write_chapters.py

"""
Takes a Met Opera json file and parses it and then writes a .inf file suitable for creating video chapters.

ffmpeg -i opera.mp4 -i opera.json.inf -map_metadata 1 -codec copy new_opera.mp4
"""

import json
import sys
from datetime import datetime

def parse_metadata(s):
    "title=Le Nozze di Figaro: Opening credits|artists=|featured=false|isFree=false|ISRC=USY3P0803132"
    "title=Act I: La vendetta . . . Tutto ancor non ho perso|artists=Paul Plishka, Wendy White, Cecilia Bartoli|featured=false|isFree=false|ISRC=USY3P0803137"

    fields = s.split("|")
    return dict(tuple(field.split("=")) for field in fields)


def get_chapters(filename):
    with  open(filename) as f:
        data = json.load(f)
    # other interesting fields:
    #   poster
    #   custom_fields has a lot of useful track metadata
    md = data['custom_fields']

    def parse_date(s):
        return datetime.strptime(s, "%m/%d/%Y")

    metadata = {
        'duration': data['duration'],
        'work':  md.get('searchtitle', "TITLE"),
        'composer':  md.get('composerlast', md.get('composer', "COMPOSER")),
        'librettist': md.get('librettist', "LIBRETTIST"),
        'conductor':  md.get('conductor', "CONDUCTOR"),
        'date':  parse_date(md.get('performancedate', "01/01/01")),
        'description': data.get('long_description', ""),
        'artists': json.loads(md.get('artistjson', "")),
        'synopsis_url': md.get('synopsis', ''),
    }
    metadata['title'] = "{composer} - {work}".format(**metadata)

    last_index = 1
    chapters = []
    for track in data['cue_points']:
        if track['type'] == 'AD':
            continue

        track_metadata = parse_metadata(track['metadata'])
        chapters.append(dict(
            index=last_index,
            start_time=int(track['time'] * 1000),
            title=track_metadata['title'],
        ))
        last_index += 1

    return chapters, metadata


def fill_ends(chapters, end):
    filled = []
    for ch0, ch1 in zip(chapters, chapters[1:]):
        ch0['end_time'] = ch1['start_time']
        filled.append(ch0)

    # deal with the last one ...
    ch = chapters[-1]
    ch['end_time'] = end
    filled.append(ch)

    return filled


def write_chapters(filename, chapters, metadata):
    # adapted from:
    #   https://medium.com/@dathanbennett/adding-chapters-to-an-mp4-file-using-ffmpeg-5e43df269687
    # see fields here:
    #   https://wiki.multimedia.cx/index.php/FFmpeg_Metadata
    # and file format here:
    #   https://ffmpeg.org/ffmpeg-formats.html#Metadata-1

    def escape(s):
        """https://ffmpeg.org/ffmpeg-formats.html#Metadata-1"""
        return s.translate(str.maketrans({
            "=": r"\=",
            ";": r"\;",
            "#": r"\#",
            '\\': r"\\"
        }))

    with open(filename, 'w') as f:
        f.write(";FFMETADATA1\n")
        f.write("title=%s\n" % escape(metadata['title']))
        f.write("artist=%s\n\n" % metadata['conductor'])
        f.write("composer=%s\n\n" % metadata['conductor'])
        f.write("year=%s\n\n" % metadata['date'].year)
        f.write("synopsis=%s\n\n" % metadata['description'])
        f.write("description=%s\n\n" % metadata['synopsis_url'])
        f.write("copyright=The Metropolitan Opera\n\n")

        for chapter in chapters:
            f.write("[CHAPTER]\n")
            f.write("TIMEBASE=1/1000\n")
            f.write("START=%s\n" % chapter['start_time'])
            f.write("END=%s\n" % chapter['end_time'])
            f.write("title=%s\n\n" % escape(chapter['title']))

        f.write("[STREAM]\n")
        f.write("title=%s\n" % escape(metadata['title']))

    print("Wrote chapter data to %s. Next: " % filename)
    print("\tffmpeg -i opera.mp4 -i %s -map_metadata 1 -codec copy out.mp4" % filename)


def write_metadata(filename, metadata):
    with open(filename, 'w', encoding="utf-8") as f:
        f.write("# {work}\n".format(**metadata))
        f.write("{composer} - {librettist}\n".format(**metadata))
        f.write("Metropolitan Opera House\n")
        f.write("{date:%A, %B %d, %Y} Telecast\n".format(**metadata))

        f.write("\n\n## Cast \n")
        for artist in metadata.get('artists', []):
            f.write("* {role}: {name}\n".format(**artist))

        f.write("\n\n## Description \n")
        f.write("{description}".format(**metadata))

        f.write("\n\nSynopsis: {synopsis_url}".format(**metadata))


def main(filename):
    chapters, metadata = get_chapters(filename)
    chapters = fill_ends(chapters, metadata['duration'])
    write_chapters(filename + '.inf', chapters, metadata)
    write_metadata(filename + '.md', metadata)


if __name__ == '__main__':
    try:
        main(sys.argv[1])
    except IndexError:
        print("usage: python write_chapters.py metadata.json")