]> jfr.im git - yt-dlp.git/blob - yt_dlp/postprocessor/embedthumbnail.py
[dash,youtube] Download live from start to end (#888)
[yt-dlp.git] / yt_dlp / postprocessor / embedthumbnail.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import base64
5 import imghdr
6 import os
7 import subprocess
8 import re
9
10 try:
11 from mutagen.flac import Picture, FLAC
12 from mutagen.mp4 import MP4, MP4Cover
13 from mutagen.oggopus import OggOpus
14 from mutagen.oggvorbis import OggVorbis
15 has_mutagen = True
16 except ImportError:
17 has_mutagen = False
18
19 from .common import PostProcessor
20 from .ffmpeg import (
21 FFmpegPostProcessor,
22 FFmpegThumbnailsConvertorPP,
23 )
24 from ..utils import (
25 check_executable,
26 encodeArgument,
27 encodeFilename,
28 error_to_compat_str,
29 Popen,
30 PostProcessingError,
31 prepend_extension,
32 shell_quote,
33 )
34
35
36 class EmbedThumbnailPPError(PostProcessingError):
37 pass
38
39
40 class EmbedThumbnailPP(FFmpegPostProcessor):
41
42 def __init__(self, downloader=None, already_have_thumbnail=False):
43 FFmpegPostProcessor.__init__(self, downloader)
44 self._already_have_thumbnail = already_have_thumbnail
45
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*[,\[]'
54 size_result = self.run_ffmpeg(filename, None, ['-hide_banner'], expected_retcodes=(1,))
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
63 def _report_run(self, exe, filename):
64 self.to_screen('%s: Adding thumbnail to "%s"' % (exe, filename))
65
66 @PostProcessor._restrict_to(images=False)
67 def run(self, info):
68 filename = info['filepath']
69 temp_filename = prepend_extension(filename, 'temp')
70
71 if not info.get('thumbnails'):
72 self.to_screen('There aren\'t any thumbnails to embed')
73 return [], info
74
75 idx = next((-i for i, t in enumerate(info['thumbnails'][::-1], 1) if t.get('filepath')), None)
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']
80 if not os.path.exists(encodeFilename(thumbnail_filename)):
81 self.report_warning('Skipping embedding the thumbnail because the file is missing.')
82 return [], info
83
84 # Correct extension for WebP file with wrong extension (see #25687, #25717)
85 convertor = FFmpegThumbnailsConvertorPP(self._downloader)
86 convertor.fixup_webp(info, idx)
87
88 original_thumbnail = thumbnail_filename = info['thumbnails'][idx]['filepath']
89
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.
94 thumbnail_ext = os.path.splitext(thumbnail_filename)[1][1:]
95 if thumbnail_ext not in ('jpg', 'jpeg', 'png'):
96 thumbnail_filename = convertor.convert_thumbnail(thumbnail_filename, 'png')
97 thumbnail_ext = 'png'
98
99 mtime = os.stat(encodeFilename(filename)).st_mtime
100
101 success = True
102 if info['ext'] == 'mp3':
103 options = [
104 '-c', 'copy', '-map', '0:0', '-map', '1:0', '-id3v2_version', '3',
105 '-metadata:s:v', 'title="Album cover"', '-metadata:s:v', 'comment="Cover (front)"']
106
107 self._report_run('ffmpeg', filename)
108 self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options)
109
110 elif info['ext'] in ['mkv', 'mka']:
111 options = ['-c', 'copy', '-map', '0', '-dn']
112
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])
123
124 self._report_run('ffmpeg', filename)
125 self.run_ffmpeg(filename, temp_filename, options)
126
127 elif info['ext'] in ['m4a', 'mp4', 'mov']:
128 prefer_atomicparsley = 'embed-thumbnail-atomicparsley' in self.get_param('compat_opts', [])
129 # Method 1: Use mutagen
130 if not has_mutagen or prefer_atomicparsley:
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.
138 f = {'jpeg': MP4Cover.FORMAT_JPEG, 'png': MP4Cover.FORMAT_PNG}[imghdr.what(thumbnail_filename)]
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
147
148 # Method 2: Use ffmpeg+ffprobe
149 if not success and not prefer_atomicparsley:
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
166
167 # Method 3: Use AtomicParsley
168 if not success:
169 success = True
170 atomicparsley = next((
171 x for x in ['AtomicParsley', 'atomicparsley']
172 if check_executable(x, ['-v'])), None)
173 if atomicparsley is None:
174 raise EmbedThumbnailPPError('AtomicParsley was not found. Please install')
175
176 cmd = [encodeFilename(atomicparsley, True),
177 encodeFilename(filename, True),
178 encodeArgument('--artwork'),
179 encodeFilename(thumbnail_filename, True),
180 encodeArgument('-o'),
181 encodeFilename(temp_filename, True)]
182 cmd += [encodeArgument(o) for o in self._configuration_args('AtomicParsley')]
183
184 self._report_run('atomicparsley', filename)
185 self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd))
186 p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
187 stdout, stderr = p.communicate_or_kill()
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
197 elif info['ext'] in ['ogg', 'opus', 'flac']:
198 if not has_mutagen:
199 raise EmbedThumbnailPPError('module mutagen was not found. Please install using `python -m pip install mutagen`')
200
201 self._report_run('mutagen', filename)
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
209 res = self._get_thumbnail_resolution(thumbnail_filename, info['thumbnails'][idx])
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')
218 f.save()
219 temp_filename = filename
220
221 else:
222 raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus/flac, m4a/mp4/mov')
223
224 if success and temp_filename != filename:
225 os.replace(temp_filename, filename)
226
227 self.try_utime(filename, mtime, mtime)
228
229 files_to_delete = [thumbnail_filename]
230 if self._already_have_thumbnail:
231 if original_thumbnail == thumbnail_filename:
232 files_to_delete = []
233 elif original_thumbnail != thumbnail_filename:
234 files_to_delete.append(original_thumbnail)
235 return files_to_delete, info