DOCX to EPUB, font subsetting quirk

jindroush · 04-01-2024, 07:26 AM

Hi,

the docx file has lowercase text uppercased by style (ie. All Caps in font settings of style).

When converting to epub, everything works, text is included in lowercase, and CSS style has
text-transform: uppercase;

BUT: When font subsetting is on, the uppercase letters are not "counted in" and therefore not included in ttf in epub file.

There are workarounds (uppercase in word, which is bleh, or not subsetting which produces unnecessarily large epubs).

Kovid, is this fixable, or should I "work around"?

Quoth · 04-01-2024, 08:31 AM

Add a separate page (with a header) with all the desired characters, convert with subset, then delete that (x)html file in the editor.

kovidgoyal · 04-01-2024, 09:01 AM

This is font subsetting in what context? If its conversion, please follow the instruxtions in: https://www.mobileread.com/forums/sh...d.php?t=186697

jindroush · 04-01-2024, 09:22 AM

Attached are original DOCX, converted EPUB and conversion report.
The font is embedded and subsetted, but only uppercase T is included, because it's in the original text.

The glyphs for All Caps and Small Caps are not present in subsetted font.

I also added EPUB with "subsetting" off, it's much longer (expected) and displays correct glyps (also expected).

kovidgoyal · 04-01-2024, 10:12 AM

Well text-transform is relatively easy to add, https://github.com/kovidgoyal/calibr...890d7580eefd1c

but font-variant is not going to happen.

jindroush · 04-01-2024, 11:15 AM

Thanks a lot for the uppercase patch!

I've included small-caps only because I had (wrong) feeling this would be the same beast.
Not the case, so I've done some digging in TTF format, I'll check how you actually parse it and if it's possible to submit a patch for this. There are 'c2sc' and 'smcp' display substitution tables which should provide the necessary glyphs for 'please don't remove me' list.

04-01-2024, 07:26 AM	#1
jindroush Connoisseur Posts: 78 Karma: 52 Join Date: Nov 2014 Device: Kindle	DOCX to EPUB, font subsetting quirk Hi, the docx file has lowercase text uppercased by style (ie. All Caps in font settings of style). When converting to epub, everything works, text is included in lowercase, and CSS style has text-transform: uppercase; BUT: When font subsetting is on, the uppercase letters are not "counted in" and therefore not included in ttf in epub file. There are workarounds (uppercase in word, which is bleh, or not subsetting which produces unnecessarily large epubs). Kovid, is this fixable, or should I "work around"?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Calibre Conversion: Inconsistent Font Output When Converting From Epub to Docx	IndiePublisher	Conversion	2	06-16-2020 02:17 AM
Conversion problem Docx to epub font failure	sinbadalexandros	Conversion	1	04-10-2020 03:17 AM
Problem with font subsetting	grumbles	Editor	17	06-20-2014 02:35 PM
Font subsetting problem	yktan	Calibre	5	09-19-2013 11:00 PM
ePub Font Subsetting	Cygfrydd	ePub	2	08-17-2010 08:53 AM

04-01-2024, 08:31 AM	#2
Quoth the rook, bossing Never. Posts: 11,173 Karma: 85874891 Join Date: Jun 2017 Location: Ireland Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper11	Add a separate page (with a header) with all the desired characters, convert with subset, then delete that (x)html file in the editor.

04-01-2024, 09:01 AM	#3
kovidgoyal creator of calibre Posts: 43,874 Karma: 22666666 Join Date: Oct 2006 Location: Mumbai, India Device: Various	This is font subsetting in what context? If its conversion, please follow the instruxtions in: https://www.mobileread.com/forums/sh...d.php?t=186697

04-01-2024, 10:12 AM	#5
kovidgoyal creator of calibre Posts: 43,874 Karma: 22666666 Join Date: Oct 2006 Location: Mumbai, India Device: Various	Well text-transform is relatively easy to add, https://github.com/kovidgoyal/calibr...890d7580eefd1c but font-variant is not going to happen.

04-01-2024, 11:15 AM	#6
jindroush Connoisseur Posts: 78 Karma: 52 Join Date: Nov 2014 Device: Kindle	Thanks a lot for the uppercase patch! I've included small-caps only because I had (wrong) feeling this would be the same beast. Not the case, so I've done some digging in TTF format, I'll check how you actually parse it and if it's possible to submit a patch for this. There are 'c2sc' and 'smcp' display substitution tables which should provide the necessary glyphs for 'please don't remove me' list.

Advert

Advert