make_srt_from_ffmpeg_ocr.py

#!/usr/bin/env python
import sys
import json
import datetime

# Feed me with output of:
# ffprobe -f lavfi -i movie=INPUT.mov,ocr -show_entries frame=pkt_dts_time:frame_tags=lavfi.ocr.text -of json
input= sys.argv[1]
id = 0
old_time = '00:00:00,000'

with open(input) as f:
    data = json.loads(f.read())
    for el in data['frames']:
        ts = (datetime.datetime.min + datetime.timedelta(seconds=float(el['pkt_dts_time']))).time()
        timestamp = '{:02}:{:02}:{:02},{:03}'.format(ts.hour, ts.minute, ts.second, ts.microsecond / 1000)
        if not old_time: old_time = timestamp
        text = unicode.strip(el['tags']['lavfi.ocr.text'])
        if text:
            id+=1
            print('{0}\n{1} --> {2}\n{3}\n'.format(id, old_time, timestamp, text))
            old_time = timestamp