add_srt_to_description.py

#!/usr/bin/env python
import sys
import os
import datetime
import time

import pysrt
from internetarchive import get_item


def download_srt(item):
    srt = list(item.get_files(formats=['SubRip']))
    if not srt:
        print('error: no srt file found for {}'.format(item.identifier))
        sys.exit(1)
    else:
        srt_fo = srt[0]
    r = srt_fo.download(silent=True)
    return srt_fo


def get_description(srt_fp, item):
    subs = pysrt.open(srt_fp)
    description = 'Closed captions transcript:\n\n'
    for x in subs:
        timestamp = str(x.start).split(',')[0]
        t = time.strptime(timestamp, '%H:%M:%S')
        ss = int(datetime.timedelta(hours=t.tm_hour,
                                    minutes=t.tm_min,
                                    seconds=t.tm_sec).total_seconds())
        link = '<a href="/details/{0}?start={1}">{2}</a>'.format(item.identifier, ss, timestamp)
        _desc = '{0}\n{1}\n\n'.format(link, x.text.encode('utf-8'))
        description += _desc
    return description


if __name__ == '__main__':
    item = get_item(sys.argv[-1])
    srt_fo = download_srt(item)
    srt_fp = os.path.join(os.getcwd(), srt_fo.name)
    description = get_description(srt_fp, item)

    if item.metadata.get('description'):
        print('warning: {} - already has a description, skipping.'.format(item.identifier))
        sys.exit(0)
    # Update description
    r = item.modify_metadata(dict(description=description.strip()))
    if not r.ok:
        print('error: {} - failed to update description.'.format(item.identifier))
        sys.exit(1)

    print('success: {}'.format(item.identifier))