splice-benchmark / covert_jsonl.py
prokajevo's picture
Add files using upload-large-folder tool
a1d4687 verified
import json
import os
from tqdm import tqdm
INPUT_JSON = "splice_segment_metadata.json"
OUTPUT_JSONL = "metadata.jsonl"
def convert_to_jsonl_ordered():
"""
Converts the nested SPLICE JSON metadata into a flat, logically ordered
JSON Lines file for a professional display in the Hugging Face Viewer.
"""
print(f"--- Converting to logically ordered '{OUTPUT_JSONL}' ---")
with open(INPUT_JSON, 'r') as f:
data = json.load(f)
with open(OUTPUT_JSONL, 'w') as f_out:
for video_info in tqdm(data, desc="Processing video tasks"):
for segment in video_info.get("segments", []):
output_record = {
"file_name": segment.get("output_path", ""),
"video_id": video_info.get("video_id", ""),
"part": segment.get("part", -1),
"label": segment.get("label", ""),
"domain": video_info.get("Domain", ""),
"class": video_info.get("class", ""),
#"subset": video_info.get("subset", ""),
"start": segment.get("start", -1.0),
"end": segment.get("end", -1.0),
#"segment_id": segment.get("segment_id", ""),
"task_duration": video_info.get("duration", -1.0),
"video_url": video_info.get("video_url", ""),
}
f_out.write(json.dumps(output_record) + '\n')
print("\n--- Conversion Complete! ---")
print(f"Successfully created '{OUTPUT_JSONL}' with a professional column order.")
if __name__ == "__main__":
convert_to_jsonl_ordered()