Update app.py

app.py

@@ -65,24 +65,16 @@ class AuditState:
 audit_state = AuditState()
 
 
-# ═══════════════════════════════════════════════════
-#  CORE AUDIT FUNCTION
-# ═══════════════════════════════════════════════════
-
 def run_audit(file, pasted_urls, domain, batch_size, timeout, delay, workers):
     if sb is None:
-        yield "❌ Supabase not connected. Set SUPABASE_URL and SUPABASE_KEY in Space secrets.", ""
+        yield "❌ Supabase not connected.", ""
         return
 
-    # Parse URLs
     urls = []
     if file is not None:
         try:
             fpath = file.name if hasattr(file, 'name') else file
-            if str(fpath).endswith('.csv'):
-                df = pd.read_csv(fpath)
-            else:
-                df = pd.read_excel(fpath)
+            df = pd.read_csv(fpath) if str(fpath).endswith('.csv') else pd.read_excel(fpath)
             url_col = None
             for col in df.columns:
                 sample = str(df[col].iloc[0]).strip().lower()
@@ -99,46 +91,42 @@ def run_audit(file, pasted_urls, domain, batch_size, timeout, delay, workers):
         urls = [u.strip() for u in pasted_urls.strip().split('\n') if u.strip().startswith('http')]
 
     if not urls:
-        yield "⚠ No valid URLs found. Upload a file or paste URLs.", ""
+        yield "⚠ No valid URLs.", ""
         return
 
-    # Deduplicate
     seen = set()
-    unique_urls = []
+    unique = []
     for u in urls:
         if u not in seen:
             seen.add(u)
-            unique_urls.append(u)
-    urls = unique_urls
+            unique.append(u)
+    urls = unique
 
     run_name = f"{domain} Audit — {datetime.now().strftime('%b %d %H:%M')} — {len(urls)} pages"
     run_id = create_run(sb, run_name, domain, len(urls), urls)
-
     audit_state.set_running(True, run_id)
     audit_state.resume()
 
     total = len(urls)
-    batch_size = int(batch_size)
-    timeout = int(timeout)
-    workers = int(workers)
+    batch_size, timeout, workers = int(batch_size), int(timeout), int(workers)
     start_time = time.time()
     batch_num = 0
-    log_lines = []
+    log = []
 
-    yield f"🚀 Started: {run_name}\n📦 {total} URLs · Batch size: {batch_size}", "▶️ Running..."
+    yield f"🚀 {run_name}\n📦 {total} URLs · Batch: {batch_size}", "▶️ Running..."
 
     try:
-        for batch_start in range(0, total, batch_size):
+        for bs in range(0, total, batch_size):
             if audit_state.is_paused():
-                completed = get_completed_count(sb, run_id)
-                update_run_status(sb, run_id, "paused", completed)
-                log_lines.append(f"⏸️ PAUSED at {completed}/{total} — resume from Past Runs tab")
+                c = get_completed_count(sb, run_id)
+                update_run_status(sb, run_id, "paused", c)
+                log.append(f"⏸️ PAUSED at {c}/{total}")
                 audit_state.set_running(False)
-                yield "\n".join(log_lines[-40:]), f"⏸️ Paused — {completed}/{total} saved"
+                yield "\n".join(log[-40:]), f"⏸️ Paused — {c}/{total}"
                 return
 
-            batch_end = min(batch_start + batch_size, total)
-            batch_urls = urls[batch_start:batch_end]
+            be = min(bs + batch_size, total)
+            batch_urls = urls[bs:be]
             batch_num += 1
             batch_results = []
 
@@ -146,72 +134,62 @@ def run_audit(file, pasted_urls, domain, batch_size, timeout, delay, workers):
                 if audit_state.is_paused():
                     if batch_results:
                         save_batch_results(sb, run_id, batch_results)
-                    completed = get_completed_count(sb, run_id)
-                    update_run_status(sb, run_id, "paused", completed)
-                    log_lines.append(f"⏸️ PAUSED at {completed}/{total}")
+                    c = get_completed_count(sb, run_id)
+                    update_run_status(sb, run_id, "paused", c)
+                    log.append(f"⏸️ PAUSED at {c}/{total}")
                     audit_state.set_running(False)
-                    yield "\n".join(log_lines[-40:]), f"⏸️ Paused — {completed}/{total} saved"
+                    yield "\n".join(log[-40:]), f"⏸️ Paused — {c}/{total}"
                     return
 
-                global_idx = batch_start + j + 1
+                gi = bs + j + 1
                 elapsed = time.time() - start_time
-                avg = elapsed / global_idx
-                eta = avg * (total - global_idx)
-                eta_str = f"{int(eta//60)}m {int(eta%60)}s" if eta > 60 else f"{eta:.0f}s"
-
-                result = audit_page(
-                    url, domain, DEFAULT_BODY_SELECTORS,
-                    suggestion_map=DEFAULT_SUGGESTION_MAP,
-                    timeout=timeout, concurrent_workers=workers,
-                )
+                eta = (elapsed / gi) * (total - gi)
+                eta_s = f"{int(eta//60)}m{int(eta%60)}s" if eta > 60 else f"{eta:.0f}s"
+
+                result = audit_page(url, domain, DEFAULT_BODY_SELECTORS,
+                    suggestion_map=DEFAULT_SUGGESTION_MAP, timeout=timeout, concurrent_workers=workers)
                 batch_results.append(result)
 
                 short = url.replace('https://www.', '').replace('https://', '')[:70]
                 if result['error']:
-                    log_lines.append(f"❌ [{global_idx}/{total}] {short} — {result['error'][:50]}")
+                    log.append(f"❌ [{gi}/{total}] {short} — {result['error'][:50]}")
                 else:
                     b = result['broken_int_count'] + result['broken_ext_count']
                     fc = result['follow_flag_count']
                     d = result['duplicate_count']
-                    flags = []
-                    if b: flags.append(f"🔴 {b} broken")
-                    if fc: flags.append(f"🟡 {fc} flags")
-                    if d: flags.append(f"🟣 {d} dups")
-                    flag_str = " · ".join(flags) if flags else "✅"
-                    log_lines.append(f"[{global_idx}/{total}] {short} — Int:{result['int_count']} Ext:{result['ext_count']} · {flag_str}")
-
-                yield "\n".join(log_lines[-40:]), f"📊 {global_idx}/{total} ({global_idx*100//total}%) · Batch {batch_num} · ETA: {eta_str}"
-
+                    fl = []
+                    if b: fl.append(f"🔴{b}broken")
+                    if fc: fl.append(f"🟡{fc}flags")
+                    if d: fl.append(f"🟣{d}dups")
+                    fs = " ".join(fl) if fl else "✅"
+                    log.append(f"[{gi}/{total}] {short} — Int:{result['int_count']} Ext:{result['ext_count']} {fs}")
+
+                yield "\n".join(log[-40:]), f"📊 {gi}/{total} ({gi*100//total}%) Batch{batch_num} ETA:{eta_s}"
                 if j < len(batch_urls) - 1:
                     time.sleep(delay)
 
-            # Save batch
             if batch_results:
                 try:
                     save_batch_results(sb, run_id, batch_results)
-                    completed = get_completed_count(sb, run_id)
-                    update_run_status(sb, run_id, "running", completed)
-                    log_lines.append(f"💾 Batch {batch_num} saved — {completed}/{total} done")
+                    c = get_completed_count(sb, run_id)
+                    update_run_status(sb, run_id, "running", c)
+                    log.append(f"💾 Batch {batch_num} saved — {c}/{total}")
                 except Exception as e:
-                    log_lines.append(f"⚠ Batch save error: {str(e)[:60]}")
-
-                yield "\n".join(log_lines[-40:]), f"💾 Saved batch {batch_num} — {min(batch_end, total)}/{total}"
+                    log.append(f"⚠ Save error: {str(e)[:60]}")
+                yield "\n".join(log[-40:]), f"💾 Batch {batch_num} saved"
                 del batch_results
 
-        # Orphan analysis
-        log_lines.append("🔍 Running orphan page analysis...")
-        yield "\n".join(log_lines[-40:]), "🔍 Orphan analysis..."
+        log.append("🔍 Orphan analysis...")
+        yield "\n".join(log[-40:]), "🔍 Orphan analysis..."
 
         all_pages = get_all_page_results(sb, run_id)
         all_results = [p['result'] for p in all_pages]
-
-        all_internal_targets = set()
-        all_page_urls = set()
+        targets, pg_urls = set(), set()
         for r in all_results:
-            all_page_urls.add(r['url'].rstrip('/').split('?')[0])
-            for link in r.get('internal_links', []):
-                all_internal_targets.add(link['url'].rstrip('/').split('?')[0])
-        orphan_pages = sorted([p for p in all_page_urls if p not in all_internal_targets])
+            pg_urls.add(r['url'].rstrip('/').split('?')[0])
+            for lk in r.get('internal_links', []):
+                targets.add(lk['url'].rstrip('/').split('?')[0])
+        orphans = sorted([p for p in pg_urls if p not in targets])
 
         summary = {
             'total_pages': len(all_results),
@@ -222,340 +200,279 @@ def run_audit(file, pasted_urls, domain, batch_size, timeout, delay, workers):
             'total_flags': sum(r.get('follow_flag_count', 0) for r in all_results),
             'total_dups': sum(r.get('duplicate_count', 0) for r in all_results),
             'total_sug': sum(len(r.get('suggestions', [])) for r in all_results),
-            'orphan_count': len(orphan_pages),
-            'orphan_urls': orphan_pages[:100],
+            'orphan_count': len(orphans),
+            'orphan_urls': orphans[:100],
         }
         update_run_status(sb, run_id, "completed", len(all_results), summary)
-
-        total_time = time.time() - start_time
-        log_lines.append(f"✅ COMPLETE! {len(all_results)} pages in {total_time:.0f}s · {len(orphan_pages)} orphans")
-        log_lines.append(f"📊 Broken: {summary['total_broken']} · Redirects: {summary['total_redirects']} · Flags: {summary['total_flags']} · Dups: {summary['total_dups']}")
-        log_lines.append("→ Go to Past Runs tab → Refresh → Generate Report")
-
+        tt = time.time() - start_time
+        log.append(f"✅ DONE! {len(all_results)} pages in {tt:.0f}s · {len(orphans)} orphans")
+        log.append(f"Broken:{summary['total_broken']} Redirects:{summary['total_redirects']} Flags:{summary['total_flags']} Dups:{summary['total_dups']}")
+        log.append("→ Past Runs → Refresh → Generate Report")
         audit_state.set_running(False)
-        yield "\n".join(log_lines[-40:]), f"✅ Complete — {len(all_results)} pages in {total_time:.0f}s"
+        yield "\n".join(log[-40:]), f"✅ Done — {len(all_results)} pages in {tt:.0f}s"
 
     except Exception as e:
-        log_lines.append(f"❌ Error: {str(e)}")
+        log.append(f"❌ {str(e)}")
         audit_state.set_running(False)
-        if run_id:
-            try:
-                completed = get_completed_count(sb, run_id)
-                update_run_status(sb, run_id, "paused", completed)
-            except:
-                pass
-        yield "\n".join(log_lines[-40:]), f"❌ Error — progress saved to Supabase"
+        try:
+            c = get_completed_count(sb, run_id)
+            update_run_status(sb, run_id, "paused", c)
+        except: pass
+        yield "\n".join(log[-40:]), "❌ Error — progress saved"
 
 
 def pause_audit():
     if audit_state.is_running():
         audit_state.request_pause()
-        return "⏸️ Pause requested — stopping after current page..."
+        return "⏸️ Stopping after current page..."
     return "No audit running."
 
 
-# ═══════════════════════════════════════════════════
-#  RESUME FUNCTION
-# ═══════════════════════════════════════════════════
-
 def resume_audit(run_id, domain, batch_size, timeout, delay, workers):
     if sb is None:
         yield "❌ Supabase not connected.", ""
         return
-
     if not run_id:
-        yield "⚠ No run selected — pick one from the dropdown above.", ""
+        yield "⚠ Select a run first.", ""
         return
 
-    all_urls_for_run = get_pending_urls(sb, run_id)
-    done_urls = get_completed_urls(sb, run_id)
-    remaining = [u for u in all_urls_for_run if u not in done_urls]
+    all_urls = get_pending_urls(sb, run_id)
+    done = get_completed_urls(sb, run_id)
+    remaining = [u for u in all_urls if u not in done]
 
     if not remaining:
-        update_run_status(sb, run_id, "completed", len(done_urls))
-        yield "✅ All pages already audited! Status updated to completed.", ""
+        update_run_status(sb, run_id, "completed", len(done))
+        yield "✅ Already complete!", ""
         return
 
-    # Get domain from run
     try:
         runs = get_all_runs(sb)
-        run_data = next((r for r in runs if r['id'] == run_id), None)
-        if run_data:
-            domain = run_data.get('domain', domain)
-    except:
-        pass
+        rd = next((r for r in runs if r['id'] == run_id), None)
+        if rd: domain = rd.get('domain', domain)
+    except: pass
 
     audit_state.set_running(True, run_id)
     audit_state.resume()
     update_run_status(sb, run_id, "running")
 
-    total = len(all_urls_for_run)
-    batch_size = int(batch_size)
-    timeout = int(timeout)
-    workers = int(workers)
+    total = len(all_urls)
+    batch_size, timeout, workers = int(batch_size), int(timeout), int(workers)
     start_time = time.time()
-    batch_num = 0
-    log_lines = [f"▶️ Resuming — {len(remaining)} pages remaining ({len(done_urls)} already done)"]
-
-    yield "\n".join(log_lines), f"📊 Resuming: {len(done_urls)}/{total}"
+    bn = 0
+    log = [f"▶️ Resuming — {len(remaining)} left ({len(done)} done)"]
+    yield "\n".join(log), f"Resuming: {len(done)}/{total}"
 
     try:
-        for batch_start in range(0, len(remaining), batch_size):
+        for bs in range(0, len(remaining), batch_size):
             if audit_state.is_paused():
-                completed = get_completed_count(sb, run_id)
-                update_run_status(sb, run_id, "paused", completed)
-                log_lines.append(f"⏸️ PAUSED at {completed}/{total}")
+                c = get_completed_count(sb, run_id)
+                update_run_status(sb, run_id, "paused", c)
+                log.append(f"⏸️ PAUSED {c}/{total}")
                 audit_state.set_running(False)
-                yield "\n".join(log_lines[-40:]), f"⏸️ Paused — {completed}/{total} saved"
+                yield "\n".join(log[-40:]), f"⏸️ Paused {c}/{total}"
                 return
 
-            batch_end = min(batch_start + batch_size, len(remaining))
-            batch_urls = remaining[batch_start:batch_end]
-            batch_num += 1
-            batch_results = []
+            be = min(bs + batch_size, len(remaining))
+            bu = remaining[bs:be]
+            bn += 1
+            br = []
 
-            for j, url in enumerate(batch_urls):
+            for j, url in enumerate(bu):
                 if audit_state.is_paused():
-                    if batch_results:
-                        save_batch_results(sb, run_id, batch_results)
-                    completed = get_completed_count(sb, run_id)
-                    update_run_status(sb, run_id, "paused", completed)
-                    log_lines.append(f"⏸️ PAUSED at {completed}/{total}")
+                    if br: save_batch_results(sb, run_id, br)
+                    c = get_completed_count(sb, run_id)
+                    update_run_status(sb, run_id, "paused", c)
+                    log.append(f"⏸️ PAUSED {c}/{total}")
                     audit_state.set_running(False)
-                    yield "\n".join(log_lines[-40:]), f"⏸️ Paused — {completed}/{total} saved"
+                    yield "\n".join(log[-40:]), f"⏸️ Paused {c}/{total}"
                     return
 
-                global_idx = len(done_urls) + batch_start + j + 1
+                gi = len(done) + bs + j + 1
                 elapsed = time.time() - start_time
-                processed = batch_start + j + 1
-                avg = elapsed / processed
-                eta = avg * (len(remaining) - processed)
-                eta_str = f"{int(eta//60)}m {int(eta%60)}s" if eta > 60 else f"{eta:.0f}s"
-
-                result = audit_page(
-                    url, domain, DEFAULT_BODY_SELECTORS,
-                    suggestion_map=DEFAULT_SUGGESTION_MAP,
-                    timeout=timeout, concurrent_workers=workers,
-                )
-                batch_results.append(result)
+                proc = bs + j + 1
+                eta = (elapsed / proc) * (len(remaining) - proc)
+                eta_s = f"{int(eta//60)}m{int(eta%60)}s" if eta > 60 else f"{eta:.0f}s"
+
+                result = audit_page(url, domain, DEFAULT_BODY_SELECTORS,
+                    suggestion_map=DEFAULT_SUGGESTION_MAP, timeout=timeout, concurrent_workers=workers)
+                br.append(result)
 
                 short = url.replace('https://www.', '').replace('https://', '')[:70]
                 if result['error']:
-                    log_lines.append(f"❌ [{global_idx}/{total}] {short}")
+                    log.append(f"❌ [{gi}/{total}] {short}")
                 else:
                     b = result['broken_int_count'] + result['broken_ext_count']
-                    flag_str = f"🔴 {b} broken" if b else "✅"
-                    log_lines.append(f"[{global_idx}/{total}] {short} · {flag_str}")
-
-                yield "\n".join(log_lines[-40:]), f"📊 {global_idx}/{total} ({global_idx*100//total}%) · ETA: {eta_str}"
-                if j < len(batch_urls) - 1:
-                    time.sleep(delay)
-
-            if batch_results:
-                save_batch_results(sb, run_id, batch_results)
-                completed = get_completed_count(sb, run_id)
-                update_run_status(sb, run_id, "running", completed)
-                log_lines.append(f"💾 Batch {batch_num} saved — {completed}/{total}")
-                del batch_results
-
-        # Orphan analysis
-        log_lines.append("🔍 Orphan analysis...")
-        yield "\n".join(log_lines[-40:]), "🔍 Orphan analysis..."
-
-        all_pages = get_all_page_results(sb, run_id)
-        all_results = [p['result'] for p in all_pages]
-
-        all_targets = set()
-        all_pg = set()
-        for r in all_results:
-            all_pg.add(r['url'].rstrip('/').split('?')[0])
-            for link in r.get('internal_links', []):
-                all_targets.add(link['url'].rstrip('/').split('?')[0])
-        orphans = sorted([p for p in all_pg if p not in all_targets])
-
-        final_summary = {
-            'total_pages': len(all_results),
-            'total_int': sum(r.get('int_count', 0) for r in all_results),
-            'total_ext': sum(r.get('ext_count', 0) for r in all_results),
-            'total_broken': sum(r.get('broken_int_count', 0) + r.get('broken_ext_count', 0) for r in all_results),
-            'total_redirects': sum(r.get('redirect_int_count', 0) + r.get('redirect_ext_count', 0) for r in all_results),
-            'total_flags': sum(r.get('follow_flag_count', 0) for r in all_results),
-            'total_dups': sum(r.get('duplicate_count', 0) for r in all_results),
-            'total_sug': sum(len(r.get('suggestions', [])) for r in all_results),
-            'orphan_count': len(orphans),
-            'orphan_urls': orphans[:100],
+                    log.append(f"[{gi}/{total}] {short} {'🔴'+str(b) if b else '✅'}")
+
+                yield "\n".join(log[-40:]), f"📊 {gi}/{total} ({gi*100//total}%) ETA:{eta_s}"
+                if j < len(bu) - 1: time.sleep(delay)
+
+            if br:
+                save_batch_results(sb, run_id, br)
+                c = get_completed_count(sb, run_id)
+                update_run_status(sb, run_id, "running", c)
+                log.append(f"💾 Batch {bn} — {c}/{total}")
+                del br
+
+        log.append("🔍 Orphan analysis...")
+        yield "\n".join(log[-40:]), "🔍 Orphans..."
+
+        ap = get_all_page_results(sb, run_id)
+        ar = [p['result'] for p in ap]
+        tgt, pg = set(), set()
+        for r in ar:
+            pg.add(r['url'].rstrip('/').split('?')[0])
+            for lk in r.get('internal_links', []): tgt.add(lk['url'].rstrip('/').split('?')[0])
+        orph = sorted([p for p in pg if p not in tgt])
+
+        fs = {
+            'total_pages': len(ar), 'total_int': sum(r.get('int_count',0) for r in ar),
+            'total_ext': sum(r.get('ext_count',0) for r in ar),
+            'total_broken': sum(r.get('broken_int_count',0)+r.get('broken_ext_count',0) for r in ar),
+            'total_redirects': sum(r.get('redirect_int_count',0)+r.get('redirect_ext_count',0) for r in ar),
+            'total_flags': sum(r.get('follow_flag_count',0) for r in ar),
+            'total_dups': sum(r.get('duplicate_count',0) for r in ar),
+            'total_sug': sum(len(r.get('suggestions',[])) for r in ar),
+            'orphan_count': len(orph), 'orphan_urls': orph[:100],
         }
-        update_run_status(sb, run_id, "completed", len(all_results), final_summary)
-
-        total_time = time.time() - start_time
-        log_lines.append(f"✅ COMPLETE! {len(all_results)} pages in {total_time:.0f}s · {len(orphans)} orphans")
+        update_run_status(sb, run_id, "completed", len(ar), fs)
+        tt = time.time() - start_time
+        log.append(f"✅ DONE! {len(ar)} pages in {tt:.0f}s · {len(orph)} orphans")
         audit_state.set_running(False)
-        yield "\n".join(log_lines[-40:]), f"✅ Complete — {len(all_results)} pages"
+        yield "\n".join(log[-40:]), f"✅ Done — {len(ar)} pages"
 
     except Exception as e:
-        log_lines.append(f"❌ Error: {str(e)}")
+        log.append(f"❌ {str(e)}")
         audit_state.set_running(False)
         try:
-            completed = get_completed_count(sb, run_id)
-            update_run_status(sb, run_id, "paused", completed)
-        except:
-            pass
-        yield "\n".join(log_lines[-40:]), "❌ Error — progress saved"
+            c = get_completed_count(sb, run_id)
+            update_run_status(sb, run_id, "paused", c)
+        except: pass
+        yield "\n".join(log[-40:]), "❌ Error"
 
 
 # ═══════════════════════════════════════════════════
-#  PAST RUNS HELPERS
+#  PAST RUNS — returns ONLY strings, no component objects
 # ═══════════════════════════════════════════════════
 
-def load_past_runs():
+def load_runs_html():
+    """Returns HTML table only."""
     if sb is None:
-        return "<p>❌ Supabase not connected</p>", gr.Dropdown(choices=[], value=None)
-
+        return "<p>❌ Supabase not connected</p>"
     runs = get_all_runs(sb)
     if not runs:
-        return "<p>No saved runs yet.</p>", gr.Dropdown(choices=[], value=None)
+        return "<p>No saved runs yet.</p>"
+
+    html = '<table style="width:100%;border-collapse:collapse;font-size:13px;">'
+    html += '<tr style="background:#f1f5f9;"><th style="padding:8px;text-align:left;">Run</th><th style="padding:8px;text-align:center;">Status</th><th style="padding:8px;text-align:center;">Pages</th><th style="padding:8px;text-align:center;">Broken</th><th style="padding:8px;text-align:center;">Flags</th><th style="padding:8px;text-align:center;">Dups</th><th style="padding:8px;text-align:center;">Orphans</th></tr>'
 
+    for r in runs:
+        s = r.get('summary', {}) or {}
+        st = r.get('status', '?')
+        sc = {'completed':'#059669','paused':'#d97706','running':'#2563eb'}.get(st,'#888')
+        sb2 = {'completed':'rgba(5,150,105,0.1)','paused':'rgba(217,119,6,0.1)','running':'rgba(37,99,235,0.1)'}.get(st,'rgba(136,136,136,0.1)')
+        cr = r.get('created_at','')[:16].replace('T',' ')
+        html += f'<tr style="border-bottom:1px solid #e2e8f0;"><td style="padding:8px;"><b>{r.get("name","?")}</b><br><span style="font-size:10px;color:#94a3b8;">{cr}</span></td><td style="padding:8px;text-align:center;"><span style="background:{sb2};color:{sc};padding:2px 8px;border-radius:10px;font-size:10px;font-weight:700;">{st.upper()}</span></td><td style="padding:8px;text-align:center;font-weight:700;">{r.get("completed_urls",0)}/{r.get("total_urls",0)}</td><td style="padding:8px;text-align:center;color:#dc2626;font-weight:700;">{s.get("total_broken","—")}</td><td style="padding:8px;text-align:center;color:#dc2626;font-weight:700;">{s.get("total_flags","—")}</td><td style="padding:8px;text-align:center;color:#db2777;font-weight:700;">{s.get("total_dups","—")}</td><td style="padding:8px;text-align:center;color:#dc2626;font-weight:700;">{s.get("orphan_count","—")}</td></tr>'
+    html += '</table>'
+    return html
+
+
+def load_runs_choices():
+    """Returns list of (label, value) tuples for dropdown."""
+    if sb is None:
+        return []
+    runs = get_all_runs(sb)
+    if not runs:
+        return []
     choices = []
     for r in runs:
-        status = r.get('status', 'unknown')
-        completed = r.get('completed_urls', 0)
-        total = r.get('total_urls', 0)
-        label = f"{r.get('name', 'Untitled')} [{status.upper()}] ({completed}/{total})"
+        st = r.get('status', '?')
+        label = f"{r.get('name','?')} [{st.upper()}] ({r.get('completed_urls',0)}/{r.get('total_urls',0)})"
         choices.append((label, r['id']))
-
-    html = '<div style="max-height:400px;overflow-y:auto;">'
-    html += '<table style="width:100%;border-collapse:collapse;font-size:13px;">'
-    html += '<tr style="background:#f1f5f9;"><th style="padding:10px;text-align:left;">Run Name</th><th style="padding:10px;text-align:center;">Status</th><th style="padding:10px;text-align:center;">Pages</th><th style="padding:10px;text-align:center;">Broken</th><th style="padding:10px;text-align:center;">Flags</th><th style="padding:10px;text-align:center;">Dups</th><th style="padding:10px;text-align:center;">Orphans</th></tr>'
-
-    for r in runs:
-        summary = r.get('summary', {}) or {}
-        status = r.get('status', 'unknown')
-        s_colors = {'completed': '#059669', 'paused': '#d97706', 'running': '#2563eb'}
-        s_bg = {'completed': 'rgba(5,150,105,0.1)', 'paused': 'rgba(217,119,6,0.1)', 'running': 'rgba(37,99,235,0.1)'}
-        color = s_colors.get(status, '#888')
-        bg = s_bg.get(status, 'rgba(136,136,136,0.1)')
-        created = r.get('created_at', '')[:16].replace('T', ' ')
-
-        html += f'''<tr style="border-bottom:1px solid #e2e8f0;">
-            <td style="padding:10px;"><b>{r.get('name','Untitled')}</b><br><span style="font-size:11px;color:#94a3b8;">{created}</span></td>
-            <td style="padding:10px;text-align:center;"><span style="background:{bg};color:{color};padding:3px 10px;border-radius:12px;font-size:11px;font-weight:700;">{status.upper()}</span></td>
-            <td style="padding:10px;text-align:center;font-weight:700;">{r.get('completed_urls',0)}/{r.get('total_urls',0)}</td>
-            <td style="padding:10px;text-align:center;color:#dc2626;font-weight:700;">{summary.get('total_broken','—')}</td>
-            <td style="padding:10px;text-align:center;color:#dc2626;font-weight:700;">{summary.get('total_flags','—')}</td>
-            <td style="padding:10px;text-align:center;color:#db2777;font-weight:700;">{summary.get('total_dups','—')}</td>
-            <td style="padding:10px;text-align:center;color:#dc2626;font-weight:700;">{summary.get('orphan_count','—')}</td>
-        </tr>'''
-
-    html += '</table></div>'
-    return html, gr.Dropdown(choices=choices, value=choices[0][1] if choices else None)
+    return choices
 
 
 def generate_report_for_run(run_id, domain):
     if sb is None or not run_id:
-        return None, "❌ No run selected or Supabase not connected."
-
+        return None, "❌ No run selected."
     try:
         run = None
-        runs = get_all_runs(sb)
-        for r in runs:
+        for r in get_all_runs(sb):
             if r['id'] == run_id:
                 run = r
                 break
-
         pages = get_all_page_results(sb, run_id)
         if not pages:
-            return None, "⚠ No page data found for this run."
-
+            return None, "⚠ No data."
         results = [p['result'] for p in pages]
-        summary = (run.get('summary', {}) or {}) if run else {}
-        orphan_urls = summary.get('orphan_urls', [])
-        report_domain = run.get('domain', domain) if run else domain
-
-        report_html = generate_report(results, orphan_urls, report_domain)
-
-        tmp = tempfile.NamedTemporaryFile(delete=False, suffix='.html', prefix='Link_Audit_')
-        tmp.write(report_html.encode('utf-8'))
+        s = (run.get('summary', {}) or {}) if run else {}
+        rh = generate_report(results, s.get('orphan_urls', []), run.get('domain', domain) if run else domain)
+        tmp = tempfile.NamedTemporaryFile(delete=False, suffix='.html', prefix='Audit_')
+        tmp.write(rh.encode('utf-8'))
         tmp.close()
-
-        return tmp.name, f"✅ Report generated — {len(results)} pages"
+        return tmp.name, f"✅ Report — {len(results)} pages"
     except Exception as e:
-        return None, f"❌ Error: {str(e)}"
+        return None, f"❌ {str(e)}"
 
 
 def generate_csv_for_run(run_id):
     if sb is None or not run_id:
         return None, "❌ No run selected."
-
     try:
         pages = get_all_page_results(sb, run_id)
         if not pages:
             return None, "⚠ No data."
-
-        rows = []
-        for p in pages:
-            r = p['result']
-            rows.append({
-                'URL': r.get('url', ''),
-                'Internal': r.get('int_count', 0),
-                'External': r.get('ext_count', 0),
-                'Broken': r.get('broken_int_count', 0) + r.get('broken_ext_count', 0),
-                'Redirects': r.get('redirect_int_count', 0) + r.get('redirect_ext_count', 0),
-                'Flags': r.get('follow_flag_count', 0),
-                'Duplicates': r.get('duplicate_count', 0),
-                'Error': r.get('error', ''),
-            })
-
-        tmp = tempfile.NamedTemporaryFile(delete=False, suffix='.csv', prefix='Audit_CSV_')
+        rows = [{'URL': p['result'].get('url',''), 'Internal': p['result'].get('int_count',0),
+                 'External': p['result'].get('ext_count',0),
+                 'Broken': p['result'].get('broken_int_count',0)+p['result'].get('broken_ext_count',0),
+                 'Redirects': p['result'].get('redirect_int_count',0)+p['result'].get('redirect_ext_count',0),
+                 'Flags': p['result'].get('follow_flag_count',0),
+                 'Dups': p['result'].get('duplicate_count',0)} for p in pages]
+        tmp = tempfile.NamedTemporaryFile(delete=False, suffix='.csv', prefix='Audit_')
         pd.DataFrame(rows).to_csv(tmp.name, index=False)
         tmp.close()
-        return tmp.name, f"✅ CSV exported — {len(rows)} rows"
+        return tmp.name, f"✅ CSV — {len(rows)} rows"
     except Exception as e:
-        return None, f"❌ Error: {str(e)}"
+        return None, f"❌ {str(e)}"
 
 
 def delete_selected_run(run_id):
     if sb is None or not run_id:
-        return "❌ No run selected.", "<p>Click Refresh</p>", gr.Dropdown(choices=[], value=None)
+        return "❌ No run selected."
     try:
         delete_run(sb, run_id)
-        html, dropdown = load_past_runs()
-        return "🗑️ Run deleted.", html, dropdown
+        return "🗑️ Deleted. Click Refresh."
     except Exception as e:
-        return f"❌ {str(e)}", "<p>Error</p>", gr.Dropdown(choices=[], value=None)
+        return f"❌ {str(e)}"
 
 
 # ═══════════════════════════════════════════════════
-#  GRADIO UI
+#  UI
 # ═══════════════════════════════════════════════════
 
 with gr.Blocks(title="Link Audit Tool", theme=gr.themes.Soft()) as app:
 
-    gr.HTML("""
-    <div style="background:linear-gradient(135deg,#1e3a5f,#2563eb);padding:24px 28px;border-radius:12px;color:white;margin-bottom:16px;">
+    gr.HTML("""<div style="background:linear-gradient(135deg,#1e3a5f,#2563eb);padding:24px 28px;border-radius:12px;color:white;margin-bottom:16px;">
         <p style="font-size:10px;font-weight:700;letter-spacing:1.5px;text-transform:uppercase;color:#93c5fd;margin-bottom:8px;">SEO LINK AUDIT TOOL</p>
         <h1 style="margin:0 0 4px 0;font-size:24px;">🔗 Bulk Link Audit</h1>
-        <p style="margin:0;opacity:0.8;font-size:13px;">Upload URLs → batch crawl with auto-save → pause/resume anytime → generate interactive report</p>
-    </div>
-    """)
+        <p style="margin:0;opacity:0.8;font-size:13px;">Upload URLs → batch crawl → pause/resume → generate report</p></div>""")
 
-    conn_msg = "🗄️ ✅ Supabase Connected" if sb else "🗄️ ❌ Supabase Not Connected — add SUPABASE_URL and SUPABASE_KEY to Space secrets"
-    gr.Markdown(f"**{conn_msg}**")
+    conn = "🗄️ ✅ Supabase Connected" if sb else "🗄️ ❌ Not Connected"
+    gr.Markdown(f"**{conn}**")
 
     with gr.Tabs():
 
-        # ═══ TAB 1: NEW AUDIT ═══
         with gr.Tab("🔍 New Audit"):
             with gr.Row():
                 with gr.Column(scale=2):
                     file_input = gr.File(label="Upload Excel / CSV", file_types=[".xlsx", ".csv", ".xls"])
-                    pasted_urls = gr.Textbox(label="Or paste URLs (one per line)", lines=5, placeholder="https://www.example.com/blog/page1\nhttps://www.example.com/blog/page2")
-
+                    pasted_urls = gr.Textbox(label="Or paste URLs (one per line)", lines=5)
                 with gr.Column(scale=1):
                     domain_input = gr.Textbox(label="Your Domain", value="edstellar.com")
                     batch_size_input = gr.Slider(5, 50, value=25, step=5, label="Batch Size")
                     timeout_input = gr.Slider(5, 60, value=15, step=5, label="Timeout (s)")
-                    delay_input = gr.Slider(0, 5, value=1.0, step=0.5, label="Delay between pages (s)")
-                    workers_input = gr.Slider(1, 10, value=5, step=1, label="Parallel link checks")
+                    delay_input = gr.Slider(0, 5, value=1.0, step=0.5, label="Delay (s)")
+                    workers_input = gr.Slider(1, 10, value=5, step=1, label="Parallel checks")
 
             with gr.Row():
                 run_btn = gr.Button("🚀 Run Audit", variant="primary", scale=2)
@@ -564,55 +481,44 @@ with gr.Blocks(title="Link Audit Tool", theme=gr.themes.Soft()) as app:
             progress_text = gr.Textbox(label="Status", interactive=False)
             log_output = gr.Textbox(label="Audit Log", lines=20, interactive=False)
 
-            run_btn.click(
-                fn=run_audit,
+            run_btn.click(fn=run_audit,
                 inputs=[file_input, pasted_urls, domain_input, batch_size_input, timeout_input, delay_input, workers_input],
-                outputs=[log_output, progress_text],
-            )
+                outputs=[log_output, progress_text])
             pause_btn.click(fn=pause_audit, outputs=[progress_text])
 
-        # ═══ TAB 2: PAST RUNS ═══
         with gr.Tab("📁 Past Runs"):
-            refresh_btn = gr.Button("🔄 Refresh Runs", variant="secondary")
-            runs_html = gr.HTML(value="<p>Click Refresh to load past runs.</p>")
-            run_dropdown = gr.Dropdown(label="Select a Run", choices=[], interactive=True)
-
             with gr.Row():
-                report_btn = gr.Button("📊 Generate HTML Report", variant="primary")
-                csv_btn = gr.Button("📋 Export CSV", variant="secondary")
-                resume_btn = gr.Button("▶️ Resume Audit", variant="primary")
-                delete_btn = gr.Button("🗑️ Delete Run", variant="stop")
+                refresh_btn = gr.Button("🔄 Refresh", variant="secondary")
+            runs_html = gr.HTML(value="<p>Click Refresh to load.</p>")
+            run_dropdown = gr.Dropdown(label="Select Run", choices=[], interactive=True, type="value")
 
-            action_status = gr.Textbox(label="Action Status", interactive=False)
+            with gr.Row():
+                report_btn = gr.Button("📊 HTML Report", variant="primary")
+                csv_btn = gr.Button("📋 CSV", variant="secondary")
+                resume_btn = gr.Button("▶️ Resume", variant="primary")
+                delete_btn = gr.Button("🗑️ Delete", variant="stop")
 
+            action_status = gr.Textbox(label="Status", interactive=False)
             with gr.Row():
-                report_file = gr.File(label="Download Report", interactive=False)
-                csv_file = gr.File(label="Download CSV", interactive=False)
+                report_file = gr.File(label="Report Download", interactive=False)
+                csv_file = gr.File(label="CSV Download", interactive=False)
 
-            gr.Markdown("---")
-            gr.Markdown("### Resume / Pause Controls")
+            gr.Markdown("---\n### Resume Controls")
             resume_progress = gr.Textbox(label="Resume Status", interactive=False)
             resume_log = gr.Textbox(label="Resume Log", lines=15, interactive=False)
             resume_pause_btn = gr.Button("⏸️ Pause Resume", variant="stop")
 
-            # Wire events
-            refresh_btn.click(fn=load_past_runs, outputs=[runs_html, run_dropdown])
-
-            report_btn.click(
-                fn=generate_report_for_run,
-                inputs=[run_dropdown, domain_input],
-                outputs=[report_file, action_status],
-            )
+            # Refresh: update HTML and dropdown separately
+            refresh_btn.click(fn=load_runs_html, outputs=[runs_html])
+            refresh_btn.click(fn=load_runs_choices, outputs=[run_dropdown])
 
+            report_btn.click(fn=generate_report_for_run, inputs=[run_dropdown, domain_input], outputs=[report_file, action_status])
             csv_btn.click(fn=generate_csv_for_run, inputs=[run_dropdown], outputs=[csv_file, action_status])
+            delete_btn.click(fn=delete_selected_run, inputs=[run_dropdown], outputs=[action_status])
 
-            delete_btn.click(fn=delete_selected_run, inputs=[run_dropdown], outputs=[action_status, runs_html, run_dropdown])
-
-            resume_btn.click(
-                fn=resume_audit,
+            resume_btn.click(fn=resume_audit,
                 inputs=[run_dropdown, domain_input, batch_size_input, timeout_input, delay_input, workers_input],
-                outputs=[resume_log, resume_progress],
-            )
+                outputs=[resume_log, resume_progress])
             resume_pause_btn.click(fn=pause_audit, outputs=[resume_progress])