Manmeet2002/Accessibility_Project

1# parse_data.py2import json3 4# This is the big file we just added5WLASL_JSON_FILE = 'WLASL_info.json' 6OUTPUT_MAP_FILE = 'sign_video_map.json'7 8print(f"Loading data from {WLASL_JSON_FILE}...")9 10try:11    with open(WLASL_JSON_FILE, 'r') as f:12        data = json.load(f)13except FileNotFoundError:14    print(f"ERROR: Could not find {WLASL_JSON_FILE}.")15    print("Please make sure you moved and renamed the file correctly.")16    exit()17 18video_map = {}19 20# The WLASL JSON is a list of entries21for entry in data:22    word = entry['gloss']23 24    # An entry can have many video instances25    if not entry['instances']:26        continue27 28    # Let's just grab the FIRST video instance for simplicity29    first_instance = entry['instances'][0]30 31    # Get the video ID (which matches the .mp4 file name)32    video_id = first_instance.get('video_id')33 34    if word and video_id:35        # Save the word in lowercase for easy lookup36        if word.lower() not in video_map:37             video_map[word.lower()] = video_id38 39# Now, save our new, simple map40with open(OUTPUT_MAP_FILE, 'w') as f:41    json.dump(video_map, f, indent=2)42 43print(f"Success! Created {OUTPUT_MAP_FILE} with {len(video_map)} sign-to-video links.")