Fix: Deep fix for Action.com availability false positives. Advanced title normalization and clarified unavailability detection logic. Optimized ActionParser.

app/services/parsers/action_parser.py

@@ -1,9 +1,11 @@
 from bs4 import BeautifulSoup
 from app.services.parsers.base_parser import BaseParser, ProductResult
 import logging
+import re
 
 logger = logging.getLogger(__name__)
 
+
 class ActionParser(BaseParser):
     def __init__(self):
         super().__init__("action.com", "https://www.action.com")
@@ -11,55 +13,62 @@ class ActionParser(BaseParser):
     def parse_search_results(self, html: str, query: str, search_url: str) -> list[ProductResult]:
         soup = BeautifulSoup(html, "html.parser")
         results = []
-        
+
         # Action products
         # Config: a.group[href^='/fr-fr/p/']
-        
+
         links = soup.select("a[href^='/fr-fr/p/']")
-        
+
         seen_urls = set()
-        
+
         for link in links:
             try:
-                href = link.get('href')
+                href = link.get("href")
                 if not href or href in seen_urls:
                     continue
                 seen_urls.add(href)
-                
+
                 url = self.make_absolute_url(href)
-                
+
                 # Action usually has the card content inside the link
                 title = None
                 title_el = link.select_one("[class*='title'], h3, h4")
                 if title_el:
                     title = title_el.get_text(strip=True)
                 else:
-                    title = link.get_text(strip=True)
-                
+                    # Fallback to specific spans or remove price info from text
+                    # Often Action cards have a .title span
+                    title = link.get_text(" ", strip=True)
+                    # Remove price-like patterns from title if we had to get whole link text
+                    title = re.sub(r"\d+,\d+\s*€.*$", "", title)
+                    title = re.sub(r"\d+\s*€.*$", "", title)
+
                 if not title:
                     continue
 
                 img_url = self.extract_image_url(link)
-                
+
                 price = None
                 price_el = link.select_one("[class*='price']")
                 if price_el:
                     price = self.parse_price_text(price_el.get_text())
-                
-                results.append(ProductResult(
-                    title=title,
-                    url=url,
-                    source="Action",
-                    price=price,
-                    currency="EUR",
-                    in_stock=True,
-                    image_url=img_url,
-                    snippet=f"Product from Action"
-                ))
-                
+
+                results.append(
+                    ProductResult(
+                        title=title,
+                        url=url,
+                        source="Action",
+                        price=price,
+                        currency="EUR",
+                        in_stock=True,
+                        image_url=img_url,
+                        snippet=f"Product from Action",
+                    )
+                )
+
             except Exception as e:
                 logger.error(f"Error parsing Action product: {e}")
                 continue
-                
+
         logger.info(f"ActionParser found {len(results)} results")
         return results

app/services/scheduler_service.py

@@ -54,10 +54,17 @@ def _normalize_title(title: str) -> str:
     title = re.sub(r"([a-zA-Z])(\d)", r"\1 \2", title)
     title = re.sub(r"(\d)([a-zA-Z])", r"\1 \2", title)
 
-    # Remove extra spaces and punctuation that might differ between sites
+    # Remove non-word characters (punctuation) early to simplify regex
     title = re.sub(r"[^\w\s]", " ", title)
     title = " ".join(title.split())
 
+    # Remove technical metadata at the end (Action specific noise)
+    # Match various patterns like "0 09 pce 342" or "100 pièces" etc.
+    title = re.sub(r"\s+\d+\s+\d+\s+pce.*$", "", title)
+    title = re.sub(r"\s+\d+\s+pièces.*$", "", title)
+    title = re.sub(r"\s+\d+\s+€.*$", "", title)
+    title = re.sub(r"\s+\d+\s+\d+\s+€.*$", "", title)
+
     return title.strip()
 
 
@@ -363,6 +370,28 @@ async def process_item_check(item_id: int):
                     f"Stored: '{item_data['name']}', Page title: '{page_title}'"
                 )
 
+                # For Action.com, if we haven't confirmed unavailability above,
+                # don't mark as unavailable yet, just mark as potential replacement error
+                # unless the title is almost completely different
+                if is_action:
+                    # Check if at least some significant words match
+                    norm_stored = _normalize_title(item_data["name"])
+                    norm_page = _normalize_title(page_title)
+                    stored_words = set(norm_stored.split())
+                    page_words = set(norm_page.split())
+                    intersection = stored_words & page_words
+                    if len(intersection) >= 2 or (len(stored_words) <= 3 and len(intersection) >= 1):
+                        logger.warning(
+                            f"Borderline title match for Action.com item {item_id}. Keeping as available but marking error."
+                        )
+                        await loop.run_in_executor(
+                            None,
+                            _update_db_error,
+                            item_id,
+                            f"Titre différent mais produit probablement correct : {page_title}",
+                        )
+                        return
+
             await loop.run_in_executor(
                 None,
                 _update_db_product_unavailable,