From 601bd995596f810f9b1d31e1bbf693c95bf8f38e Mon Sep 17 00:00:00 2001
From: Maciej Pienczyn <maciej.pienczyn@inpi.pl>
Date: Sat, 21 Feb 2026 10:24:55 +0100
Subject: [PATCH] feat: remember rejected candidates, skip in future bulk
 discovery

- Bulk discovery skips companies with any candidate (including rejected)
- Single discovery skips URLs from previously rejected domains
- Dashboard shows list of companies rejected by admin with note
  that they won't be re-searched in bulk mode

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 blueprints/admin/routes_data_quality.py      | 24 ++++++++++++++++++++
 blueprints/admin/routes_website_discovery.py |  4 ++--
 services/website_discovery_service.py        |  9 +++++++-
 templates/admin/data_quality_dashboard.html  | 14 ++++++++++++
 4 files changed, 48 insertions(+), 3 deletions(-)

diff --git a/blueprints/admin/routes_data_quality.py b/blueprints/admin/routes_data_quality.py
index 76a5307..aba3691 100644
--- a/blueprints/admin/routes_data_quality.py
+++ b/blueprints/admin/routes_data_quality.py
@@ -260,6 +260,29 @@ def admin_data_quality():
                 ),
             })
 
+        # Companies with rejected candidates (already reviewed)
+        rejected_company_ids = set(
+            r[0] for r in db.query(WebsiteDiscoveryCandidate.company_id).filter(
+                WebsiteDiscoveryCandidate.status == 'rejected'
+            ).distinct().all()
+        )
+        # Exclude companies that also have pending/accepted candidates
+        active_candidate_ids = set(
+            r[0] for r in db.query(WebsiteDiscoveryCandidate.company_id).filter(
+                WebsiteDiscoveryCandidate.status.in_(['pending', 'accepted'])
+            ).distinct().all()
+        )
+        only_rejected_ids = rejected_company_ids - active_candidate_ids
+        rejected_companies = []
+        for cid in only_rejected_ids:
+            comp = company_map.get(cid)
+            if comp and not comp.website:
+                rejected_companies.append({
+                    'company_name': comp.name,
+                    'company_id': cid,
+                })
+        rejected_companies.sort(key=lambda x: x['company_name'])
+
         # Count companies without website
         companies_without_website = sum(1 for c in companies_table if not c['website'])
 
@@ -273,6 +296,7 @@ def admin_data_quality():
             companies_table=companies_table,
             available_data=available_data,
             discovery_data=discovery_data,
+            rejected_companies=rejected_companies,
             companies_without_website=companies_without_website,
             now=now,
         )
diff --git a/blueprints/admin/routes_website_discovery.py b/blueprints/admin/routes_website_discovery.py
index c06c0a0..84b5d24 100644
--- a/blueprints/admin/routes_website_discovery.py
+++ b/blueprints/admin/routes_website_discovery.py
@@ -80,10 +80,10 @@ def discover_websites_bulk():
         _save_job(job_id, job)
         db = SessionLocal()
         try:
-            # Skip companies that already have a pending/accepted candidate
+            # Skip companies that already have any candidate (pending/accepted/rejected)
             already_have = set(
                 r[0] for r in db.query(WebsiteDiscoveryCandidate.company_id).filter(
-                    WebsiteDiscoveryCandidate.status.in_(['pending', 'accepted'])
+                    WebsiteDiscoveryCandidate.status.in_(['pending', 'accepted', 'rejected'])
                 ).distinct().all()
             )
 
diff --git a/services/website_discovery_service.py b/services/website_discovery_service.py
index f512163..f05aabc 100644
--- a/services/website_discovery_service.py
+++ b/services/website_discovery_service.py
@@ -288,12 +288,19 @@ class WebsiteDiscoveryService:
                 if domain.startswith('www.'):
                     domain = domain[4:]
 
-                # Check for existing candidate with this URL
+                # Check for existing candidate (exact URL or same domain rejected)
                 existing = db.query(WebsiteDiscoveryCandidate).filter_by(
                     company_id=company.id, candidate_url=url
                 ).first()
                 if existing:
                     continue
+                rejected_domain = db.query(WebsiteDiscoveryCandidate).filter(
+                    WebsiteDiscoveryCandidate.company_id == company.id,
+                    WebsiteDiscoveryCandidate.candidate_domain == domain,
+                    WebsiteDiscoveryCandidate.status == 'rejected',
+                ).first()
+                if rejected_domain:
+                    continue
 
                 # Fetch root + common subpages for verification data
                 all_text = ''
diff --git a/templates/admin/data_quality_dashboard.html b/templates/admin/data_quality_dashboard.html
index 4dbcadc..7273648 100644
--- a/templates/admin/data_quality_dashboard.html
+++ b/templates/admin/data_quality_dashboard.html
@@ -750,6 +750,20 @@
     {% else %}
     <p style="color: var(--text-secondary); font-size: var(--font-size-sm);">Brak kandydatów. Kliknij "Szukaj WWW" aby uruchomić wyszukiwanie.</p>
     {% endif %}
+
+    {% if rejected_companies %}
+    <div style="margin-top: var(--spacing-lg); padding: var(--spacing-md); background: #fef2f2; border: 1px solid #fecaca; border-radius: var(--radius);">
+        <span style="font-size: var(--font-size-sm); color: #991b1b; font-weight: 500;">
+            Odrzucone przez admina ({{ rejected_companies|length }}):
+        </span>
+        <span style="font-size: var(--font-size-xs); color: #b91c1c;">
+            {% for rc in rejected_companies %}{{ rc.company_name }}{% if not loop.last %}, {% endif %}{% endfor %}
+        </span>
+        <span style="font-size: var(--font-size-xs); color: #991b1b; display: block; margin-top: 4px;">
+            Te firmy nie będą ponownie wyszukiwane w trybie zbiorczym.
+        </span>
+    </div>
+    {% endif %}
 </div>
 
 <!-- Bulk Discovery Modal -->