MangAI v1.15 released
New to version 1.15:
-AI improvements to MPM and MSM interpretation
Improved detection of string chapters (chapters with names instead of numbers, such as “Omake”).
Improved detection of special pages within string and special chapters.
Better ordering of string and special chapters.
Added automatic conflict resolution of string chapters.
Find pages that don’t have a number in the filename but should be considered page #1.
Better handling of preceding chapter sort. This is used in rare instances to help sort pages by page number regardless of chapter numbering.
Various logic tweaks.
-Various fixes and improvements
No longer allows output_folder to be the parent (one folder up) of source in order to prevent potential conflicts.
Added a few non-English chapter delimiters: chapitre, capítulo, capitulo.
I focused on the AI of handling special and string chapters. These are tough to deal with and occur more frequently than you might expect. A lot of work went into this which is why the update took so long.
The addition of non-English chapter delimiters made a huge difference when I was debugging a problem someone sent me. In that particular case, MangAI was coming back with “chapitre” as a string chapter multiple times. All MangAI needed to know is that “chapitre” means “chapter” and it started working correctly. I hope in the next update to allow the user to specify their own volume, chapter, and page delimiters to better support the languages used in their filenames. Or you can email me which terms I should add to the code.
One feature I didn’t get to finish was intelligent ePub table-of-contents. MangAI currently only uses separate folder names to determine bookmarks in a toc. It would be much better to leverage the data collected during MPM and MSM analysis to set the best bookmarks possible. The more I worked on this though, the more I realized how difficult this is going to be. I unfortunately have to put this requested feature on hold for the time being.
Download the latest version from the MangAI project page.
