Update parser.py

parser.py

@@ -1,33 +1,65 @@
+import re
 from bs4 import BeautifulSoup
 
 def parse_html(html: str, final_url: str, word: str) -> str:
-    soup = BeautifulSoup(html, 'lxml')
-    full_text = soup.get_text(separator=' ', strip=True).lower()
-    is_direct = '/username/' in final_url and '?query=' not in final_url
+    """
+    Highly resilient HTML payload engine optimized for processing
+    both direct page DOM structures and AJAX search response grids safely.
+    """
+    if not html or not html.strip():
+        return 'AVAILABLE'
 
-    if is_direct:
+    soup = BeautifulSoup(html, 'html.parser')
+    
+    # Pre-process raw text buffers cleanly
+    full_text = " ".join(soup.get_text(separator=' ', strip=True).split()).lower()
+    word_clean = word.strip().replace("@", "").lower()
+    
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # APPROACH 1: DIRECT PROFILE RESOLUTION SIGNATURES
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # Safely checks for structural components if the payload is a full profile DOM
+    if '/username/' in final_url.lower() and '?query=' not in final_url.lower():
         status_tag = soup.find(class_='tm-section-header-status')
         if status_tag:
             t = status_tag.get_text(strip=True).lower()
             if 'sold' in t: return 'SOLD'
             if 'taken' in t: return 'TAKEN'
             if 'available' in t: return 'AVAILABLE'
-            if 'on auction' in t: return 'ON_AUCTION'
-            if 'for sale' in t: return 'FOR_SALE'
-        
-        if 'is already taken' in full_text:
+            if 'on auction' in t or 'bidding' in t: return 'ON_AUCTION'
+            if 'sale' in t or 'purchase' in t: return 'FOR_SALE'
+
+        if 'is already taken' in full_text or 'make an offer' in full_text:
             return 'TAKEN'
 
-    rows = soup.find_all('tr')
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # APPROACH 2: AJAX DATA GRID / TABLE ROW INTERCEPTIONS
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # Iterate through all functional rows or element dividers cleanly
+    rows = soup.find_all(['tr', 'div', 'a'], class_=re.compile(re.escape('row') + r'|tm-table-row|tm-row', re.IGNORECASE)) or soup.find_all('tr')
+    
     for row in rows:
-        text = row.get_text(separator=' ', strip=True).lower()
-        if f"@{word}" in text.split():
-            if 'unavailable' in text and 'unknown' in text: return 'UNAVAILABLE'
-            if 'on auction' in text: return 'ON_AUCTION'
-            if 'for sale' in text and 'not for sale' not in text: return 'FOR_SALE'
-            if 'available' in text: return 'AVAILABLE'
-            if 'sold' in text: return 'SOLD'
-            if 'taken' in text: return 'TAKEN'
-            if 'unavailable' in text: return 'UNAVAILABLE'
+        row_text = " ".join(row.get_text(separator=' ', strip=True).split()).lower()
+        
+        # Performance check mapping boundaries cleanly using exact string containment
+        if f"@{word_clean}" in row_text or word_clean in row_text:
+            # Check explicit validation statuses matching your matrix boundaries
+            if 'on auction' in row_text or 'bidding' in row_text: return 'ON_AUCTION'
+            if 'for sale' in row_text and 'not for sale' not in row_text: return 'FOR_SALE'
+            if 'available' in row_text: return 'AVAILABLE'
+            if 'sold' in row_text: return 'SOLD'
+            if 'taken' in row_text or 'make an offer' in row_text: return 'TAKEN'
+            if 'unavailable' in row_text: return 'UNAVAILABLE'
+
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # APPROACH 3: GLOBAL FAILSAFE PATTERN RESOLUTION
+    # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
+    # Ultimate boundary backup scan across text layers if explicit row bindings fail
+    if "on auction" in full_text: return 'ON_AUCTION'
+    if "for sale" in full_text and "not for sale" not in full_text: return 'FOR_SALE'
+    if "sold" in full_text: return 'SOLD'
+    if "taken" in full_text or "make an offer" in full_text: return 'TAKEN'
+    if "unavailable" in full_text: return 'UNAVAILABLE'
+    if "available" in full_text: return 'AVAILABLE'
 
     return 'UNCERTAIN'