def test_read_from_file(self):
os.write(
self.script_description, """1
00:00:17,500 --> 00:00:18,870
Yeah, really!
2
00:00:17,500 --> 00:00:18,870
3
00:00:17,500 --> 00:00:18,870
House number
35
4
00:00:21,250 --> 00:00:22,750
Serves you right.""")
parsed = SrtScript.from_file(self.script_path).events
self.assertEquals(17.5, parsed[0].start)
self.assertEquals(18.87, parsed[0].end)
self.assertEquals("Yeah, really!", parsed[0].text)
self.assertEquals(17.5, parsed[1].start)
self.assertEquals(18.87, parsed[1].end)
self.assertEquals("", parsed[1].text)
self.assertEquals(17.5, parsed[2].start)
self.assertEquals(18.87, parsed[2].end)
self.assertEquals("House number\n35", parsed[2].text)
self.assertEquals(21.25, parsed[3].start)
self.assertEquals(22.75, parsed[3].end)
self.assertEquals("Serves you right.", parsed[3].text)
def test_read_from_file(self):
os.write(self.script_description, """1
00:00:17,500 --> 00:00:18,870
Yeah, really!
2
00:00:17,500 --> 00:00:18,870
3
00:00:17,500 --> 00:00:18,870
House number
35
4
00:00:21,250 --> 00:00:22,750
Serves you right.""")
parsed = SrtScript.from_file(self.script_path).events
self.assertEquals(17.5, parsed[0].start)
self.assertEquals(18.87, parsed[0].end)
self.assertEquals("Yeah, really!", parsed[0].text)
self.assertEquals(17.5, parsed[1].start)
self.assertEquals(18.87, parsed[1].end)
self.assertEquals("", parsed[1].text)
self.assertEquals(17.5, parsed[2].start)
self.assertEquals(18.87, parsed[2].end)
self.assertEquals("House number\n35", parsed[2].text)
self.assertEquals(21.25, parsed[3].start)
self.assertEquals(22.75, parsed[3].end)
self.assertEquals("Serves you right.", parsed[3].text)
def test_write_to_file(self):
events = [
SrtEvent.from_string(SINGLE_LINE_SRT_EVENT),
SrtEvent.from_string(MULTILINE_SRT_EVENT)
]
SrtScript(events).save_to_file(self.script_path)
with open(self.script_path) as script:
text = script.read()
self.assertEquals(SINGLE_LINE_SRT_EVENT + "\n\n" + MULTILINE_SRT_EVENT,
text)
def run(args):
ignore_chapters = args.chapters_file is not None and args.chapters_file.lower() == 'none'
write_plot = plot_enabled and args.plot_path
if write_plot:
plt.clf()
plt.ylabel('Shift, seconds')
plt.xlabel('Event index')
# first part should do all possible validation and should NOT take significant amount of time
check_file_exists(args.source, 'Source')
check_file_exists(args.destination, 'Destination')
check_file_exists(args.src_timecodes, 'Source timecodes')
check_file_exists(args.dst_timecodes, 'Source timecodes')
check_file_exists(args.script_file, 'Script')
if not ignore_chapters:
check_file_exists(args.chapters_file, 'Chapters')
if args.src_keyframes not in ('auto', 'make'):
check_file_exists(args.src_keyframes, 'Source keyframes')
if args.dst_keyframes not in ('auto', 'make'):
check_file_exists(args.dst_keyframes, 'Destination keyframes')
if (args.src_timecodes and args.src_fps) or (args.dst_timecodes and args.dst_fps):
raise SushiError('Both fps and timecodes file cannot be specified at the same time')
src_demuxer = Demuxer(args.source)
dst_demuxer = Demuxer(args.destination)
if src_demuxer.is_wav and not args.script_file:
raise SushiError("Script file isn't specified")
if (args.src_keyframes and not args.dst_keyframes) or (args.dst_keyframes and not args.src_keyframes):
raise SushiError('Either none or both of src and dst keyframes should be provided')
create_directory_if_not_exists(args.temp_dir)
# selecting source audio
if src_demuxer.is_wav:
src_audio_path = args.source
else:
src_audio_path = format_full_path(args.temp_dir, args.source, '.sushi.wav')
src_demuxer.set_audio(stream_idx=args.src_audio_idx, output_path=src_audio_path, sample_rate=args.sample_rate)
# selecting destination audio
if dst_demuxer.is_wav:
dst_audio_path = args.destination
else:
dst_audio_path = format_full_path(args.temp_dir, args.destination, '.sushi.wav')
dst_demuxer.set_audio(stream_idx=args.dst_audio_idx, output_path=dst_audio_path, sample_rate=args.sample_rate)
# selecting source subtitles
if args.script_file:
src_script_path = args.script_file
else:
stype = src_demuxer.get_subs_type(args.src_script_idx)
src_script_path = format_full_path(args.temp_dir, args.source, '.sushi'+ stype)
src_demuxer.set_script(stream_idx=args.src_script_idx, output_path=src_script_path)
script_extension = get_extension(src_script_path)
if script_extension not in ('.ass', '.srt'):
raise SushiError('Unknown script type')
# selection destination subtitles
if args.output_script:
dst_script_path = args.output_script
dst_script_extension = get_extension(args.output_script)
if dst_script_extension != script_extension:
raise SushiError("Source and destination script file types don't match ({0} vs {1})"
.format(script_extension, dst_script_extension))
else:
dst_script_path = format_full_path(args.temp_dir, args.destination, '.sushi' + script_extension)
# selecting chapters
if args.grouping and not ignore_chapters:
if args.chapters_file:
if get_extension(args.chapters_file) == '.xml':
chapter_times = chapters.get_xml_start_times(args.chapters_file)
else:
chapter_times = chapters.get_ogm_start_times(args.chapters_file)
elif not src_demuxer.is_wav:
chapter_times = src_demuxer.chapters
output_path = format_full_path(args.temp_dir, src_demuxer.path, ".sushi.chapters.txt")
src_demuxer.set_chapters(output_path)
else:
chapter_times = []
else:
chapter_times = []
# selecting keyframes and timecodes
if args.src_keyframes:
def select_keyframes(file_arg, demuxer):
auto_file = format_full_path(args.temp_dir, demuxer.path, '.sushi.keyframes.txt')
if file_arg in ('auto', 'make'):
if file_arg == 'make' or not os.path.exists(auto_file):
if not demuxer.has_video:
raise SushiError("Cannot make keyframes for {0} because it doesn't have any video!"
.format(demuxer.path))
demuxer.set_keyframes(output_path=auto_file)
return auto_file
else:
return file_arg
def select_timecodes(external_file, fps_arg, demuxer):
if external_file:
return external_file
elif fps_arg:
return None
elif demuxer.has_video:
path = format_full_path(args.temp_dir, demuxer.path, '.sushi.timecodes.txt')
demuxer.set_timecodes(output_path=path)
return path
else:
raise SushiError('Fps, timecodes or video files must be provided if keyframes are used')
src_keyframes_file = select_keyframes(args.src_keyframes, src_demuxer)
dst_keyframes_file = select_keyframes(args.dst_keyframes, dst_demuxer)
src_timecodes_file = select_timecodes(args.src_timecodes, args.src_fps, src_demuxer)
dst_timecodes_file = select_timecodes(args.dst_timecodes, args.dst_fps, dst_demuxer)
# after this point nothing should fail so it's safe to start slow operations
# like running the actual demuxing
src_demuxer.demux()
dst_demuxer.demux()
try:
if args.src_keyframes:
src_timecodes = Timecodes.cfr(args.src_fps) if args.src_fps else Timecodes.from_file(src_timecodes_file)
src_keytimes = [src_timecodes.get_frame_time(f) for f in keyframes.parse_keyframes(src_keyframes_file)]
dst_timecodes = Timecodes.cfr(args.dst_fps) if args.dst_fps else Timecodes.from_file(dst_timecodes_file)
dst_keytimes = [dst_timecodes.get_frame_time(f) for f in keyframes.parse_keyframes(dst_keyframes_file)]
script = AssScript.from_file(src_script_path) if script_extension == '.ass' else SrtScript.from_file(src_script_path)
script.sort_by_time()
src_stream = WavStream(src_audio_path, sample_rate=args.sample_rate, sample_type=args.sample_type)
dst_stream = WavStream(dst_audio_path, sample_rate=args.sample_rate, sample_type=args.sample_type)
search_groups = prepare_search_groups(script.events,
source_duration=src_stream.duration_seconds,
chapter_times=chapter_times,
max_ts_duration=args.max_ts_duration,
max_ts_distance=args.max_ts_distance)
calculate_shifts(src_stream, dst_stream, search_groups,
normal_window=args.window,
max_window=args.max_window,
rewind_thresh=args.rewind_thresh if args.grouping else 0)
events = script.events
if write_plot:
plt.plot([x.shift for x in events], label='From audio')
if args.grouping:
if not ignore_chapters and chapter_times:
groups = groups_from_chapters(events, chapter_times)
for g in groups:
fix_near_borders(g)
smooth_events([x for x in g if not x.linked], args.smooth_radius)
groups = split_broken_groups(groups)
else:
fix_near_borders(events)
smooth_events([x for x in events if not x.linked], args.smooth_radius)
groups = detect_groups(events)
if write_plot:
plt.plot([x.shift for x in events], label='Borders fixed')
for g in groups:
start_shift = g[0].shift
end_shift = g[-1].shift
avg_shift = average_shifts(g)
logging.info(u'Group (start: {0}, end: {1}, lines: {2}), '
u'shifts (start: {3}, end: {4}, average: {5})'
.format(format_time(g[0].start), format_time(g[-1].end), len(g), start_shift, end_shift,
avg_shift))
if args.src_keyframes:
for e in (x for x in events if x.linked):
e.resolve_link()
for g in groups:
snap_groups_to_keyframes(g, chapter_times, args.max_ts_duration, args.max_ts_distance, src_keytimes,
dst_keytimes, src_timecodes, dst_timecodes, args.max_kf_distance, args.kf_mode)
else:
fix_near_borders(events)
if write_plot:
plt.plot([x.shift for x in events], label='Borders fixed')
if args.src_keyframes:
for e in (x for x in events if x.linked):
e.resolve_link()
snap_groups_to_keyframes(events, chapter_times, args.max_ts_duration, args.max_ts_distance, src_keytimes,
dst_keytimes, src_timecodes, dst_timecodes, args.max_kf_distance, args.kf_mode)
for event in events:
event.apply_shift()
script.save_to_file(dst_script_path)
if write_plot:
plt.plot([x.shift + (x._start_shift + x._end_shift)/2.0 for x in events], label='After correction')
plt.legend(fontsize=5, frameon=False, fancybox=False)
plt.savefig(args.plot_path, dpi=300)
finally:
if args.cleanup:
src_demuxer.cleanup()
dst_demuxer.cleanup()
def run(args):
ignore_chapters = args.chapters_file is not None and args.chapters_file.lower(
) == 'none'
write_plot = plot_enabled and args.plot_path
if write_plot:
plt.clf()
plt.ylabel('Shift, seconds')
plt.xlabel('Event index')
# first part should do all possible validation and should NOT take significant amount of time
check_file_exists(args.source, 'Source')
check_file_exists(args.destination, 'Destination')
check_file_exists(args.src_timecodes, 'Source timecodes')
check_file_exists(args.dst_timecodes, 'Source timecodes')
check_file_exists(args.script_file, 'Script')
if not ignore_chapters:
check_file_exists(args.chapters_file, 'Chapters')
if args.src_keyframes not in ('auto', 'make'):
check_file_exists(args.src_keyframes, 'Source keyframes')
if args.dst_keyframes not in ('auto', 'make'):
check_file_exists(args.dst_keyframes, 'Destination keyframes')
if (args.src_timecodes and args.src_fps) or (args.dst_timecodes
and args.dst_fps):
raise SushiError(
'Both fps and timecodes file cannot be specified at the same time')
src_demuxer = Demuxer(args.source)
dst_demuxer = Demuxer(args.destination)
if src_demuxer.is_wav and not args.script_file:
raise SushiError("Script file isn't specified")
if (args.src_keyframes
and not args.dst_keyframes) or (args.dst_keyframes
and not args.src_keyframes):
raise SushiError(
'Either none or both of src and dst keyframes should be provided')
create_directory_if_not_exists(args.temp_dir)
# selecting source audio
if src_demuxer.is_wav:
src_audio_path = args.source
else:
src_audio_path = format_full_path(args.temp_dir, args.source,
'.sushi.wav')
src_demuxer.set_audio(stream_idx=args.src_audio_idx,
output_path=src_audio_path,
sample_rate=args.sample_rate)
# selecting destination audio
if dst_demuxer.is_wav:
dst_audio_path = args.destination
else:
dst_audio_path = format_full_path(args.temp_dir, args.destination,
'.sushi.wav')
dst_demuxer.set_audio(stream_idx=args.dst_audio_idx,
output_path=dst_audio_path,
sample_rate=args.sample_rate)
# selecting source subtitles
if args.script_file:
src_script_path = args.script_file
else:
stype = src_demuxer.get_subs_type(args.src_script_idx)
src_script_path = format_full_path(args.temp_dir, args.source,
'.sushi' + stype)
src_demuxer.set_script(stream_idx=args.src_script_idx,
output_path=src_script_path)
script_extension = get_extension(src_script_path)
if script_extension not in ('.ass', '.srt'):
raise SushiError('Unknown script type')
# selection destination subtitles
if args.output_script:
dst_script_path = args.output_script
dst_script_extension = get_extension(args.output_script)
if dst_script_extension != script_extension:
raise SushiError(
"Source and destination script file types don't match ({0} vs {1})"
.format(script_extension, dst_script_extension))
else:
dst_script_path = format_full_path(args.temp_dir, args.destination,
'.sushi' + script_extension)
# selecting chapters
if args.grouping and not ignore_chapters:
if args.chapters_file:
if get_extension(args.chapters_file) == '.xml':
chapter_times = chapters.get_xml_start_times(
args.chapters_file)
else:
chapter_times = chapters.get_ogm_start_times(
args.chapters_file)
elif not src_demuxer.is_wav:
chapter_times = src_demuxer.chapters
output_path = format_full_path(args.temp_dir, src_demuxer.path,
".sushi.chapters.txt")
src_demuxer.set_chapters(output_path)
else:
chapter_times = []
else:
chapter_times = []
# selecting keyframes and timecodes
if args.src_keyframes:
def select_keyframes(file_arg, demuxer):
auto_file = format_full_path(args.temp_dir, demuxer.path,
'.sushi.keyframes.txt')
if file_arg in ('auto', 'make'):
if file_arg == 'make' or not os.path.exists(auto_file):
if not demuxer.has_video:
raise SushiError(
"Cannot make keyframes for {0} because it doesn't have any video!"
.format(demuxer.path))
demuxer.set_keyframes(output_path=auto_file)
return auto_file
else:
return file_arg
def select_timecodes(external_file, fps_arg, demuxer):
if external_file:
return external_file
elif fps_arg:
return None
elif demuxer.has_video:
path = format_full_path(args.temp_dir, demuxer.path,
'.sushi.timecodes.txt')
demuxer.set_timecodes(output_path=path)
return path
else:
raise SushiError(
'Fps, timecodes or video files must be provided if keyframes are used'
)
src_keyframes_file = select_keyframes(args.src_keyframes, src_demuxer)
dst_keyframes_file = select_keyframes(args.dst_keyframes, dst_demuxer)
src_timecodes_file = select_timecodes(args.src_timecodes, args.src_fps,
src_demuxer)
dst_timecodes_file = select_timecodes(args.dst_timecodes, args.dst_fps,
dst_demuxer)
# after this point nothing should fail so it's safe to start slow operations
# like running the actual demuxing
src_demuxer.demux()
dst_demuxer.demux()
try:
if args.src_keyframes:
src_timecodes = Timecodes.cfr(
args.src_fps) if args.src_fps else Timecodes.from_file(
src_timecodes_file)
src_keytimes = [
src_timecodes.get_frame_time(f)
for f in parse_keyframes(src_keyframes_file)
]
dst_timecodes = Timecodes.cfr(
args.dst_fps) if args.dst_fps else Timecodes.from_file(
dst_timecodes_file)
dst_keytimes = [
dst_timecodes.get_frame_time(f)
for f in parse_keyframes(dst_keyframes_file)
]
script = AssScript.from_file(
src_script_path
) if script_extension == '.ass' else SrtScript.from_file(
src_script_path)
script.sort_by_time()
src_stream = WavStream(src_audio_path,
sample_rate=args.sample_rate,
sample_type=args.sample_type)
dst_stream = WavStream(dst_audio_path,
sample_rate=args.sample_rate,
sample_type=args.sample_type)
calculate_shifts(
src_stream,
dst_stream,
script.events,
chapter_times=chapter_times,
window=args.window,
max_window=args.max_window,
rewind_thresh=args.rewind_thresh if args.grouping else 0,
max_ts_duration=args.max_ts_duration,
max_ts_distance=args.max_ts_distance)
events = script.events
if write_plot:
plt.plot([x.shift for x in events], label='From audio')
if args.grouping:
if not ignore_chapters and chapter_times:
groups = groups_from_chapters(events, chapter_times)
for g in groups:
fix_near_borders(g)
smooth_events([x for x in g if not x.linked],
args.smooth_radius)
groups = split_broken_groups(groups, args.min_group_size)
else:
fix_near_borders(events)
smooth_events([x for x in events if not x.linked],
args.smooth_radius)
groups = detect_groups(events, args.min_group_size)
if write_plot:
plt.plot([x.shift for x in events], label='Borders fixed')
for g in groups:
start_shift = g[0].shift
end_shift = g[-1].shift
avg_shift = average_shifts(g)
logging.info(
u'Group (start: {0}, end: {1}, lines: {2}), '
u'shifts (start: {3}, end: {4}, average: {5})'.format(
format_time(g[0].start), format_time(g[-1].end),
len(g), start_shift, end_shift, avg_shift))
if args.src_keyframes:
for e in (x for x in events if x.linked):
e.resolve_link()
for g in groups:
snap_groups_to_keyframes(
g, chapter_times, args.max_ts_duration,
args.max_ts_distance, src_keytimes, dst_keytimes,
src_timecodes, dst_timecodes, args.max_kf_distance,
args.kf_mode)
if args.write_avs:
write_shift_avs(dst_script_path + '.avs', groups,
src_audio_path, dst_audio_path)
else:
fix_near_borders(events)
if write_plot:
plt.plot([x.shift for x in events], label='Borders fixed')
if args.src_keyframes:
for e in (x for x in events if x.linked):
e.resolve_link()
snap_groups_to_keyframes(events, chapter_times,
args.max_ts_duration,
args.max_ts_distance, src_keytimes,
dst_keytimes, src_timecodes,
dst_timecodes, args.max_kf_distance,
args.kf_mode)
for event in events:
event.apply_shift()
script.save_to_file(dst_script_path)
if write_plot:
plt.plot([
x.shift + (x._start_shift + x._end_shift) / 2.0 for x in events
],
label='After correction')
plt.legend(fontsize=5, frameon=False, fancybox=False)
plt.savefig(args.plot_path, dpi=300)
finally:
if args.cleanup:
src_demuxer.cleanup()
dst_demuxer.cleanup()