Conversion aborts when using Calibre "Subset Fonts"

[GrannyGrump]

I've searched the forum, and seems that nobody is having a problem but me. I installed the newest version of Calibre portable (my first upgrade since v 8.38) with the intention of using the Subset Fonts feature.

My first attempt, converting epub to epub, the conversion completed, but deleted all four embedded fonts and @font-face statements, but retained the references in the CSS styles.

I reimported the book to Calibre. The next attempt, the conversion aborted. The traceback said it subsetted the fonts ok, but aborted upon deleting the FOURTH AND FINAL original font, with an error code of "1".

I thought perhaps that font was corrupt, so deleted it in Sigil, deleted the book from Calibre and re-imported. This time the conversion aborted at the point of deleting the THIRD AND FINAL font, with an error code of "1".

Before I run myself crazier deleting fonts and running the conversion multiple times, I am begging for help.

I'm attaching the traceback message.


I hope somebody has good advice for me, it will win you gold stars in heaven!

[JSWolf]

Before you try to subset the font, does the ePub actually pass validation? I ask this because of...

Decreased the font OEBPS/Fonts/VeraMono.ttf to 57.9% of its original size
The font OEBPS/Fonts/VeraMono.ttf is unused. Removing it.
The font OEBPS/Fonts/VeraMono.ttf is unused. Removing it.

From what I see, Calibre is having a problem because of a problem in your code. Calibre should not be subsetting an unused font. If your code validates, post your CSS and OPF. If it does not validate and once the errors are fixed so it does validate and Calibre still has trouble, then post the CSS and OPF.

[kovidgoyal]

The file you are trying to convert appears to have multiple conflicting references to the same font file (multiple @font-face rules referencinf the same physical font) which is confusing calibre.

[GrannyGrump]

it says the same about all the embedded fonts. It subsets each one, then deletes. I assumed it was *keeping the subsets* and deleting the **original full-size** fonts. Did I assume wrongly how this operates?

Decreased the font OEBPS/Fonts/Artifika-Regular.ttf to 50.8% of its original size
Decreased the font OEBPS/Fonts/charlotte_small_caps_let.ttf to 48.6% of its original size
Decreased the font OEBPS/Fonts/VeraMono.ttf to 57.9% of its original size

The font OEBPS/Fonts/Artifika-Regular.ttf is unused. Removing it.
The font OEBPS/Fonts/charlotte_small_caps_let.ttf is unused. Removing it.
The font OEBPS/Fonts/VeraMono.ttf is unused. Removing it.
The font OEBPS/Fonts/VeraMono.ttf is unused. Removing it.

The validation has warnings in it (which I expected) about anchor tags pointing to non-existant files, but no ERROR messages. Would those warnings actually make the conversion abort upon deletion of a font. I usually make sure to clean up for validation, but didn't think these particular errors would be disastrous for this fast experiment. And the fast has turned into long and involved.

[GrannyGrump]

@Kovid, YES! that was the problem, I had @font-face rules on an in-line CSS for one of the xhtml files. (I didn't know that would cause trouble.) I have now done a successful conversion, using the subsetting feature.

Thank you for the speedy reply. You ROCK, and Calibre is the greatest thing since sliced bread.

[JSWolf]

delete post

[JSWolf]

Originally Posted by grannyGrumpy

@Kovid, YES! that was the problem, I had @font-face rules on an in-line CSS for one of the xhtml files. (I didn't know that would cause trouble.) I have now done a successful conversion, using the subsetting feature.

Thank you for the speedy reply. You ROCK, and Calibre is the greatest thing since sliced bread.

Now, did you take the subsetted fonts from the converted ePub and replace the fonts in the original ePub and edit the CSS & OPF if any of the font files were deleted for not being in use?

[GrannyGrump]

@JSWolf -- Yup, loaded the subsetted fonts, fixed the coding, and reviewed in ADE. Everything seems to be working well. I just have to quit being lazy and fix my anchor tags, and all will be well. I still have to do a page by page check to make sure there are no font failures in any small spots, but it is all looking good.

And all because you made the comment on another thread about subsetting with Calibre. Thanks!

[JSWolf]

Originally Posted by grannyGrumpy

@JSWolf -- Yup, loaded the subsetted fonts, fixed the coding, and reviewed in ADE. Everything seems to be working well. I just have to quit being lazy and fix my anchor tags, and all will be well. I still have to do a page by page check to make sure there are no font failures in any small spots, but it is all looking good.

And all because you made the comment on another thread about subsetting with Calibre. Thanks!

You are very welcome.

Because of subsetting, I've been able to take ePub that were quite large with all the fonts and reduce them considerably. I never bothered to check them. I just loaded onto my 650 and read. I did not see any problems.

[Sabardeyn]

A bit off topic, but would subsetting be a potential problem for foreign readers since letters/characters might be missing? I cannot think of an exact case, but something like in one language a certain character is used (ex: a) but in another it's a accented/unicode letter (ex: ã)?

[theducks]

Originally Posted by Sabardeyn

A bit off topic, but would subsetting be a potential problem for foreign readers since letters/characters might be missing? I cannot think of an exact case, but something like in one language a certain character is used (ex: a) but in another it's a accented/unicode letter (ex: ã)?

I would expect that subsetting would be be for the books language/charset