yt_dlp/postprocessor/embedthumbnail.py

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