diff options
Diffstat (limited to 'youtube_dlc/postprocessor')
| -rw-r--r-- | youtube_dlc/postprocessor/embedthumbnail.py | 126 | ||||
| -rw-r--r-- | youtube_dlc/postprocessor/ffmpeg.py | 98 |
2 files changed, 173 insertions, 51 deletions
diff --git a/youtube_dlc/postprocessor/embedthumbnail.py b/youtube_dlc/postprocessor/embedthumbnail.py index 98a3531f1..bad005cca 100644 --- a/youtube_dlc/postprocessor/embedthumbnail.py +++ b/youtube_dlc/postprocessor/embedthumbnail.py @@ -4,6 +4,15 @@ from __future__ import unicode_literals import os import subprocess +import struct +import re +import base64 + +try: + import mutagen + _has_mutagen = True +except ImportError: + _has_mutagen = False from .ffmpeg import FFmpegPostProcessor @@ -11,11 +20,12 @@ from ..utils import ( check_executable, encodeArgument, encodeFilename, + error_to_compat_str, PostProcessingError, prepend_extension, + process_communicate_or_kill, replace_extension, shell_quote, - process_communicate_or_kill, ) @@ -73,6 +83,7 @@ class EmbedThumbnailPP(FFmpegPostProcessor): # Rename back to unescaped for further processing os.rename(encodeFilename(escaped_thumbnail_jpg_filename), encodeFilename(thumbnail_jpg_filename)) thumbnail_filename = thumbnail_jpg_filename + thumbnail_ext = 'jpg' success = True if info['ext'] == 'mp3': @@ -83,47 +94,92 @@ class EmbedThumbnailPP(FFmpegPostProcessor): self.to_screen('Adding thumbnail to "%s"' % filename) self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options) - elif info['ext'] == 'mkv': - options = [ - '-c', 'copy', '-map', '0', '-dn', '-attach', thumbnail_filename, - '-metadata:s:t', 'mimetype=image/jpeg', '-metadata:s:t', 'filename=cover.jpg'] - - self.to_screen('Adding thumbnail to "%s"' % filename) - self.run_ffmpeg_multiple_files([filename], temp_filename, options) - - elif info['ext'] in ['m4a', 'mp4']: - if not check_executable('AtomicParsley', ['-v']): - raise EmbedThumbnailPPError('AtomicParsley was not found. Please install.') + elif info['ext'] in ['mkv', 'mka']: + options = ['-c', 'copy', '-map', '0', '-dn'] - cmd = [encodeFilename('AtomicParsley', True), - encodeFilename(filename, True), - encodeArgument('--artwork'), - encodeFilename(thumbnail_filename, True), - encodeArgument('-o'), - encodeFilename(temp_filename, True)] - cmd += [encodeArgument(o) for o in self._configuration_args(exe='AtomicParsley')] + mimetype = 'image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg') + old_stream, new_stream = self.get_stream_number( + filename, ('tags', 'mimetype'), mimetype) + if old_stream is not None: + options.extend(['-map', '-0:%d' % old_stream]) + new_stream -= 1 + options.extend([ + '-attach', thumbnail_filename, + '-metadata:s:%d' % new_stream, 'mimetype=%s' % mimetype, + '-metadata:s:%d' % new_stream, 'filename=cover.%s' % thumbnail_ext]) self.to_screen('Adding thumbnail to "%s"' % filename) - self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd)) - - p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) - stdout, stderr = process_communicate_or_kill(p) - - if p.returncode != 0: - msg = stderr.decode('utf-8', 'replace').strip() - raise EmbedThumbnailPPError(msg) - # for formats that don't support thumbnails (like 3gp) AtomicParsley - # won't create to the temporary file - if b'No changes' in stdout: - self.report_warning('The file format doesn\'t support embedding a thumbnail') - success = False + self.run_ffmpeg(filename, temp_filename, options) + + elif info['ext'] in ['m4a', 'mp4', 'mov']: + try: + options = ['-c', 'copy', '-map', '0', '-dn', '-map', '1'] + + old_stream, new_stream = self.get_stream_number( + filename, ('disposition', 'attached_pic'), 1) + if old_stream is not None: + options.extend(['-map', '-0:%d' % old_stream]) + new_stream -= 1 + options.extend(['-disposition:%s' % new_stream, 'attached_pic']) + + self.to_screen('Adding thumbnail to "%s"' % filename) + self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options) + + except PostProcessingError as err: + self.report_warning('unable to embed using ffprobe & ffmpeg; %s' % error_to_compat_str(err)) + if not check_executable('AtomicParsley', ['-v']): + raise EmbedThumbnailPPError('AtomicParsley was not found. Please install.') + + cmd = [encodeFilename('AtomicParsley', True), + encodeFilename(filename, True), + encodeArgument('--artwork'), + encodeFilename(thumbnail_filename, True), + encodeArgument('-o'), + encodeFilename(temp_filename, True)] + cmd += [encodeArgument(o) for o in self._configuration_args(exe='AtomicParsley')] + + self.to_screen('Adding thumbnail to "%s"' % filename) + self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd)) + p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + stdout, stderr = process_communicate_or_kill(p) + if p.returncode != 0: + msg = stderr.decode('utf-8', 'replace').strip() + raise EmbedThumbnailPPError(msg) + # for formats that don't support thumbnails (like 3gp) AtomicParsley + # won't create to the temporary file + if b'No changes' in stdout: + self.report_warning('The file format doesn\'t support embedding a thumbnail') + success = False + + elif info['ext'] in ['ogg', 'opus']: + if not _has_mutagen: + raise EmbedThumbnailPPError('module mutagen was not found. Please install.') + size_regex = r',\s*(?P<w>\d+)x(?P<h>\d+)\s*[,\[]' + size_result = self.run_ffmpeg_multiple_files([thumbnail_filename], '', ['-hide_banner']) + mobj = re.search(size_regex, size_result) + width, height = int(mobj.group('w')), int(mobj.group('h')) + mimetype = ('image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg')).encode('ascii') + + # https://xiph.org/flac/format.html#metadata_block_picture + data = bytearray() + data += struct.pack('>II', 3, len(mimetype)) + data += mimetype + data += struct.pack('>IIIIII', 0, width, height, 8, 0, os.stat(thumbnail_filename).st_size) # 32 if png else 24 + + fin = open(thumbnail_filename, "rb") + data += fin.read() + fin.close() + + temp_filename = filename + f = mutagen.File(temp_filename) + f.tags['METADATA_BLOCK_PICTURE'] = base64.b64encode(data).decode('ascii') + f.save() else: - raise EmbedThumbnailPPError('Only mp3, mkv, m4a and mp4 are supported for thumbnail embedding for now.') + raise EmbedThumbnailPPError('Supported filetypes for thumbnail embedding are: mp3, mkv/mka, ogg/opus, m4a/mp4/mov') - if success: + if success and temp_filename != filename: os.remove(encodeFilename(filename)) os.rename(encodeFilename(temp_filename), encodeFilename(filename)) - files_to_delete = [] if self._already_have_thumbnail else [thumbnail_filename] return files_to_delete, info diff --git a/youtube_dlc/postprocessor/ffmpeg.py b/youtube_dlc/postprocessor/ffmpeg.py index 18696a932..f2be0f415 100644 --- a/youtube_dlc/postprocessor/ffmpeg.py +++ b/youtube_dlc/postprocessor/ffmpeg.py @@ -5,6 +5,7 @@ import os import subprocess import time import re +import json from .common import AudioConversionError, PostProcessor @@ -20,8 +21,9 @@ from ..utils import ( subtitles_filename, dfxp2srt, ISO639Utils, - replace_extension, process_communicate_or_kill, + replace_extension, + traverse_dict, ) @@ -201,6 +203,37 @@ class FFmpegPostProcessor(PostProcessor): return mobj.group(1) return None + def get_metadata_object(self, path, opts=[]): + if self.probe_basename != 'ffprobe': + if self.probe_available: + self.report_warning('Only ffprobe is supported for metadata extraction') + raise PostProcessingError('ffprobe not found. Please install.') + self.check_version() + + cmd = [ + encodeFilename(self.probe_executable, True), + encodeArgument('-hide_banner'), + encodeArgument('-show_format'), + encodeArgument('-show_streams'), + encodeArgument('-print_format'), + encodeArgument('json'), + ] + + cmd += opts + cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True)) + if self._downloader.params.get('verbose', False): + self._downloader.to_screen('[debug] ffprobe command line: %s' % shell_quote(cmd)) + p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) + stdout, stderr = p.communicate() + return json.loads(stdout.decode('utf-8', 'replace')) + + def get_stream_number(self, path, keys, value): + streams = self.get_metadata_object(path)['streams'] + num = next( + (i for i, stream in enumerate(streams) if traverse_dict(stream, keys, casesense=False) == value), + None) + return num, len(streams) + def run_ffmpeg_multiple_files(self, input_paths, out_path, opts): self.check_version() @@ -227,10 +260,12 @@ class FFmpegPostProcessor(PostProcessor): p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) stdout, stderr = process_communicate_or_kill(p) if p.returncode != 0: - stderr = stderr.decode('utf-8', 'replace') - msg = stderr.strip().split('\n')[-1] - raise FFmpegPostProcessorError(msg) + stderr = stderr.decode('utf-8', 'replace').strip() + if self._downloader.params.get('verbose', False): + self.report_error(stderr) + raise FFmpegPostProcessorError(stderr.split('\n')[-1]) self.try_utime(out_path, oldest_mtime, oldest_mtime) + return stderr.decode('utf-8', 'replace') def run_ffmpeg(self, path, out_path, opts): self.run_ffmpeg_multiple_files([path], out_path, opts) @@ -240,6 +275,8 @@ class FFmpegPostProcessor(PostProcessor): # interprets that as a protocol) or can start with '-' (-- is broken in # ffmpeg, see https://ffmpeg.org/trac/ffmpeg/ticket/2127 for details) # Also leave '-' intact in order not to break streaming to stdout. + if fn.startswith(('http://', 'https://')): + return fn return 'file:' + fn if fn != '-' else fn @@ -349,21 +386,35 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor): class FFmpegVideoRemuxerPP(FFmpegPostProcessor): def __init__(self, downloader=None, preferedformat=None): super(FFmpegVideoRemuxerPP, self).__init__(downloader) - self._preferedformat = preferedformat + self._preferedformats = preferedformat.lower().split('/') def run(self, information): path = information['filepath'] - if information['ext'] == self._preferedformat: - self.to_screen('Not remuxing video file %s - already is in target format %s' % (path, self._preferedformat)) + sourceext, targetext = information['ext'].lower(), None + for pair in self._preferedformats: + kv = pair.split('>') + if len(kv) == 1 or kv[0].strip() == sourceext: + targetext = kv[-1].strip() + break + + _skip_msg = ( + 'could not find a mapping for %s' if not targetext + else 'already is in target format %s' if sourceext == targetext + else None) + if _skip_msg: + self.to_screen('Not remuxing media file %s - %s' % (path, _skip_msg % sourceext)) return [], information + options = ['-c', 'copy', '-map', '0', '-dn'] - prefix, sep, ext = path.rpartition('.') - outpath = prefix + sep + self._preferedformat - self.to_screen('Remuxing video from %s to %s, Destination: ' % (information['ext'], self._preferedformat) + outpath) + if targetext in ['mp4', 'm4a', 'mov']: + options.extend(['-movflags', '+faststart']) + prefix, sep, oldext = path.rpartition('.') + outpath = prefix + sep + targetext + self.to_screen('Remuxing video from %s to %s; Destination: %s' % (sourceext, targetext, outpath)) self.run_ffmpeg(path, outpath, options) information['filepath'] = outpath - information['format'] = self._preferedformat - information['ext'] = self._preferedformat + information['format'] = targetext + information['ext'] = targetext return [path], information @@ -406,18 +457,22 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): sub_langs = [] sub_filenames = [] webm_vtt_warn = False + mp4_ass_warn = False for lang, sub_info in subtitles.items(): sub_ext = sub_info['ext'] if sub_ext == 'json': - self.to_screen('JSON subtitles cannot be embedded') + self.report_warning('JSON subtitles cannot be embedded') elif ext != 'webm' or ext == 'webm' and sub_ext == 'vtt': sub_langs.append(lang) sub_filenames.append(subtitles_filename(filename, lang, sub_ext, ext)) else: if not webm_vtt_warn and ext == 'webm' and sub_ext != 'vtt': webm_vtt_warn = True - self.to_screen('Only WebVTT subtitles can be embedded in webm files') + self.report_warning('Only WebVTT subtitles can be embedded in webm files') + if not mp4_ass_warn and ext == 'mp4' and sub_ext == 'ass': + mp4_ass_warn = True + self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues') if not sub_langs: return [], information @@ -441,7 +496,7 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): opts.extend(['-metadata:s:s:%d' % i, 'language=%s' % lang_code]) temp_filename = prepend_extension(filename, 'temp') - self.to_screen('Embedding subtitles in \'%s\'' % filename) + self.to_screen('Embedding subtitles in "%s"' % filename) self.run_ffmpeg_multiple_files(input_files, temp_filename, opts) os.remove(encodeFilename(filename)) os.rename(encodeFilename(temp_filename), encodeFilename(filename)) @@ -471,7 +526,6 @@ class FFmpegMetadataPP(FFmpegPostProcessor): # 1. https://kdenlive.org/en/project/adding-meta-data-to-mp4-video/ # 2. https://wiki.multimedia.cx/index.php/FFmpeg_Metadata # 3. https://kodi.wiki/view/Video_file_tagging - # 4. http://atomicparsley.sourceforge.net/mpeg-4files.html add('title', ('track', 'title')) add('date', 'upload_date') @@ -524,6 +578,18 @@ class FFmpegMetadataPP(FFmpegPostProcessor): in_filenames.append(metadata_filename) options.extend(['-map_metadata', '1']) + if '__infojson_filepath' in info and info['ext'] in ('mkv', 'mka'): + old_stream, new_stream = self.get_stream_number( + filename, ('tags', 'mimetype'), 'application/json') + if old_stream is not None: + options.extend(['-map', '-0:%d' % old_stream]) + new_stream -= 1 + + options.extend([ + '-attach', info['__infojson_filepath'], + '-metadata:s:%d' % new_stream, 'mimetype=application/json' + ]) + self.to_screen('Adding metadata to \'%s\'' % filename) self.run_ffmpeg_multiple_files(in_filenames, temp_filename, options) if chapters: |