]> jfr.im git - yt-dlp.git/blame - yt_dlp/postprocessor/embedthumbnail.py
Use `os.replace` where applicable (#793)
[yt-dlp.git] / yt_dlp / postprocessor / embedthumbnail.py
CommitLineData
dcdb292f 1# coding: utf-8
ddbed364 2from __future__ import unicode_literals
3
95131b21 4import base64
5import imghdr
ddbed364 6import os
95131b21 7import subprocess
06167fbb 8import re
06167fbb 9
10try:
95131b21 11 from mutagen.flac import Picture, FLAC
acdecdfa 12 from mutagen.mp4 import MP4, MP4Cover
13 from mutagen.oggopus import OggOpus
14 from mutagen.oggvorbis import OggVorbis
06ff212d 15 has_mutagen = True
06167fbb 16except ImportError:
06ff212d 17 has_mutagen = False
ddbed364 18
8326b00a 19from .common import PostProcessor
8fa43c73 20from .ffmpeg import (
21 FFmpegPostProcessor,
22 FFmpegThumbnailsConvertorPP,
23)
ddbed364 24from ..utils import (
25 check_executable,
2cc6d135 26 encodeArgument,
ddbed364 27 encodeFilename,
06167fbb 28 error_to_compat_str,
ddbed364 29 PostProcessingError,
30 prepend_extension,
06167fbb 31 process_communicate_or_kill,
f5b1bca9 32 shell_quote,
ddbed364 33)
34
35
36class EmbedThumbnailPPError(PostProcessingError):
37 pass
38
39
31fd9c76 40class EmbedThumbnailPP(FFmpegPostProcessor):
1b77b347 41
8e595397 42 def __init__(self, downloader=None, already_have_thumbnail=False):
8fa43c73 43 FFmpegPostProcessor.__init__(self, downloader)
8e595397
YCH
44 self._already_have_thumbnail = already_have_thumbnail
45
95131b21 46 def _get_thumbnail_resolution(self, filename, thumbnail_dict):
47 def guess():
48 width, height = thumbnail_dict.get('width'), thumbnail_dict.get('height')
49 if width and height:
50 return width, height
51
52 try:
53 size_regex = r',\s*(?P<w>\d+)x(?P<h>\d+)\s*[,\[]'
00034c14 54 size_result = self.run_ffmpeg(filename, None, ['-hide_banner'], expected_retcodes=(1,))
95131b21 55 mobj = re.search(size_regex, size_result)
56 if mobj is None:
57 return guess()
58 except PostProcessingError as err:
59 self.report_warning('unable to find the thumbnail resolution; %s' % error_to_compat_str(err))
60 return guess()
61 return int(mobj.group('w')), int(mobj.group('h'))
62
acdecdfa 63 def _report_run(self, exe, filename):
64 self.to_screen('%s: Adding thumbnail to "%s"' % (exe, filename))
65
8326b00a 66 @PostProcessor._restrict_to(images=False)
ddbed364 67 def run(self, info):
68 filename = info['filepath']
69 temp_filename = prepend_extension(filename, 'temp')
ddbed364 70
8e595397 71 if not info.get('thumbnails'):
1b77b347 72 self.to_screen('There aren\'t any thumbnails to embed')
b5cbe3d6 73 return [], info
ddbed364 74
337e0c62 75 idx = next((-i for i, t in enumerate(info['thumbnails'][::-1], 1) if t.get('filepath')), None)
885cc0b7 76 if idx is None:
77 self.to_screen('There are no thumbnails on disk')
78 return [], info
79 thumbnail_filename = info['thumbnails'][idx]['filepath']
c33a8639 80 if not os.path.exists(encodeFilename(thumbnail_filename)):
f446cc66 81 self.report_warning('Skipping embedding the thumbnail because the file is missing.')
c33a8639
YCH
82 return [], info
83
bff857a8 84 # Correct extension for WebP file with wrong extension (see #25687, #25717)
8fa43c73 85 convertor = FFmpegThumbnailsConvertorPP(self._downloader)
885cc0b7 86 convertor.fixup_webp(info, idx)
8fa43c73 87
885cc0b7 88 original_thumbnail = thumbnail_filename = info['thumbnails'][idx]['filepath']
bff857a8 89
a927acb1 90 # Convert unsupported thumbnail formats to PNG (see #25687, #25717)
91 # Original behavior was to convert to JPG, but since JPG is a lossy
92 # format, there will be some additional data loss.
93 # PNG, on the other hand, is lossless.
4d224a30 94 thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
15a4fd53 95 if thumbnail_ext not in ('jpg', 'jpeg', 'png'):
a927acb1 96 thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
97 thumbnail_ext = 'png'
777d5a45 98
ca879745 99 mtime = os.stat(encodeFilename(filename)).st_mtime
100
67002a5a 101 success = True
8e2915d7 102 if info['ext'] == 'mp3':
92995e62 103 options = [
c76eb41b 104 '-c', 'copy', '-map', '0:0', '-map', '1:0', '-id3v2_version', '3',
e51f368c 105 '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (front)"']
ddbed364 106
acdecdfa 107 self._report_run('ffmpeg', filename)
bb8ca1d1 108 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
ddbed364 109
06167fbb 110 elif info['ext'] in ['mkv', 'mka']:
111 options = ['-c', 'copy', '-map', '0', '-dn']
ddbed364 112
06167fbb 113 mimetype = 'image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg')
114 old_stream, new_stream = self.get_stream_number(
115 filename, ('tags', 'mimetype'), mimetype)
116 if old_stream is not None:
117 options.extend(['-map', '-0:%d' % old_stream])
118 new_stream -= 1
119 options.extend([
120 '-attach', thumbnail_filename,
121 '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
122 '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])
ddbed364 123
acdecdfa 124 self._report_run('ffmpeg', filename)
06167fbb 125 self.run_ffmpeg(filename, temp_filename, options)
126
127 elif info['ext'] in ['m4a', 'mp4', 'mov']:
e858a9d6 128 prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
acdecdfa 129 # Method 1: Use mutagen
e858a9d6 130 if not has_mutagen or prefer_atomicparsley:
acdecdfa 131 success = False
132 else:
133 try:
134 self._report_run('mutagen', filename)
135 meta = MP4(filename)
136 # NOTE: the 'covr' atom is a non-standard MPEG-4 atom,
137 # Apple iTunes 'M4A' files include the 'moov.udta.meta.ilst' atom.
f17c7022 138 f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
acdecdfa 139 with open(thumbnail_filename, 'rb') as thumbfile:
140 thumb_data = thumbfile.read()
141 meta.tags['covr'] = [MP4Cover(data=thumb_data, imageformat=f)]
142 meta.save()
143 temp_filename = filename
144 except Exception as err:
145 self.report_warning('unable to embed using mutagen; %s' % error_to_compat_str(err))
146 success = False
06167fbb 147
acdecdfa 148 # Method 2: Use ffmpeg+ffprobe
e858a9d6 149 if not success and not prefer_atomicparsley:
acdecdfa 150 success = True
151 try:
152 options = ['-c', 'copy', '-map', '0', '-dn', '-map', '1']
153
154 old_stream, new_stream = self.get_stream_number(
155 filename, ('disposition', 'attached_pic'), 1)
156 if old_stream is not None:
157 options.extend(['-map', '-0:%d' % old_stream])
158 new_stream -= 1
159 options.extend(['-disposition:%s' % new_stream, 'attached_pic'])
160
161 self._report_run('ffmpeg', filename)
162 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
163 except PostProcessingError as err:
164 self.report_warning('unable to embed using ffprobe & ffmpeg; %s' % error_to_compat_str(err))
165 success = False
06167fbb 166
acdecdfa 167 # Method 3: Use AtomicParsley
168 if not success:
169 success = True
bc2ca1bb 170 atomicparsley = next((
171 x for x in ['AtomicParsley', 'atomicparsley']
172 if check_executable(x, ['-v'])), None)
173 if atomicparsley is None:
beb4b92a 174 raise EmbedThumbnailPPError('AtomicParsley was not found. Please install')
06167fbb 175
bc2ca1bb 176 cmd = [encodeFilename(atomicparsley, True),
06167fbb 177 encodeFilename(filename, True),
178 encodeArgument('--artwork'),
179 encodeFilename(thumbnail_filename, True),
180 encodeArgument('-o'),
181 encodeFilename(temp_filename, True)]
e92caff5 182 cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]
06167fbb 183
acdecdfa 184 self._report_run('atomicparsley', filename)
06167fbb 185 self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
186 p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
187 stdout, stderr = process_communicate_or_kill(p)
188 if p.returncode != 0:
189 msg = stderr.decode('utf-8', 'replace').strip()
190 raise EmbedThumbnailPPError(msg)
191 # for formats that don't support thumbnails (like 3gp) AtomicParsley
192 # won't create to the temporary file
193 if b'No changes' in stdout:
194 self.report_warning('The file format doesn\'t support embedding a thumbnail')
195 success = False
196
95131b21 197 elif info['ext'] in ['ogg', 'opus', 'flac']:
06ff212d 198 if not has_mutagen:
e38df8f9 199 raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python -m pip install mutagen`')
06167fbb 200
acdecdfa 201 self._report_run('mutagen', filename)
95131b21 202 f = {'opus': OggOpus, 'flac': FLAC, 'ogg': OggVorbis}[info['ext']](filename)
203
204 pic = Picture()
205 pic.mime = 'image/%s' % imghdr.what(thumbnail_filename)
206 with open(thumbnail_filename, 'rb') as thumbfile:
207 pic.data = thumbfile.read()
208 pic.type = 3 # front cover
885cc0b7 209 res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
95131b21 210 if res is not None:
211 pic.width, pic.height = res
212
213 if info['ext'] == 'flac':
214 f.add_picture(pic)
215 else:
216 # https://wiki.xiph.org/VorbisComment#METADATA_BLOCK_PICTURE
217 f['METADATA_BLOCK_PICTURE'] = base64.b64encode(pic.write()).decode('ascii')
06167fbb 218 f.save()
acdecdfa 219 temp_filename = filename
67002a5a 220
ddbed364 221 else:
95131b21 222 raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/mov')
ddbed364 223
06167fbb 224 if success and temp_filename != filename:
d75201a8 225 os.replace(temp_filename, filename)
2e339f59 226
ca879745 227 self.try_utime(filename, mtime, mtime)
228
2e339f59 229 files_to_delete = [thumbnail_filename]
de6000d9 230 if self._already_have_thumbnail:
2e339f59 231 if original_thumbnail == thumbnail_filename:
232 files_to_delete = []
0e004051 233 elif original_thumbnail != thumbnail_filename:
234 files_to_delete.append(original_thumbnail)
67002a5a 235 return files_to_delete, info