[ModifyChapters] fixes (See desc)

* [docs] Fix typo
* Do not enable `sponskrub` by default
* Fix `--force-keyframes-at-cuts`
* Don't embed subtitles if the video has been cut. Previously, running `--remove-chapters` with `--embed-subs` multiple times caused repeated cuts and out-of-sync subtitles
* Store `_real_duration` to prevent running ffprobe multiple times

3 files changed, 29 insertions, 15 deletions

diff --git a/yt_dlp/options.py b/yt_dlp/options.py
index 505160cec..c2d7a74ff 100644
--- a/yt_dlp/options.py
+++ b/yt_dlp/options.py
@@ -1394,7 +1394,7 @@ def parseOpts(overrideArguments=None):
             'SponsorBlock categories to create chapters for, separated by commas. '
             'Available categories are all, %s. You can prefix the category with a "-" to exempt it. '
             'See https://wiki.sponsor.ajay.app/index.php/Segment_Categories for description of the categories. '
-            'Eg: --sponsorblock-query all,-preview' % ', '.join(SponsorBlockPP.CATEGORIES.keys())))
+            'Eg: --sponsorblock-mark all,-preview' % ', '.join(SponsorBlockPP.CATEGORIES.keys())))
     sponsorblock.add_option(
         '--sponsorblock-remove', metavar='CATS',
         dest='sponsorblock_remove', default=set(), action='callback', type='str',
@@ -1421,7 +1421,7 @@ def parseOpts(overrideArguments=None):
 
     sponsorblock.add_option(
         '--sponskrub',
-        action='store_true', dest='sponskrub', default=None,
+        action='store_true', dest='sponskrub', default=False,
         help=optparse.SUPPRESS_HELP)
     sponsorblock.add_option(
         '--no-sponskrub',
diff --git a/yt_dlp/postprocessor/ffmpeg.py b/yt_dlp/postprocessor/ffmpeg.py
index 806334645..25488e58b 100644
--- a/yt_dlp/postprocessor/ffmpeg.py
+++ b/yt_dlp/postprocessor/ffmpeg.py
@@ -15,6 +15,7 @@ from ..utils import (
     dfxp2srt,
     encodeArgument,
     encodeFilename,
+    float_or_none,
     get_exe_version,
     is_outdated_version,
     ISO639Utils,
@@ -233,6 +234,23 @@ class FFmpegPostProcessor(PostProcessor):
             None)
         return num, len(streams)
 
+    def _get_real_video_duration(self, info, 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']:
+                raise PostProcessingError('ffprobe returned empty duration')
+        except PostProcessingError as e:
+            if fatal:
+                raise PostProcessingError(f'Unable to determine video duration; {e}')
+        return info.setdefault('_real_duration', None)
+
+    def _duration_mismatch(self, d1, d2):
+        if not d1 or not d2:
+            return None
+        return abs(d1 - d2) > 1
+
     def run_ffmpeg_multiple_files(self, input_paths, out_path, opts, **kwargs):
         return self.real_run_ffmpeg(
             [(path, []) for path in input_paths],
@@ -528,6 +546,10 @@ class FFmpegEmbedSubtitlePP(FFmpegPostProcessor):
             return [], information
 
         filename = information['filepath']
+        if self._duration_mismatch(
+                self._get_real_video_duration(information, False), information['duration']):
+            self.to_screen(f'Skipping {self.pp_key()} since the real and expected durations mismatch')
+            return [], information
 
         ext = information['ext']
         sub_langs, sub_names, sub_filenames = [], [], []
diff --git a/yt_dlp/postprocessor/modify_chapters.py b/yt_dlp/postprocessor/modify_chapters.py
index 3d6493b68..9a7ba8eff 100644
--- a/yt_dlp/postprocessor/modify_chapters.py
+++ b/yt_dlp/postprocessor/modify_chapters.py
@@ -9,11 +9,9 @@ from .ffmpeg import (
 )
 from .sponsorblock import SponsorBlockPP
 from ..utils import (
-    float_or_none,
     orderedSet,
     PostProcessingError,
     prepend_extension,
-    traverse_obj,
 )
 
 
@@ -37,7 +35,7 @@ class ModifyChaptersPP(FFmpegPostProcessor):
         if not chapters and not sponsor_chapters:
             return [], info
 
-        real_duration = self._get_real_video_duration(info['filepath'])
+        real_duration = self._get_real_video_duration(info)
         if not chapters:
             chapters = [{'start_time': 0, 'end_time': real_duration, 'title': info['title']}]
 
@@ -45,8 +43,8 @@ class ModifyChaptersPP(FFmpegPostProcessor):
         if not cuts:
             return [], info
 
-        if abs(real_duration - info['duration']) > 1:
-            if abs(real_duration - info['chapters'][-1]['end_time']) < 1:
+        if self._duration_mismatch(real_duration, info.get('duration')):
+            if not self._duration_mismatch(real_duration, info['chapters'][-1]['end_time']):
                 self.to_screen(f'Skipping {self.pp_key()} since the video appears to be already cut')
                 return [], info
             if not info.get('__real_download'):
@@ -72,6 +70,7 @@ class ModifyChaptersPP(FFmpegPostProcessor):
             os.replace(out_file, in_file)
             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):
@@ -101,13 +100,6 @@ class ModifyChaptersPP(FFmpegPostProcessor):
 
         return chapters, sponsor_chapters
 
-    def _get_real_video_duration(self, filename):
-        duration = float_or_none(
-            traverse_obj(self.get_metadata_object(filename), ('format', 'duration')))
-        if duration is None:
-            raise PostProcessingError('ffprobe returned empty duration')
-        return duration
-
     def _get_supported_subs(self, info):
         for sub in (info.get('requested_subtitles') or {}).values():
             sub_file = sub.get('filepath')
@@ -311,7 +303,7 @@ class ModifyChaptersPP(FFmpegPostProcessor):
         in_file = filename
         out_file = prepend_extension(in_file, 'temp')
         if force_keyframes:
-            in_file = self.force_keyframes(in_file, (t for r in ranges_to_cut for t in r))
+            in_file = self.force_keyframes(in_file, (t for c in ranges_to_cut for t in (c['start_time'], c['end_time'])))
         self.to_screen(f'Removing chapters from {filename}')
         self.concat_files([in_file] * len(concat_opts), out_file, concat_opts)
         if in_file != filename: