonen)
@@ -1034,20 +1050,110 @@ async function scrapeMemberUntersuchungen(
     + `?search=1&searchid_mitgliedschaften=${idMitgliedschaft}&id_personen=${idPersonen}`
     + `&id_mitgliedschaften=${idMitgliedschaft}&searchid_personen=${idPersonen}&searchid_maskmode=`;
 
-  const result = await navigateAndGetTableRows(frame, url);
-  if (!result) return [];
+  // Always dump for diagnosis when debug is on
+  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(`    → Untersuchungen ERROR page: ${landed}`);
+    await dumpHtml(frame, `untersuchungen_error_StNr${standesbuchNr}`);
+    return [];
+  }
+
+  // Show all rows
+  await selectAlleAnzeige(frame);
+
+  // Dump HTML for diagnosis (always when debug enabled)
+  await dumpHtml(frame, `untersuchungen_StNr${standesbuchNr}`);
+
+  // Try to navigate to history/detail view if available
+  // FDISK may show only the most recent per exam type on the list page.
+  // Look for a "Verlauf" or "Detail" or "Alle anzeigen" link/button
+  const hasHistoryLink = await frame.evaluate(() => {
+    const links = Array.from(document.querySelectorAll('a, input[type="button"], button'));
+    for (const el of links) {
+      const text = (el.textContent || '').toLowerCase();
+      const title = (el.getAttribute('title') || '').toLowerCase();
+      if (text.includes('verlauf') || text.includes('historie') || text.includes('alle anzeigen')
+        || title.includes('verlauf') || title.includes('historie')) {
+        return (el as HTMLElement).id || (el as HTMLAnchorElement).href || text;
+      }
+    }
+    return null;
+  }).catch(() => null);
+
+  if (hasHistoryLink) {
+    log(`    → Found history link: ${hasHistoryLink}`);
+  }
+
+  // Parse the table using navigateAndGetTableRows logic (reuse existing page state)
+  // Re-collect rows from the already-loaded page
+  const allRows = await frame.evaluate(() => {
+    const results: Array<{ cells: string[]; tableClass: string }> = [];
+    for (const table of Array.from(document.querySelectorAll('table'))) {
+      const cls = table.className || '';
+      for (const tr of Array.from(table.querySelectorAll('tbody tr, tr'))) {
+        if (tr.closest('table') !== table) continue;
+        const tds = Array.from(tr.querySelectorAll('td'));
+        if (tds.length < 2) continue;
+        results.push({
+          tableClass: cls,
+          cells: tds.map(td => {
+            const input = td.querySelector('input[type="text"], input:not([type])') as HTMLInputElement | null;
+            if (input) return input.value?.trim() ?? '';
+            const sel = td.querySelector('select') as HTMLSelectElement | null;
+            if (sel) {
+              const opt = sel.options[sel.selectedIndex];
+              return (opt?.text || opt?.value || '').trim();
+            }
+            const anchor = td.querySelector('a');
+            const atitle = anchor?.getAttribute('title')?.trim();
+            if (atitle) return atitle;
+            return td.textContent?.trim() ?? '';
+          }),
+        });
+      }
+    }
+    return results;
+  }).catch(() => [] as Array<{ cells: string[]; tableClass: string }>);
+
+  const fdcRows = allRows.filter(r => r.tableClass.includes('FdcLayList'));
+  const resultRows = fdcRows.length > 0 ? fdcRows : allRows;
+  const mapped = resultRows.map(r => ({
+    cells: r.cells.map(c => c.replace(/\u00A0/g, ' ').trim()),
+  }));
+
+  // Find date column
+  const datePattern = /^\d{2}\.\d{2}\.\d{4}$/;
+  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;
+  }
+
+  const dataRows = dateColIdx >= 0
+    ? mapped.filter(r => datePattern.test(r.cells[dateColIdx] ?? ''))
+    : [];
+
+  log(`    → Untersuchungen: ${allRows.length} total rows, ${dataRows.length} data rows (date in col ${dateColIdx})`);
+
+  if (dataRows.length === 0) {
+    await dumpHtml(frame, `untersuchungen_empty_StNr${standesbuchNr}`);
+  }
 
-  const { rows, dateColIdx } = result;
   const results: FdiskUntersuchung[] = [];
-  for (const row of rows) {
-    // Collect non-empty values from columns after the date column
+  for (const row of dataRows) {
     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;