fix: correct nested data extraction from SEOAuditor results

OnPageSEOResult uses nested objects (meta_tags.title, images.total_images,
structured_data.has_structured_data). TechnicalSEOResult uses robots_txt.exists,
sitemap.exists, canonical.has_canonical. Fixed all field access paths.
Extracted DB save logic to _save_audit_to_db() for clarity.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

blueprints/admin/routes_portal_seo.py

@@ -14,7 +14,6 @@ import time as time_module
 from datetime import datetime, date
 from decimal import Decimal
 
-import requests
 from flask import (
     abort, render_template, request, redirect, url_for,
     flash, Response, stream_with_context
@@ -61,6 +60,93 @@ def _get_auditor():
     return SEOAuditor()
 
 
+def _save_audit_to_db(result, notes, user_email):
+    """Extract fields from SEOAuditor result dict and save to database.
+
+    Result structure (after to_dict()):
+    - result['onpage']['meta_tags']['title'] / ['description']
+    - result['onpage']['images']['total_images'] / ['images_without_alt']
+    - result['onpage']['structured_data']['has_structured_data']
+    - result['onpage']['open_graph']['og_title'] (non-None = has OG)
+    - result['technical']['robots_txt']['exists']
+    - result['technical']['sitemap']['exists']
+    - result['technical']['canonical']['has_canonical']
+    - result['pagespeed']['scores']['performance'] etc.
+    - result['pagespeed']['core_web_vitals']['lcp_ms'] etc.
+    """
+    onpage = result.get('onpage', {})
+    tech = result.get('technical', {})
+    ps = result.get('pagespeed', {})
+    scores = result.get('scores', {})
+
+    # On-page nested
+    meta_tags = onpage.get('meta_tags', {})
+    images = onpage.get('images', {})
+    structured = onpage.get('structured_data', {})
+    og = onpage.get('open_graph', {})
+
+    # Technical nested
+    robots = tech.get('robots_txt', {})
+    sitemap = tech.get('sitemap', {})
+    canonical = tech.get('canonical', {})
+    indexability = tech.get('indexability', {})
+
+    # PageSpeed nested
+    cwv = ps.get('core_web_vitals', {})
+
+    # Security headers (from technical redirect chain response headers)
+    # These are stored separately in audit_company result
+    sec = result.get('security_headers', {})
+
+    db = SessionLocal()
+    try:
+        audit = PortalSEOAudit(
+            audited_at=datetime.now(),
+            url=PORTAL_URL,
+            # PageSpeed scores
+            pagespeed_performance=scores.get('pagespeed_performance'),
+            pagespeed_seo=scores.get('pagespeed_seo'),
+            pagespeed_accessibility=scores.get('pagespeed_accessibility'),
+            pagespeed_best_practices=scores.get('pagespeed_best_practices'),
+            # Core Web Vitals
+            lcp_ms=cwv.get('lcp_ms'),
+            fcp_ms=cwv.get('fcp_ms'),
+            cls=cwv.get('cls'),
+            tbt_ms=cwv.get('tbt_ms'),
+            speed_index_ms=cwv.get('speed_index_ms'),
+            # On-page checks
+            has_meta_title=bool(meta_tags.get('title')),
+            has_meta_description=bool(meta_tags.get('description')),
+            has_canonical=canonical.get('has_canonical', False),
+            has_robots_txt=robots.get('exists', False),
+            has_sitemap=sitemap.get('exists', False),
+            has_structured_data=structured.get('has_structured_data', False),
+            has_og_tags=bool(og.get('og_title')),
+            has_ssl=result.get('final_url', '').startswith('https'),
+            is_mobile_friendly=indexability.get('is_indexable'),
+            # Security headers
+            has_hsts=sec.get('has_hsts'),
+            has_csp=sec.get('has_csp'),
+            has_x_frame=sec.get('has_x_frame_options'),
+            has_x_content_type=sec.get('has_x_content_type'),
+            # Content metrics
+            image_count=images.get('total_images'),
+            images_without_alt=images.get('images_without_alt'),
+            # Full data
+            full_results=_make_json_safe(result),
+            notes=notes,
+            created_by=user_email
+        )
+        db.add(audit)
+        db.commit()
+        return audit.id
+    except Exception:
+        db.rollback()
+        raise
+    finally:
+        db.close()
+
+
 @bp.route('/portal-seo')
 @login_required
 def admin_portal_seo():
@@ -85,32 +171,25 @@ def admin_portal_seo():
 @bp.route('/portal-seo/run/stream')
 @login_required
 def admin_portal_seo_run_stream():
-    """SSE endpoint for streaming portal SEO audit progress step by step."""
+    """SSE endpoint for streaming portal SEO audit progress step by step.
+
+    Runs the same audit pipeline as SEOAuditor.audit_company() but yields
+    progress events between each step.
+    """
     if not is_audit_owner():
         abort(404)
 
     notes = request.args.get('notes', '')
     user_email = current_user.email
 
-    STEPS = [
-        (1, 'Inicjalizacja audytora SEO...'),
-        (2, 'Pobieranie strony nordabiznes.pl...'),
-        (3, 'Analiza on-page SEO...'),
-        (4, 'Sprawdzanie techniczne (robots.txt, sitemap, SSL)...'),
-        (5, 'PageSpeed Insights API...'),
-        (6, 'Analiza Local SEO...'),
-        (7, 'Sprawdzanie cytowań (citations)...'),
-        (8, 'Sprawdzanie aktualności treści...'),
-        (9, 'Zapisywanie wyników...'),
-    ]
-    TOTAL = len(STEPS)
+    TOTAL = 9
 
     def generate():
         try:
             # Step 1: Init auditor
             yield _sse_event({
                 'step': 1, 'total': TOTAL,
-                'message': STEPS[0][1], 'status': 'running'
+                'message': 'Inicjalizacja audytora SEO...', 'status': 'running'
             })
 
             auditor = _get_auditor()
@@ -133,6 +212,7 @@ def admin_portal_seo_run_stream():
                 'audit_date': datetime.now(),
                 'website_url': PORTAL_URL,
                 'errors': [],
+                'scores': {},
             }
             html_content = None
             final_url = PORTAL_URL
@@ -140,7 +220,7 @@ def admin_portal_seo_run_stream():
             # Step 2: Fetch page
             yield _sse_event({
                 'step': 2, 'total': TOTAL,
-                'message': STEPS[1][1], 'status': 'running'
+                'message': 'Pobieranie strony nordabiznes.pl...', 'status': 'running'
             })
 
             try:
@@ -150,49 +230,65 @@ def admin_portal_seo_run_stream():
                 )
                 load_ms = int((time_module.time() - start) * 1000)
                 final_url = resp.url
+                result['http_status'] = resp.status_code
+                result['load_time_ms'] = load_ms
+                result['final_url'] = final_url
 
                 if resp.status_code == 200:
                     if resp.encoding and resp.encoding.lower() == 'iso-8859-1':
                         resp.encoding = resp.apparent_encoding
                     html_content = resp.text
-                    result['http_status'] = 200
-                    result['load_time_ms'] = load_ms
-                    result['final_url'] = final_url
                     yield _sse_event({
                         'step': 2, 'total': TOTAL,
-                        'message': f'Strona pobrana ({load_ms}ms)',
+                        'message': f'Strona pobrana ({load_ms}ms, {len(html_content)//1024}KB)',
                         'status': 'done'
                     })
                 else:
                     result['errors'].append(f'HTTP {resp.status_code}')
                     yield _sse_event({
                         'step': 2, 'total': TOTAL,
-                        'message': f'HTTP {resp.status_code}',
-                        'status': 'warning'
+                        'message': f'HTTP {resp.status_code}', 'status': 'warning'
                     })
             except Exception as e:
                 result['errors'].append(str(e)[:100])
                 yield _sse_event({
                     'step': 2, 'total': TOTAL,
-                    'message': f'Błąd pobierania: {str(e)[:80]}',
-                    'status': 'error'
+                    'message': f'Błąd: {str(e)[:80]}', 'status': 'error'
                 })
 
             # Step 3: On-page analysis
             yield _sse_event({
                 'step': 3, 'total': TOTAL,
-                'message': STEPS[2][1], 'status': 'running'
+                'message': 'Analiza on-page SEO (meta tagi, nagłówki, obrazy)...',
+                'status': 'running'
             })
 
             if html_content:
                 try:
-                    onpage = auditor.onpage_analyzer.analyze_html(
+                    onpage_obj = auditor.onpage_analyzer.analyze_html(
                         html_content, base_url=final_url
                     )
-                    result['onpage'] = onpage.to_dict()
+                    onpage_dict = onpage_obj.to_dict()
+                    result['onpage'] = onpage_dict
+
+                    # Build summary
+                    mt = onpage_dict.get('meta_tags', {})
+                    imgs = onpage_dict.get('images', {})
+                    sd = onpage_dict.get('structured_data', {})
+                    title = mt.get('title', '')
+                    title_short = (title[:35] + '...') if title and len(title) > 35 else title
+                    parts = []
+                    if title:
+                        parts.append(f'title="{title_short}"')
+                    parts.append(f'{imgs.get("total_images", 0)} obrazów')
+                    if imgs.get('images_without_alt'):
+                        parts.append(f'{imgs["images_without_alt"]} bez alt')
+                    if sd.get('has_structured_data'):
+                        parts.append('Schema.org')
+
                     yield _sse_event({
                         'step': 3, 'total': TOTAL,
-                        'message': f'On-page: title="{onpage.meta_title[:40]}..."' if onpage.meta_title else 'On-page: brak meta title',
+                        'message': f'On-page: {", ".join(parts)}',
                         'status': 'done'
                     })
                 except Exception as e:
@@ -205,30 +301,45 @@ def admin_portal_seo_run_stream():
             else:
                 yield _sse_event({
                     'step': 3, 'total': TOTAL,
-                    'message': 'Pominięto (brak HTML)',
-                    'status': 'skipped'
+                    'message': 'Pominięto (brak HTML)', 'status': 'skipped'
                 })
 
             # Step 4: Technical SEO
             yield _sse_event({
                 'step': 4, 'total': TOTAL,
-                'message': STEPS[3][1], 'status': 'running'
+                'message': 'Sprawdzanie techniczne (robots.txt, sitemap, SSL, canonical)...',
+                'status': 'running'
             })
 
             try:
-                tech_result = auditor.technical_checker.check_url(final_url)
-                result['technical'] = tech_result.to_dict()
-                checks = []
-                td = tech_result.to_dict()
-                if td.get('has_robots_txt'):
-                    checks.append('robots.txt')
-                if td.get('has_sitemap'):
-                    checks.append('sitemap')
-                if td.get('has_ssl'):
-                    checks.append('SSL')
+                tech_obj = auditor.technical_checker.check_url(final_url)
+                tech_dict = tech_obj.to_dict()
+                result['technical'] = tech_dict
+
+                checks_ok = []
+                checks_fail = []
+                if tech_dict.get('robots_txt', {}).get('exists'):
+                    checks_ok.append('robots.txt')
+                else:
+                    checks_fail.append('robots.txt')
+                if tech_dict.get('sitemap', {}).get('exists'):
+                    checks_ok.append('sitemap')
+                else:
+                    checks_fail.append('sitemap')
+                if tech_dict.get('canonical', {}).get('has_canonical'):
+                    checks_ok.append('canonical')
+                else:
+                    checks_fail.append('canonical')
+
+                msg_parts = []
+                if checks_ok:
+                    msg_parts.append(f'OK: {", ".join(checks_ok)}')
+                if checks_fail:
+                    msg_parts.append(f'Brak: {", ".join(checks_fail)}')
+
                 yield _sse_event({
                     'step': 4, 'total': TOTAL,
-                    'message': f'Technical: {", ".join(checks) if checks else "brak kluczowych elementów"}',
+                    'message': f'Technical: {" | ".join(msg_parts)}',
                     'status': 'done'
                 })
             except Exception as e:
@@ -242,7 +353,8 @@ def admin_portal_seo_run_stream():
             # Step 5: PageSpeed Insights
             yield _sse_event({
                 'step': 5, 'total': TOTAL,
-                'message': STEPS[4][1], 'status': 'running'
+                'message': 'PageSpeed Insights API (może potrwać do 30s)...',
+                'status': 'running'
             })
 
             try:
@@ -261,14 +373,19 @@ def admin_portal_seo_run_stream():
                     }
                     yield _sse_event({
                         'step': 5, 'total': TOTAL,
-                        'message': f'PageSpeed: Perf={ps_result.scores.performance}, SEO={ps_result.scores.seo}',
+                        'message': (
+                            f'Perf={ps_result.scores.performance}, '
+                            f'SEO={ps_result.scores.seo}, '
+                            f'A11y={ps_result.scores.accessibility}, '
+                            f'BP={ps_result.scores.best_practices}'
+                        ),
                         'status': 'done'
                     })
                 else:
                     result['errors'].append('PageSpeed API quota exceeded')
                     yield _sse_event({
                         'step': 5, 'total': TOTAL,
-                        'message': 'Limit API wyczerpany — pominięto',
+                        'message': 'Limit API wyczerpany',
                         'status': 'warning'
                     })
             except Exception as e:
@@ -282,7 +399,8 @@ def admin_portal_seo_run_stream():
             # Step 6: Local SEO
             yield _sse_event({
                 'step': 6, 'total': TOTAL,
-                'message': STEPS[5][1], 'status': 'running'
+                'message': 'Analiza Local SEO (NAP, Google Maps, lokalne słowa kluczowe)...',
+                'status': 'running'
             })
 
             if html_content:
@@ -294,7 +412,7 @@ def admin_portal_seo_run_stream():
                     score = local_seo.get('local_seo_score', 0)
                     yield _sse_event({
                         'step': 6, 'total': TOTAL,
-                        'message': f'Local SEO score: {score}',
+                        'message': f'Local SEO score: {score}/100',
                         'status': 'done'
                     })
                 except Exception as e:
@@ -307,14 +425,14 @@ def admin_portal_seo_run_stream():
             else:
                 yield _sse_event({
                     'step': 6, 'total': TOTAL,
-                    'message': 'Pominięto (brak HTML)',
-                    'status': 'skipped'
+                    'message': 'Pominięto (brak HTML)', 'status': 'skipped'
                 })
 
             # Step 7: Citations
             yield _sse_event({
                 'step': 7, 'total': TOTAL,
-                'message': STEPS[6][1], 'status': 'running'
+                'message': 'Sprawdzanie cytowań w katalogach (Google, Yelp, Facebook)...',
+                'status': 'running'
             })
 
             try:
@@ -339,7 +457,8 @@ def admin_portal_seo_run_stream():
             # Step 8: Content freshness
             yield _sse_event({
                 'step': 8, 'total': TOTAL,
-                'message': STEPS[7][1], 'status': 'running'
+                'message': 'Sprawdzanie aktualności treści...',
+                'status': 'running'
             })
 
             try:
@@ -350,7 +469,7 @@ def admin_portal_seo_run_stream():
                 fscore = freshness.get('content_freshness_score', 0)
                 yield _sse_event({
                     'step': 8, 'total': TOTAL,
-                    'message': f'Aktualność treści: {fscore}',
+                    'message': f'Aktualność treści: {fscore}/100',
                     'status': 'done'
                 })
             except Exception as e:
@@ -364,54 +483,12 @@ def admin_portal_seo_run_stream():
             # Step 9: Save to DB
             yield _sse_event({
                 'step': 9, 'total': TOTAL,
-                'message': STEPS[8][1], 'status': 'running'
+                'message': 'Zapisywanie wyników do bazy danych...',
+                'status': 'running'
             })
 
-            # Extract data for DB columns
-            ps = result.get('pagespeed', {})
-            ps_scores = result.get('scores', {})
-            cwv = ps.get('core_web_vitals', {})
-            tech = result.get('technical', {})
-            onpage = result.get('onpage', {})
-            sec = tech.get('security_headers', {})
-
-            db = SessionLocal()
             try:
-                audit = PortalSEOAudit(
-                    audited_at=datetime.now(),
-                    url=PORTAL_URL,
-                    pagespeed_performance=ps_scores.get('pagespeed_performance'),
-                    pagespeed_seo=ps_scores.get('pagespeed_seo'),
-                    pagespeed_accessibility=ps_scores.get('pagespeed_accessibility'),
-                    pagespeed_best_practices=ps_scores.get('pagespeed_best_practices'),
-                    lcp_ms=cwv.get('lcp_ms'),
-                    fcp_ms=cwv.get('fcp_ms'),
-                    cls=cwv.get('cls'),
-                    tbt_ms=cwv.get('tbt_ms'),
-                    speed_index_ms=cwv.get('speed_index_ms'),
-                    has_meta_title=bool(onpage.get('meta_title')),
-                    has_meta_description=bool(onpage.get('meta_description')),
-                    has_canonical=tech.get('has_canonical'),
-                    has_robots_txt=tech.get('has_robots_txt'),
-                    has_sitemap=tech.get('has_sitemap'),
-                    has_structured_data=onpage.get('has_structured_data'),
-                    has_og_tags=onpage.get('has_og_tags'),
-                    has_ssl=tech.get('has_ssl'),
-                    is_mobile_friendly=tech.get('is_mobile_friendly'),
-                    has_hsts=sec.get('has_hsts'),
-                    has_csp=sec.get('has_csp'),
-                    has_x_frame=sec.get('has_x_frame_options'),
-                    has_x_content_type=sec.get('has_x_content_type'),
-                    page_size_bytes=onpage.get('page_size_bytes'),
-                    image_count=onpage.get('total_images'),
-                    images_without_alt=onpage.get('images_without_alt'),
-                    full_results=_make_json_safe(result),
-                    notes=notes,
-                    created_by=user_email
-                )
-                db.add(audit)
-                db.commit()
-                audit_id = audit.id
+                audit_id = _save_audit_to_db(result, notes, user_email)
 
                 yield _sse_event({
                     'step': 9, 'total': TOTAL,
@@ -419,17 +496,15 @@ def admin_portal_seo_run_stream():
                     'status': 'done'
                 })
 
-                # Send complete event
                 yield _sse_event({
                     'status': 'complete',
                     'audit_id': audit_id,
-                    'performance': ps_scores.get('pagespeed_performance'),
-                    'seo': ps_scores.get('pagespeed_seo'),
+                    'performance': result.get('scores', {}).get('pagespeed_performance'),
+                    'seo': result.get('scores', {}).get('pagespeed_seo'),
                     'errors': result.get('errors', []),
                 })
 
             except Exception as e:
-                db.rollback()
                 logger.error(f'Portal SEO save failed: {e}', exc_info=True)
                 yield _sse_event({
                     'step': 9, 'total': TOTAL,
@@ -437,8 +512,6 @@ def admin_portal_seo_run_stream():
                     'status': 'error'
                 })
                 yield _sse_event({'status': 'error', 'message': str(e)[:200]})
-            finally:
-                db.close()
 
         except Exception as e:
             logger.error(f'Portal SEO audit stream failed: {e}', exc_info=True)