]> jfr.im git - yt-dlp.git/blame - yt_dlp/postprocessor/embedthumbnail.py
[dependencies] Create module with all dependency imports
[yt-dlp.git] / yt_dlp / postprocessor / embedthumbnail.py
CommitLineData
95131b21 1import base64
2import imghdr
ddbed364 3import os
06167fbb 4import re
f8271158 5import subprocess
06167fbb 6
8326b00a 7from .common import PostProcessor
f8271158 8from .ffmpeg import FFmpegPostProcessor, FFmpegThumbnailsConvertorPP
9b8ee23b 9from ..dependencies import mutagen
ddbed364 10from ..utils import (
f8271158 11 Popen,
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)
20
9b8ee23b 21if 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
28class EmbedThumbnailPPError(PostProcessingError):
29 pass
30
31
31fd9c76 32class 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
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
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
a927acb1 82 # Convert unsupported thumbnail formats to PNG (see #25687, #25717)
83 # Original behavior was to convert to JPG, but since JPG is a lossy
84 # format, there will be some additional data loss.
85 # PNG, on the other hand, is lossless.
4d224a30 86 thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
15a4fd53 87 if thumbnail_ext not in ('jpg', 'jpeg', 'png'):
a927acb1 88 thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
89 thumbnail_ext = 'png'
777d5a45 90
ca879745 91 mtime = os.stat(encodeFilename(filename)).st_mtime
92
67002a5a 93 success = True
8e2915d7 94 if info['ext'] == 'mp3':
92995e62 95 options = [
f8942946 96 '-c', 'copy', '-map', '0:0', '-map', '1:0', '-write_id3v1', '1', '-id3v2_version', '3',
e51f368c 97 '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (front)"']
ddbed364 98
acdecdfa 99 self._report_run('ffmpeg', filename)
bb8ca1d1 100 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
ddbed364 101
06167fbb 102 elif info['ext'] in ['mkv', 'mka']:
397235c5 103 options = list(self.stream_copy_opts())
ddbed364 104
06167fbb 105 mimetype = 'image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg')
106 old_stream, new_stream = self.get_stream_number(
107 filename, ('tags', 'mimetype'), mimetype)
108 if old_stream is not None:
109 options.extend(['-map', '-0:%d' % old_stream])
110 new_stream -= 1
111 options.extend([
112 '-attach', thumbnail_filename,
113 '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
114 '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])
ddbed364 115
acdecdfa 116 self._report_run('ffmpeg', filename)
06167fbb 117 self.run_ffmpeg(filename, temp_filename, options)
118
119 elif info['ext'] in ['m4a', 'mp4', 'mov']:
e858a9d6 120 prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
acdecdfa 121 # Method 1: Use mutagen
9b8ee23b 122 if not mutagen or prefer_atomicparsley:
acdecdfa 123 success = False
124 else:
125 try:
126 self._report_run('mutagen', filename)
127 meta = MP4(filename)
128 # NOTE: the 'covr' atom is a non-standard MPEG-4 atom,
129 # Apple iTunes 'M4A' files include the 'moov.udta.meta.ilst' atom.
f17c7022 130 f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
acdecdfa 131 with open(thumbnail_filename, 'rb') as thumbfile:
132 thumb_data = thumbfile.read()
133 meta.tags['covr'] = [MP4Cover(data=thumb_data, imageformat=f)]
134 meta.save()
135 temp_filename = filename
136 except Exception as err:
137 self.report_warning('unable to embed using mutagen; %s' % error_to_compat_str(err))
138 success = False
06167fbb 139
77cee0f1 140 # Method 2: Use AtomicParsley
141 if not success:
142 success = True
143 atomicparsley = next((
144 x for x in ['AtomicParsley', 'atomicparsley']
145 if check_executable(x, ['-v'])), None)
146 if atomicparsley is None:
147 self.to_screen('Neither mutagen nor AtomicParsley was found. Falling back to ffmpeg')
148 success = False
149 else:
150 if not prefer_atomicparsley:
151 self.to_screen('mutagen was not found. Falling back to AtomicParsley')
152 cmd = [encodeFilename(atomicparsley, True),
153 encodeFilename(filename, True),
154 encodeArgument('--artwork'),
155 encodeFilename(thumbnail_filename, True),
156 encodeArgument('-o'),
157 encodeFilename(temp_filename, True)]
158 cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]
159
160 self._report_run('atomicparsley', filename)
161 self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
162 p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
163 stdout, stderr = p.communicate_or_kill()
164 if p.returncode != 0:
165 msg = stderr.decode('utf-8', 'replace').strip()
166 self.report_warning(f'Unable to embed thumbnails using AtomicParsley; {msg}')
167 # for formats that don't support thumbnails (like 3gp) AtomicParsley
168 # won't create to the temporary file
169 if b'No changes' in stdout:
170 self.report_warning('The file format doesn\'t support embedding a thumbnail')
171 success = False
172
173 # Method 3: Use ffmpeg+ffprobe
174 # Thumbnails attached using this method doesn't show up as cover in some cases
175 # See https://github.com/yt-dlp/yt-dlp/issues/2125, https://github.com/yt-dlp/yt-dlp/issues/411
176 if not success:
acdecdfa 177 success = True
178 try:
397235c5 179 options = [*self.stream_copy_opts(), '-map', '1']
acdecdfa 180
181 old_stream, new_stream = self.get_stream_number(
182 filename, ('disposition', 'attached_pic'), 1)
183 if old_stream is not None:
184 options.extend(['-map', '-0:%d' % old_stream])
185 new_stream -= 1
186 options.extend(['-disposition:%s' % new_stream, 'attached_pic'])
187
188 self._report_run('ffmpeg', filename)
189 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
190 except PostProcessingError as err:
06167fbb 191 success = False
77cee0f1 192 raise EmbedThumbnailPPError(f'Unable to embed using ffprobe & ffmpeg; {err}')
06167fbb 193
95131b21 194 elif info['ext'] in ['ogg', 'opus', 'flac']:
9b8ee23b 195 if not mutagen:
e38df8f9 196 raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python -m pip install mutagen`')
06167fbb 197
acdecdfa 198 self._report_run('mutagen', filename)
95131b21 199 f = {'opus': OggOpus, 'flac': FLAC, 'ogg': OggVorbis}[info['ext']](filename)
200
201 pic = Picture()
202 pic.mime = 'image/%s' % imghdr.what(thumbnail_filename)
203 with open(thumbnail_filename, 'rb') as thumbfile:
204 pic.data = thumbfile.read()
205 pic.type = 3 # front cover
885cc0b7 206 res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
95131b21 207 if res is not None:
208 pic.width, pic.height = res
209
210 if info['ext'] == 'flac':
211 f.add_picture(pic)
212 else:
213 # https://wiki.xiph.org/VorbisComment#METADATA_BLOCK_PICTURE
214 f['METADATA_BLOCK_PICTURE'] = base64.b64encode(pic.write()).decode('ascii')
06167fbb 215 f.save()
acdecdfa 216 temp_filename = filename
67002a5a 217
ddbed364 218 else:
95131b21 219 raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/mov')
ddbed364 220
06167fbb 221 if success and temp_filename != filename:
d75201a8 222 os.replace(temp_filename, filename)
2e339f59 223
ca879745 224 self.try_utime(filename, mtime, mtime)
225
2e339f59 226 files_to_delete = [thumbnail_filename]
de6000d9 227 if self._already_have_thumbnail:
2e339f59 228 if original_thumbnail == thumbnail_filename:
229 files_to_delete = []
0e004051 230 elif original_thumbnail != thumbnail_filename:
231 files_to_delete.append(original_thumbnail)
67002a5a 232 return files_to_delete, info