Splitting

[mrmikel]

When I tried to split my document at the start of Contents, it split at a different place, the start of the list of tables. When I went back to try to split it where I wanted to I got:

calibre, version 1.24.0
ERROR: Unhandled exception: <b>IndexError</b>:tuple index out of range

calibre 1.24 [64bit] isfrozen: True is64bit: True
Windows-7-6.1.7601-SP1 Windows ('64bit', 'WindowsPE')
('Windows', '7', '6.1.7601')
Python 2.7.5
Windows: ('7', '6.1.7601', 'SP1', 'Multiprocessor Free')
Traceback (most recent call last):
File "site-packages\calibre\gui2\tweak_book\boss.py", line 72, in ans
File "site-packages\calibre\gui2\tweak_book\boss.py", line 889, in split_requested
File "site-packages\calibre\ebooks\oeb\polish\split.py", line 172, in split
File "site-packages\calibre\ebooks\oeb\polish\toc.py", line 358, in node_from_loc
IndexError: tuple index out of range

[kovidgoyal]

Run fix html or beautify before splitting.

[mrmikel]

I ran beautify and it worked just the same way. Ditto with fix html.

I had run check book multiple times to get rid of all the errors before.

[kovidgoyal]

Hmm, ok I will take a look at it when I have a moment.

[kovidgoyal]

Your problem is caused by having <h3> inside <p> which is illegal in HTML and causes the block counting to get thrown off. See

<p>
<h3>The Authors</h3>
</p>

Fix that and you will be fine. Unfortunately, because that file is valid XML (but not valid HTML) the Fix HTML tool does not catch and fix that error, since trying to detect invalid html inside valid XML is rather expensive.

[mrmikel]

Thanks. This is the price of dealing with website copied HTML- no headings and shoehorning them in.

Is there a chance of being able to do these one at time from the code window side without having to do xpath?

[kovidgoyal]

xpath is not that hard, just add

id="split-here"

to the tag you want to split at and then in the xpath dialog click the wizard and type id in the "Having attribute" box and split-here in the "With value" box.

[mrmikel]

I did it. I had to leave out h3 in the top box, which I assumed I needed.

It certainly is easy enough to put in the breaks this way. And they will go where I put them, regardless of whatever funky code there is.

I'm going to run the html/xml defective code through epubcheck to see if it catches it.

Later, it did catch it, but it takes some thought to figure out what they are getting at.

[kovidgoyal]

I've added some code to make the editor catch invalid markup like that and auto-fix it during splitting. https://github.com/kovidgoyal/calibr...7c87ed082083ef

[mrmikel]

Thank you very much. I have just brought in yet another web site book. It was offered as html or pdf. I figured an pdf conversion would be even worse!

BTW, your check book messages are generally more useful than Flight Crew or epubcheck. They nearly always get you closer to the "scene of the crime"!

[kovidgoyal]

Originally Posted by mrmikel

BTW, your check book messages are generally more useful than Flight Crew or epubcheck. They nearly always get you closer to the "scene of the crime.!

Indeed, one of my main design principles for Check Book is to have it output only useful messages, as far as possible. And have it automatically fix trivial bookkeeping mistakes so that the human can concentrate on the real problems.

I find epubcheck and similar schema based validation tools drown the signal in the noise.