Merge pull request #181 from R0m1k3/claude/rebuild-search-amazon-01LBnzN1wfNMe1EG9HcpMCdK

fix: Improve site mapping and relax GenericParser filtering

app/services/improved_search_service.py

@@ -620,14 +620,45 @@ async def search_products(
         results=[],
     )
 
-    # 2. Map DB sites to Config keys
+    # 2. Map DB sites to Config keys with improved matching
     site_keys = []
     for site in active_sites:
+        matched = False
+
+        # Normalize domain for comparison (remove www., e-, etc.)
+        normalized_domain = site.domain.lower().replace("www.", "").strip()
+
         for key in SITE_CONFIGS.keys():
-            if key in site.domain or site.domain in key:
+            normalized_key = key.lower().replace("www.", "").strip()
+
+            # Try multiple matching strategies:
+            # 1. Exact match
+            if normalized_domain == normalized_key:
                 site_keys.append(key)
+                matched = True
+                logger.debug(f"✓ Matched {site.domain} → {key} (exact)")
                 break
 
+            # 2. Key contains domain or domain contains key
+            if normalized_key in normalized_domain or normalized_domain in normalized_key:
+                site_keys.append(key)
+                matched = True
+                logger.debug(f"✓ Matched {site.domain} → {key} (contains)")
+                break
+
+            # 3. Remove prefixes like "e-" and try again
+            domain_without_prefix = normalized_domain.replace("e-", "").replace("la-", "")
+            key_without_prefix = normalized_key.replace("e-", "").replace("la-", "")
+
+            if domain_without_prefix == key_without_prefix:
+                site_keys.append(key)
+                matched = True
+                logger.debug(f"✓ Matched {site.domain} → {key} (without prefix)")
+                break
+
+        if not matched:
+            logger.warning(f"⚠️ No config found for site: {site.domain} (id={site.id})")
+
     # 3. Execute searches and stream results
     generators = [ImprovedSearchService.search_site_generator(key, query) for key in site_keys]
 

app/services/parsers/generic_parser.py

@@ -94,11 +94,30 @@ class GenericParser(BaseParser):
                         title = img.get("alt", "")
 
                 if not title or len(title) < 3:
+                    self.logger.debug(f"Skipped: empty or too short title")
                     continue
 
-                # Filter by query (relaxed: at least one word match)
-                if not self.filter_by_query(title, query, strict=False):
-                    continue
+                self.logger.debug(f"Extracted title: '{title[:60]}'...")
+
+                # RELAXED FILTERING: Only filter out if title is completely unrelated
+                # Accept product if:
+                # 1. Query is empty/short
+                # 2. At least one query word is in title (case-insensitive)
+                # 3. Title length is reasonable (not just numbers/symbols)
+                if query and len(query) > 2:
+                    # Very permissive: keep if ANY query word matches OR title is substantial
+                    query_words = [w.lower() for w in query.split() if len(w) > 2]
+                    title_lower = title.lower()
+
+                    if query_words:
+                        # Check if at least one word matches
+                        has_match = any(word in title_lower for word in query_words)
+
+                        # If no match and title is very short, skip
+                        # But if title is substantial (>15 chars), keep it anyway (might be relevant)
+                        if not has_match and len(title) < 15:
+                            self.logger.debug(f"Filtered: '{title[:40]}' - no query match and too short")
+                            continue
 
                 # Extract image using configured selector
                 image_url = None