Validazione

src/vtt_utils.py

@@ -95,6 +95,49 @@ def validate_vtt(vtt_content: str) -> Tuple[str, str]:
                     "warning",
                 )
 
+        # Check for punctuation followed by lowercase
+        last_char = None
+        last_line_num = None
+
+        for i, line in enumerate(lines):
+            if "-->" not in line:
+                continue
+
+            # Get text lines for this cue
+            j = i + 1
+            while j < len(lines):
+                content_line = lines[j]
+                if "-->" in content_line:
+                    break
+                if content_line.strip() == "":
+                    break
+
+                # Process text line
+                # Remove speaker tag for validation
+                clean_text = re.sub(r"<v\s+[^>]+>", "", content_line).strip()
+
+                if clean_text:
+                    # Check internal to the line
+                    match = re.search(r"([.!?])\s+([a-z])", clean_text)
+                    if match:
+                        return (
+                            f"🟡 Warning: Punctuation followed by lowercase at line {j + 1}",
+                            "warning",
+                        )
+
+                    # Check across boundary
+                    if last_char and last_char in ".!?":
+                        if clean_text[0].islower():
+                            return (
+                                f"🟡 Warning: Punctuation followed by lowercase across lines {last_line_num} and {j + 1}",
+                                "warning",
+                            )
+
+                    last_char = clean_text[-1]
+                    last_line_num = j + 1
+
+                j += 1
+
         return "🟢 Valid", "success"
     except Exception as e:
         return f"🔴 Validation error: {str(e)}", "error"