[yt-dlp.git] / yt_dlp / postprocessor / embedthumbnail.py

import base64
import os
import re
import subprocess

from .common import PostProcessor
from .ffmpeg import FFmpegPostProcessor, FFmpegThumbnailsConvertorPP
from ..compat import imghdr
from ..dependencies import mutagen
from ..utils import (
    Popen,
    PostProcessingError,
    check_executable,
    encodeArgument,
    encodeFilename,
    error_to_compat_str,
    prepend_extension,
    shell_quote,
)

if mutagen:
    from mutagen.flac import FLAC, Picture
    from mutagen.mp4 import MP4, MP4Cover
    from mutagen.oggopus import OggOpus
    from mutagen.oggvorbis import OggVorbis


class EmbedThumbnailPPError(PostProcessingError):
    pass


class EmbedThumbnailPP(FFmpegPostProcessor):

    def __init__(self, downloader=None, already_have_thumbnail=False):
        FFmpegPostProcessor.__init__(self, downloader)
        self._already_have_thumbnail = already_have_thumbnail

    def _get_thumbnail_resolution(self, filename, thumbnail_dict):
        def guess():
            width, height = thumbnail_dict.get('width'), thumbnail_dict.get('height')
            if width and height:
                return width, height

        try:
            size_regex = r',\s*(?P<w>\d+)x(?P<h>\d+)\s*[,\[]'
            size_result = self.run_ffmpeg(filename, None, ['-hide_banner'], expected_retcodes=(1,))
            mobj = re.search(size_regex, size_result)
            if mobj is None:
                return guess()
        except PostProcessingError as err:
            self.report_warning('unable to find the thumbnail resolution; %s' % error_to_compat_str(err))
            return guess()
        return int(mobj.group('w')), int(mobj.group('h'))

    def _report_run(self, exe, filename):
        self.to_screen(f'{exe}: Adding thumbnail to "{filename}"')

    @PostProcessor._restrict_to(images=False)
    def run(self, info):
        filename = info['filepath']
        temp_filename = prepend_extension(filename, 'temp')

        if not info.get('thumbnails'):
            self.to_screen('There aren\'t any thumbnails to embed')
            return [], info

        idx = next((-i for i, t in enumerate(info['thumbnails'][::-1], 1) if t.get('filepath')), None)
        if idx is None:
            self.to_screen('There are no thumbnails on disk')
            return [], info
        thumbnail_filename = info['thumbnails'][idx]['filepath']
        if not os.path.exists(encodeFilename(thumbnail_filename)):
            self.report_warning('Skipping embedding the thumbnail because the file is missing.')
            return [], info

        # Correct extension for WebP file with wrong extension (see #25687, #25717)
        convertor = FFmpegThumbnailsConvertorPP(self._downloader)
        convertor.fixup_webp(info, idx)

        original_thumbnail = thumbnail_filename = info['thumbnails'][idx]['filepath']

        # Convert unsupported thumbnail formats (see #25687, #25717)
        # PNG is preferred since JPEG is lossy
        thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
        if info['ext'] not in ('mkv', 'mka') and thumbnail_ext not in ('jpg', 'jpeg', 'png'):
            thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
            thumbnail_ext = 'png'

        mtime = os.stat(encodeFilename(filename)).st_mtime

        success = True
        if info['ext'] == 'mp3':
            options = [
                '-c', 'copy', '-map', '0:0', '-map', '1:0', '-write_id3v1', '1', '-id3v2_version', '3',
                '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment=Cover (front)']

            self._report_run('ffmpeg', filename)
            self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)

        elif info['ext'] in ['mkv', 'mka']:
            options = list(self.stream_copy_opts())

            mimetype = f'image/{thumbnail_ext.replace("jpg", "jpeg")}'
            old_stream, new_stream = self.get_stream_number(
                filename, ('tags', 'mimetype'), mimetype)
            if old_stream is not None:
                options.extend(['-map', '-0:%d' % old_stream])
                new_stream -= 1
            options.extend([
                '-attach', self._ffmpeg_filename_argument(thumbnail_filename),
                '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
                '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])

            self._report_run('ffmpeg', filename)
            self.run_ffmpeg(filename, temp_filename, options)

        elif info['ext'] in ['m4a', 'mp4', 'm4v', 'mov']:
            prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
            # Method 1: Use mutagen
            if not mutagen or prefer_atomicparsley:
                success = False
            else:
                try:
                    self._report_run('mutagen', filename)
                    meta = MP4(filename)
                    # NOTE: the 'covr' atom is a non-standard MPEG-4 atom,
                    # Apple iTunes 'M4A' files include the 'moov.udta.meta.ilst' atom.
                    f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
                    with open(thumbnail_filename, 'rb') as thumbfile:
                        thumb_data = thumbfile.read()
                    meta.tags['covr'] = [MP4Cover(data=thumb_data, imageformat=f)]
                    meta.save()
                    temp_filename = filename
                except Exception as err:
                    self.report_warning('unable to embed using mutagen; %s' % error_to_compat_str(err))
                    success = False

            # Method 2: Use AtomicParsley
            if not success:
                success = True
                atomicparsley = next((
                    # libatomicparsley.so : See https://github.com/xibr/ytdlp-lazy/issues/1
                    x for x in ['AtomicParsley', 'atomicparsley', 'libatomicparsley.so']
                    if check_executable(x, ['-v'])), None)
                if atomicparsley is None:
                    self.to_screen('Neither mutagen nor AtomicParsley was found. Falling back to ffmpeg')
                    success = False
                else:
                    if not prefer_atomicparsley:
                        self.to_screen('mutagen was not found. Falling back to AtomicParsley')
                    cmd = [encodeFilename(atomicparsley, True),
                           encodeFilename(filename, True),
                           encodeArgument('--artwork'),
                           encodeFilename(thumbnail_filename, True),
                           encodeArgument('-o'),
                           encodeFilename(temp_filename, True)]
                    cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]

                    self._report_run('atomicparsley', filename)
                    self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
                    stdout, stderr, returncode = Popen.run(cmd, text=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
                    if returncode:
                        self.report_warning(f'Unable to embed thumbnails using AtomicParsley; {stderr.strip()}')
                    # for formats that don't support thumbnails (like 3gp) AtomicParsley
                    # won't create to the temporary file
                    if 'No changes' in stdout:
                        self.report_warning('The file format doesn\'t support embedding a thumbnail')
                        success = False

            # Method 3: Use ffmpeg+ffprobe
            # Thumbnails attached using this method doesn't show up as cover in some cases
            # See https://github.com/yt-dlp/yt-dlp/issues/2125, https://github.com/yt-dlp/yt-dlp/issues/411
            if not success:
                success = True
                try:
                    options = [*self.stream_copy_opts(), '-map', '1']

                    old_stream, new_stream = self.get_stream_number(
                        filename, ('disposition', 'attached_pic'), 1)
                    if old_stream is not None:
                        options.extend(['-map', '-0:%d' % old_stream])
                        new_stream -= 1
                    options.extend(['-disposition:%s' % new_stream, 'attached_pic'])

                    self._report_run('ffmpeg', filename)
                    self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
                except PostProcessingError as err:
                    success = False
                    raise EmbedThumbnailPPError(f'Unable to embed using ffprobe & ffmpeg; {err}')

        elif info['ext'] in ['ogg', 'opus', 'flac']:
            if not mutagen:
                raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python3 -m pip install mutagen`')

            self._report_run('mutagen', filename)
            f = {'opus': OggOpus, 'flac': FLAC, 'ogg': OggVorbis}[info['ext']](filename)

            pic = Picture()
            pic.mime = 'image/%s' % imghdr.what(thumbnail_filename)
            with open(thumbnail_filename, 'rb') as thumbfile:
                pic.data = thumbfile.read()
            pic.type = 3  # front cover
            res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
            if res is not None:
                pic.width, pic.height = res

            if info['ext'] == 'flac':
                f.add_picture(pic)
            else:
                # https://wiki.xiph.org/VorbisComment#METADATA_BLOCK_PICTURE
                f['METADATA_BLOCK_PICTURE'] = base64.b64encode(pic.write()).decode('ascii')
            f.save()
            temp_filename = filename

        else:
            raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/m4v/mov')

        if success and temp_filename != filename:
            os.replace(temp_filename, filename)

        self.try_utime(filename, mtime, mtime)
        converted = original_thumbnail != thumbnail_filename
        self._delete_downloaded_files(
            thumbnail_filename if converted or not self._already_have_thumbnail else None,
            original_thumbnail if converted and not self._already_have_thumbnail else None,
            info=info)
        return [], info
Commit	Line	Data
95131b21	1	import base64
ddbed364	2	import os
06167fbb	3	import re
f8271158	4	import subprocess
06167fbb	5
8326b00a	6	from .common import PostProcessor
f8271158	7	from .ffmpeg import FFmpegPostProcessor, FFmpegThumbnailsConvertorPP
5792c950	8	from ..compat import imghdr
9b8ee23b	9	from ..dependencies import mutagen
ddbed364	10	from ..utils import (
f8271158	11	Popen,
f8271158	12	PostProcessingError,
ddbed364	13	check_executable,
2cc6d135	14	encodeArgument,
ddbed364	15	encodeFilename,
06167fbb	16	error_to_compat_str,
ddbed364	17	prepend_extension,
f5b1bca9	18	shell_quote,
ddbed364	19	)
ddbed364	20
9b8ee23b	21	if mutagen:
	22	from mutagen.flac import FLAC, Picture
	23	from mutagen.mp4 import MP4, MP4Cover
	24	from mutagen.oggopus import OggOpus
	25	from mutagen.oggvorbis import OggVorbis
	26
ddbed364	27
	28	class EmbedThumbnailPPError(PostProcessingError):
	29	pass
	30
	31
31fd9c76	32	class EmbedThumbnailPP(FFmpegPostProcessor):
1b77b347	33
8e595397	34	def __init__(self, downloader=None, already_have_thumbnail=False):
8fa43c73	35	FFmpegPostProcessor.__init__(self, downloader)
8e595397 YCH	36	self._already_have_thumbnail = already_have_thumbnail
8e595397 YCH	37
95131b21	38	def _get_thumbnail_resolution(self, filename, thumbnail_dict):
	39	def guess():
	40	width, height = thumbnail_dict.get('width'), thumbnail_dict.get('height')
	41	if width and height:
	42	return width, height
	43
	44	try:
	45	size_regex = r',\s(?P<w>\d+)x(?P<h>\d+)\s[,\[]'
00034c14	46	size_result = self.run_ffmpeg(filename, None, ['-hide_banner'], expected_retcodes=(1,))
95131b21	47	mobj = re.search(size_regex, size_result)
	48	if mobj is None:
	49	return guess()
	50	except PostProcessingError as err:
	51	self.report_warning('unable to find the thumbnail resolution; %s' % error_to_compat_str(err))
	52	return guess()
	53	return int(mobj.group('w')), int(mobj.group('h'))
	54
acdecdfa	55	def _report_run(self, exe, filename):
86e5f3ed	56	self.to_screen(f'{exe}: Adding thumbnail to "{filename}"')
acdecdfa	57
8326b00a	58	@PostProcessor._restrict_to(images=False)
ddbed364	59	def run(self, info):
	60	filename = info['filepath']
	61	temp_filename = prepend_extension(filename, 'temp')
ddbed364	62
8e595397	63	if not info.get('thumbnails'):
1b77b347	64	self.to_screen('There aren\'t any thumbnails to embed')
b5cbe3d6	65	return [], info
ddbed364	66
337e0c62	67	idx = next((-i for i, t in enumerate(info['thumbnails'][::-1], 1) if t.get('filepath')), None)
885cc0b7	68	if idx is None:
	69	self.to_screen('There are no thumbnails on disk')
	70	return [], info
	71	thumbnail_filename = info['thumbnails'][idx]['filepath']
c33a8639	72	if not os.path.exists(encodeFilename(thumbnail_filename)):
f446cc66	73	self.report_warning('Skipping embedding the thumbnail because the file is missing.')
c33a8639 YCH	74	return [], info
c33a8639 YCH	75
bff857a8	76	# Correct extension for WebP file with wrong extension (see #25687, #25717)
8fa43c73	77	convertor = FFmpegThumbnailsConvertorPP(self._downloader)
885cc0b7	78	convertor.fixup_webp(info, idx)
8fa43c73	79
885cc0b7	80	original_thumbnail = thumbnail_filename = info['thumbnails'][idx]['filepath']
bff857a8	81
7774db5b ES	82	# Convert unsupported thumbnail formats (see #25687, #25717)
7774db5b ES	83	# PNG is preferred since JPEG is lossy
1d485a1a	84	thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
7774db5b	85	if info['ext'] not in ('mkv', 'mka') and thumbnail_ext not in ('jpg', 'jpeg', 'png'):
a927acb1	86	thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
a927acb1	87	thumbnail_ext = 'png'
777d5a45	88
ca879745	89	mtime = os.stat(encodeFilename(filename)).st_mtime
ca879745	90
67002a5a	91	success = True
8e2915d7	92	if info['ext'] == 'mp3':
92995e62	93	options = [
f8942946	94	'-c', 'copy', '-map', '0:0', '-map', '1:0', '-write_id3v1', '1', '-id3v2_version', '3',
e02e6d86	95	'-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment=Cover (front)']
ddbed364	96
acdecdfa	97	self._report_run('ffmpeg', filename)
bb8ca1d1	98	self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
ddbed364	99
06167fbb	100	elif info['ext'] in ['mkv', 'mka']:
397235c5	101	options = list(self.stream_copy_opts())
ddbed364	102
1d485a1a	103	mimetype = f'image/{thumbnail_ext.replace("jpg", "jpeg")}'
06167fbb	104	old_stream, new_stream = self.get_stream_number(
	105	filename, ('tags', 'mimetype'), mimetype)
	106	if old_stream is not None:
	107	options.extend(['-map', '-0:%d' % old_stream])
	108	new_stream -= 1
	109	options.extend([
0f0875ed	110	'-attach', self._ffmpeg_filename_argument(thumbnail_filename),
06167fbb	111	'-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
06167fbb	112	'-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])
ddbed364	113
acdecdfa	114	self._report_run('ffmpeg', filename)
06167fbb	115	self.run_ffmpeg(filename, temp_filename, options)
06167fbb	116
8a4cd12c	117	elif info['ext'] in ['m4a', 'mp4', 'm4v', 'mov']:
e858a9d6	118	prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
acdecdfa	119	# Method 1: Use mutagen
9b8ee23b	120	if not mutagen or prefer_atomicparsley:
acdecdfa	121	success = False
	122	else:
	123	try:
	124	self._report_run('mutagen', filename)
	125	meta = MP4(filename)
	126	# NOTE: the 'covr' atom is a non-standard MPEG-4 atom,
	127	# Apple iTunes 'M4A' files include the 'moov.udta.meta.ilst' atom.
f17c7022	128	f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
acdecdfa	129	with open(thumbnail_filename, 'rb') as thumbfile:
	130	thumb_data = thumbfile.read()
	131	meta.tags['covr'] = [MP4Cover(data=thumb_data, imageformat=f)]
	132	meta.save()
	133	temp_filename = filename
	134	except Exception as err:
	135	self.report_warning('unable to embed using mutagen; %s' % error_to_compat_str(err))
	136	success = False
06167fbb	137
77cee0f1	138	# Method 2: Use AtomicParsley
	139	if not success:
	140	success = True
	141	atomicparsley = next((
b5e9a641	142	# libatomicparsley.so : See https://github.com/xibr/ytdlp-lazy/issues/1
b5e9a641	143	x for x in ['AtomicParsley', 'atomicparsley', 'libatomicparsley.so']
77cee0f1	144	if check_executable(x, ['-v'])), None)
	145	if atomicparsley is None:
	146	self.to_screen('Neither mutagen nor AtomicParsley was found. Falling back to ffmpeg')
	147	success = False
	148	else:
	149	if not prefer_atomicparsley:
	150	self.to_screen('mutagen was not found. Falling back to AtomicParsley')
	151	cmd = [encodeFilename(atomicparsley, True),
	152	encodeFilename(filename, True),
	153	encodeArgument('--artwork'),
	154	encodeFilename(thumbnail_filename, True),
	155	encodeArgument('-o'),
	156	encodeFilename(temp_filename, True)]
	157	cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]
	158
	159	self._report_run('atomicparsley', filename)
	160	self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
f0c9fb96	161	stdout, stderr, returncode = Popen.run(cmd, text=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
	162	if returncode:
	163	self.report_warning(f'Unable to embed thumbnails using AtomicParsley; {stderr.strip()}')
77cee0f1	164	# for formats that don't support thumbnails (like 3gp) AtomicParsley
77cee0f1	165	# won't create to the temporary file
f0c9fb96	166	if 'No changes' in stdout:
77cee0f1	167	self.report_warning('The file format doesn\'t support embedding a thumbnail')
	168	success = False
	169
	170	# Method 3: Use ffmpeg+ffprobe
	171	# Thumbnails attached using this method doesn't show up as cover in some cases
	172	# See https://github.com/yt-dlp/yt-dlp/issues/2125, https://github.com/yt-dlp/yt-dlp/issues/411
	173	if not success:
acdecdfa	174	success = True
acdecdfa	175	try:
397235c5	176	options = [*self.stream_copy_opts(), '-map', '1']
acdecdfa	177
	178	old_stream, new_stream = self.get_stream_number(
	179	filename, ('disposition', 'attached_pic'), 1)
	180	if old_stream is not None:
	181	options.extend(['-map', '-0:%d' % old_stream])
	182	new_stream -= 1
	183	options.extend(['-disposition:%s' % new_stream, 'attached_pic'])
	184
	185	self._report_run('ffmpeg', filename)
	186	self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
	187	except PostProcessingError as err:
06167fbb	188	success = False
77cee0f1	189	raise EmbedThumbnailPPError(f'Unable to embed using ffprobe & ffmpeg; {err}')
06167fbb	190
95131b21	191	elif info['ext'] in ['ogg', 'opus', 'flac']:
9b8ee23b	192	if not mutagen:
47ab66db	193	raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python3 -m pip install mutagen`')
06167fbb	194
acdecdfa	195	self._report_run('mutagen', filename)
95131b21	196	f = {'opus': OggOpus, 'flac': FLAC, 'ogg': OggVorbis}[info['ext']](filename)
	197
	198	pic = Picture()
	199	pic.mime = 'image/%s' % imghdr.what(thumbnail_filename)
	200	with open(thumbnail_filename, 'rb') as thumbfile:
	201	pic.data = thumbfile.read()
	202	pic.type = 3 # front cover
885cc0b7	203	res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
95131b21	204	if res is not None:
	205	pic.width, pic.height = res
	206
	207	if info['ext'] == 'flac':
	208	f.add_picture(pic)
	209	else:
	210	# https://wiki.xiph.org/VorbisComment#METADATA_BLOCK_PICTURE
	211	f['METADATA_BLOCK_PICTURE'] = base64.b64encode(pic.write()).decode('ascii')
06167fbb	212	f.save()
acdecdfa	213	temp_filename = filename
67002a5a	214
ddbed364	215	else:
8a4cd12c	216	raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/m4v/mov')
ddbed364	217
06167fbb	218	if success and temp_filename != filename:
d75201a8	219	os.replace(temp_filename, filename)
2e339f59	220
ca879745	221	self.try_utime(filename, mtime, mtime)
43d7f5a5	222	converted = original_thumbnail != thumbnail_filename
	223	self._delete_downloaded_files(
	224	thumbnail_filename if converted or not self._already_have_thumbnail else None,
	225	original_thumbnail if converted and not self._already_have_thumbnail else None,
	226	info=info)
	227	return [], info