making webwatcher better hopefully

.local/bin/webwatcher.py

@@ -93,15 +93,16 @@ def main():
         sys.exit(3)
 
     try:
-        matches = re.findall(args.pattern, html, flags)
+        # Capture surrounding context so multiple "Beherit" items are treated separately
-        # If the regex has groups, re.findall returns tuples; normalize to strings
+        context_window = 120  # number of characters around each match to include
-        norm = []
+        unique = set()
-        for m in matches:
+
-            if isinstance(m, tuple):
+        for match in re.finditer(args.pattern, html, flags):
-                norm.append(" | ".join(m))
+            start = max(match.start() - context_window, 0)
-            else:
+            end = min(match.end() + context_window, len(html))
-                norm.append(m)
+            snippet = html[start:end].replace("\n", " ")
-        unique = set(norm)
+            unique.add(snippet.strip())
+
     except re.error as e:
         log_line(log_file, f"ERROR regex '{args.pattern}': {e}")
         sys.exit(4)