fix: Strict adult candidate validation preventing search engine query relaxation to non-person stock photos

server.py

@@ -423,12 +423,57 @@ def scrape_babepedia_images(name, limit=4):
         pass
     return results
 
+def clean_performer_search_name(raw_name):
+    cleaned = re.sub(r'\s*[-–—]\s*.*$', '', raw_name or '')
+    cleaned = re.sub(r'\(.*?\)', '', cleaned)
+    cleaned = re.sub(r'[\(\)\[\]\{\}]', '', cleaned)
+    cleaned = re.sub(r'[\s\.\-_]+', ' ', cleaned).strip()
+    return cleaned
+
+def is_valid_adult_candidate(img_url, title, desc, performer_name):
+    clean_name = clean_performer_search_name(performer_name).lower()
+    name_tokens = [tok for tok in clean_name.split() if len(tok) > 2]
+    if not name_tokens:
+        return False
+
+    url_lower = (img_url or '').lower()
+    text_lower = f"{img_url} {title} {desc}".lower()
+
+    # 1. At least one significant token of the performer's name MUST appear in the URL, title or description
+    if not any(tok in text_lower for tok in name_tokens):
+        return False
+
+    # 2. Blacklist non-person, nature, stock photo, and unrelated domains
+    blacklist_domains = [
+        'pxhere.com', 'pixabay.com', 'freepik.com', 'alamy.com', 'wikipedia.org',
+        'wikimedia.org', 'infoescola.com', 'ufrgs.br', 'todojujuy.com', 'gettyimages.com',
+        'shutterstock.com', 'tripadvisor.com', 'etsy.com', 'indiamike.com', 'toolstop.co.uk',
+        'homedepot.com', 'youtube.com', 'disney', 'marvel', 'cartoon', 'anime', 'uhrcenter.de'
+    ]
+    if any(b in url_lower for b in blacklist_domains):
+        return False
+
+    # 3. Blacklist non-human junk keywords (frogs, tools, shrines, boats, etc.)
+    junk_keywords = [
+        'sapo', 'frog', 'toad', 'shrine', 'temple', 'power-tool', 'drill', 'tractor',
+        'boat', 'yacht', 'car-parts', 'engine', 'railroad', 'railway', 'hardware',
+        'wildlife', 'amphibian', 'reptile', 'jewelry', 'schmuck', 'panties', 'costume'
+    ]
+    if any(j in text_lower for j in junk_keywords):
+        return False
+
+    return True
+
 def search_performer_candidate_images(name, aliases=None, limit=6):
+    clean_name = clean_performer_search_name(name)
+    if not clean_name or len(clean_name) < 3:
+        return []
+
     results = []
     seen = set()
 
     # 1. Primary: Direct Babepedia Adult Database Scrape
-    babe_imgs = scrape_babepedia_images(name, limit=4)
+    babe_imgs = scrape_babepedia_images(clean_name, limit=4)
     for it in babe_imgs:
         if it['url'] not in seen:
             seen.add(it['url'])
@@ -439,18 +484,19 @@ def search_performer_candidate_images(name, aliases=None, limit=6):
 
     # 2. Secondary: Adult Database Targeted Queries on Bing
     queries = [
-        f'"{name}" site:babepedia.com',
+        f'"{clean_name}" site:babepedia.com',
-        f'"{name}" site:freeones.com',
+        f'"{clean_name}" site:freeones.com',
-        f'"{name}" site:iafd.com',
+        f'"{clean_name}" site:iafd.com',
-        f'"{name}" site:boobpedia.com',
+        f'"{clean_name}" site:boobpedia.com',
-        f'"{name}" adult star headshot portrait',
+        f'"{clean_name}" adult star portrait',
-        f'"{name}" adult photoshoot portrait'
+        f'"{clean_name}" adult model photoshoot'
     ]
     if aliases and isinstance(aliases, list):
         for a in aliases[:2]:
-            if a and a.lower() != name.lower():
+            clean_a = clean_performer_search_name(a)
-                queries.append(f'"{a}" site:babepedia.com')
+            if clean_a and clean_a.lower() != clean_name.lower():
-                queries.append(f'"{a}" adult star portrait')
+                queries.append(f'"{clean_a}" site:babepedia.com')
+                queries.append(f'"{clean_a}" adult star portrait')
 
     for q in queries:
         url = f"https://www.bing.com/images/search?q={urllib.parse.quote(q)}&FORM=HDRSC2"
@@ -464,14 +510,23 @@ def search_performer_candidate_images(name, aliases=None, limit=6):
         )
         try:
             with urllib.request.urlopen(req, timeout=6) as res:
-                html = res.read().decode('utf-8', errors='ignore')
+                page_html = res.read().decode('utf-8', errors='ignore')
-                murls = re.findall(r'&quot;murl&quot;:&quot;(http[^&]+)&quot;', html)
+                raw_blocks = re.findall(r'class=\"iusc\"[^\>]*m=\"([^\"]+)\"', page_html)
-                turls = re.findall(r'&quot;turl&quot;:&quot;(http[^&]+)&quot;', html)
+                for raw_b in raw_blocks:
-                for i, u in enumerate(murls):
+                    unesc = urllib.parse.unquote(raw_b).replace('&quot;', '"')
-                    clean_u = u.replace(r'\/', '/')
+                    murl_m = re.search(r'\"murl\":\"([^\"]+)\"', unesc)
-                    if clean_u not in seen and not any(bad in clean_u.lower() for bad in ['youtube.com', 'wikimedia.org', 'wikipedia.org', 'disney', 'marvel', 'cartoon', 'anime']):
+                    turl_m = re.search(r'\"turl\":\"([^\"]+)\"', unesc)
+                    title_m = re.search(r'\"t\":\"([^\"]+)\"', unesc)
+                    desc_m = re.search(r'\"desc\":\"([^\"]+)\"', unesc)
+
+                    if murl_m:
+                        clean_u = murl_m.group(1).replace(r'\/', '/')
+                        t = turl_m.group(1).replace(r'\/', '/') if turl_m else clean_u
+                        title = title_m.group(1) if title_m else ''
+                        desc = desc_m.group(1) if desc_m else ''
+
+                        if clean_u not in seen and is_valid_adult_candidate(clean_u, title, desc, clean_name):
                             seen.add(clean_u)
-                        t = turls[i].replace(r'\/', '/') if i < len(turls) else clean_u
                             results.append({'url': clean_u, 'thumbnail': t, 'source': 'Adult Web'})
                             if len(results) >= limit:
                                 break