diff options
author | Jesús <heckyel@hyperbola.info> | 2022-04-06 03:37:17 +0800 |
---|---|---|
committer | Jesús <heckyel@hyperbola.info> | 2022-04-06 03:37:17 +0800 |
commit | 1e5a50b71d8f0eae6007bedc329eecb24bb5aba3 (patch) | |
tree | a8611cda6596391cb6fb645e1469dcd356b63924 /hypervideo_dl/postprocessor | |
parent | f52fb3bceeb9d22b5106c1796fecec474a0cc138 (diff) | |
download | hypervideo-1e5a50b71d8f0eae6007bedc329eecb24bb5aba3.tar.lz hypervideo-1e5a50b71d8f0eae6007bedc329eecb24bb5aba3.tar.xz hypervideo-1e5a50b71d8f0eae6007bedc329eecb24bb5aba3.zip |
update from upstream
Diffstat (limited to 'hypervideo_dl/postprocessor')
-rw-r--r-- | hypervideo_dl/postprocessor/__init__.py | 8 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/common.py | 44 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/embedthumbnail.py | 77 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/exec.py | 21 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/ffmpeg.py | 519 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/metadataparser.py | 29 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/modify_chapters.py | 22 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/sponskrub.py | 13 | ||||
-rw-r--r-- | hypervideo_dl/postprocessor/sponsorblock.py | 37 |
9 files changed, 498 insertions, 272 deletions
diff --git a/hypervideo_dl/postprocessor/__init__.py b/hypervideo_dl/postprocessor/__init__.py index 07c87b7..e47631e 100644 --- a/hypervideo_dl/postprocessor/__init__.py +++ b/hypervideo_dl/postprocessor/__init__.py @@ -2,12 +2,16 @@ from ..utils import load_plugins +from .common import PostProcessor from .embedthumbnail import EmbedThumbnailPP from .exec import ExecPP, ExecAfterDownloadPP from .ffmpeg import ( FFmpegPostProcessor, + FFmpegCopyStreamPP, + FFmpegConcatPP, FFmpegEmbedSubtitlePP, FFmpegExtractAudioPP, + FFmpegFixupDuplicateMoovPP, FFmpegFixupDurationPP, FFmpegFixupStretchedPP, FFmpegFixupTimestampPP, @@ -39,5 +43,5 @@ def get_postprocessor(key): return globals()[key + 'PP'] -__all__ = [name for name in globals().keys() if name.endswith('IE')] -__all__.append('FFmpegPostProcessor') +__all__ = [name for name in globals().keys() if name.endswith('PP')] +__all__.extend(('PostProcessor', 'FFmpegPostProcessor')) diff --git a/hypervideo_dl/postprocessor/common.py b/hypervideo_dl/postprocessor/common.py index b491afb..3899646 100644 --- a/hypervideo_dl/postprocessor/common.py +++ b/hypervideo_dl/postprocessor/common.py @@ -1,14 +1,19 @@ from __future__ import unicode_literals -import copy import functools +import itertools +import json import os +import time +import urllib.error -from ..compat import compat_str from ..utils import ( _configuration_args, encodeFilename, + network_exceptions, PostProcessingError, + sanitized_Request, + write_string, ) @@ -17,7 +22,7 @@ class PostProcessorMetaClass(type): def run_wrapper(func): @functools.wraps(func) def run(self, info, *args, **kwargs): - info_copy = copy.deepcopy(self._copy_infodict(info)) + info_copy = self._copy_infodict(info) self._hook_progress({'status': 'started'}, info_copy) ret = func(self, info, *args, **kwargs) if ret is not None: @@ -63,7 +68,7 @@ class PostProcessor(metaclass=PostProcessorMetaClass): @classmethod def pp_key(cls): name = cls.__name__[:-2] - return compat_str(name[6:]) if name[:6].lower() == 'ffmpeg' else name + return name[6:] if name[:6].lower() == 'ffmpeg' else name def to_screen(self, text, prefix=True, *args, **kwargs): tag = '[%s] ' % self.PP_NAME if prefix else '' @@ -74,6 +79,11 @@ class PostProcessor(metaclass=PostProcessorMetaClass): if self._downloader: return self._downloader.report_warning(text, *args, **kwargs) + def deprecation_warning(self, text): + if self._downloader: + return self._downloader.deprecation_warning(text) + write_string(f'DeprecationWarning: {text}') + def report_error(self, text, *args, **kwargs): # Exists only for compatibility. Do not use if self._downloader: @@ -98,12 +108,14 @@ class PostProcessor(metaclass=PostProcessorMetaClass): return getattr(self._downloader, '_copy_infodict', dict)(info_dict) @staticmethod - def _restrict_to(*, video=True, audio=True, images=True): + def _restrict_to(*, video=True, audio=True, images=True, simulated=True): allowed = {'video': video, 'audio': audio, 'images': images} def decorator(func): @functools.wraps(func) def wrapper(self, info): + if not simulated and (self.get_param('simulate') or self.get_param('skip_download')): + return [], info format_type = ( 'video' if info.get('vcodec') != 'none' else 'audio' if info.get('acodec') != 'none' @@ -173,6 +185,28 @@ class PostProcessor(metaclass=PostProcessorMetaClass): progress_template.get('postprocess-title') or 'hypervideo %(progress._default_template)s', progress_dict)) + def _download_json(self, url, *, expected_http_errors=(404,)): + # While this is not an extractor, it behaves similar to one and + # so obey extractor_retries and sleep_interval_requests + max_retries = self.get_param('extractor_retries', 3) + sleep_interval = self.get_param('sleep_interval_requests') or 0 + + self.write_debug(f'{self.PP_NAME} query: {url}') + for retries in itertools.count(): + try: + rsp = self._downloader.urlopen(sanitized_Request(url)) + return json.loads(rsp.read().decode(rsp.info().get_param('charset') or 'utf-8')) + except network_exceptions as e: + if isinstance(e, urllib.error.HTTPError) and e.code in expected_http_errors: + return None + if retries < max_retries: + self.report_warning(f'{e}. Retrying...') + if sleep_interval > 0: + self.to_screen(f'Sleeping {sleep_interval} seconds ...') + time.sleep(sleep_interval) + continue + raise PostProcessingError(f'Unable to communicate with {self.PP_NAME} API: {e}') + class AudioConversionError(PostProcessingError): pass diff --git a/hypervideo_dl/postprocessor/embedthumbnail.py b/hypervideo_dl/postprocessor/embedthumbnail.py index 3139a63..815221d 100644 --- a/hypervideo_dl/postprocessor/embedthumbnail.py +++ b/hypervideo_dl/postprocessor/embedthumbnail.py @@ -26,9 +26,9 @@ from ..utils import ( encodeArgument, encodeFilename, error_to_compat_str, + Popen, PostProcessingError, prepend_extension, - process_communicate_or_kill, shell_quote, ) @@ -108,7 +108,7 @@ class EmbedThumbnailPP(FFmpegPostProcessor): self.run_ffmpeg_multiple_files([filename, thumbnail_filename], temp_filename, options) elif info['ext'] in ['mkv', 'mka']: - options = ['-c', 'copy', '-map', '0', '-dn'] + options = list(self.stream_copy_opts()) mimetype = 'image/%s' % ('png' if thumbnail_ext == 'png' else 'jpeg') old_stream, new_stream = self.get_stream_number( @@ -145,11 +145,46 @@ class EmbedThumbnailPP(FFmpegPostProcessor): self.report_warning('unable to embed using mutagen; %s' % error_to_compat_str(err)) success = False - # Method 2: Use ffmpeg+ffprobe - if not success and not prefer_atomicparsley: + # Method 2: Use AtomicParsley + if not success: + success = True + atomicparsley = next(( + x for x in ['AtomicParsley', 'atomicparsley'] + if check_executable(x, ['-v'])), None) + if atomicparsley is None: + self.to_screen('Neither mutagen nor AtomicParsley was found. Falling back to ffmpeg') + success = False + else: + if not prefer_atomicparsley: + self.to_screen('mutagen was not found. Falling back to AtomicParsley') + 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('AtomicParsley')] + + self._report_run('atomicparsley', filename) + self.write_debug('AtomicParsley command line: %s' % shell_quote(cmd)) + p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + stdout, stderr = p.communicate_or_kill() + if p.returncode != 0: + msg = stderr.decode('utf-8', 'replace').strip() + self.report_warning(f'Unable to embed thumbnails using AtomicParsley; {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 + + # Method 3: Use ffmpeg+ffprobe + # Thumbnails attached using this method doesn't show up as cover in some cases + # See https://github.com/hypervideo/hypervideo/issues/2125, https://github.com/hypervideo/hypervideo/issues/411 + if not success: success = True try: - options = ['-c', 'copy', '-map', '0', '-dn', '-map', '1'] + options = [*self.stream_copy_opts(), '-map', '1'] old_stream, new_stream = self.get_stream_number( filename, ('disposition', 'attached_pic'), 1) @@ -161,38 +196,8 @@ class EmbedThumbnailPP(FFmpegPostProcessor): self._report_run('ffmpeg', 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)) - success = False - - # Method 3: Use AtomicParsley - if not success: - success = True - atomicparsley = next(( - x for x in ['AtomicParsley', 'atomicparsley'] - if check_executable(x, ['-v'])), None) - if atomicparsley is None: - 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('AtomicParsley')] - - self._report_run('atomicparsley', 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 + raise EmbedThumbnailPPError(f'Unable to embed using ffprobe & ffmpeg; {err}') elif info['ext'] in ['ogg', 'opus', 'flac']: if not has_mutagen: diff --git a/hypervideo_dl/postprocessor/exec.py b/hypervideo_dl/postprocessor/exec.py index 7a3cb49..c0bd6df 100644 --- a/hypervideo_dl/postprocessor/exec.py +++ b/hypervideo_dl/postprocessor/exec.py @@ -22,11 +22,13 @@ class ExecPP(PostProcessor): if tmpl_dict: # if there are no replacements, tmpl_dict = {} return self._downloader.escape_outtmpl(tmpl) % tmpl_dict - # If no replacements are found, replace {} for backard compatibility - if '{}' not in cmd: - cmd += ' {}' - return cmd.replace('{}', compat_shlex_quote( - info.get('filepath') or info['_filename'])) + filepath = info.get('filepath', info.get('_filename')) + # If video, and no replacements are found, replace {} for backard compatibility + if filepath: + if '{}' not in cmd: + cmd += ' {}' + cmd = cmd.replace('{}', compat_shlex_quote(filepath)) + return cmd def run(self, info): for tmpl in self.exec_cmd: @@ -38,5 +40,10 @@ class ExecPP(PostProcessor): return [], info -class ExecAfterDownloadPP(ExecPP): # for backward compatibility - pass +# Deprecated +class ExecAfterDownloadPP(ExecPP): + def __init__(self, *args, **kwargs): + super().__init__(*args, **kwargs) + self.deprecation_warning( + 'hypervideo_dl.postprocessor.ExecAfterDownloadPP is deprecated ' + 'and may be removed in a future version. Use hypervideo_dl.postprocessor.ExecPP instead') diff --git a/hypervideo_dl/postprocessor/ffmpeg.py b/hypervideo_dl/postprocessor/ffmpeg.py index a6d6d78..3e6edcf 100644 --- a/hypervideo_dl/postprocessor/ffmpeg.py +++ b/hypervideo_dl/postprocessor/ffmpeg.py @@ -1,5 +1,6 @@ from __future__ import unicode_literals +import collections import io import itertools import os @@ -12,21 +13,24 @@ from .common import AudioConversionError, PostProcessor from ..compat import compat_str from ..utils import ( + determine_ext, dfxp2srt, encodeArgument, encodeFilename, float_or_none, - get_exe_version, + _get_exe_version_output, + detect_exe_version, is_outdated_version, ISO639Utils, orderedSet, + Popen, PostProcessingError, prepend_extension, - process_communicate_or_kill, replace_extension, shell_quote, traverse_obj, variadic, + write_json_file, ) @@ -41,6 +45,7 @@ EXT_TO_OUT_FORMATS = { 'ts': 'mpegts', 'wma': 'asf', 'wmv': 'asf', + 'vtt': 'webvtt', } ACODECS = { 'mp3': 'libmp3lame', @@ -50,6 +55,7 @@ ACODECS = { 'opus': 'libopus', 'vorbis': 'libvorbis', 'wav': None, + 'alac': None, } @@ -74,15 +80,25 @@ class FFmpegPostProcessor(PostProcessor): self.report_warning(warning) @staticmethod + def get_versions_and_features(downloader=None): + pp = FFmpegPostProcessor(downloader) + return pp._versions, pp._features + + @staticmethod def get_versions(downloader=None): - return FFmpegPostProcessor(downloader)._versions + return FFmpegPostProcessor.get_versions_and_features(downloader)[0] + + _version_cache, _features_cache = {}, {} def _determine_executables(self): programs = ['avprobe', 'avconv', 'ffmpeg', 'ffprobe'] - prefer_ffmpeg = True - def get_ffmpeg_version(path): - ver = get_exe_version(path, args=['-version']) + def get_ffmpeg_version(path, prog): + if path in self._version_cache: + self._versions[prog], self._features = self._version_cache[path], self._features_cache.get(path, {}) + return + out = _get_exe_version_output(path, ['-bsfs'], to_screen=self.write_debug) + ver = detect_exe_version(out) if out else False if ver: regexs = [ r'(?:\d+:)?([0-9.]+)-[0-9]+ubuntu[0-9.]+$', # Ubuntu, see [1] @@ -93,60 +109,66 @@ class FFmpegPostProcessor(PostProcessor): mobj = re.match(regex, ver) if mobj: ver = mobj.group(1) - return ver + self._versions[prog] = self._version_cache[path] = ver + if prog != 'ffmpeg' or not out: + return + + mobj = re.search(r'(?m)^\s+libavformat\s+(?:[0-9. ]+)\s+/\s+(?P<runtime>[0-9. ]+)', out) + lavf_runtime_version = mobj.group('runtime').replace(' ', '') if mobj else None + self._features = self._features_cache[path] = { + 'fdk': '--enable-libfdk-aac' in out, + 'setts': 'setts' in out.splitlines(), + 'needs_adtstoasc': is_outdated_version(lavf_runtime_version, '57.56.100', False), + } self.basename = None self.probe_basename = None - self._paths = None self._versions = None - if self._downloader: - prefer_ffmpeg = self.get_param('prefer_ffmpeg', True) - location = self.get_param('ffmpeg_location') - if location is not None: - if not os.path.exists(location): - self.report_warning( - 'ffmpeg-location %s does not exist! ' - 'Continuing without ffmpeg.' % (location)) - self._versions = {} - return - elif os.path.isdir(location): - dirname, basename = location, None - else: - basename = os.path.splitext(os.path.basename(location))[0] - basename = next((p for p in programs if basename.startswith(p)), 'ffmpeg') - dirname = os.path.dirname(os.path.abspath(location)) - if basename in ('ffmpeg', 'ffprobe'): - prefer_ffmpeg = True - - self._paths = dict( - (p, os.path.join(dirname, p)) for p in programs) - if basename: - self._paths[basename] = location - self._versions = dict( - (p, get_ffmpeg_version(self._paths[p])) for p in programs) - if self._versions is None: - self._versions = dict( - (p, get_ffmpeg_version(p)) for p in programs) - self._paths = dict((p, p) for p in programs) + self._features = {} - if prefer_ffmpeg is False: - prefs = ('avconv', 'ffmpeg') + prefer_ffmpeg = self.get_param('prefer_ffmpeg', True) + location = self.get_param('ffmpeg_location') + if location is None: + self._paths = {p: p for p in programs} else: - prefs = ('ffmpeg', 'avconv') - for p in prefs: - if self._versions[p]: - self.basename = p - break - + if not os.path.exists(location): + self.report_warning( + 'ffmpeg-location %s does not exist! ' + 'Continuing without ffmpeg.' % (location)) + self._versions = {} + return + elif os.path.isdir(location): + dirname, basename = location, None + else: + basename = os.path.splitext(os.path.basename(location))[0] + basename = next((p for p in programs if basename.startswith(p)), 'ffmpeg') + dirname = os.path.dirname(os.path.abspath(location)) + if basename in ('ffmpeg', 'ffprobe'): + prefer_ffmpeg = True + + self._paths = dict( + (p, os.path.join(dirname, p)) for p in programs) + if basename: + self._paths[basename] = location + + self._versions = {} + executables = {'basename': ('ffmpeg', 'avconv'), 'probe_basename': ('ffprobe', 'avprobe')} if prefer_ffmpeg is False: - prefs = ('avprobe', 'ffprobe') - else: - prefs = ('ffprobe', 'avprobe') - for p in prefs: - if self._versions[p]: - self.probe_basename = p - break + executables = {k: v[::-1] for k, v in executables.items()} + for var, prefs in executables.items(): + for p in prefs: + get_ffmpeg_version(self._paths[p], p) + if self._versions[p]: + setattr(self, var, p) + break + + if self.basename == 'avconv': + self.deprecation_warning( + 'Support for avconv is deprecated and may be removed in a future version. Use ffmpeg instead') + if self.probe_basename == 'avprobe': + self.deprecation_warning( + 'Support for avprobe is deprecated and may be removed in a future version. Use ffprobe instead') @property def available(self): @@ -164,6 +186,18 @@ class FFmpegPostProcessor(PostProcessor): def probe_executable(self): return self._paths[self.probe_basename] + @staticmethod + def stream_copy_opts(copy=True, *, ext=None): + yield from ('-map', '0') + # Don't copy Apple TV chapters track, bin_data + # See https://github.com/hypervideo/hypervideo/issues/2, #19042, #19024, https://trac.ffmpeg.org/ticket/6016 + yield from ('-dn', '-ignore_unknown') + if copy: + yield from ('-c', 'copy') + # For some reason, '-c copy -map 0' is not enough to copy subtitles + if ext in ('mp4', 'mov'): + yield from ('-c:s', 'mov_text') + def get_audio_codec(self, path): if not self.probe_available and not self.available: raise PostProcessingError('ffprobe and ffmpeg not found. Please install or provide the path using --ffmpeg-location') @@ -178,10 +212,8 @@ class FFmpegPostProcessor(PostProcessor): encodeArgument('-i')] cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True)) self.write_debug('%s command line: %s' % (self.basename, shell_quote(cmd))) - handle = subprocess.Popen( - cmd, stderr=subprocess.PIPE, - stdout=subprocess.PIPE, stdin=subprocess.PIPE) - stdout_data, stderr_data = process_communicate_or_kill(handle) + handle = Popen(cmd, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE) + stdout_data, stderr_data = handle.communicate_or_kill() expected_ret = 0 if self.probe_available else 1 if handle.wait() != expected_ret: return None @@ -223,7 +255,7 @@ class FFmpegPostProcessor(PostProcessor): cmd += opts cmd.append(encodeFilename(self._ffmpeg_filename_argument(path), True)) self.write_debug('ffprobe command line: %s' % shell_quote(cmd)) - p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) + p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) stdout, stderr = p.communicate() return json.loads(stdout.decode('utf-8', 'replace')) @@ -234,22 +266,23 @@ class FFmpegPostProcessor(PostProcessor): None) return num, len(streams) - def _get_real_video_duration(self, info, fatal=True): + def _get_real_video_duration(self, filepath, fatal=True): try: - if '_real_duration' not in info: - info['_real_duration'] = float_or_none( - traverse_obj(self.get_metadata_object(info['filepath']), ('format', 'duration'))) - if not info['_real_duration']: + duration = float_or_none( + traverse_obj(self.get_metadata_object(filepath), ('format', 'duration'))) + if not duration: raise PostProcessingError('ffprobe returned empty duration') + return duration except PostProcessingError as e: if fatal: - raise PostProcessingError(f'Unable to determine video duration; {e}') - return info.setdefault('_real_duration', None) + raise PostProcessingError(f'Unable to determine video duration: {e.msg}') def _duration_mismatch(self, d1, d2): if not d1 or not d2: return None - return abs(d1 - d2) > 1 + # The duration is often only known to nearest second. So there can be <1sec disparity natually. + # Further excuse an additional <1sec difference. + return abs(d1 - d2) > 2 def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs): return self.real_run_ffmpeg( @@ -269,8 +302,10 @@ class FFmpegPostProcessor(PostProcessor): def make_args(file, args, name, number): keys = ['_%s%d' % (name, number), '_%s' % name] - if name == 'o' and number == 1: - keys.append('') + if name == 'o': + args += ['-movflags', '+faststart'] + if number == 1: + keys.append('') args += self._configuration_args(self.basename, keys) if name == 'i': args.append('-i') @@ -284,8 +319,8 @@ class FFmpegPostProcessor(PostProcessor): for i, (path, opts) in enumerate(path_opts) if path) self.write_debug('ffmpeg command line: %s' % shell_quote(cmd)) - p = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) - stdout, stderr = process_communicate_or_kill(p) + p = Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE, stdin=subprocess.PIPE) + stdout, stderr = p.communicate_or_kill() if p.returncode not in variadic(expected_retcodes): stderr = stderr.decode('utf-8', 'replace').strip() self.write_debug(stderr) @@ -324,8 +359,9 @@ class FFmpegPostProcessor(PostProcessor): timestamps = timestamps[1:] keyframe_file = prepend_extension(filename, 'keyframes.temp') self.to_screen(f'Re-encoding "{filename}" with appropriate keyframes') - self.run_ffmpeg(filename, keyframe_file, ['-force_key_frames', ','.join( - f'{t:.6f}' for t in timestamps)]) + self.run_ffmpeg(filename, keyframe_file, [ + *self.stream_copy_opts(False, ext=determine_ext(filename)), + '-force_key_frames', ','.join(f'{t:.6f}' for t in timestamps)]) return keyframe_file def concat_files(self, in_files, out_file, concat_opts=None): @@ -340,17 +376,12 @@ class FFmpegPostProcessor(PostProcessor): with open(concat_file, 'wt', encoding='utf-8') as f: f.writelines(self._concat_spec(in_files, concat_opts)) - out_flags = ['-c', 'copy'] - if out_file.rpartition('.')[-1] in ('mp4', 'mov'): - # For some reason, '-c copy' is not enough to copy subtitles - out_flags.extend(['-c:s', 'mov_text', '-movflags', '+faststart']) + out_flags = list(self.stream_copy_opts(ext=determine_ext(out_file))) - try: - self.real_run_ffmpeg( - [(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])], - [(out_file, out_flags)]) - finally: - os.remove(concat_file) + self.real_run_ffmpeg( + [(concat_file, ['-hide_banner', '-nostdin', '-f', 'concat', '-safe', '0'])], + [(out_file, out_flags)]) + os.remove(concat_file) @classmethod def _concat_spec(cls, in_files, concat_opts=None): @@ -367,14 +398,36 @@ class FFmpegPostProcessor(PostProcessor): class FFmpegExtractAudioPP(FFmpegPostProcessor): COMMON_AUDIO_EXTS = ('wav', 'flac', 'm4a', 'aiff', 'mp3', 'ogg', 'mka', 'opus', 'wma') - SUPPORTED_EXTS = ('best', 'aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav') + SUPPORTED_EXTS = ('aac', 'flac', 'mp3', 'm4a', 'opus', 'vorbis', 'wav', 'alac') def __init__(self, downloader=None, preferredcodec=None, preferredquality=None, nopostoverwrites=False): FFmpegPostProcessor.__init__(self, downloader) self._preferredcodec = preferredcodec or 'best' - self._preferredquality = preferredquality + self._preferredquality = float_or_none(preferredquality) self._nopostoverwrites = nopostoverwrites + def _quality_args(self, codec): + if self._preferredquality is None: + return [] + elif self._preferredquality > 10: + return ['-b:a', f'{self._preferredquality}k'] + + limits = { + 'libmp3lame': (10, 0), + 'libvorbis': (0, 10), + # FFmpeg's AAC encoder does not have an upper limit for the value of -q:a. + # Experimentally, with values over 4, bitrate changes were minimal or non-existent + 'aac': (0.1, 4), + 'libfdk_aac': (1, 5), + }.get(codec) + if not limits: + return [] + + q = limits[1] + (limits[0] - limits[1]) * (self._preferredquality / 10) + if codec == 'libfdk_aac': + return ['-vbr', f'{int(q)}'] + return ['-q:a', f'{q}'] + def run_ffmpeg(self, path, out_path, codec, more_opts): if codec is None: acodec_opts = [] @@ -388,7 +441,7 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor): @PostProcessor._restrict_to(images=False) def run(self, information): - path = information['filepath'] + orig_path = path = information['filepath'] orig_ext = information['ext'] if self._preferredcodec == 'best' and orig_ext in self.COMMON_AUDIO_EXTS: @@ -414,69 +467,74 @@ class FFmpegExtractAudioPP(FFmpegPostProcessor): more_opts = ['-f', 'adts'] if filecodec == 'vorbis': extension = 'ogg' + elif filecodec == 'alac': + acodec = None + extension = 'm4a' + more_opts += ['-acodec', 'alac'] else: # MP3 otherwise. acodec = 'libmp3lame' extension = 'mp3' - more_opts = [] - if self._preferredquality is not None: - if int(self._preferredquality) < 10: - more_opts += ['-q:a', self._preferredquality] - else: - more_opts += ['-b:a', self._preferredquality + 'k'] + more_opts = self._quality_args(acodec) else: # We convert the audio (lossy if codec is lossy) acodec = ACODECS[self._preferredcodec] + if acodec == 'aac' and self._features.get('fdk'): + acodec = 'libfdk_aac' extension = self._preferredcodec - more_opts = [] - if self._preferredquality is not None: - # The opus codec doesn't support the -aq option - if int(self._preferredquality) < 10 and extension != 'opus': - more_opts += ['-q:a', self._preferredquality] - else: - more_opts += ['-b:a', self._preferredquality + 'k'] + more_opts = self._quality_args(acodec) if self._preferredcodec == 'aac': more_opts += ['-f', 'adts'] - if self._preferredcodec == 'm4a': + elif self._preferredcodec == 'm4a': more_opts += ['-bsf:a', 'aac_adtstoasc'] - if self._preferredcodec == 'vorbis': + elif self._preferredcodec == 'vorbis': extension = 'ogg' - if self._preferredcodec == 'wav': + elif self._preferredcodec == 'wav': extension = 'wav' more_opts += ['-f', 'wav'] + elif self._preferredcodec == 'alac': + extension = 'm4a' + more_opts += ['-acodec', 'alac'] prefix, sep, ext = path.rpartition('.') # not os.path.splitext, since the latter does not work on unicode in all setups - new_path = prefix + sep + extension - - information['filepath'] = new_path - information['ext'] = extension - - # If we download foo.mp3 and convert it to... foo.mp3, then don't delete foo.mp3, silly. - if (new_path == path - or (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)))): + temp_path = new_path = prefix + sep + extension + + if new_path == path: + if acodec == 'copy': + self.to_screen(f'File is already in target format {self._preferredcodec}, skipping') + return [], information + orig_path = prepend_extension(path, 'orig') + temp_path = prepend_extension(path, 'temp') + if (self._nopostoverwrites and os.path.exists(encodeFilename(new_path)) + and os.path.exists(encodeFilename(orig_path))): self.to_screen('Post-process file %s exists, skipping' % new_path) return [], information try: - self.to_screen('Destination: ' + new_path) - self.run_ffmpeg(path, new_path, acodec, more_opts) + self.to_screen(f'Destination: {new_path}') + self.run_ffmpeg(path, temp_path, acodec, more_opts) except AudioConversionError as e: raise PostProcessingError( 'audio conversion failed: ' + e.msg) except Exception: raise PostProcessingError('error running ' + self.basename) + os.replace(path, orig_path) + os.replace(temp_path, new_path) + information['filepath'] = new_path + information['ext'] = extension + # Try to update the date time for extracted audio file. if information.get('filetime') is not None: self.try_utime( new_path, time.time(), information['filetime'], errnote='Cannot update utime of audio file') - return [path], information + return [orig_path], information class FFmpegVideoConvertorPP(FFmpegPostProcessor): - SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mp3', 'mka', 'm4a', 'ogg', 'opus') + SUPPORTED_EXTS = ('mp4', 'mkv', 'flv', 'webm', 'mov', 'avi', 'mka', 'ogg', *FFmpegExtractAudioPP.SUPPORTED_EXTS) FORMAT_RE = re.compile(r'{0}(?:/{0})*$'.format(r'(?:\w+>)?(?:%s)' % '|'.join(SUPPORTED_EXTS))) _ACTION = 'converting' @@ -492,9 +550,9 @@ class FFmpegVideoConvertorPP(FFmpegPostProcessor): @staticmethod def _options(target_ext): + yield from FFmpegPostProcessor.stream_copy_opts(False) if target_ext == 'avi': - return ['-c:v', 'libxvid', '-vtag', 'XVID'] - return [] + yield from ('-c:v', 'libxvid', '-vtag', 'XVID') @PostProcessor._restrict_to(images=False) def run(self, info): @@ -505,7 +563,7 @@ class FFmpegVideoConvertorPP(FFmpegPostProcessor): else f'already is in target format {source_ext}' if source_ext == target_ext else None) if _skip_msg: - self.to_screen(f'Not {self._ACTION} media file {filename!r}; {_skip_msg}') + self.to_screen(f'Not {self._ACTION} media file "{filename}"; {_skip_msg}') return [], info outpath = replace_extension(filename, target_ext, source_ext) @@ -522,10 +580,7 @@ class FFmpegVideoRemuxerPP(FFmpegVideoConvertorPP): @staticmethod def _options(target_ext): - options = ['-c', 'copy', '-map', '0', '-dn'] - if target_ext in ['mp4', 'm4a', 'mov']: - options.extend(['-movflags', '+faststart']) - return options + return FFmpegPostProcessor.stream_copy_opts() class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): @@ -534,22 +589,28 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): self._already_have_subtitle = already_have_subtitle @PostProcessor._restrict_to(images=False) - def run(self, information): - if information['ext'] not in ('mp4', 'webm', 'mkv'): + def run(self, info): + if info['ext'] not in ('mp4', 'webm', 'mkv'): self.to_screen('Subtitles can only be embedded in mp4, webm or mkv files') - return [], information - subtitles = information.get('requested_subtitles') + return [], info + subtitles = info.get('requested_subtitles') if not subtitles: self.to_screen('There aren\'t any subtitles to embed') - return [], information + return [], info + + filename = info['filepath'] - filename = information['filepath'] - if information.get('duration') and self._duration_mismatch( - self._get_real_video_duration(information, False), information['duration']): + # Disabled temporarily. There needs to be a way to overide this + # in case of duration actually mismatching in extractor + # See: https://github.com/hypervideo/hypervideo/issues/1870, https://github.com/hypervideo/hypervideo/issues/1385 + ''' + if info.get('duration') and not info.get('__real_download') and self._duration_mismatch( + self._get_real_video_duration(filename, False), info['duration']): self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch') - return [], information + return [], info + ''' - ext = information['ext'] + ext = info['ext'] sub_langs, sub_names, sub_filenames = [], [], [] webm_vtt_warn = False mp4_ass_warn = False @@ -574,21 +635,16 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): self.report_warning('ASS subtitles cannot be properly embedded in mp4 files; expect issues') if not sub_langs: - return [], information + return [], info input_files = [filename] + sub_filenames opts = [ - '-c', 'copy', '-map', '0', '-dn', + *self.stream_copy_opts(ext=info['ext']), # Don't copy the existing subtitles, we may be running the # postprocessor a second time '-map', '-0:s', - # Don't copy Apple TV chapters track, bin_data (see #19042, #19024, - # https://trac.ffmpeg.org/ticket/6016) - '-map', '-0:d', ] - if information['ext'] == 'mp4': - opts += ['-c:s', 'mov_text'] for i, (lang, name) in enumerate(zip(sub_langs, sub_names)): opts.extend(['-map', '%d:0' % (i + 1)]) lang_code = ISO639Utils.short2long(lang) or lang @@ -603,34 +659,44 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor): os.replace(temp_filename, filename) files_to_delete = [] if self._already_have_subtitle else sub_filenames - return files_to_delete, information + return files_to_delete, info class FFmpegMetadataPP(FFmpegPostProcessor): - def __init__(self, downloader, add_metadata=True, add_chapters=True): + def __init__(self, downloader, add_metadata=True, add_chapters=True, add_infojson='if_exists'): FFmpegPostProcessor.__init__(self, downloader) self._add_metadata = add_metadata self._add_chapters = add_chapters + self._add_infojson = add_infojson @staticmethod def _options(target_ext): - yield from ('-map', '0', '-dn') - if target_ext == 'm4a': + audio_only = target_ext == 'm4a' + yield from FFmpegPostProcessor.stream_copy_opts(not audio_only) + if audio_only: yield from ('-vn', '-acodec', 'copy') - else: - yield from ('-c', 'copy') @PostProcessor._restrict_to(images=False) def run(self, info): filename, metadata_filename = info['filepath'], None - options = [] + files_to_delete, options = [], [] if self._add_chapters and info.get('chapters'): metadata_filename = replace_extension(filename, 'meta') options.extend(self._get_chapter_opts(info['chapters'], metadata_filename)) + files_to_delete.append(metadata_filename) if self._add_metadata: options.extend(self._get_metadata_opts(info)) + if self._add_infojson: + if info['ext'] in ('mkv', 'mka'): + infojson_filename = info.get('infojson_filename') + options.extend(self._get_infojson_opts(info, infojson_filename)) + if not infojson_filename: + files_to_delete.append(info.get('infojson_filename')) + elif self._add_infojson is True: + self.to_screen('The info-json can only be attached to mkv/mka files') + if not options: self.to_screen('There isn\'t any metadata to add') return [], info @@ -640,8 +706,8 @@ class FFmpegMetadataPP(FFmpegPostProcessor): self.run_ffmpeg_multiple_files( (filename, metadata_filename), temp_filename, itertools.chain(self._options(info['ext']), *options)) - if metadata_filename: - os.remove(metadata_filename) + for file in filter(None, files_to_delete): + os.remove(file) # Don't obey --keep-files os.replace(temp_filename, filename) return [], info @@ -663,15 +729,15 @@ class FFmpegMetadataPP(FFmpegPostProcessor): yield ('-map_metadata', '1') def _get_metadata_opts(self, info): - metadata = {} - meta_prefix = 'meta_' + meta_prefix = 'meta' + metadata = collections.defaultdict(dict) def add(meta_list, info_list=None): value = next(( - str(info[key]) for key in [meta_prefix] + list(variadic(info_list or meta_list)) + str(info[key]) for key in [f'{meta_prefix}_'] + list(variadic(info_list or meta_list)) if info.get(key) is not None), None) if value not in ('', None): - metadata.update({meta_f: value for meta_f in variadic(meta_list)}) + metadata['common'].update({meta_f: value for meta_f in variadic(meta_list)}) # See [1-4] for some info on media metadata/metadata supported # by ffmpeg. @@ -693,32 +759,50 @@ class FFmpegMetadataPP(FFmpegPostProcessor): add('season_number') add('episode_id', ('episode', 'episode_id')) add('episode_sort', 'episode_number') + if 'embed-metadata' in self.get_param('compat_opts', []): + add('comment', 'description') + metadata['common'].pop('synopsis', None) + meta_regex = rf'{re.escape(meta_prefix)}(?P<i>\d+)?_(?P<key>.+)' for key, value in info.items(): - if value is not None and key != meta_prefix and key.startswith(meta_prefix): - metadata[key[len(meta_prefix):]] = value + mobj = re.fullmatch(meta_regex, key) + if value is not None and mobj: + metadata[mobj.group('i') or 'common'][mobj.group('key')] = value - for name, value in metadata.items(): + for name, value in metadata['common'].items(): yield ('-metadata', f'{name}={value}') stream_idx = 0 for fmt in info.get('requested_formats') or []: stream_count = 2 if 'none' not in (fmt.get('vcodec'), fmt.get('acodec')) else 1 - if fmt.get('language'): - lang = ISO639Utils.short2long(fmt['language']) or fmt['language'] - for i in range(stream_count): - yield ('-metadata:s:%d' % (stream_idx + i), 'language=%s' % lang) + lang = ISO639Utils.short2long(fmt.get('language') or '') or fmt.get('language') + for i in range(stream_idx, stream_idx + stream_count): + if lang: + metadata[str(i)].setdefault('language', lang) + for name, value in metadata[str(i)].items(): + yield (f'-metadata:s:{i}', f'{name}={value}') stream_idx += stream_count - if ('no-attach-info-json' not in self.get_param('compat_opts', []) - and '__infojson_filename' in info and info['ext'] in ('mkv', 'mka')): - old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json') - if old_stream is not None: - yield ('-map', '-0:%d' % old_stream) - new_stream -= 1 + def _get_infojson_opts(self, info, infofn): + if not infofn or not os.path.exists(infofn): + if self._add_infojson is not True: + return + infofn = infofn or '%s.temp' % ( + self._downloader.prepare_filename(info, 'infojson') + or replace_extension(self._downloader.prepare_filename(info), 'info.json', info['ext'])) + if not self._downloader._ensure_dir_exists(infofn): + return + self.write_debug(f'Writing info-json to: {infofn}') + write_json_file(self._downloader.sanitize_info(info, self.get_param('clean_infojson', True)), infofn) + info['infojson_filename'] = infofn + + old_stream, new_stream = self.get_stream_number(info['filepath'], ('tags', 'mimetype'), 'application/json') + if old_stream is not None: + yield ('-map', '-0:%d' % old_stream) + new_stream -= 1 - yield ('-attach', info['__infojson_filename'], - '-metadata:s:%d' % new_stream, 'mimetype=application/json') + yield ('-attach', infofn, + '-metadata:s:%d' % new_stream, 'mimetype=application/json') class FFmpegMergerPP(FFmpegPostProcessor): @@ -775,7 +859,7 @@ class FFmpegFixupStretchedPP(FFmpegFixupPostProcessor): stretched_ratio = info.get('stretched_ratio') if stretched_ratio not in (None, 1): self._fixup('Fixing aspect ratio', info['filepath'], [ - '-c', 'copy', '-map', '0', '-dn', '-aspect', '%f' % stretched_ratio]) + *self.stream_copy_opts(), '-aspect', '%f' % stretched_ratio]) return [], info @@ -783,17 +867,27 @@ class FFmpegFixupM4aPP(FFmpegFixupPostProcessor): @PostProcessor._restrict_to(images=False, video=False) def run(self, info): if info.get('container') == 'm4a_dash': - self._fixup('Correcting container', info['filepath'], [ - '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4']) + self._fixup('Correcting container', info['filepath'], [*self.stream_copy_opts(), '-f', 'mp4']) return [], info class FFmpegFixupM3u8PP(FFmpegFixupPostProcessor): + def _needs_fixup(self, info): + yield info['ext'] in ('mp4', 'm4a') + yield info['protocol'].startswith('m3u8') + try: + metadata = self.get_metadata_object(info['filepath']) + except PostProcessingError as e: + self.report_warning(f'Unable to extract metadata: {e.msg}') + yield True + else: + yield traverse_obj(metadata, ('format', 'format_name'), casesense=False) == 'mpegts' + @PostProcessor._restrict_to(images=False) def run(self, info): - if self.get_audio_codec(info['filepath']) == 'aac': - self._fixup('Fixing malformed AAC bitstream', info['filepath'], [ - '-c', 'copy', '-map', '0', '-dn', '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']) + if all(self._needs_fixup(info)): + self._fixup('Fixing MPEG-TS in MP4 container', info['filepath'], [ + *self.stream_copy_opts(), '-f', 'mp4', '-bsf:a', 'aac_adtstoasc']) return [], info @@ -807,25 +901,34 @@ class FFmpegFixupTimestampPP(FFmpegFixupPostProcessor): @PostProcessor._restrict_to(images=False) def run(self, info): - required_version = '4.4' - if is_outdated_version(self._versions[self.basename], required_version): + if not self._features.get('setts'): self.report_warning( 'A re-encode is needed to fix timestamps in older versions of ffmpeg. ' - f'Please install ffmpeg {required_version} or later to fixup without re-encoding') + 'Please install ffmpeg 4.4 or later to fixup without re-encoding') opts = ['-vf', 'setpts=PTS-STARTPTS'] else: opts = ['-c', 'copy', '-bsf', 'setts=ts=TS-STARTPTS'] - self._fixup('Fixing frame timestamp', info['filepath'], opts + ['-map', '0', '-dn', '-ss', self.trim]) + self._fixup('Fixing frame timestamp', info['filepath'], opts + [*self.stream_copy_opts(False), '-ss', self.trim]) return [], info -class FFmpegFixupDurationPP(FFmpegFixupPostProcessor): +class FFmpegCopyStreamPP(FFmpegFixupPostProcessor): + MESSAGE = 'Copying stream' + @PostProcessor._restrict_to(images=False) def run(self, info): - self._fixup('Fixing video duration', info['filepath'], ['-c', 'copy', '-map', '0', '-dn']) + self._fixup(self.MESSAGE, info['filepath'], self.stream_copy_opts()) return [], info +class FFmpegFixupDurationPP(FFmpegCopyStreamPP): + MESSAGE = 'Fixing video duration' + + +class FFmpegFixupDuplicateMoovPP(FFmpegCopyStreamPP): + MESSAGE = 'Fixing duplicate MOOV atoms' + + class FFmpegSubtitlesConvertorPP(FFmpegPostProcessor): SUPPORTED_EXTS = ('srt', 'vtt', 'ass', 'lrc') @@ -942,14 +1045,14 @@ class FFmpegSplitChaptersPP(FFmpegPostProcessor): self.to_screen('Splitting video by chapters; %d chapters found' % len(chapters)) for idx, chapter in enumerate(chapters): destination, opts = self._ffmpeg_args_for_chapter(idx + 1, chapter, info) - self.real_run_ffmpeg([(in_file, opts)], [(destination, ['-c', 'copy'])]) + self.real_run_ffmpeg([(in_file, opts)], [(destination, self.stream_copy_opts())]) if in_file != info['filepath']: os.remove(in_file) return [], info class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor): - SUPPORTED_EXTS = ('jpg', 'png') + SUPPORTED_EXTS = ('jpg', 'png', 'webp') def __init__(self, downloader=None, format=None): super(FFmpegThumbnailsConvertorPP, self).__init__(downloader) @@ -993,12 +1096,12 @@ class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor): files_to_delete = [] has_thumbnail = False - for idx, thumbnail_dict in enumerate(info['thumbnails']): - if 'filepath' not in thumbnail_dict: + for idx, thumbnail_dict in enumerate(info.get('thumbnails') or []): + original_thumbnail = thumbnail_dict.get('filepath') + if not original_thumbnail: continue has_thumbnail = True self.fixup_webp(info, idx) - original_thumbnail = thumbnail_dict['filepath'] _, thumbnail_ext = os.path.splitext(original_thumbnail) if thumbnail_ext: thumbnail_ext = thumbnail_ext[1:].lower() @@ -1015,3 +1118,57 @@ class FFmpegThumbnailsConvertorPP(FFmpegPostProcessor): if not has_thumbnail: self.to_screen('There aren\'t any thumbnails to convert') return files_to_delete, info + + +class FFmpegConcatPP(FFmpegPostProcessor): + def __init__(self, downloader, only_multi_video=False): + self._only_multi_video = only_multi_video + super().__init__(downloader) + + def _get_codecs(self, file): + codecs = traverse_obj(self.get_metadata_object(file), ('streams', ..., 'codec_name')) + self.write_debug(f'Codecs = {", ".join(codecs)}') + return tuple(codecs) + + def concat_files(self, in_files, out_file): + if not self._downloader._ensure_dir_exists(out_file): + return + if len(in_files) == 1: + if os.path.realpath(in_files[0]) != os.path.realpath(out_file): + self.to_screen(f'Moving "{in_files[0]}" to "{out_file}"') + os.replace(in_files[0], out_file) + return [] + + if len(set(map(self._get_codecs, in_files))) > 1: + raise PostProcessingError( + 'The files have different streams/codecs and cannot be concatenated. ' + 'Either select different formats or --recode-video them to a common format') + + self.to_screen(f'Concatenating {len(in_files)} files; Destination: {out_file}') + super().concat_files(in_files, out_file) + return in_files + + @PostProcessor._restrict_to(images=False, simulated=False) + def run(self, info): + entries = info.get('entries') or [] + if not any(entries) or (self._only_multi_video and info['_type'] != 'multi_video'): + return [], info + elif traverse_obj(entries, (..., 'requested_downloads', lambda _, v: len(v) > 1)): + raise PostProcessingError('Concatenation is not supported when downloading multiple separate formats') + + in_files = traverse_obj(entries, (..., 'requested_downloads', 0, 'filepath')) or [] + if len(in_files) < len(entries): + raise PostProcessingError('Aborting concatenation because some downloads failed') + + ie_copy = self._downloader._playlist_infodict(info) + exts = traverse_obj(entries, (..., 'requested_downloads', 0, 'ext'), (..., 'ext')) + ie_copy['ext'] = exts[0] if len(set(exts)) == 1 else 'mkv' + out_file = self._downloader.prepare_filename(ie_copy, 'pl_video') + + files_to_delete = self.concat_files(in_files, out_file) + + info['requested_downloads'] = [{ + 'filepath': out_file, + 'ext': ie_copy['ext'], + }] + return files_to_delete, info diff --git a/hypervideo_dl/postprocessor/metadataparser.py b/hypervideo_dl/postprocessor/metadataparser.py index 96aac9b..01ee6c1 100644 --- a/hypervideo_dl/postprocessor/metadataparser.py +++ b/hypervideo_dl/postprocessor/metadataparser.py @@ -1,5 +1,4 @@ import re - from enum import Enum from .common import PostProcessor @@ -16,7 +15,7 @@ class MetadataParserPP(PostProcessor): for f in actions: action = f[0] assert isinstance(action, self.Actions) - self._actions.append(getattr(self, action._value_)(*f[1:])) + self._actions.append(getattr(self, action.value)(*f[1:])) @classmethod def validate_action(cls, action, *data): @@ -26,12 +25,17 @@ class MetadataParserPP(PostProcessor): ''' if not isinstance(action, cls.Actions): raise ValueError(f'{action!r} is not a valid action') - getattr(cls, action._value_)(cls, *data) + getattr(cls, action.value)(cls, *data) # So this can raise error to validate @staticmethod def field_to_template(tmpl): if re.match(r'[a-zA-Z_]+$', tmpl): return f'%({tmpl})s' + + from ..YoutubeDL import YoutubeDL + err = YoutubeDL.validate_outtmpl(tmpl) + if err: + raise err return tmpl @staticmethod @@ -66,7 +70,7 @@ class MetadataParserPP(PostProcessor): self.write_debug(f'Searching for {out_re.pattern!r} in {template!r}') match = out_re.search(data_to_parse) if match is None: - self.report_warning(f'Could not interpret {inp!r} as {out!r}') + self.to_screen(f'Could not interpret {inp!r} as {out!r}') return for attribute, value in match.groupdict().items(): info[attribute] = value @@ -80,7 +84,7 @@ class MetadataParserPP(PostProcessor): def f(info): val = info.get(field) if val is None: - self.report_warning(f'Video does not have a {field}') + self.to_screen(f'Video does not have a {field}') return elif not isinstance(val, str): self.report_warning(f'Cannot replace in field {field} since it is a {type(val).__name__}') @@ -99,18 +103,23 @@ class MetadataParserPP(PostProcessor): class MetadataFromFieldPP(MetadataParserPP): @classmethod def to_action(cls, f): - match = re.match(r'(?P<in>.*?)(?<!\\):(?P<out>.+)$', f) + match = re.match(r'(?s)(?P<in>.*?)(?<!\\):(?P<out>.+)$', f) if match is None: raise ValueError(f'it should be FROM:TO, not {f!r}') return ( cls.Actions.INTERPRET, match.group('in').replace('\\:', ':'), - match.group('out')) + match.group('out'), + ) def __init__(self, downloader, formats): - MetadataParserPP.__init__(self, downloader, [self.to_action(f) for f in formats]) + super().__init__(downloader, [self.to_action(f) for f in formats]) -class MetadataFromTitlePP(MetadataParserPP): # for backward compatibility +# Deprecated +class MetadataFromTitlePP(MetadataParserPP): def __init__(self, downloader, titleformat): - MetadataParserPP.__init__(self, downloader, [(self.Actions.INTERPRET, 'title', titleformat)]) + super().__init__(downloader, [(self.Actions.INTERPRET, 'title', titleformat)]) + self.deprecation_warning( + 'hypervideo_dl.postprocessor.MetadataFromTitlePP is deprecated ' + 'and may be removed in a future version. Use hypervideo_dl.postprocessor.MetadataFromFieldPP instead') diff --git a/hypervideo_dl/postprocessor/modify_chapters.py b/hypervideo_dl/postprocessor/modify_chapters.py index a0818c4..22506bc 100644 --- a/hypervideo_dl/postprocessor/modify_chapters.py +++ b/hypervideo_dl/postprocessor/modify_chapters.py @@ -24,19 +24,21 @@ class ModifyChaptersPP(FFmpegPostProcessor): *, sponsorblock_chapter_title=DEFAULT_SPONSORBLOCK_CHAPTER_TITLE, force_keyframes=False): FFmpegPostProcessor.__init__(self, downloader) self._remove_chapters_patterns = set(remove_chapters_patterns or []) - self._remove_sponsor_segments = set(remove_sponsor_segments or []) + self._remove_sponsor_segments = set(remove_sponsor_segments or []) - set(SponsorBlockPP.POI_CATEGORIES.keys()) self._ranges_to_remove = set(remove_ranges or []) self._sponsorblock_chapter_title = sponsorblock_chapter_title self._force_keyframes = force_keyframes @PostProcessor._restrict_to(images=False) def run(self, info): + # Chapters must be preserved intact when downloading multiple formats of the same video. chapters, sponsor_chapters = self._mark_chapters_to_remove( - info.get('chapters') or [], info.get('sponsorblock_chapters') or []) + copy.deepcopy(info.get('chapters')) or [], + copy.deepcopy(info.get('sponsorblock_chapters')) or []) if not chapters and not sponsor_chapters: return [], info - real_duration = self._get_real_video_duration(info) + real_duration = self._get_real_video_duration(info['filepath']) if not chapters: chapters = [{'start_time': 0, 'end_time': real_duration, 'title': info['title']}] @@ -55,6 +57,7 @@ class ModifyChaptersPP(FFmpegPostProcessor): self.write_debug('Expected and actual durations mismatch') concat_opts = self._make_concat_opts(cuts, real_duration) + self.write_debug('Concat spec = %s' % ', '.join(f'{c.get("inpoint", 0.0)}-{c.get("outpoint", "inf")}' for c in concat_opts)) def remove_chapters(file, is_sub): return file, self.remove_chapters(file, cuts, concat_opts, self._force_keyframes and not is_sub) @@ -65,12 +68,13 @@ class ModifyChaptersPP(FFmpegPostProcessor): # Renaming should only happen after all files are processed files_to_remove = [] for in_file, out_file in in_out_files: + mtime = os.stat(in_file).st_mtime uncut_file = prepend_extension(in_file, 'uncut') os.replace(in_file, uncut_file) os.replace(out_file, in_file) + self.try_utime(in_file, mtime, mtime) files_to_remove.append(uncut_file) - info['_real_duration'] = info['chapters'][-1]['end_time'] return files_to_remove, info def _mark_chapters_to_remove(self, chapters, sponsor_chapters): @@ -126,7 +130,7 @@ class ModifyChaptersPP(FFmpegPostProcessor): cuts = [] def append_cut(c): - assert 'remove' in c + assert 'remove' in c, 'Not a cut is appended to cuts' last_to_cut = cuts[-1] if cuts else None if last_to_cut and last_to_cut['end_time'] >= c['start_time']: last_to_cut['end_time'] = max(last_to_cut['end_time'], c['end_time']) @@ -154,7 +158,7 @@ class ModifyChaptersPP(FFmpegPostProcessor): new_chapters = [] def append_chapter(c): - assert 'remove' not in c + assert 'remove' not in c, 'Cut is appended to chapters' length = c['end_time'] - c['start_time'] - excess_duration(c) # Chapter is completely covered by cuts or sponsors. if length <= 0: @@ -237,7 +241,7 @@ class ModifyChaptersPP(FFmpegPostProcessor): heapq.heappush(chapters, (c['start_time'], i, c)) # (normal, sponsor) and (sponsor, sponsor) else: - assert '_categories' in c + assert '_categories' in c, 'Normal chapters overlap' cur_chapter['_was_cut'] = True c['_was_cut'] = True # Push the part after the sponsor to PQ. @@ -301,7 +305,7 @@ class ModifyChaptersPP(FFmpegPostProcessor): 'name': SponsorBlockPP.CATEGORIES[category], 'category_names': [SponsorBlockPP.CATEGORIES[c] for c in cats] }) - c['title'] = self._downloader.evaluate_outtmpl(self._sponsorblock_chapter_title, c) + c['title'] = self._downloader.evaluate_outtmpl(self._sponsorblock_chapter_title, c.copy()) # Merge identically named sponsors. if (new_chapters and 'categories' in new_chapters[-1] and new_chapters[-1]['title'] == c['title']): @@ -331,6 +335,6 @@ class ModifyChaptersPP(FFmpegPostProcessor): continue opts[-1]['outpoint'] = f'{s["start_time"]:.6f}' # Do not create 0 duration chunk at the end. - if s['end_time'] != duration: + if s['end_time'] < duration: opts.append({'inpoint': f'{s["end_time"]:.6f}'}) return opts diff --git a/hypervideo_dl/postprocessor/sponskrub.py b/hypervideo_dl/postprocessor/sponskrub.py index 932555a..400cbcc 100644 --- a/hypervideo_dl/postprocessor/sponskrub.py +++ b/hypervideo_dl/postprocessor/sponskrub.py @@ -11,9 +11,9 @@ from ..utils import ( encodeFilename, shell_quote, str_or_none, + Popen, PostProcessingError, prepend_extension, - process_communicate_or_kill, ) @@ -22,13 +22,18 @@ class SponSkrubPP(PostProcessor): _temp_ext = 'spons' _exe_name = 'sponskrub' - def __init__(self, downloader, path='', args=None, ignoreerror=False, cut=False, force=False): + def __init__(self, downloader, path='', args=None, ignoreerror=False, cut=False, force=False, _from_cli=False): PostProcessor.__init__(self, downloader) self.force = force self.cutout = cut self.args = str_or_none(args) or '' # For backward compatibility self.path = self.get_exe(path) + if not _from_cli: + self.deprecation_warning( + 'hypervideo_dl.postprocessor.SponSkrubPP support is deprecated and may be removed in a future version. ' + 'Use hypervideo_dl.postprocessor.SponsorBlock and hypervideo_dl.postprocessor.ModifyChaptersPP instead') + if not ignoreerror and self.path is None: if path: raise PostProcessingError('sponskrub not found in "%s"' % path) @@ -81,8 +86,8 @@ class SponSkrubPP(PostProcessor): self.write_debug('sponskrub command line: %s' % shell_quote(cmd)) pipe = None if self.get_param('verbose') else subprocess.PIPE - p = subprocess.Popen(cmd, stdout=pipe) - stdout = process_communicate_or_kill(p)[0] + p = Popen(cmd, stdout=pipe) + stdout = p.communicate_or_kill()[0] if p.returncode == 0: os.replace(temp_filename, filename) diff --git a/hypervideo_dl/postprocessor/sponsorblock.py b/hypervideo_dl/postprocessor/sponsorblock.py index 7265a9d..7943014 100644 --- a/hypervideo_dl/postprocessor/sponsorblock.py +++ b/hypervideo_dl/postprocessor/sponsorblock.py @@ -1,25 +1,29 @@ +from hashlib import sha256 import json import re -from hashlib import sha256 from .ffmpeg import FFmpegPostProcessor -from ..compat import compat_urllib_parse_urlencode, compat_HTTPError -from ..utils import PostProcessingError, network_exceptions, sanitized_Request +from ..compat import compat_urllib_parse_urlencode class SponsorBlockPP(FFmpegPostProcessor): - + # https://wiki.sponsor.ajay.app/w/Types EXTRACTORS = { 'Youtube': 'YouTube', } + POI_CATEGORIES = { + 'poi_highlight': 'Highlight', + } CATEGORIES = { 'sponsor': 'Sponsor', 'intro': 'Intermission/Intro Animation', 'outro': 'Endcards/Credits', 'selfpromo': 'Unpaid/Self Promotion', - 'interaction': 'Interaction Reminder', 'preview': 'Preview/Recap', - 'music_offtopic': 'Non-Music Section' + 'filler': 'Filler Tangent', + 'interaction': 'Interaction Reminder', + 'music_offtopic': 'Non-Music Section', + **POI_CATEGORIES, } def __init__(self, downloader, categories=None, api='https://sponsor.ajay.app'): @@ -33,6 +37,7 @@ class SponsorBlockPP(FFmpegPostProcessor): self.to_screen(f'SponsorBlock is not supported for {extractor}') return [], info + self.to_screen('Fetching SponsorBlock segments') info['sponsorblock_chapters'] = self._get_sponsor_chapters(info, info['duration']) return [], info @@ -41,9 +46,15 @@ class SponsorBlockPP(FFmpegPostProcessor): def duration_filter(s): start_end = s['segment'] + # Ignore entire video segments (https://wiki.sponsor.ajay.app/w/Types). + if start_end == (0, 0): + return False # Ignore milliseconds difference at the start. if start_end[0] <= 1: start_end[0] = 0 + # Make POI chapters 1 sec so that we can properly mark them + if s['category'] in self.POI_CATEGORIES.keys(): + start_end[1] += 1 # Ignore milliseconds difference at the end. # Never allow the segment to exceed the video. if duration and duration - start_end[1] <= 1: @@ -78,19 +89,9 @@ class SponsorBlockPP(FFmpegPostProcessor): url = f'{self._API_URL}/api/skipSegments/{hash[:4]}?' + compat_urllib_parse_urlencode({ 'service': service, 'categories': json.dumps(self._categories), + 'actionTypes': json.dumps(['skip', 'poi']) }) - for d in self._get_json(url): + for d in self._download_json(url) or []: if d['videoID'] == video_id: return d['segments'] return [] - - def _get_json(self, url): - self.write_debug(f'SponsorBlock query: {url}') - try: - rsp = self._downloader.urlopen(sanitized_Request(url)) - except network_exceptions as e: - if isinstance(e, compat_HTTPError) and e.code == 404: - return [] - raise PostProcessingError(f'Unable to communicate with SponsorBlock API - {e}') - - return json.loads(rsp.read().decode(rsp.info().get_param('charset') or 'utf-8')) |