EPUB output

[mikeywilliams25]

file sent in Ticket #1095

[HardBoot]

Originally Posted by kovidgoyal

The problem is that the reader cannot handle EPUB files that have individual components larger than 300K, so calibre tries to split these files. It tries to split them on tag boundaries like page-breaks, heading tags, <p> tags etc. If your lit file contains a txt file instead of HTML (or an HTML file with just one tag with a lot of text), calibre wont find any tags and will therefore fail to split. Basically, this should only happen for some txt based LIT files. So raising tickets on these kinds of files is pointless.

It's really weird that you LRF file are so large. try running lrf2lrs on them and seeing what is taking up all the space. The text itself or embedded images or embedded fonts.

The next release of calibre will have a control in the bulk metadata edit to delete a format.

Ok thanks for the information and advice. I must say 300k sounds ridicuously restrictive but what do I know. Deciding what to split obviously takes a lot of processing power, some of the conversions can take 20mins and my poor laptop is burning up.

I'll try lrf2lrs and and report back.

The bulk delete will be very handy (deleting them one at a time was getting rather tedious!)

BTW I find the button to create the author sort very useful. It would be handy to have it on the bulk metadata edit as well if that's possible.

[kovidgoyal]

actually i decided to implement splitting for <pre> tags that contain a lot of text as well. So a number of your files should work in the future.

[kovidgoyal]

Originally Posted by mikeywilliams25

I converted a .lit file to epub and to lrf (calibre-0.4.88)
The epub doesn't resize when viewed on the reader, S,M or L are all the same size.
The lrf resizes fine.
Is this something in my setup or a feature of epubs?

I notice that scientific american also doesnt resize when in epub format.

Turns out that Adobe DE cant resize the fonts when some font sizes are specified in absolute values. Probably because it uses an obsolete HTML rendering engine.

[HardBoot]

Originally Posted by kovidgoyal

It's really weird that you LRF file are so large. try running lrf2lrs on them and seeing what is taking up all the space. The text itself or embedded images or embedded fonts.

I did as you said and attached is a screen capture of the files it produced.

I've no idea what conclusion to draw from this information so hopefully it's meaningful to you!

[kovidgoyal]

its the embedded fonts that are causing the large files. Disable font embedding and you will be fine.

[HardBoot]

Originally Posted by kovidgoyal

its the embedded fonts that are causing the large files. Disable font embedding and you will be fine.

Excellent thanks...that's fixed it.

[mikeywilliams25]

Originally Posted by kovidgoyal

Turns out that Adobe DE cant resize the fonts when some font sizes are specified in absolute values. Probably because it uses an obsolete HTML rendering engine.

I take it that it is therefore better to use lrf for files with specified font sizes - or will the epub conversion be able at some time in the future to take account of this in the way that the lrf conversion does.

[dynabook]

How do you specify fonts in relative sizes? Say 10 pt vs. 12 Pt? or 10pt, 12 pt, 18 pt?

How about in a CSS style sheet?

[kovidgoyal]

@mikeywilliams

This is already worked around in 0.4.89

@dynabook

pt sizes are not relative, they are absolute. You can use the override CSS and the base font size settings to change font sizes.

Note that for those of you that downloaded 0.4.89 in the last 12 hrs. There is a regression in the windows version that is preventing epub output from working. Please re-download the windows installer as I have now fixed the bug.

[Peto]

Hello there, Kovid,

I converted the pdf again using 0.89 both to lrf and epub. The dark color background and the page numbers are gone in epub, so that is solved.

Page jumps are gone, though, along with all the format except capitals and cursive, and the original header is treated as if it was main text in the original, so it repeatedly appears intertwined with the text.

Paragraphs are gone too, so I get text lines cut in the middle all over.

[kovidgoyal]

Originally Posted by Peto

Hello there, Kovid,

Page jumps are gone, though, along with all the format except capitals and cursive, and the original header is treated as if it was main text in the original, so it repeatedly appears intertwined with the text.

Paragraphs are gone too, so I get text lines cut in the middle all over.

Are you saying the problem is in both LRF and EPUB?

[JeffElkins]

I'm at 972 ebooks converted to epub. About 2.5K books to go. I've had minimal problems with this batch convert!

[kovidgoyal]

Originally Posted by JeffElkins

I'm at 972 ebooks converted to epub. About 2.5K books to go. I've had minimal problems with this batch convert!

Cool, the code is maturing rapidly

[mikeywilliams25]

OK - tried 0.4.89 (thank you for such speedy updates)
Some problems.
The lit file i sent you wouldn't convert to epub - lots of errors. No doubt it is not a well formed lit file, but it does read ok in lit reader tho the fonts are fixed size.

Scientific American crashed the reader on formatting even when transferred using eBook library. Looking at the file using eBook library it had very odd pages and font sizes.

Tried converting an html zipped file to epub - the font was much too small, so tried upping the base font size to 200%. The font was now a reasonable size but the lines were much too close together - the line feed or whatever it is called had not been increased.

Good luck!