Update app.py

app.py

@@ -165,11 +165,14 @@ Take the following input conversation, and reformat it so that:
 - No timestamps, names, parentheses, or extra formatting, no chapter names, no special characters beside ":" 
 - No blank lines
 - Do not invent or change the content
-
+- you are not allowed to use anywhere in the text the character +#-*<>"()[]
 Example output:
 Speaker 1: Hello and welcome.
 Speaker 2: Thanks! Glad to be here.
-
+Speaker 1: ...
+Speaker 2: ...
+Speaker 1: ...
+Speaker 2: ...
 Now format the following:
 {raw_text}
 """
@@ -237,21 +240,28 @@ Now format the following:
         add_log("⚠️ No valid JSON found, creating fallback")
         return self.create_fallback_podcast(text)
         
+    def normalize_speaker_lines(self,text: str) -> str:
+        """Normalize lines to 'Speaker 1: text' format based on presence of 1 or 2 and a ':' or '-'."""
+        # Convert markdown and bracketed formats to 'Speaker X: ...'
+        text = re.sub(
+            r'(?i)^.*?([12])[^a-zA-Z0-9]*[:\-]\s*',
+            lambda m: f"Speaker {m.group(1)}: ",
+            text,
+            flags=re.MULTILINE
+        )
+        return text    
+        
     def conversation_to_json(self, text: str) -> Dict:
 
         """Convert speaker-formatted text to podcast JSON structure"""
         # Allow leading whitespace and enforce full line match
-        cleaned_lines = []
-        for line in text.splitlines():
-            if re.match(r'^\s*Speaker\s*[12]\s*:', line.strip()):
-                cleaned_lines.append(line.strip())
-
-        podcast = []
-        for line in cleaned_lines:
-            match = re.match(r'^Speaker\s*([12])\s*:\s*(.+)', line)
-            if match:
-                speaker, content = match.groups()
-                podcast.append({"speaker": int(speaker), "line": content.strip()})
+        """Convert speaker-formatted text to podcast JSON structure"""
+        text = self.normalize_speaker_lines(text)
+
+         
+        # Match strict "Speaker X: ..." lines only
+        lines = re.findall(r'^Speaker\s+([12]):\s*(.+)', text, flags=re.MULTILINE)
+        podcast = [{"speaker": int(s), "line": l.strip()} for s, l in lines]
 
         return {
             "topic": "Generated from Input",