Automation scripts made for fixing Hidive subtitles and other functions.
These are the first batch of scripts. There are a few more I haven't finished cleaning up yet, which I might add later. I'm not an experienced programmer, so constructive criticism is appreciated.
These scripts are first and foremost made for SubsPlus+ releases and Hidive subtitles, so don't expect the more specific scripts to work outside of that context.
Generate chapters by matching themes downloaded from AnimeThemes to the episode.
It creates a .themes folder with the downloaded themes for future runs and charts showing where the themes matched in the episode. Chapters will not be generated if no matches or more than 2 themes are matched, or 2 themes are in the same half of the episode. Themes tagged with Transition or Over on animethemes will not be downloaded and non-conventional themes like Oshi no Ko will likely not work as intended.
pip install -r requirements.txt
If you want frame-perfect chapters using --snap, keyframe generation requires:
- https://github.com/vapoursynth/vapoursynth
- https://github.com/FFMS/ffms2 (Install in vapoursynth plugins)
- https://github.com/dubhater/vapoursynth-scxvid (Install in vapoursynth plugins)
$ python Auto_Chap.py --help
usage: Auto_Chap.py [-h] --input INPUT [--search-name SEARCH_NAME] [--year YEAR] [--snap [SNAP]]
[--work-path WORK_PATH] [--output OUTPUT] [--delete-themes] [--charts]
Automatic anime chapter generator using AnimeThemes.
options:
-h, --help show this help message and exit
--input INPUT, -i INPUT
Video/Audio file.
--search-name SEARCH_NAME, -s SEARCH_NAME
Search to pass to animethemes.moe Example: Spy Classroom Season 2. To only use themes that are
already downloaded, don't add this argument.
--year YEAR Release year to help filter the search. Put the negative number to allow that year or later.
--snap [SNAP] Milisecond window to snap to nearest keyframe for frame-perfect chapters. Efficiently
generates necessary keyframes from video. Defaults to 1000ms if no value added. Values higher
than about 1000 currently crash.
--work-path WORK_PATH, -w WORK_PATH
Place to create a .themes folder for storing persistant information per series. Defaults to
where the episode is.
--output OUTPUT, -o OUTPUT
Output chapter file. Defaults to where the episode is.
--delete-themes, -d Delete the themes and charts after running.
--charts, -c Make charts of where themes are matched in the episode. They can almost double processing time
in some cases though.Generate chapters for an episode.
python Auto_Chap.py -i "Dangers in My Heart - 01.mkv" -s "Dangers in My Heart Season 1"
Generate chapters without a persistant .themes folder.
python Auto_Chap.py -i "Dangers in My Heart - 01.mkv" -s "Dangers in My Heart Season 1" -d
Work in a project folder and output to chapter path.
python Auto_Chap.py -i "Dangers in My Heart - 01.mkv" -s "Dangers in My Heart Season 1" -w "Projects/DMH" -o "Projects/DMH/Chapters/01.chapter.txt"
Run with themes predownloaded (in .ogg and in .themes folder).
python Auto_Chap.py -i "Dangers in My Heart - 01.mkv"
Filter search for shows that released on or after 2023.
python Auto_Chap.py -i "Shangri-la Frontier - 01.mkv" -s "Shangri-la frontier Season 1" --year -2023
Snap to nearest keyframe within 1000ms for frame-perfect chapters. Snap values that are too high don't work for some reason but 1000 should be plenty.
python Auto_Chap.py -i "Dangers in My Heart - 01.mkv" -s "Dangers in My Heart Season 1" --snap 1000
Chapter names can be changed at the top of the script.
PRE_OP = "Prologue"
OPENING = "Opening"
EPISODE = "Episode"
ENDING = "Ending"
POST_ED = "Epilogue"Snap chapter file to nearest keyframe using existing scxvid keyframes. Chapter file must be in simple chapter format. Useful if you already have keyframes but if not then use the keyframe generation in Auto_Chap.
$ python Chapter_Snapper.py -h
usage: Chapter_Snapper.py [-h] --input INPUT --keyframes KEYFRAMES [--output OUTPUT] [--snap-ms SNAP_MS]
[--fps FPS]
Snap chapters to nearest keyframe.
options:
-h, --help show this help message and exit
--input INPUT, -i INPUT
Chapter file. Must be in simple format.
--keyframes KEYFRAMES, -kf KEYFRAMES
SCXvid keyframes. Try to have minimal mkv delay or it might not line up.
--output OUTPUT, -o OUTPUT
Output chapter file. Defaults to where input is.
--snap-ms SNAP_MS, -s SNAP_MS
How many milliseconds to consider snapping to. Defaults to 1000ms.
--fps FPS FPS of the video. Defaults to 23.976.Generate with increased snap window.
python Chapter_Snapper.py -i "autochap.txt" -kf "keyframes.txt" -s 2000 -o "Project/chapter_snapped.txt"
Change outline color of a line when it is overlapping another. Supports top and bottom track but only if they are defined in styles not inline. Works well for Crunchyroll and Hidive shows.
python Overlap_Blue.py infile.ass outfile.ass
Change settings at the top of the script. Specify dialogue font if you don't want it to color an alternative dialogue font.
DIALOGUE_FONT = None
COLOR_HEX = "&H743E15&"Relavant for multi-downloader-nx and hidive-downloader-nx. Split and combine lines so that each line has its own event. This prevents subtitles from shifting positions. Note: Some orders will be reversed to preserve rendering position.
python Hidive_Splitter.py infile.ass outfile.ass
Capitalize and fix stutters:
- F-f-find -> F-F-Find
- Sh-she -> Sh-She
- W-when -> Wh-When
- S..so -> S..So (Can be configure to be S-So or S... So)
python P-Proper_Stutter.py infile.ass outfile.ass
Designed for hidive scripts from multi-downloader-nx.
- Add fade to song styles
- Fix em-dash
- Fix triple dialogue lines
- !? -> ?!
- Increase layers on non 'caption' styles
- Fix incorrectly styled 'caption's
python Regex_Stuff.py infile.ass outfile.ass
- hiisi13/Audio Offset for finding themes within an episode
- AnimeThemes api for getting themes
- iamevn ass parsing code
- tp7/Prass keyframe snapping code for Chapter_Snapper
- Myaamori/keyframes.py inspiration for auto_chap keyframes generation
- FichteFoll/snap_scenechanges.py functions and logic for converting frames to time, etc. in auto_chap snapping