yt_de_study/main.py

import os
import textwrap
from pprint import pprint

TIME_SEP = " --> "
DECIMAL_SEP = '.'
HMS_SEP = ':'

FONT_SIZE = 18
WORD_BREAK = 60

MAX_ITER = 0
SHOULD_REMOVE_TEMPS = False
SHOULD_CUT_VIDEOS = True
SHOULD_COPY_CODEC = False

# FINAL_TITLE = "plastic"
# DE_VIDEO = "https://www.youtube.com/watch?v=mhmpeIyG0uM"
# EN_VIDEO = "https://www.youtube.com/watch?v=RS7IzU2VJIQ"
#
# FINAL_TITLE = "atomic_bombs"
# DE_VIDEO = "https://www.youtube.com/watch?v=NBeeRVkTeJg"
# EN_VIDEO = "https://www.youtube.com/watch?v=JyECrGp-Sw8"

FINAL_TITLE = "fusion_power"
DE_VIDEO = "https://www.youtube.com/watch?v=lj4IC70kDIU"
EN_VIDEO = "https://www.youtube.com/watch?v=mZsaaturR6E"


def ts_to_sec(ts):
    hms, ms = ts.split(DECIMAL_SEP)
    h, m, s = hms.split(HMS_SEP)
    return int(h) * 3600 + int(m) * 60 + int(s) + int(ms) / 1000


def parse_subs(filename):
    with open(filename) as f:
        subs, start_ts, end_ts, in_text, text = [], "", "", False, ""
        for line in f.readlines():
            if TIME_SEP in line:
                if start_ts:
                    subs.append((ts_to_sec(start_ts), ts_to_sec(end_ts), text))
                start_ts, end_ts = line.split(TIME_SEP)
                in_text, text = True, ""
                continue
            if in_text:
                text += line.strip() + " "
        return subs


def run_os_cmd(command):
    print(command)
    output = os.popen(command).read()
    return output


def cut_videos(idx, en_start, en_end, de_start, de_end, en_txt, de_txt):
    print("cut_{:02d}: en: {:6.2f} -> {:6.2f}, de: {:6.2f} -> {:6.2f}".format(
        idx, en_start, en_end, de_start, de_end,
    ))
    with open('tmp/concat_{}.txt'.format(idx), 'w+') as f:
        f.write('file de_{}.mp4\n'.format(idx))
        f.write('file en_{}.mp4\n'.format(idx))
        f.write('file de_{}.mp4\n'.format(idx))
    run_os_cmd("ffmpeg -y -i en.mp4 -ss {en_start} -to {en_end} tmp/en_{idx}.mp4 2>logs/01_ffmpeg_cut_en_{idx}.log".format(
        en_start=en_start, en_end=en_end, idx=idx))
    run_os_cmd("ffmpeg -y -i de.mp4 -ss {de_start} -to {de_end} tmp/de_{idx}.mp4 2>logs/01_ffmpeg_cut_en_{idx}.log".format(
        de_start=de_start, de_end=de_end, idx=idx))
    if SHOULD_COPY_CODEC:
        run_os_cmd(
            """ffmpeg -y -f concat -safe 0 -i tmp/concat_{idx}.txt -c copy tmp/deende_{idx}.mp4 2>logs/02_ffmpeg_concat_{idx}_copy.log""".format(
                idx=idx))
    else:
        run_os_cmd(
            """ffmpeg -y -f concat -safe 0 -i tmp/concat_{idx}.txt tmp/deende_{idx}.mp4 2>logs/02_ffmpeg_concat_{idx}_nocopy.log""".format(
                idx=idx))
    en_txt_spl = '\n'.join(textwrap.wrap(en_txt, WORD_BREAK, break_long_words=False))
    de_txt_spl = '\n'.join(textwrap.wrap(de_txt, WORD_BREAK, break_long_words=False))
    text = "{}\n\n\n{}".format(de_txt_spl, en_txt_spl)
    with open('tmp/text_{}.txt'.format(idx), 'w+') as f:
        f.write(text)
    run_os_cmd(
        """ffmpeg -y -i tmp/deende_{idx}.mp4 -vf drawtext="textfile=tmp/text_{idx}.txt: fontcolor=white: fontsize={fontsize}: box=1: boxcolor=black@0.5: boxborderw=5: x=(w-text_w)/2: y=(h-text_h)/2" -c:a copy tmp/deende_{idx}_s.mp4 2>logs/03_ffmpeg_sub_{idx}.log""".format(
            idx=idx, text=text, fontsize=FONT_SIZE))


def download_vids_and_subs():
    if SHOULD_REMOVE_TEMPS:
        run_os_cmd("rm en.mp4 de.mp4 subs.en.vtt subs.de.vtt")
    run_os_cmd("mkdir -p logs")
    run_os_cmd("mkdir -p tmp")
    run_os_cmd("""youtube-dl -f worst --output="de.%(ext)s" {}""".format(DE_VIDEO))
    run_os_cmd("""youtube-dl -f worst --output="en.%(ext)s" {}""".format(EN_VIDEO))
    run_os_cmd(
        """youtube-dl --output="subs.%(ext)s" --write-sub --sub-format vtt --sub-lang de --skip-download {}""".format(
            DE_VIDEO))
    run_os_cmd(
        """youtube-dl --output="subs.%(ext)s" --write-sub --sub-format vtt --sub-lang en --skip-download {}""".format(
            EN_VIDEO))


def time_subs():
    de_subs = parse_subs('subs.de.vtt')
    en_subs = parse_subs('subs.en.vtt')
    latest_de_idx = 0
    iterable = en_subs[:MAX_ITER] if MAX_ITER else en_subs
    for i, en_sub in enumerate(iterable):
        if latest_de_idx >= len(de_subs):
            print("{:6.2f} {:6.2f} {:>6} {:>6} {:>90} {:>90}".format(
                en_sub[0], en_sub[1], '', '',
                en_sub[2], '',
            ))
            continue
        else:
            print("{:6.2f} {:6.2f} {:6.2f} {:6.2f} {:>90} {:>90}".format(
                en_sub[0], en_sub[1], de_subs[latest_de_idx][0], de_subs[latest_de_idx][1],
                en_sub[2], de_subs[latest_de_idx][2],
            ))
            de_first_start = de_subs[latest_de_idx][0]
            de_txt = de_subs[latest_de_idx][2]
            latest_de_idx += 1
            en_sub_start, en_sub_end = en_sub[0], en_sub[1]
            de_last_end = de_subs[latest_de_idx - 1][1]
            while latest_de_idx < len(de_subs):
                de_next_sub_start, de_next_sub_end, txt = de_subs[latest_de_idx]
                if de_next_sub_end < en_sub_end:
                    latest_de_idx += 1
                    de_txt += txt
                    print("{:>6} {:>6} {:6.2f} {:6.2f} {:>90} {:>90}".format(
                        '', '', de_next_sub_start, de_next_sub_end,
                        '', txt,
                    ))
                else:
                    de_last_end = de_subs[latest_de_idx - 1][1]
                    break
        if SHOULD_CUT_VIDEOS:
            cut_videos(i, en_sub[0], en_sub[1], de_first_start, de_last_end, en_sub[2], de_txt)
    return len(iterable)


def concat_all_videos(max_iters):
    with open('tmp/concat.txt', 'w+') as f:
        for idx in range(max_iters):
            f.write('file deende_{}_s.mp4\n'.format(idx))
    if SHOULD_COPY_CODEC:
        run_os_cmd(
            """ffmpeg -y -f concat -safe 0 -i tmp/concat.txt -c copy {final_title}.mp4 2>logs/04_ffmpeg_all_copy.log""".format(
                final_title=FINAL_TITLE))
    else:
        run_os_cmd(
            """ffmpeg -y -f concat -safe 0 -i tmp/concat.txt {final_title}.mp4 2>logs/04_ffmpeg_all_nocopy.log""".format(
                final_title=FINAL_TITLE))
    if SHOULD_REMOVE_TEMPS:
        run_os_cmd("rm -r tmp/")
        for idx in range(MAX_ITER):
            run_os_cmd("rm deende_{idx}.mp4".format(idx=idx))
            run_os_cmd("rm deende_{idx}_s.mp4".format(idx=idx))


def main():
    # download_vids_and_subs()
    max_iters = time_subs()
    concat_all_videos(max_iters)


if __name__ == "__main__":
    main()