youtube_dlc/postprocessor/embedthumbnail.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 import os
   6 import subprocess
   7 import struct
   8 import re
   9 import base64
  10
  11 try:
  12     import mutagen
  13     _has_mutagen = True
  14 except ImportError:
  15     _has_mutagen = False
  16
  17 from .ffmpeg import FFmpegPostProcessor
  18
  19 from ..utils import (
  20     check_executable,
  21     encodeArgument,
  22     encodeFilename,
  23     error_to_compat_str,
  24     PostProcessingError,
  25     prepend_extension,
  26     process_communicate_or_kill,
  27     replace_extension,
  28     shell_quote,
  29 )
  30
  31
  32 class EmbedThumbnailPPError(PostProcessingError):
  33     pass
  34
  35
  36 class EmbedThumbnailPP(FFmpegPostProcessor):
  37
  38     def __init__(self, downloader=None, already_have_thumbnail=False):
  39         super(EmbedThumbnailPP, self).__init__(downloader)
  40         self._already_have_thumbnail = already_have_thumbnail
  41
  42     def run(self, info):
  43         filename = info['filepath']
  44         temp_filename = prepend_extension(filename, 'temp')
  45         files_to_delete = []
  46
  47         if not info.get('thumbnails'):
  48             self.to_screen('There aren\'t any thumbnails to embed')
  49             return [], info
  50
  51         thumbnail_filename = info['thumbnails'][-1]['filename']
  52
  53         if not os.path.exists(encodeFilename(thumbnail_filename)):
  54             self.report_warning('Skipping embedding the thumbnail because the file is missing.')
  55             return [], info
  56
  57         def is_webp(path):
  58             with open(encodeFilename(path), 'rb') as f:
  59                 b = f.read(12)
  60             return b[0:4] == b'RIFF' and b[8:] == b'WEBP'
  61
  62         # Correct extension for WebP file with wrong extension (see #25687, #25717)
  63         _, thumbnail_ext = os.path.splitext(thumbnail_filename)
  64         if thumbnail_ext:
  65             thumbnail_ext = thumbnail_ext[1:].lower()
  66             if thumbnail_ext != 'webp' and is_webp(thumbnail_filename):
  67                 self.to_screen('Correcting extension to webp and escaping path for thumbnail "%s"' % thumbnail_filename)
  68                 thumbnail_webp_filename = replace_extension(thumbnail_filename, 'webp')
  69                 os.rename(encodeFilename(thumbnail_filename), encodeFilename(thumbnail_webp_filename))
  70                 thumbnail_filename = thumbnail_webp_filename
  71                 thumbnail_ext = 'webp'
  72
  73         # Convert unsupported thumbnail formats to JPEG (see #25687, #25717)
  74         if thumbnail_ext not in ['jpg', 'png']:
  75             # NB: % is supposed to be escaped with %% but this does not work
  76             # for input files so working around with standard substitution
  77             escaped_thumbnail_filename = thumbnail_filename.replace('%', '#')
  78             os.rename(encodeFilename(thumbnail_filename), encodeFilename(escaped_thumbnail_filename))
  79             escaped_thumbnail_jpg_filename = replace_extension(escaped_thumbnail_filename, 'jpg')
  80             self.to_screen('Converting thumbnail "%s" to JPEG' % escaped_thumbnail_filename)
  81             self.run_ffmpeg(escaped_thumbnail_filename, escaped_thumbnail_jpg_filename, ['-bsf:v', 'mjpeg2jpeg'])
  82             files_to_delete.append(escaped_thumbnail_filename)
  83             thumbnail_jpg_filename = replace_extension(thumbnail_filename, 'jpg')
  84             # Rename back to unescaped for further processing
  85             os.rename(encodeFilename(escaped_thumbnail_jpg_filename), encodeFilename(thumbnail_jpg_filename))
  86             thumbnail_filename = thumbnail_jpg_filename
  87             thumbnail_ext = 'jpg'
  88
  89         success = True
  90         if info['ext'] == 'mp3':
  91             options = [
  92                 '-c', 'copy', '-map', '0:0', '-map', '1:0', '-id3v2_version', '3',
  93                 '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (front)"']
  94
  95             self.to_screen('Adding thumbnail to "%s"' % filename)
  96             self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
  97
  98         elif info['ext'] in ['mkv', 'mka']:
  99             options = ['-c', 'copy', '-map', '0', '-dn']
 100
 101             mimetype = 'image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg')
 102             old_stream, new_stream = self.get_stream_number(
 103                 filename, ('tags', 'mimetype'), mimetype)
 104             if old_stream is not None:
 105                 options.extend(['-map', '-0:%d' % old_stream])
 106                 new_stream -= 1
 107             options.extend([
 108                 '-attach', thumbnail_filename,
 109                 '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype,
 110                 '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext])
 111
 112             self.to_screen('Adding thumbnail to "%s"' % filename)
 113             self.run_ffmpeg(filename, temp_filename, options)
 114
 115         elif info['ext'] in ['m4a', 'mp4', 'mov']:
 116             try:
 117                 options = ['-c', 'copy', '-map', '0', '-dn', '-map', '1']
 118
 119                 old_stream, new_stream = self.get_stream_number(
 120                     filename, ('disposition', 'attached_pic'), 1)
 121                 if old_stream is not None:
 122                     options.extend(['-map', '-0:%d' % old_stream])
 123                     new_stream -= 1
 124                 options.extend(['-disposition:%s' % new_stream, 'attached_pic'])
 125
 126                 self.to_screen('Adding thumbnail to "%s"' % filename)
 127                 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
 128
 129             except PostProcessingError as err:
 130                 self.report_warning('unable to embed using ffprobe & ffmpeg; %s' % error_to_compat_str(err))
 131                 if not check_executable('AtomicParsley', ['-v']):
 132                     raise EmbedThumbnailPPError('AtomicParsley was not found. Please install.')
 133
 134                 cmd = [encodeFilename('AtomicParsley', True),
 135                        encodeFilename(filename, True),
 136                        encodeArgument('--artwork'),
 137                        encodeFilename(thumbnail_filename, True),
 138                        encodeArgument('-o'),
 139                        encodeFilename(temp_filename, True)]
 140                 cmd += [encodeArgument(o) for o in self._configuration_args(exe='AtomicParsley')]
 141
 142                 self.to_screen('Adding thumbnail to "%s"' % filename)
 143                 self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
 144                 p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
 145                 stdout, stderr = process_communicate_or_kill(p)
 146                 if p.returncode != 0:
 147                     msg = stderr.decode('utf-8', 'replace').strip()
 148                     raise EmbedThumbnailPPError(msg)
 149                 # for formats that don't support thumbnails (like 3gp) AtomicParsley
 150                 # won't create to the temporary file
 151                 if b'No changes' in stdout:
 152                     self.report_warning('The file format doesn\'t support embedding a thumbnail')
 153                     success = False
 154
 155         elif info['ext'] in ['ogg', 'opus']:
 156             if not _has_mutagen:
 157                 raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python -m pip install mutagen`')
 158             self.to_screen('Adding thumbnail to "%s"' % filename)
 159
 160             size_regex = r',\s*(?P<w>\d+)x(?P<h>\d+)\s*[,\[]'
 161             size_result = self.run_ffmpeg(thumbnail_filename, thumbnail_filename, ['-hide_banner'])
 162             mobj = re.search(size_regex, size_result)
 163             width, height = int(mobj.group('w')), int(mobj.group('h'))
 164             mimetype = ('image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg')).encode('ascii')
 165
 166             # https://xiph.org/flac/format.html#metadata_block_picture
 167             data = bytearray()
 168             data += struct.pack('>II', 3, len(mimetype))
 169             data += mimetype
 170             data += struct.pack('>IIIIII', 0, width, height, 8, 0, os.stat(thumbnail_filename).st_size)  # 32 if png else 24
 171
 172             fin = open(thumbnail_filename, "rb")
 173             data += fin.read()
 174             fin.close()
 175
 176             temp_filename = filename
 177             f = mutagen.File(temp_filename)
 178             f.tags['METADATA_BLOCK_PICTURE'] = base64.b64encode(data).decode('ascii')
 179             f.save()
 180
 181         else:
 182             raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus, m4a/mp4/mov')
 183
 184         if success and temp_filename != filename:
 185             os.remove(encodeFilename(filename))
 186             os.rename(encodeFilename(temp_filename), encodeFilename(filename))
 187         if self._already_have_thumbnail:
 188             info['__files_to_move'][thumbnail_filename] = replace_extension(
 189                 info['__thumbnail_filename'], os.path.splitext(thumbnail_filename)[1][1:])
 190         else:
 191             files_to_delete.append(thumbnail_filename)
 192         return files_to_delete, info