Debug: Check sentence existence before HTML escaping

chatgpt.py

@@ -1213,6 +1213,9 @@ with gr.Blocks() as app:
                 if not base:
                     return gr.update(value="<pre>No input data</pre>")
                 
+                print(f"[Highlight] Input text length: {len(base)} chars")
+                print(f"[Highlight] Input text first 100 chars: {base[:100]}")
+                
                 # Escape HTML
                 html_text = base.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;')
                 
@@ -1234,6 +1237,16 @@ with gr.Blocks() as app:
                             print(f"[Highlight]   Match #{idx+1}: Empty sentence, skipping")
                             continue
                         
+                        # Check if sentence exists in original text BEFORE escaping
+                        if sentence not in base:
+                            print(f"[Highlight]   Match #{idx+1}: Sentence not in original text")
+                            print(f"[Highlight]     Looking for: {sentence[:80]}")
+                            # Try to find partial match
+                            sentence_parts = sentence.split('\n')[0]  # Try first line
+                            if sentence_parts in base:
+                                print(f"[Highlight]     Found partial match: {sentence_parts[:60]}")
+                            continue
+                        
                         # Escape HTML in sentence
                         escaped_sentence = sentence.replace('&', '&amp;').replace('<', '&lt;').replace('>', '&gt;')