detecting chapters in plain txt

[morgon]

Hi,

I would like to convert a plain (ascii) text file to epub and generate a toc.

As this is a text file xpath does not make much sense (at least to me), but I can construct a regext that allows chapters to be detected.

How can I teach calibre to use this regex to detect chapters when converting plain text to epub?

many thanks!

[theducks]

The default xpath looks for these keywords:
(chapter|book|section|part)\s+)|((prolog|prologue| epilogue)

[porphyry5]

Originally Posted by morgon

Hi,

I would like to convert a plain (ascii) text file to epub and generate a toc.

As this is a text file xpath does not make much sense (at least to me), but I can construct a regext that allows chapters to be detected.

How can I teach calibre to use this regex to detect chapters when converting plain text to epub?

many thanks!

I think the controlling factor is any line of text that does NOT end with a period, or other sentence ender, will generate a bolded chapter title line on a fresh page and an entry in the auto-generated TOC
I know the above is true for lines consisting only of a number, or the word Chapter or Part followed by a single word or a number.
If the last character of such a line is a period it is treated as any other sentence in the text.

[kovidgoyal]

you can use a regex inside an xpath expression. And as documented here: http://manual.calibre-ebook.com/conv...l#introduction

you need to look at the html produced by the intermediate steps in the conversion to figure out what xpath to use.

[eschwartz]

Originally Posted by porphyry5

I think the controlling factor is any line of text that does NOT end with a period, or other sentence ender, will generate a bolded chapter title line on a fresh page and an entry in the auto-generated TOC
I know the above is true for lines consisting only of a number, or the word Chapter or Part followed by a single word or a number.
If the last character of such a line is a period it is treated as any other sentence in the text.

That's more to do with the markdown-to-HTML step (if applicable) of the conversion.
The xpath sees an h1 tag.

[porphyry5]

Originally Posted by eschwartz

That's more to do with the markdown-to-HTML step (if applicable) of the conversion.
The xpath sees an h1 tag.

The original poster said

Originally Posted by morgon

As this is a text file xpath does not make much sense (at least to me), but I can construct a regext that allows chapters to be detected.

How can I teach calibre to use this regex to detect chapters when converting plain text to epub?

That is, he was asking how to do it without using xpath.

[eschwartz]

Originally Posted by porphyry5

The original poster said

That is, he was asking how to do it without using xpath.

No, that is in fact a crystal-clear example of "asking how to do it using xpath".
On account of xpath is where you would put a regex.



This is beside the fact that your mention of heuristics operating on the text does not, in fact, help unless one is willing to rewrite the book.

As I said -- that is markdown-to-html.
Converting from one format to another.

You have to have the book conform to those heuristics, not the other way around.

[porphyry5]

Originally Posted by eschwartz

No, that is in fact a crystal-clear example of "asking how to do it using xpath".

Apparently we are arguing over the meaning of @morgon's statement "As this is a text file xpath does not make much sense...". You tell him how to do what he wants using xpath, I tell him how he can structure his plain text, without using xpath, to achieve the same result.

As has been said before, there is more than one way to skin a cat.