1 files changed, 118 insertions, 0 deletions

diff --git a/highlight.py b/highlight.py
new file mode 100644
index 0000000..361a0d2
--- /dev/null
+++ b/highlight.py
@@ -0,0 +1,118 @@
+import sys
+import json
+import re
+from datetime import timedelta
+
+# ANSI escape codes for terminal colors
+class Colors:
+    GREEN = '\033[92m'  # Highlight for valid ad text
+    RED = '\033[91m'    # Highlight for invalid target ranges
+    BLUE = '\033[94m'   # Header color
+    RESET = '\033[0m'   # Reset to default color
+
+def timestamp_to_seconds(ts_str: str) -> float:
+    """Converts HH:MM:SS.mmm string to total seconds. Returns -1.0 on failure."""
+    try:
+        parts = ts_str.split(':')
+        h, m = int(parts[0]), int(parts[1])
+        s_ms = parts[2].split('.')
+        s, ms = int(s_ms[0]), int(s_ms[1])
+        return h * 3600 + m * 60 + s + ms / 1000.0
+    except (ValueError, IndexError):
+        return -1.0
+
+def visualize_entry(entry: dict):
+    """Parses and prints a single entry from the jsonl file with highlighting."""
+    entry_id = entry.get("id", "N/A")
+    full_text = entry.get("text", "")
+    target_str = entry.get("target", "")
+
+    print(f"{Colors.BLUE}{'='*80}\nID: {entry_id}{Colors.RESET}")
+
+    # --- 1. Parse the transcript from the 'text' field ---
+    transcript_lines = []
+    # Regex to find the start of the actual transcript text
+    transcript_match = re.search(r"Transcript:\n(.*)", full_text, re.DOTALL)
+    if not transcript_match:
+        print(f"{Colors.RED}Could not find transcript text in entry.{Colors.RESET}")
+        return
+
+    raw_transcript = transcript_match.group(1).strip()
+    line_pattern = re.compile(r"\[(\d{2}:\d{2}:\d{2}\.\d{3}) --> (\d{2}:\d{2}:\d{2}\.\d{3})\](.*)")
+    
+    min_transcript_time, max_transcript_time = float('inf'), float('-inf')
+
+    for line in raw_transcript.split('\n'):
+        match = line_pattern.match(line)
+        if match:
+            start_str, end_str, text = match.groups()
+            start_sec = timestamp_to_seconds(start_str)
+            end_sec = timestamp_to_seconds(end_str)
+            transcript_lines.append({"start": start_sec, "end": end_sec, "line": line})
+            min_transcript_time = min(min_transcript_time, start_sec)
+            max_transcript_time = max(max_transcript_time, end_sec)
+
+    # --- 2. Parse and validate the target ranges ---
+    valid_target_ranges = []
+    has_invalid_target = False
+    
+    if target_str:
+        for target_line in target_str.strip().split('\n'):
+            if '-' not in target_line:
+                has_invalid_target = True
+                continue
+            
+            start_target_str, end_target_str = target_line.split('-', 1)
+            start_target_sec = timestamp_to_seconds(start_target_str.strip())
+            end_target_sec = timestamp_to_seconds(end_target_str.strip())
+
+            # Validation checks
+            is_valid_format = start_target_sec != -1.0 and end_target_sec != -1.0
+            is_valid_range = is_valid_format and start_target_sec < end_target_sec
+            # Check if the target overlaps with the transcript's time frame at all
+            is_overlapping = is_valid_range and start_target_sec < max_transcript_time and end_target_sec > min_transcript_time
+
+            if is_overlapping:
+                valid_target_ranges.append({"start": start_target_sec, "end": end_target_sec})
+            else:
+                has_invalid_target = True
+    
+    # --- 3. Print results with highlighting ---
+    print(f"TARGET: ", end="")
+    if has_invalid_target:
+        print(f"{Colors.RED}{target_str or 'EMPTY'}{Colors.RESET}")
+    else:
+        print(f"{Colors.GREEN}{target_str or 'EMPTY'}{Colors.RESET}")
+    print("-" * 30)
+
+    for line_info in transcript_lines:
+        is_ad_line = False
+        for target_range in valid_target_ranges:
+            # Check for overlap between the line's time and the target's time
+            if line_info['start'] < target_range['end'] and line_info['end'] > target_range['start']:
+                is_ad_line = True
+                break
+        
+        if is_ad_line:
+            print(f"{Colors.GREEN}{line_info['line']}{Colors.RESET}")
+        else:
+            print(line_info['line'])
+
+def main(filepath: str):
+    try:
+        with open(filepath, 'r', encoding='utf-8') as f:
+            for line in f:
+                try:
+                    entry = json.loads(line)
+                    visualize_entry(entry)
+                except json.JSONDecodeError:
+                    print(f"{Colors.RED}Error: Could not decode JSON line: {line.strip()}{Colors.RESET}")
+    except FileNotFoundError:
+        print(f"{Colors.RED}Error: File not found at {filepath}{Colors.RESET}")
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print("Usage: python highlight.py <path_to_labels.jsonl>")
+        sys.exit(1)
+    
+    main(sys.argv[1])