]> jfr.im git - yt-dlp.git/blobdiff - yt_dlp/YoutubeDL.py
[utils] Improve `traverse_obj`
[yt-dlp.git] / yt_dlp / YoutubeDL.py
index 8e058485a505b9647eb7c77b12d3b986e7c6edf2..3dfab69b2dcf20843922ae1eff2bacbdd9979009 100644 (file)
@@ -1,4 +1,4 @@
-#!/usr/bin/env python
+#!/usr/bin/env python3
 # coding: utf-8
 
 from __future__ import absolute_import, unicode_literals
@@ -20,6 +20,7 @@
 import shutil
 import subprocess
 import sys
+import tempfile
 import time
 import tokenize
 import traceback
     float_or_none,
     format_bytes,
     format_field,
-    FORMAT_RE,
+    STR_FORMAT_RE,
     formatSeconds,
     GeoRestrictedError,
+    HEADRequest,
     int_or_none,
     iri_to_uri,
     ISO3166Utils,
+    LazyList,
     locked_file,
     make_dir,
     make_HTTPS_handler,
@@ -85,7 +88,6 @@
     preferredencoding,
     prepend_extension,
     process_communicate_or_kill,
-    random_uuidv4,
     register_socks_protocols,
     RejectedVideoReached,
     render_table,
     str_or_none,
     strftime_or_none,
     subtitles_filename,
+    ThrottledDownload,
     to_high_limit_path,
-    traverse_dict,
+    traverse_obj,
+    try_get,
     UnavailableVideoError,
     url_basename,
     version_tuple,
 )
 from .downloader.rtmp import rtmpdump_version
 from .postprocessor import (
+    get_postprocessor,
+    FFmpegFixupDurationPP,
     FFmpegFixupM3u8PP,
     FFmpegFixupM4aPP,
     FFmpegFixupStretchedPP,
+    FFmpegFixupTimestampPP,
     FFmpegMergerPP,
     FFmpegPostProcessor,
-    # FFmpegSubtitlesConvertorPP,
-    get_postprocessor,
     MoveFilesAfterDownloadPP,
 )
 from .version import __version__
@@ -204,6 +209,9 @@ class YoutubeDL(object):
                        into a single file
     allow_multiple_audio_streams:   Allow multiple audio streams to be merged
                        into a single file
+    check_formats      Whether to test if the formats are downloadable.
+                       Can be True (check all), False (check none)
+                       or None (check only if requested by extractor)
     paths:             Dictionary of output paths. The allowed keys are 'home'
                        'temp' and the keys of OUTTMPL_TYPES (in utils.py)
     outtmpl:           Dictionary of templates for output names. Allowed keys
@@ -387,17 +395,15 @@ class YoutubeDL(object):
                        if True, otherwise use ffmpeg/avconv if False, otherwise
                        use downloader suggested by extractor if None.
     compat_opts:       Compatibility options. See "Differences in default behavior".
-                       Note that only format-sort, format-spec, no-live-chat,
-                       no-attach-info-json, playlist-index, list-formats,
-                       no-direct-merge, no-youtube-channel-redirect,
-                       and no-youtube-unavailable-videos works when used via the API
+                       The following options do not work when used through the API:
+                       filename, abort-on-error, multistreams, no-live-chat,
+                       no-playlist-metafiles. Refer __init__.py for their implementation
 
     The following parameters are not used by YoutubeDL itself, they are used by
     the downloader (see yt_dlp/downloader/common.py):
-    nopart, updatetime, buffersize, ratelimit, min_filesize, max_filesize, test,
-    noresizebuffer, retries, continuedl, noprogress, consoletitle,
-    xattr_set_filesize, external_downloader_args, hls_use_mpegts,
-    http_chunk_size.
+    nopart, updatetime, buffersize, ratelimit, throttledratelimit, min_filesize,
+    max_filesize, test, noresizebuffer, retries, continuedl, noprogress, consoletitle,
+    xattr_set_filesize, external_downloader_args, hls_use_mpegts, http_chunk_size.
 
     The following options are used by the post processors:
     prefer_ffmpeg:     If False, use avconv instead of ffmpeg if both are available,
@@ -415,11 +421,16 @@ class YoutubeDL(object):
     dynamic_mpd:       Whether to process dynamic DASH manifests (default: True)
     hls_split_discontinuity: Split HLS playlists to different formats at
                        discontinuities such as ad breaks (default: False)
-    youtube_include_dash_manifest: If True (default), DASH manifests and related
+    extractor_args:    A dictionary of arguments to be passed to the extractors.
+                       See "EXTRACTOR ARGUMENTS" for details.
+                       Eg: {'youtube': {'skip': ['dash', 'hls']}}
+    youtube_include_dash_manifest: Deprecated - Use extractor_args instead.
+                       If True (default), DASH manifests and related
                        data will be downloaded and processed by extractor.
                        You can reduce network I/O by disabling it if you don't
                        care about DASH. (only for youtube)
-    youtube_include_hls_manifest: If True (default), HLS manifests and related
+    youtube_include_hls_manifest: Deprecated - Use extractor_args instead.
+                       If True (default), HLS manifests and related
                        data will be downloaded and processed by extractor.
                        You can reduce network I/O by disabling it if you don't
                        care about HLS. (only for youtube)
@@ -439,7 +450,7 @@ class YoutubeDL(object):
     params = None
     _ies = []
     _pps = {'pre_process': [], 'before_dl': [], 'after_move': [], 'post_process': []}
-    __prepare_filename_warned = False
+    _reported_warnings = set()
     _first_webpage_request = True
     _download_retcode = None
     _num_downloads = None
@@ -454,7 +465,7 @@ def __init__(self, params=None, auto_init=True):
         self._ies = []
         self._ies_instances = {}
         self._pps = {'pre_process': [], 'before_dl': [], 'after_move': [], 'post_process': []}
-        self.__prepare_filename_warned = False
+        self._reported_warnings = set()
         self._first_webpage_request = True
         self._post_hooks = []
         self._progress_hooks = []
@@ -471,8 +482,7 @@ def __init__(self, params=None, auto_init=True):
 
         if sys.version_info < (3, 6):
             self.report_warning(
-                'Support for Python version %d.%d have been deprecated and will break in future versions of yt-dlp! '
-                'Update to Python 3.6 or above' % sys.version_info[:2])
+                'Python version %d.%d is not supported! Please update to Python 3.6 or above' % sys.version_info[:2])
 
         def check_deprecated(param, option, suggestion):
             if self.params.get(param) is not None:
@@ -538,6 +548,11 @@ def check_deprecated(param, option, suggestion):
 
         self.outtmpl_dict = self.parse_outtmpl()
 
+        # Creating format selector here allows us to catch syntax errors before the extraction
+        self.format_selector = (
+            None if self.params.get('format') is None
+            else self.build_format_selector(self.params['format']))
+
         self._setup_opener()
 
         """Preload the archive, if any is specified"""
@@ -563,14 +578,9 @@ def preload_download_archive(fn):
             self.add_default_info_extractors()
 
         for pp_def_raw in self.params.get('postprocessors', []):
-            pp_class = get_postprocessor(pp_def_raw['key'])
             pp_def = dict(pp_def_raw)
-            del pp_def['key']
-            if 'when' in pp_def:
-                when = pp_def['when']
-                del pp_def['when']
-            else:
-                when = 'post_process'
+            when = pp_def.pop('when', 'post_process')
+            pp_class = get_postprocessor(pp_def.pop('key'))
             pp = pp_class(self, **compat_kwargs(pp_def))
             self.add_post_processor(pp, when=when)
 
@@ -657,12 +667,10 @@ def to_stdout(self, message, skip_eol=False, quiet=False):
         """Print message to stdout"""
         if self.params.get('logger'):
             self.params['logger'].debug(message)
-        elif not quiet:
-            message = self._bidi_workaround(message)
-            terminator = ['\n', ''][skip_eol]
-            output = message + terminator
-
-            self._write_string(output, self._screen_file)
+        elif not quiet or self.params.get('verbose'):
+            self._write_string(
+                '%s%s' % (self._bidi_workaround(message), ('' if skip_eol else '\n')),
+                self._err_file if quiet else self._screen_file)
 
     def to_stderr(self, message):
         """Print message to stderr"""
@@ -670,9 +678,7 @@ def to_stderr(self, message):
         if self.params.get('logger'):
             self.params['logger'].error(message)
         else:
-            message = self._bidi_workaround(message)
-            output = message + '\n'
-            self._write_string(output, self._err_file)
+            self._write_string('%s\n' % self._bidi_workaround(message), self._err_file)
 
     def to_console_title(self, message):
         if not self.params.get('consoletitle', False):
@@ -749,11 +755,15 @@ def to_screen(self, message, skip_eol=False):
         self.to_stdout(
             message, skip_eol, quiet=self.params.get('quiet', False))
 
-    def report_warning(self, message):
+    def report_warning(self, message, only_once=False):
         '''
         Print the message to stderr, it will be prefixed with 'WARNING:'
         If stderr is a tty file the 'WARNING:' will be colored
         '''
+        if only_once:
+            if message in self._reported_warnings:
+                return
+            self._reported_warnings.add(message)
         if self.params.get('logger') is not None:
             self.params['logger'].warning(message)
         else:
@@ -816,134 +826,154 @@ def parse_outtmpl(self):
                     'Put  from __future__ import unicode_literals  at the top of your code file or consider switching to Python 3.x.')
         return outtmpl_dict
 
+    def get_output_path(self, dir_type='', filename=None):
+        paths = self.params.get('paths', {})
+        assert isinstance(paths, dict)
+        path = os.path.join(
+            expand_path(paths.get('home', '').strip()),
+            expand_path(paths.get(dir_type, '').strip()) if dir_type else '',
+            filename or '')
+
+        # Temporary fix for #4787
+        # 'Treat' all problem characters by passing filename through preferredencoding
+        # to workaround encoding issues with subprocess on python2 @ Windows
+        if sys.version_info < (3, 0) and sys.platform == 'win32':
+            path = encodeFilename(path, True).decode(preferredencoding())
+        return sanitize_path(path, force=self.params.get('windowsfilenames'))
+
+    @staticmethod
+    def validate_outtmpl(tmpl):
+        ''' @return None or Exception object '''
+        try:
+            re.sub(
+                STR_FORMAT_RE.format(''),
+                lambda mobj: ('%' if not mobj.group('has_key') else '') + mobj.group(0),
+                tmpl
+            ) % collections.defaultdict(int)
+            return None
+        except ValueError as err:
+            return err
+
     def prepare_outtmpl(self, outtmpl, info_dict, sanitize=None):
         """ Make the template and info_dict suitable for substitution (outtmpl % info_dict)"""
-        template_dict = dict(info_dict)
+        info_dict = dict(info_dict)
         na = self.params.get('outtmpl_na_placeholder', 'NA')
 
-        # duration_string
-        template_dict['duration_string'] = (  # %(duration>%H-%M-%S)s is wrong if duration > 24hrs
+        info_dict['duration_string'] = (  # %(duration>%H-%M-%S)s is wrong if duration > 24hrs
             formatSeconds(info_dict['duration'], '-' if sanitize else ':')
             if info_dict.get('duration', None) is not None
             else None)
-
-        # epoch
-        template_dict['epoch'] = int(time.time())
-
-        # autonumber
-        autonumber_size = self.params.get('autonumber_size')
-        if autonumber_size is None:
-            autonumber_size = 5
-        template_dict['autonumber'] = self.params.get('autonumber_start', 1) - 1 + self._num_downloads
-
-        # resolution if not defined
-        if template_dict.get('resolution') is None:
-            if template_dict.get('width') and template_dict.get('height'):
-                template_dict['resolution'] = '%dx%d' % (template_dict['width'], template_dict['height'])
-            elif template_dict.get('height'):
-                template_dict['resolution'] = '%sp' % template_dict['height']
-            elif template_dict.get('width'):
-                template_dict['resolution'] = '%dx?' % template_dict['width']
+        info_dict['epoch'] = int(time.time())
+        info_dict['autonumber'] = self.params.get('autonumber_start', 1) - 1 + self._num_downloads
+        if info_dict.get('resolution') is None:
+            info_dict['resolution'] = self.format_resolution(info_dict, default=None)
 
         # For fields playlist_index and autonumber convert all occurrences
         # of %(field)s to %(field)0Nd for backward compatibility
         field_size_compat_map = {
-            'playlist_index': len(str(template_dict.get('_last_playlist_index') or '')),
-            'autonumber': autonumber_size,
+            'playlist_index': len(str(info_dict.get('_last_playlist_index') or '')),
+            'autonumber': self.params.get('autonumber_size') or 5,
+        }
+
+        TMPL_DICT = {}
+        EXTERNAL_FORMAT_RE = re.compile(STR_FORMAT_RE.format('[^)]*'))
+        MATH_FUNCTIONS = {
+            '+': float.__add__,
+            '-': float.__sub__,
         }
-        FIELD_SIZE_COMPAT_RE = r'(?<!%)%\((?P<field>autonumber|playlist_index)\)s'
-        mobj = re.search(FIELD_SIZE_COMPAT_RE, outtmpl)
-        if mobj:
-            outtmpl = re.sub(
-                FIELD_SIZE_COMPAT_RE,
-                r'%%(\1)0%dd' % field_size_compat_map[mobj.group('field')],
-                outtmpl)
-
-        numeric_fields = list(self._NUMERIC_FIELDS)
-        if sanitize is None:
-            sanitize = lambda k, v: v
-
-        EXTERNAL_FORMAT_RE = FORMAT_RE.format('(?P<key>[^)]*)')
         # Field is of the form key1.key2...
         # where keys (except first) can be string, int or slice
-        FIELD_RE = r'\w+(?:\.(?:\w+|[-\d]*(?::[-\d]*){0,2}))*'
+        FIELD_RE = r'\w+(?:\.(?:\w+|{num}|{num}?(?::{num}?){{1,2}}))*'.format(num=r'(?:-?\d+)')
+        MATH_FIELD_RE = r'''{field}|{num}'''.format(field=FIELD_RE, num=r'-?\d+(?:.\d+)?')
+        MATH_OPERATORS_RE = r'(?:%s)' % '|'.join(map(re.escape, MATH_FUNCTIONS.keys()))
         INTERNAL_FORMAT_RE = re.compile(r'''(?x)
             (?P<negate>-)?
-            (?P<fields>{0})
-            (?P<maths>(?:[-+]-?(?:\d+(?:\.\d+)?|{0}))*)
+            (?P<fields>{field})
+            (?P<maths>(?:{math_op}{math_field})*)
             (?:>(?P<strf_format>.+?))?
             (?:\|(?P<default>.*?))?
-            $'''.format(FIELD_RE))
-        MATH_OPERATORS_RE = re.compile(r'(?<![-+])([-+])')
-        MATH_FUNCTIONS = {
-            '+': float.__add__,
-            '-': float.__sub__,
-        }
-        for outer_mobj in re.finditer(EXTERNAL_FORMAT_RE, outtmpl):
-            final_key = outer_mobj.group('key')
-            str_type = outer_mobj.group('type')
-            value = None
-            mobj = re.match(INTERNAL_FORMAT_RE, final_key)
-            if mobj is not None:
-                mobj = mobj.groupdict()
-                # Object traversal
-                fields = mobj['fields'].split('.')
-                value = traverse_dict(template_dict, fields)
-                # Negative
-                if mobj['negate']:
-                    value = float_or_none(value)
-                    if value is not None:
-                        value *= -1
-                # Do maths
-                if mobj['maths']:
-                    value = float_or_none(value)
+            $'''.format(field=FIELD_RE, math_op=MATH_OPERATORS_RE, math_field=MATH_FIELD_RE))
+
+        get_key = lambda k: traverse_obj(
+            info_dict, k.split('.'), is_user_input=True, traverse_string=True)
+
+        def get_value(mdict):
+            # Object traversal
+            value = get_key(mdict['fields'])
+            # Negative
+            if mdict['negate']:
+                value = float_or_none(value)
+                if value is not None:
+                    value *= -1
+            # Do maths
+            offset_key = mdict['maths']
+            if offset_key:
+                value = float_or_none(value)
+                operator = None
+                while offset_key:
+                    item = re.match(
+                        MATH_FIELD_RE if operator else MATH_OPERATORS_RE,
+                        offset_key).group(0)
+                    offset_key = offset_key[len(item):]
+                    if operator is None:
+                        operator = MATH_FUNCTIONS[item]
+                        continue
+                    item, multiplier = (item[1:], -1) if item[0] == '-' else (item, 1)
+                    offset = float_or_none(item)
+                    if offset is None:
+                        offset = float_or_none(get_key(item))
+                    try:
+                        value = operator(value, multiplier * offset)
+                    except (TypeError, ZeroDivisionError):
+                        return None
                     operator = None
-                    for item in MATH_OPERATORS_RE.split(mobj['maths'])[1:]:
-                        if item == '':
-                            value = None
-                        if value is None:
-                            break
-                        if operator:
-                            item, multiplier = (item[1:], -1) if item[0] == '-' else (item, 1)
-                            offset = float_or_none(item)
-                            if offset is None:
-                                offset = float_or_none(traverse_dict(template_dict, item.split('.')))
-                            try:
-                                value = operator(value, multiplier * offset)
-                            except (TypeError, ZeroDivisionError):
-                                value = None
-                            operator = None
-                        else:
-                            operator = MATH_FUNCTIONS[item]
-                # Datetime formatting
-                if mobj['strf_format']:
-                    value = strftime_or_none(value, mobj['strf_format'])
-                # Set default
-                if value is None and mobj['default'] is not None:
-                    value = mobj['default']
-            # Sanitize
-            if str_type in 'crs' and value is not None:  # string
-                value = sanitize('%{}'.format(str_type) % fields[-1], value)
-            else:  # numeric
-                numeric_fields.append(final_key)
+            # Datetime formatting
+            if mdict['strf_format']:
+                value = strftime_or_none(value, mdict['strf_format'])
+
+            return value
+
+        def create_key(outer_mobj):
+            if not outer_mobj.group('has_key'):
+                return '%{}'.format(outer_mobj.group(0))
+
+            key = outer_mobj.group('key')
+            fmt = outer_mobj.group('format')
+            mobj = re.match(INTERNAL_FORMAT_RE, key)
+            if mobj is None:
+                value, default, mobj = None, na, {'fields': ''}
+            else:
+                mobj = mobj.groupdict()
+                default = mobj['default'] if mobj['default'] is not None else na
+                value = get_value(mobj)
+
+            if fmt == 's' and value is not None and key in field_size_compat_map.keys():
+                fmt = '0{:d}d'.format(field_size_compat_map[key])
+
+            value = default if value is None else value
+
+            if fmt == 'c':
+                value = compat_str(value)
+                if value is None:
+                    value, fmt = default, 's'
+                else:
+                    value = value[0]
+            elif fmt[-1] not in 'rs':  # numeric
                 value = float_or_none(value)
-            if value is not None:
-                template_dict[final_key] = value
-
-        # Missing numeric fields used together with integer presentation types
-        # in format specification will break the argument substitution since
-        # string NA placeholder is returned for missing fields. We will patch
-        # output template for missing fields to meet string presentation type.
-        for numeric_field in numeric_fields:
-            if template_dict.get(numeric_field) is None:
-                outtmpl = re.sub(
-                    FORMAT_RE.format(re.escape(numeric_field)),
-                    r'%({0})s'.format(numeric_field), outtmpl)
-
-        template_dict = collections.defaultdict(lambda: na, (
-            (k, v if isinstance(v, compat_numeric_types) else sanitize(k, v))
-            for k, v in template_dict.items() if v is not None))
-        return outtmpl, template_dict
+                if value is None:
+                    value, fmt = default, 's'
+            if sanitize:
+                if fmt[-1] == 'r':
+                    # If value is an object, sanitize might convert it to a string
+                    # So we convert it to repr first
+                    value, fmt = repr(value), '%ss' % fmt[:-1]
+                if fmt[-1] in 'csr':
+                    value = sanitize(mobj['fields'].split('.')[-1], value)
+            key += '\0%s' % fmt
+            TMPL_DICT[key] = value
+            return '%({key}){fmt}'.format(key=key, fmt=fmt)
+
+        return EXTERNAL_FORMAT_RE.sub(create_key, outtmpl), TMPL_DICT
 
     def _prepare_filename(self, info_dict, tmpl_type='default'):
         try:
@@ -969,7 +999,7 @@ def _prepare_filename(self, info_dict, tmpl_type='default'):
 
             force_ext = OUTTMPL_TYPES.get(tmpl_type)
             if force_ext is not None:
-                filename = replace_extension(filename, force_ext, template_dict.get('ext'))
+                filename = replace_extension(filename, force_ext, info_dict.get('ext'))
 
             # https://github.com/blackjack4494/youtube-dlc/issues/85
             trim_file_name = self.params.get('trim_file_name', False)
@@ -988,35 +1018,23 @@ def _prepare_filename(self, info_dict, tmpl_type='default'):
 
     def prepare_filename(self, info_dict, dir_type='', warn=False):
         """Generate the output filename."""
-        paths = self.params.get('paths', {})
-        assert isinstance(paths, dict)
+
         filename = self._prepare_filename(info_dict, dir_type or 'default')
 
-        if warn and not self.__prepare_filename_warned:
-            if not paths:
+        if warn:
+            if not self.params.get('paths'):
                 pass
             elif filename == '-':
-                self.report_warning('--paths is ignored when an outputting to stdout')
+                self.report_warning('--paths is ignored when an outputting to stdout', only_once=True)
             elif os.path.isabs(filename):
-                self.report_warning('--paths is ignored since an absolute path is given in output template')
+                self.report_warning('--paths is ignored since an absolute path is given in output template', only_once=True)
             self.__prepare_filename_warned = True
         if filename == '-' or not filename:
             return filename
 
-        homepath = expand_path(paths.get('home', '').strip())
-        assert isinstance(homepath, compat_str)
-        subdir = expand_path(paths.get(dir_type, '').strip()) if dir_type else ''
-        assert isinstance(subdir, compat_str)
-        path = os.path.join(homepath, subdir, filename)
+        return self.get_output_path(dir_type, filename)
 
-        # Temporary fix for #4787
-        # 'Treat' all problem characters by passing filename through preferredencoding
-        # to workaround encoding issues with subprocess on python2 @ Windows
-        if sys.version_info < (3, 0) and sys.platform == 'win32':
-            path = encodeFilename(path, True).decode(preferredencoding())
-        return sanitize_path(path, force=self.params.get('windowsfilenames'))
-
-    def _match_entry(self, info_dict, incomplete=False):
+    def _match_entry(self, info_dict, incomplete=False, silent=False):
         """ Returns None if the file should be downloaded """
 
         video_title = info_dict.get('title', info_dict.get('id', 'video'))
@@ -1064,7 +1082,8 @@ def check_filter():
             reason = check_filter()
             break_opt, break_err = 'break_on_reject', RejectedVideoReached
         if reason is not None:
-            self.to_screen('[download] ' + reason)
+            if not silent:
+                self.to_screen('[download] ' + reason)
             if self.params.get(break_opt, False):
                 raise break_err()
         return reason
@@ -1137,6 +1156,10 @@ def wrapper(self, *args, **kwargs):
                 self.report_error(msg)
             except ExtractorError as e:  # An error we somewhat expected
                 self.report_error(compat_str(e), e.format_traceback())
+            except ThrottledDownload:
+                self.to_stderr('\r')
+                self.report_warning('The download speed is below throttle limit. Re-extracting data')
+                return wrapper(self, *args, **kwargs)
             except (MaxDownloadsReached, ExistingVideoReached, RejectedVideoReached):
                 raise
             except Exception as e:
@@ -1164,12 +1187,17 @@ def __extract_info(self, url, ie, download, extra_info, process):
             return ie_result
 
     def add_default_extra_info(self, ie_result, ie, url):
-        self.add_extra_info(ie_result, {
-            'extractor': ie.IE_NAME,
-            'webpage_url': url,
-            'webpage_url_basename': url_basename(url),
-            'extractor_key': ie.ie_key(),
-        })
+        if url is not None:
+            self.add_extra_info(ie_result, {
+                'webpage_url': url,
+                'original_url': url,
+                'webpage_url_basename': url_basename(url),
+            })
+        if ie is not None:
+            self.add_extra_info(ie_result, {
+                'extractor': ie.IE_NAME,
+                'extractor_key': ie.ie_key(),
+            })
 
     def process_ie_result(self, ie_result, download=True, extra_info={}):
         """
@@ -1186,7 +1214,11 @@ def process_ie_result(self, ie_result, download=True, extra_info={}):
             extract_flat = self.params.get('extract_flat', False)
             if ((extract_flat == 'in_playlist' and 'playlist' in extra_info)
                     or extract_flat is True):
-                self.__forced_printings(ie_result, self.prepare_filename(ie_result), incomplete=True)
+                info_copy = ie_result.copy()
+                self.add_extra_info(info_copy, extra_info)
+                ie = try_get(ie_result.get('ie_key'), self.get_info_extractor)
+                self.add_default_extra_info(info_copy, ie, ie_result['url'])
+                self.__forced_printings(info_copy, self.prepare_filename(info_copy), incomplete=True)
                 return ie_result
 
         if result_type == 'video':
@@ -1309,7 +1341,7 @@ def fill_missing_entries(entries, indexes):
 
         playlist_results = []
 
-        playliststart = self.params.get('playliststart', 1) - 1
+        playliststart = self.params.get('playliststart', 1)
         playlistend = self.params.get('playlistend')
         # For backwards compatibility, interpret -1 as whole list
         if playlistend == -1:
@@ -1329,50 +1361,48 @@ def iter_playlistitems(format):
             playlistitems = orderedSet(iter_playlistitems(playlistitems_str))
 
         ie_entries = ie_result['entries']
-
-        def make_playlistitems_entries(list_ie_entries):
-            num_entries = len(list_ie_entries)
-            for i in playlistitems:
-                if -num_entries < i <= num_entries:
-                    yield list_ie_entries[i - 1]
-                elif incomplete_entries:
+        msg = (
+            'Downloading %d videos' if not isinstance(ie_entries, list)
+            else 'Collected %d videos; downloading %%d of them' % len(ie_entries))
+        if not isinstance(ie_entries, (list, PagedList)):
+            ie_entries = LazyList(ie_entries)
+
+        def get_entry(i):
+            return YoutubeDL.__handle_extraction_exceptions(
+                lambda self, i: ie_entries[i - 1]
+            )(self, i)
+
+        entries = []
+        for i in playlistitems or itertools.count(playliststart):
+            if playlistitems is None and playlistend is not None and playlistend < i:
+                break
+            entry = None
+            try:
+                entry = get_entry(i)
+                if entry is None:
                     raise EntryNotInPlaylist()
+            except (IndexError, EntryNotInPlaylist):
+                if incomplete_entries:
+                    raise EntryNotInPlaylist()
+                elif not playlistitems:
+                    break
+            entries.append(entry)
+            try:
+                if entry is not None:
+                    self._match_entry(entry, incomplete=True, silent=True)
+            except (ExistingVideoReached, RejectedVideoReached):
+                break
+        ie_result['entries'] = entries
 
-        if isinstance(ie_entries, list):
-            n_all_entries = len(ie_entries)
-            if playlistitems:
-                entries = list(make_playlistitems_entries(ie_entries))
-            else:
-                entries = ie_entries[playliststart:playlistend]
-            n_entries = len(entries)
-            msg = 'Collected %d videos; downloading %d of them' % (n_all_entries, n_entries)
-        elif isinstance(ie_entries, PagedList):
-            if playlistitems:
-                entries = []
-                for item in playlistitems:
-                    entries.extend(ie_entries.getslice(
-                        item - 1, item
-                    ))
-            else:
-                entries = ie_entries.getslice(
-                    playliststart, playlistend)
-            n_entries = len(entries)
-            msg = 'Downloading %d videos' % n_entries
-        else:  # iterable
-            if playlistitems:
-                entries = list(make_playlistitems_entries(list(itertools.islice(
-                    ie_entries, 0, max(playlistitems)))))
-            else:
-                entries = list(itertools.islice(
-                    ie_entries, playliststart, playlistend))
-            n_entries = len(entries)
-            msg = 'Downloading %d videos' % n_entries
+        # Save playlist_index before re-ordering
+        entries = [
+            ((playlistitems[i - 1] if playlistitems else i), entry)
+            for i, entry in enumerate(entries, 1)
+            if entry is not None]
+        n_entries = len(entries)
 
-        if any((entry is None for entry in entries)):
-            raise EntryNotInPlaylist()
         if not playlistitems and (playliststart or playlistend):
-            playlistitems = list(range(1 + playliststart, 1 + playliststart + len(entries)))
-        ie_result['entries'] = entries
+            playlistitems = list(range(playliststart, playliststart + n_entries))
         ie_result['requested_entries'] = playlistitems
 
         if self.params.get('allow_playlist_files', True):
@@ -1419,11 +1449,6 @@ def make_playlistitems_entries(list_ie_entries):
                         self.report_error('Cannot write playlist description file ' + descfn)
                         return
 
-        # Save playlist_index before re-ordering
-        entries = [
-            ((playlistitems[i - 1] if playlistitems else i), entry)
-            for i, entry in enumerate(entries, 1)]
-
         if self.params.get('playlistreverse', False):
             entries = entries[::-1]
         if self.params.get('playlistrandom', False):
@@ -1431,7 +1456,7 @@ def make_playlistitems_entries(list_ie_entries):
 
         x_forwarded_for = ie_result.get('__x_forwarded_for_ip')
 
-        self.to_screen('[%s] playlist %s: %s' % (ie_result['extractor'], playlist, msg))
+        self.to_screen('[%s] playlist %s: %s' % (ie_result['extractor'], playlist, msg % n_entries))
         failures = 0
         max_failures = self.params.get('skip_playlist_after_errors') or float('inf')
         for i, entry_tuple in enumerate(entries, 1):
@@ -1492,12 +1517,11 @@ def _build_format_filter(self, filter_spec):
             '!=': operator.ne,
         }
         operator_rex = re.compile(r'''(?x)\s*
-            (?P<key>width|height|tbr|abr|vbr|asr|filesize|filesize_approx|fps)
-            \s*(?P<op>%s)(?P<none_inclusive>\s*\?)?\s*
-            (?P<value>[0-9.]+(?:[kKmMgGtTpPeEzZyY]i?[Bb]?)?)
-            $
+            (?P<key>width|height|tbr|abr|vbr|asr|filesize|filesize_approx|fps)\s*
+            (?P<op>%s)(?P<none_inclusive>\s*\?)?\s*
+            (?P<value>[0-9.]+(?:[kKmMgGtTpPeEzZyY]i?[Bb]?)?)\s*
             ''' % '|'.join(map(re.escape, OPERATORS.keys())))
-        m = operator_rex.search(filter_spec)
+        m = operator_rex.fullmatch(filter_spec)
         if m:
             try:
                 comparison_value = int(m.group('value'))
@@ -1518,13 +1542,12 @@ def _build_format_filter(self, filter_spec):
                 '$=': lambda attr, value: attr.endswith(value),
                 '*=': lambda attr, value: value in attr,
             }
-            str_operator_rex = re.compile(r'''(?x)
-                \s*(?P<key>[a-zA-Z0-9._-]+)
-                \s*(?P<negation>!\s*)?(?P<op>%s)(?P<none_inclusive>\s*\?)?
-                \s*(?P<value>[a-zA-Z0-9._-]+)
-                \s*$
+            str_operator_rex = re.compile(r'''(?x)\s*
+                (?P<key>[a-zA-Z0-9._-]+)\s*
+                (?P<negation>!\s*)?(?P<op>%s)(?P<none_inclusive>\s*\?)?\s*
+                (?P<value>[a-zA-Z0-9._-]+)\s*
                 ''' % '|'.join(map(re.escape, STR_OPERATORS.keys())))
-            m = str_operator_rex.search(filter_spec)
+            m = str_operator_rex.fullmatch(filter_spec)
             if m:
                 comparison_value = m.group('value')
                 str_op = STR_OPERATORS[m.group('op')]
@@ -1534,7 +1557,7 @@ def _build_format_filter(self, filter_spec):
                     op = str_op
 
         if not m:
-            raise ValueError('Invalid filter specification %r' % filter_spec)
+            raise SyntaxError('Invalid filter specification %r' % filter_spec)
 
         def _filter(f):
             actual_value = f.get(m.group('key'))
@@ -1689,9 +1712,12 @@ def _merge(formats_pair):
             formats_info.extend(format_2.get('requested_formats', (format_2,)))
 
             if not allow_multiple_streams['video'] or not allow_multiple_streams['audio']:
-                get_no_more = {"video": False, "audio": False}
+                get_no_more = {'video': False, 'audio': False}
                 for (i, fmt_info) in enumerate(formats_info):
-                    for aud_vid in ["audio", "video"]:
+                    if fmt_info.get('acodec') == fmt_info.get('vcodec') == 'none':
+                        formats_info.pop(i)
+                        continue
+                    for aud_vid in ['audio', 'video']:
                         if not allow_multiple_streams[aud_vid] and fmt_info.get(aud_vid[0] + 'codec') != 'none':
                             if get_no_more[aud_vid]:
                                 formats_info.pop(i)
@@ -1742,21 +1768,26 @@ def _merge(formats_pair):
             return new_dict
 
         def _check_formats(formats):
+            if not check_formats:
+                yield from formats
+                return
             for f in formats:
                 self.to_screen('[info] Testing format %s' % f['format_id'])
-                paths = self.params.get('paths', {})
-                temp_file = os.path.join(
-                    expand_path(paths.get('home', '').strip()),
-                    expand_path(paths.get('temp', '').strip()),
-                    'ytdl.%s.f%s.check-format' % (random_uuidv4(), f['format_id']))
+                temp_file = tempfile.NamedTemporaryFile(
+                    suffix='.tmp', delete=False,
+                    dir=self.get_output_path('temp') or None)
+                temp_file.close()
                 try:
-                    dl, _ = self.dl(temp_file, f, test=True)
-                except (ExtractorError, IOError, OSError, ValueError) + network_exceptions:
-                    dl = False
+                    success, _ = self.dl(temp_file.name, f, test=True)
+                except (DownloadError, IOError, OSError, ValueError) + network_exceptions:
+                    success = False
                 finally:
-                    if os.path.exists(temp_file):
-                        os.remove(temp_file)
-                if dl:
+                    if os.path.exists(temp_file.name):
+                        try:
+                            os.remove(temp_file.name)
+                        except OSError:
+                            self.report_warning('Unable to delete temporary file "%s"' % temp_file.name)
+                if success:
                     yield f
                 else:
                     self.to_screen('[info] Unable to download format %s. Skipping...' % f['format_id'])
@@ -1767,8 +1798,7 @@ def _build_selector_function(selector):
 
                 def selector_function(ctx):
                     for f in fs:
-                        for format in f(ctx):
-                            yield format
+                        yield from f(ctx)
                 return selector_function
 
             elif selector.type == GROUP:  # ()
@@ -1784,17 +1814,21 @@ def selector_function(ctx):
                             return picked_formats
                     return []
 
+            elif selector.type == MERGE:  # +
+                selector_1, selector_2 = map(_build_selector_function, selector.selector)
+
+                def selector_function(ctx):
+                    for pair in itertools.product(
+                            selector_1(copy.deepcopy(ctx)), selector_2(copy.deepcopy(ctx))):
+                        yield _merge(pair)
+
             elif selector.type == SINGLE:  # atom
                 format_spec = selector.selector or 'best'
 
                 # TODO: Add allvideo, allaudio etc by generalizing the code with best/worst selector
                 if format_spec == 'all':
                     def selector_function(ctx):
-                        formats = list(ctx['formats'])
-                        if check_formats:
-                            formats = _check_formats(formats)
-                        for f in formats:
-                            yield f
+                        yield from _check_formats(ctx['formats'])
                 elif format_spec == 'mergeall':
                     def selector_function(ctx):
                         formats = list(_check_formats(ctx['formats']))
@@ -1818,14 +1852,16 @@ def selector_function(ctx):
                         format_modified = mobj.group('mod') is not None
 
                         format_fallback = not format_type and not format_modified  # for b, w
-                        filter_f = (
+                        _filter_f = (
                             (lambda f: f.get('%scodec' % format_type) != 'none')
                             if format_type and format_modified  # bv*, ba*, wv*, wa*
                             else (lambda f: f.get('%scodec' % not_format_type) == 'none')
                             if format_type  # bv, ba, wv, wa
                             else (lambda f: f.get('vcodec') != 'none' and f.get('acodec') != 'none')
                             if not format_modified  # b, w
-                            else None)  # b*, w*
+                            else lambda f: True)  # b*, w*
+                        filter_f = lambda f: _filter_f(f) and (
+                            f.get('vcodec') != 'none' or f.get('acodec') != 'none')
                     else:
                         filter_f = ((lambda f: f.get('ext') == format_spec)
                                     if format_spec in ['mp4', 'flv', 'webm', '3gp', 'm4a', 'mp3', 'ogg', 'aac', 'wav']  # extension
@@ -1833,29 +1869,17 @@ def selector_function(ctx):
 
                     def selector_function(ctx):
                         formats = list(ctx['formats'])
-                        if not formats:
-                            return
                         matches = list(filter(filter_f, formats)) if filter_f is not None else formats
                         if format_fallback and ctx['incomplete_formats'] and not matches:
                             # for extractors with incomplete formats (audio only (soundcloud)
                             # or video only (imgur)) best/worst will fallback to
                             # best/worst {video,audio}-only format
                             matches = formats
-                        if format_reverse:
-                            matches = matches[::-1]
-                        if check_formats:
-                            matches = list(itertools.islice(_check_formats(matches), format_idx))
-                        n = len(matches)
-                        if -n <= format_idx - 1 < n:
+                        matches = LazyList(_check_formats(matches[::-1 if format_reverse else 1]))
+                        try:
                             yield matches[format_idx - 1]
-
-            elif selector.type == MERGE:        # +
-                selector_1, selector_2 = map(_build_selector_function, selector.selector)
-
-                def selector_function(ctx):
-                    for pair in itertools.product(
-                            selector_1(copy.deepcopy(ctx)), selector_2(copy.deepcopy(ctx))):
-                        yield _merge(pair)
+                        except IndexError:
+                            return
 
             filters = [self._build_format_filter(f) for f in selector.filters]
 
@@ -1918,8 +1942,7 @@ def _calc_cookies(self, info_dict):
         self.cookiejar.add_cookie_header(pr)
         return pr.get_header('Cookie')
 
-    @staticmethod
-    def _sanitize_thumbnails(info_dict):
+    def _sanitize_thumbnails(self, info_dict):
         thumbnails = info_dict.get('thumbnails')
         if thumbnails is None:
             thumbnail = info_dict.get('thumbnail')
@@ -1932,12 +1955,40 @@ def _sanitize_thumbnails(info_dict):
                 t.get('height') if t.get('height') is not None else -1,
                 t.get('id') if t.get('id') is not None else '',
                 t.get('url')))
+
+            def thumbnail_tester():
+                if self.params.get('check_formats'):
+                    test_all = True
+                    to_screen = lambda msg: self.to_screen(f'[info] {msg}')
+                else:
+                    test_all = False
+                    to_screen = self.write_debug
+
+                def test_thumbnail(t):
+                    if not test_all and not t.get('_test_url'):
+                        return True
+                    to_screen('Testing thumbnail %s' % t['id'])
+                    try:
+                        self.urlopen(HEADRequest(t['url']))
+                    except network_exceptions as err:
+                        to_screen('Unable to connect to thumbnail %s URL "%s" - %s. Skipping...' % (
+                            t['id'], t['url'], error_to_compat_str(err)))
+                        return False
+                    return True
+
+                return test_thumbnail
+
             for i, t in enumerate(thumbnails):
-                t['url'] = sanitize_url(t['url'])
-                if t.get('width') and t.get('height'):
-                    t['resolution'] = '%dx%d' % (t['width'], t['height'])
                 if t.get('id') is None:
                     t['id'] = '%d' % i
+                if t.get('width') and t.get('height'):
+                    t['resolution'] = '%dx%d' % (t['width'], t['height'])
+                t['url'] = sanitize_url(t['url'])
+
+            if self.params.get('check_formats') is not False:
+                info_dict['thumbnails'] = LazyList(filter(thumbnail_tester(), thumbnails[::-1])).reverse()
+            else:
+                info_dict['thumbnails'] = thumbnails
 
     def process_video_result(self, info_dict, download=True):
         assert info_dict.get('_type', 'video') == 'video'
@@ -1977,10 +2028,6 @@ def sanitize_numeric_fields(info):
 
         self._sanitize_thumbnails(info_dict)
 
-        if self.params.get('list_thumbnails'):
-            self.list_thumbnails(info_dict)
-            return
-
         thumbnail = info_dict.get('thumbnail')
         thumbnails = info_dict.get('thumbnails')
         if thumbnail:
@@ -2023,13 +2070,6 @@ def sanitize_numeric_fields(info):
         automatic_captions = info_dict.get('automatic_captions')
         subtitles = info_dict.get('subtitles')
 
-        if self.params.get('listsubtitles', False):
-            if 'automatic_captions' in info_dict:
-                self.list_subtitles(
-                    info_dict['id'], automatic_captions, 'automatic captions')
-            self.list_subtitles(info_dict['id'], subtitles, 'subtitles')
-            return
-
         info_dict['requested_subtitles'] = self.process_subtitles(
             info_dict['id'], subtitles, automatic_captions)
 
@@ -2117,18 +2157,27 @@ def is_wellformed(f):
 
         info_dict, _ = self.pre_process(info_dict)
 
-        if self.params.get('listformats'):
-            if not info_dict.get('formats'):
-                raise ExtractorError('No video formats found', expected=True)
-            self.list_formats(info_dict)
+        list_only = self.params.get('list_thumbnails') or self.params.get('listformats') or self.params.get('listsubtitles')
+        if list_only:
+            self.__forced_printings(info_dict, self.prepare_filename(info_dict), incomplete=True)
+            if self.params.get('list_thumbnails'):
+                self.list_thumbnails(info_dict)
+            if self.params.get('listformats'):
+                if not info_dict.get('formats'):
+                    raise ExtractorError('No video formats found', expected=True)
+                self.list_formats(info_dict)
+            if self.params.get('listsubtitles'):
+                if 'automatic_captions' in info_dict:
+                    self.list_subtitles(
+                        info_dict['id'], automatic_captions, 'automatic captions')
+                self.list_subtitles(info_dict['id'], subtitles, 'subtitles')
             return
 
-        req_format = self.params.get('format')
-        if req_format is None:
+        format_selector = self.format_selector
+        if format_selector is None:
             req_format = self._default_format_spec(info_dict, download=download)
             self.write_debug('Default format spec: %s' % req_format)
-
-        format_selector = self.build_format_selector(req_format)
+            format_selector = self.build_format_selector(req_format)
 
         # While in format selection we may need to have an access to the original
         # format set in order to calculate some metrics or do some processing.
@@ -2162,6 +2211,8 @@ def is_wellformed(f):
                 raise ExtractorError('Requested format is not available', expected=True)
             else:
                 self.report_warning('Requested format is not available')
+                # Process what we can, even without any available formats.
+                self.process_info(dict(info_dict))
         elif download:
             self.to_screen(
                 '[info] %s: Downloading %d format(s): %s' % (
@@ -2326,7 +2377,7 @@ def process_info(self, info_dict):
         # TODO: backward compatibility, to be removed
         info_dict['fulltitle'] = info_dict['title']
 
-        if 'format' not in info_dict:
+        if 'format' not in info_dict and 'ext' in info_dict:
             info_dict['format'] = info_dict['ext']
 
         if self._match_entry(info_dict) is not None:
@@ -2341,7 +2392,7 @@ def process_info(self, info_dict):
         files_to_move = {}
 
         # Forced printings
-        self.__forced_printings(info_dict, full_filename, incomplete=False)
+        self.__forced_printings(info_dict, full_filename, incomplete=('format' not in info_dict))
 
         if self.params.get('simulate', False):
             if self.params.get('force_write_download_archive', False):
@@ -2566,17 +2617,10 @@ def compatible_formats(formats):
 
                     requested_formats = info_dict['requested_formats']
                     old_ext = info_dict['ext']
-                    if self.params.get('merge_output_format') is None:
-                        if not compatible_formats(requested_formats):
-                            info_dict['ext'] = 'mkv'
-                            self.report_warning(
-                                'Requested formats are incompatible for merge and will be merged into mkv.')
-                        if (info_dict['ext'] == 'webm'
-                                and self.params.get('writethumbnail', False)
-                                and info_dict.get('thumbnails')):
-                            info_dict['ext'] = 'mkv'
-                            self.report_warning(
-                                'webm doesn\'t support embedding a thumbnail, mkv will be used.')
+                    if self.params.get('merge_output_format') is None and not compatible_formats(requested_formats):
+                        info_dict['ext'] = 'mkv'
+                        self.report_warning(
+                            'Requested formats are incompatible for merge and will be merged into mkv.')
 
                     def correct_ext(filename):
                         filename_real_ext = os.path.splitext(filename)[1][1:]
@@ -2662,65 +2706,53 @@ def correct_ext(filename):
                 return
 
             if success and full_filename != '-':
-                # Fixup content
-                fixup_policy = self.params.get('fixup')
-                if fixup_policy is None:
-                    fixup_policy = 'detect_or_warn'
-
-                INSTALL_FFMPEG_MESSAGE = 'Install ffmpeg to fix this automatically.'
-
-                stretched_ratio = info_dict.get('stretched_ratio')
-                if stretched_ratio is not None and stretched_ratio != 1:
-                    if fixup_policy == 'warn':
-                        self.report_warning('%s: Non-uniform pixel ratio (%s)' % (
-                            info_dict['id'], stretched_ratio))
-                    elif fixup_policy == 'detect_or_warn':
-                        stretched_pp = FFmpegFixupStretchedPP(self)
-                        if stretched_pp.available:
-                            info_dict['__postprocessors'].append(stretched_pp)
-                        else:
-                            self.report_warning(
-                                '%s: Non-uniform pixel ratio (%s). %s'
-                                % (info_dict['id'], stretched_ratio, INSTALL_FFMPEG_MESSAGE))
-                    else:
-                        assert fixup_policy in ('ignore', 'never')
 
-                if (info_dict.get('requested_formats') is None
-                        and info_dict.get('container') == 'm4a_dash'
-                        and info_dict.get('ext') == 'm4a'):
-                    if fixup_policy == 'warn':
-                        self.report_warning(
-                            '%s: writing DASH m4a. '
-                            'Only some players support this container.'
-                            % info_dict['id'])
-                    elif fixup_policy == 'detect_or_warn':
-                        fixup_pp = FFmpegFixupM4aPP(self)
-                        if fixup_pp.available:
-                            info_dict['__postprocessors'].append(fixup_pp)
-                        else:
-                            self.report_warning(
-                                '%s: writing DASH m4a. '
-                                'Only some players support this container. %s'
-                                % (info_dict['id'], INSTALL_FFMPEG_MESSAGE))
-                    else:
-                        assert fixup_policy in ('ignore', 'never')
-
-                if ('protocol' in info_dict
-                        and get_suitable_downloader(info_dict, self.params).__name__ == 'HlsFD'):
-                    if fixup_policy == 'warn':
-                        self.report_warning('%s: malformed AAC bitstream detected.' % (
-                            info_dict['id']))
-                    elif fixup_policy == 'detect_or_warn':
-                        fixup_pp = FFmpegFixupM3u8PP(self)
-                        if fixup_pp.available:
-                            info_dict['__postprocessors'].append(fixup_pp)
-                        else:
-                            self.report_warning(
-                                '%s: malformed AAC bitstream detected. %s'
-                                % (info_dict['id'], INSTALL_FFMPEG_MESSAGE))
-                    else:
-                        assert fixup_policy in ('ignore', 'never')
+                def fixup():
+                    do_fixup = True
+                    fixup_policy = self.params.get('fixup')
+                    vid = info_dict['id']
 
+                    if fixup_policy in ('ignore', 'never'):
+                        return
+                    elif fixup_policy == 'warn':
+                        do_fixup = False
+                    elif fixup_policy != 'force':
+                        assert fixup_policy in ('detect_or_warn', None)
+                        if not info_dict.get('__real_download'):
+                            do_fixup = False
+
+                    def ffmpeg_fixup(cndn, msg, cls):
+                        if not cndn:
+                            return
+                        if not do_fixup:
+                            self.report_warning(f'{vid}: {msg}')
+                            return
+                        pp = cls(self)
+                        if pp.available:
+                            info_dict['__postprocessors'].append(pp)
+                        else:
+                            self.report_warning(f'{vid}: {msg}. Install ffmpeg to fix this automatically')
+
+                    stretched_ratio = info_dict.get('stretched_ratio')
+                    ffmpeg_fixup(
+                        stretched_ratio not in (1, None),
+                        f'Non-uniform pixel ratio {stretched_ratio}',
+                        FFmpegFixupStretchedPP)
+
+                    ffmpeg_fixup(
+                        (info_dict.get('requested_formats') is None
+                         and info_dict.get('container') == 'm4a_dash'
+                         and info_dict.get('ext') == 'm4a'),
+                        'writing DASH m4a. Only some players support this container',
+                        FFmpegFixupM4aPP)
+
+                    downloader = (get_suitable_downloader(info_dict, self.params).__name__
+                                  if 'protocol' in info_dict else None)
+                    ffmpeg_fixup(downloader == 'HlsFD', 'malformed AAC bitstream detected', FFmpegFixupM3u8PP)
+                    ffmpeg_fixup(downloader == 'WebSocketFragmentFD', 'malformed timestamps detected', FFmpegFixupTimestampPP)
+                    ffmpeg_fixup(downloader == 'WebSocketFragmentFD', 'malformed duration detected', FFmpegFixupDurationPP)
+
+                fixup()
                 try:
                     info_dict = self.post_process(dl_filename, info_dict, files_to_move)
                 except PostProcessingError as err:
@@ -2780,7 +2812,7 @@ def download_with_info_file(self, info_filename):
             info = self.filter_requested_info(json.loads('\n'.join(f)), self.params.get('clean_infojson', True))
         try:
             self.process_ie_result(info, download=True)
-        except (DownloadError, EntryNotInPlaylist):
+        except (DownloadError, EntryNotInPlaylist, ThrottledDownload):
             webpage_url = info.get('webpage_url')
             if webpage_url is not None:
                 self.report_warning('The info failed to download, trying with "%s"' % webpage_url)
@@ -2791,19 +2823,20 @@ def download_with_info_file(self, info_filename):
 
     @staticmethod
     def filter_requested_info(info_dict, actually_filter=True):
-        info_dict.pop('__original_infodict', None)  # Always remove this
-        if not actually_filter:
+        remove_keys = ['__original_infodict']  # Always remove this since this may contain a copy of the entire dict
+        keep_keys = ['_type'],  # Always keep this to facilitate load-info-json
+        if actually_filter:
+            remove_keys += ('requested_formats', 'requested_subtitles', 'requested_entries', 'filepath', 'entries', 'original_url')
+            empty_values = (None, {}, [], set(), tuple())
+            reject = lambda k, v: k not in keep_keys and (
+                k.startswith('_') or k in remove_keys or v in empty_values)
+        else:
             info_dict['epoch'] = int(time.time())
-            return info_dict
-        exceptions = {
-            'remove': ['requested_formats', 'requested_subtitles', 'requested_entries', 'filepath', 'entries'],
-            'keep': ['_type'],
-        }
-        keep_key = lambda k: k in exceptions['keep'] or not (k.startswith('_') or k in exceptions['remove'])
+            reject = lambda k, v: k in remove_keys
         filter_fn = lambda obj: (
-            list(map(filter_fn, obj)) if isinstance(obj, (list, tuple))
+            list(map(filter_fn, obj)) if isinstance(obj, (LazyList, list, tuple, set))
             else obj if not isinstance(obj, dict)
-            else dict((k, filter_fn(v)) for k, v in obj.items() if keep_key(k)))
+            else dict((k, filter_fn(v)) for k, v in obj.items() if not reject(k, v)))
         return filter_fn(info_dict)
 
     def run_pp(self, pp, infodict):
@@ -2912,6 +2945,8 @@ def record_download_archive(self, info_dict):
     @staticmethod
     def format_resolution(format, default='unknown'):
         if format.get('vcodec') == 'none':
+            if format.get('acodec') == 'none':
+                return 'images'
             return 'audio only'
         if format.get('resolution') is not None:
             return format['resolution']
@@ -2981,22 +3016,11 @@ def _format_note(self, fdict):
             res += '~' + format_bytes(fdict['filesize_approx'])
         return res
 
-    def _format_note_table(self, f):
-        def join_fields(*vargs):
-            return ', '.join((val for val in vargs if val != ''))
-
-        return join_fields(
-            'UNSUPPORTED' if f.get('ext') in ('f4f', 'f4m') else '',
-            format_field(f, 'language', '[%s]'),
-            format_field(f, 'format_note'),
-            format_field(f, 'container', ignore=(None, f.get('ext'))),
-            format_field(f, 'asr', '%5dHz'))
-
     def list_formats(self, info_dict):
         formats = info_dict.get('formats', [info_dict])
         new_format = (
             'list-formats' not in self.params.get('compat_opts', [])
-            and self.params.get('list_formats_as_table', True) is not False)
+            and self.params.get('listformats_table', True) is not False)
         if new_format:
             table = [
                 [
@@ -3014,11 +3038,15 @@ def list_formats(self, info_dict):
                     format_field(f, 'acodec', default='unknown').replace('none', ''),
                     format_field(f, 'abr', '%3dk'),
                     format_field(f, 'asr', '%5dHz'),
-                    self._format_note_table(f)]
-                for f in formats
-                if f.get('preference') is None or f['preference'] >= -1000]
+                    ', '.join(filter(None, (
+                        'UNSUPPORTED' if f.get('ext') in ('f4f', 'f4m') else '',
+                        format_field(f, 'language', '[%s]'),
+                        format_field(f, 'format_note'),
+                        format_field(f, 'container', ignore=(None, f.get('ext'))),
+                        format_field(f, 'asr', '%5dHz')))),
+                ] for f in formats if f.get('preference') is None or f['preference'] >= -1000]
             header_line = ['ID', 'EXT', 'RESOLUTION', 'FPS', '|', ' FILESIZE', '  TBR', 'PROTO',
-                           '|', 'VCODEC', '  VBR', 'ACODEC', ' ABR', ' ASR', 'NOTE']
+                           '|', 'VCODEC', '  VBR', 'ACODEC', ' ABR', ' ASR', 'MORE INFO']
         else:
             table = [
                 [
@@ -3031,22 +3059,19 @@ def list_formats(self, info_dict):
             header_line = ['format code', 'extension', 'resolution', 'note']
 
         self.to_screen(
-            '[info] Available formats for %s:\n%s' % (info_dict['id'], render_table(
-                header_line,
-                table,
-                delim=new_format,
-                extraGap=(0 if new_format else 1),
-                hideEmpty=new_format)))
+            '[info] Available formats for %s:' % info_dict['id'])
+        self.to_stdout(render_table(
+            header_line, table, delim=new_format, extraGap=(0 if new_format else 1), hideEmpty=new_format))
 
     def list_thumbnails(self, info_dict):
-        thumbnails = info_dict.get('thumbnails')
+        thumbnails = list(info_dict.get('thumbnails'))
         if not thumbnails:
             self.to_screen('[info] No thumbnails present for %s' % info_dict['id'])
             return
 
         self.to_screen(
             '[info] Thumbnails for %s:' % info_dict['id'])
-        self.to_screen(render_table(
+        self.to_stdout(render_table(
             ['ID', 'width', 'height', 'URL'],
             [[t['id'], t.get('width', 'unknown'), t.get('height', 'unknown'), t['url']] for t in thumbnails]))
 
@@ -3058,12 +3083,12 @@ def list_subtitles(self, video_id, subtitles, name='subtitles'):
             'Available %s for %s:' % (name, video_id))
 
         def _row(lang, formats):
-            exts, names = zip(*((f['ext'], f.get('name', 'unknown')) for f in reversed(formats)))
+            exts, names = zip(*((f['ext'], f.get('name') or 'unknown') for f in reversed(formats)))
             if len(set(names)) == 1:
                 names = [] if names[0] == 'unknown' else names[:1]
             return [lang, ', '.join(names), ', '.join(exts)]
 
-        self.to_screen(render_table(
+        self.to_stdout(render_table(
             ['Language', 'Name', 'Formats'],
             [_row(lang, formats) for lang, formats in subtitles.items()],
             hideEmpty=True))
@@ -3241,14 +3266,15 @@ def _write_thumbnails(self, info_dict, filename):  # return the extensions
         multiple = write_all and len(thumbnails) > 1
 
         ret = []
-        for t in thumbnails[::1 if write_all else -1]:
+        for t in thumbnails[::-1]:
             thumb_ext = determine_ext(t['url'], 'jpg')
             suffix = '%s.' % t['id'] if multiple else ''
             thumb_display_id = '%s ' % t['id'] if multiple else ''
-            t['filepath'] = thumb_filename = replace_extension(filename, suffix + thumb_ext, info_dict.get('ext'))
+            thumb_filename = replace_extension(filename, suffix + thumb_ext, info_dict.get('ext'))
 
             if not self.params.get('overwrites', True) and os.path.exists(encodeFilename(thumb_filename)):
                 ret.append(suffix + thumb_ext)
+                t['filepath'] = thumb_filename
                 self.to_screen('[%s] %s: Thumbnail %sis already present' %
                                (info_dict['extractor'], info_dict['id'], thumb_display_id))
             else:
@@ -3261,6 +3287,7 @@ def _write_thumbnails(self, info_dict, filename):  # return the extensions
                     ret.append(suffix + thumb_ext)
                     self.to_screen('[%s] %s: Writing thumbnail %sto: %s' %
                                    (info_dict['extractor'], info_dict['id'], thumb_display_id, thumb_filename))
+                    t['filepath'] = thumb_filename
                 except network_exceptions as err:
                     self.report_warning('Unable to download thumbnail "%s": %s' %
                                         (t['url'], error_to_compat_str(err)))