Trouble generating working chapters for Kindle

[fishmech]

When creating books out of multiple webpages, I often end up having problems getting the table of contents data to work correctly on my Kindle Paperwhite and Voyage. Specifically, it often results in the reader falsely reporting "chapter complete" for the time left in each chapter (which is typically a single original web page) rather than the appropriate reading time. Additionally, the display that normally shows the title of the chapter or the title of the heading in the TOC will instead just display the top entry in the table of contents, when it's working incorrectly.

All this happens despite the actual text rendering fine on the device and being able to at least go between "chapters" correctly through the menu. I don't really understand how to troubleshoot this.

[eschwartz]

Can you attach a sample AZW3 with this problem?

[fishmech]

Originally Posted by eschwartz

Can you attach a sample AZW3 with this problem?

Yeah, this is an AZW3 that has the problem:
http://fishmech.net/Creepypasta.com%...pasta.com.azw3

And this is an AZW3 that doesn't have the problem, and has the same content as the second half of the first file:
http://fishmech.net/Creepypasta.com%...pasta.com.azw3

Edit: Just to be clear, these both originate from Microsoft Word 2010 DOCX documents, converted to AZW3 using Calibre. I also do some conversion from AZW3 to EPUB for editing things, since while editing AZW3 files directly it'll randomly lose the attribute that allows you to "share" text from the Kindle itself to twitter/goodreads/facebook.

[kovidgoyal]

This is almost certainly caused by something in the trailing bits at the end of each text record. Unfortunately, that is a part of the mobi format that is the least well understood (it is an absolute nightmare that took me days of reverse engineering). Amazon's engineering really, really sucks. As such the effort to fix issues in it is really not worth it for me. Patches are welcome, but be warned it is not going to be simple to fix.

One thing to check in your input files is that the ToC entries are in correct order and dont have duplicates. That is every toc entry points to some location in the file strictly after the previous entry.

[fishmech]

Originally Posted by kovidgoyal

This is almost certainly caused by something in the trailing bits at the end of each text record. Unfortunately, that is a part of the mobi format that is the least well understood (it is an absolute nightmare that took me days of reverse engineering). Amazon's engineering really, really sucks. As such the effort to fix issues in it is really not worth it for me. Patches are welcome, but be warned it is not going to be simple to fix.

One thing to check in your input files is that the ToC entries are in correct order and dont have duplicates. That is every toc entry points to some location in the file strictly after the previous entry.

Where should I look in the text records to try to fix it manually? Would it be in the body text files themselves or in metadata.opf and toc.ncx?

[kovidgoyal]

This is not something you can fix manually. It's in the binary trailer at the end of every text record in a MOBI/azw3 file.

https://wiki.mobileread.com/wiki/MOBI#Trailing_entries

[fishmech]

Originally Posted by kovidgoyal

This is not something you can fix manually. It's in the binary trailer at the end of every text record in a MOBI/azw3 file.

https://wiki.mobileread.com/wiki/MOBI#Trailing_entries

Ah, ok. Thank you for the response.

[eschwartz]

The ToC entries are of course in the ToC.ncx. You can also use the ToC editor in the calibre ebook-editor.