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     error_to_compat_str,
  17     prepend_extension,
  18     shell_quote,
  19 )
  20
  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
  27
  28 class EmbedThumbnailPPError(PostProcessingError):
  29     pass
  30
  31
  32 class EmbedThumbnailPP(FFmpegPostProcessor):
  33
  34     def __init__(self, downloader=None, already_have_thumbnail=False):
  35         FFmpegPostProcessor.__init__(self, downloader)
  36         self._already_have_thumbnail = already_have_thumbnail
  37
  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*[,\[]'
  46             size_result = self.run_ffmpeg(filename, None, ['-hide_banner'], expected_retcodes=(1,))
  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
  55     def _report_run(self, exe, filename):
  56         self.to_screen(f'{exe}: Adding thumbnail to "{filename}"')
  57
  58     @PostProcessor._restrict_to(images=False)
  59     def run(self, info):
  60         filename = info['filepath']
  61         temp_filename = prepend_extension(filename, 'temp')
  62
  63         if not info.get('thumbnails'):
  64             self.to_screen('There aren\'t any thumbnails to embed')
  65             return [], info
  66
  67         idx = next((-i for i, t in enumerate(info['thumbnails'][::-1], 1) if t.get('filepath')), None)
  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']
  72         if not os.path.exists(encodeFilename(thumbnail_filename)):
  73             self.report_warning('Skipping embedding the thumbnail because the file is missing.')
  74             return [], info
  75
  76         # Correct extension for WebP file with wrong extension (see #25687, #25717)
  77         convertor = FFmpegThumbnailsConvertorPP(self._downloader)
  78         convertor.fixup_webp(info, idx)
  79
  80         original_thumbnail = thumbnail_filename = info['thumbnails'][idx]['filepath']
  81
  82         # Convert unsupported thumbnail formats (see #25687, #25717)
  83         # PNG is preferred since JPEG is lossy
  84         thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
  85         if info['ext'] not in ('mkv', 'mka') and thumbnail_ext not in ('jpg', 'jpeg', 'png'):
  86             thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
  87             thumbnail_ext = 'png'
  88
  89         mtime = os.stat(encodeFilename(filename)).st_mtime
  90
  91         success = True
  92         if info['ext'] == 'mp3':
  93             options = [
  94                 '-c', 'copy', '-map', '0:0', '-map', '1:0', '-write_id3v1', '1', '-id3v2_version', '3',
  95                 '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (front)"']
  96
  97             self._report_run('ffmpeg', filename)
  98             self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
  99
 100         elif info['ext'] in ['mkv', 'mka']:
 101             options = list(self.stream_copy_opts())
 102
 103             mimetype = f'image/{thumbnail_ext.replace("jpg", "jpeg")}'
 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([
 110                 '-attach', thumbnail_filename,
 111                 '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
 112                 '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])
 113
 114             self._report_run('ffmpeg', filename)
 115             self.run_ffmpeg(filename, temp_filename, options)
 116
 117         elif info['ext'] in ['m4a', 'mp4', 'mov']:
 118             prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
 119             # Method 1: Use mutagen
 120             if not mutagen or prefer_atomicparsley:
 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.
 128                     f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
 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
 137
 138             # Method 2: Use AtomicParsley
 139             if not success:
 140                 success = True
 141                 atomicparsley = next((
 142                     x for x in ['AtomicParsley', 'atomicparsley']
 143                     if check_executable(x, ['-v'])), None)
 144                 if atomicparsley is None:
 145                     self.to_screen('Neither mutagen nor AtomicParsley was found. Falling back to ffmpeg')
 146                     success = False
 147                 else:
 148                     if not prefer_atomicparsley:
 149                         self.to_screen('mutagen was not found. Falling back to AtomicParsley')
 150                     cmd = [encodeFilename(atomicparsley, True),
 151                            encodeFilename(filename, True),
 152                            encodeArgument('--artwork'),
 153                            encodeFilename(thumbnail_filename, True),
 154                            encodeArgument('-o'),
 155                            encodeFilename(temp_filename, True)]
 156                     cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]
 157
 158                     self._report_run('atomicparsley', filename)
 159                     self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
 160                     p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 161                     stdout, stderr = p.communicate_or_kill()
 162                     if p.returncode != 0:
 163                         msg = stderr.decode('utf-8', 'replace').strip()
 164                         self.report_warning(f'Unable to embed thumbnails using AtomicParsley; {msg}')
 165                     # for formats that don't support thumbnails (like 3gp) AtomicParsley
 166                     # won't create to the temporary file
 167                     if b'No changes' in stdout:
 168                         self.report_warning('The file format doesn\'t support embedding a thumbnail')
 169                         success = False
 170
 171             # Method 3: Use ffmpeg+ffprobe
 172             # Thumbnails attached using this method doesn't show up as cover in some cases
 173             # See https://github.com/yt-dlp/yt-dlp/issues/2125, https://github.com/yt-dlp/yt-dlp/issues/411
 174             if not success:
 175                 success = True
 176                 try:
 177                     options = [*self.stream_copy_opts(), '-map', '1']
 178
 179                     old_stream, new_stream = self.get_stream_number(
 180                         filename, ('disposition', 'attached_pic'), 1)
 181                     if old_stream is not None:
 182                         options.extend(['-map', '-0:%d' % old_stream])
 183                         new_stream -= 1
 184                     options.extend(['-disposition:%s' % new_stream, 'attached_pic'])
 185
 186                     self._report_run('ffmpeg', filename)
 187                     self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
 188                 except PostProcessingError as err:
 189                     success = False
 190                     raise EmbedThumbnailPPError(f'Unable to embed using ffprobe & ffmpeg; {err}')
 191
 192         elif info['ext'] in ['ogg', 'opus', 'flac']:
 193             if not mutagen:
 194                 raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python -m pip install mutagen`')
 195
 196             self._report_run('mutagen', filename)
 197             f = {'opus': OggOpus, 'flac': FLAC, 'ogg': OggVorbis}[info['ext']](filename)
 198
 199             pic = Picture()
 200             pic.mime = 'image/%s' % imghdr.what(thumbnail_filename)
 201             with open(thumbnail_filename, 'rb') as thumbfile:
 202                 pic.data = thumbfile.read()
 203             pic.type = 3  # front cover
 204             res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
 205             if res is not None:
 206                 pic.width, pic.height = res
 207
 208             if info['ext'] == 'flac':
 209                 f.add_picture(pic)
 210             else:
 211                 # https://wiki.xiph.org/VorbisComment#METADATA_BLOCK_PICTURE
 212                 f['METADATA_BLOCK_PICTURE'] = base64.b64encode(pic.write()).decode('ascii')
 213             f.save()
 214             temp_filename = filename
 215
 216         else:
 217             raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/mov')
 218
 219         if success and temp_filename != filename:
 220             os.replace(temp_filename, filename)
 221
 222         self.try_utime(filename, mtime, mtime)
 223         converted = original_thumbnail != thumbnail_filename
 224         self._delete_downloaded_files(
 225             thumbnail_filename if converted or not self._already_have_thumbnail else None,
 226             original_thumbnail if converted and not self._already_have_thumbnail else None,
 227             info=info)
 228         return [], info