hb-subset discards mark/mkmk feature? #2841

madig · 2021-02-02T23:05:06Z

madig
Feb 2, 2021

I want to subset a font while preserving anchors. How do I get hb-subset to stop discarding them?

Feb 9, 2021

So what others said. If you don't specify the mark characters, they won't be included.

What's the alternative? Include all marks that interact with specified characters? That results in a very arbitrary selection. Though I see how that would be useful for production workflows. Feel free to open an issue to discuss the semantics (and we should first add to fonttools subsetter). There has been similar requests in the past by Adam Twardoch and others...

View full answer

behdad · 2021-02-05T14:31:05Z

behdad
Feb 5, 2021
Maintainer

It should work. Unless those lookup types were not implemented (yet). Check PRs. I'll take a look next week.

0 replies

garretrieger · 2021-02-05T17:21:30Z

garretrieger
Feb 5, 2021
Maintainer

First thing, hb-subset by default drops the layout tables so if you're not already doing so add --drop-tables-=GSUB,GPOS,GDEF to your command line.

Mark to Mark lookups should be implemented currently. However there are a few PR's which fixed various issues with layout subsetting still pending so it's possible those may fix the issue. Do you have an example font and the command line for the subset command that you are running?

11 replies

qxliu76 Feb 9, 2021
Collaborator

yes all wanted glyphs need to be specified explicitly. I tried --text "́aá" and it's working fine. No ttx issue for me.

madig Feb 9, 2021
Author

Ok, now I'm scratching my head. Attaching a Unicode text file with marks at the end: gs.txt and fontTools 4.19.1. Am on HarfBuzz master, commit feeff59. Font as linked above.

(venv) /tmp $ hb-subset --text-file /tmp/gs.txt FiraSans-Regular.ttf -o out.ttf --drop-tables-=GSUB,GPOS,GDEF
(venv) /tmp $ ttx -o - out.ttf | rg "mark"
Dumping "out.ttf" to "-"...
Dumping 'GlyphOrder' table...
Dumping 'head' table...
Dumping 'hhea' table...
Dumping 'maxp' table...
Dumping 'OS/2' table...
Dumping 'hmtx' table...
Dumping 'cmap' table...
Dumping 'fpgm' table...
Dumping 'prep' table...
Dumping 'cvt ' table...
Dumping 'loca' table...
Dumping 'glyf' table...
Dumping 'name' table...
Dumping 'post' table...
Dumping 'gasp' table...
Dumping 'GDEF' table...
ERROR: An exception occurred during the decompilation of the 'GPOS' table
Traceback (most recent call last):
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/ttFont.py", line 372, in __getitem__
    return self.tables[tag]
KeyError: 'GPOS'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/ttFont.py", line 391, in __getitem__
    table.decompile(data, self)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otBase.py", line 41, in decompile
    self.table.decompile(reader, font)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otBase.py", line 640, in decompile
    table[conv.name] = conv.read(reader, font, table)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 547, in read
    table.decompile(reader, font)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otBase.py", line 636, in decompile
    table[conv.name] = conv.readArray(reader, font, table, countValue)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 167, in readArray
    l.append(self.read(reader, font, tableDict))
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 547, in read
    table.decompile(reader, font)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otBase.py", line 636, in decompile
    table[conv.name] = conv.readArray(reader, font, table, countValue)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 167, in readArray
    l.append(self.read(reader, font, tableDict))
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 547, in read
    table.decompile(reader, font)
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otBase.py", line 619, in decompile
    conv = conv.getConverter(reader.tableTag,
  File "/tmp/venv/lib64/python3.9/site-packages/fontTools/ttLib/tables/otConverters.py", line 596, in getConverter
    tableClass = self.lookupTypes[tableType][lookupType]
KeyError: (34, 'SubTable', 'Lookup', 'LookupList')
Dumping 'GPOS' table...
Dumping 'GSUB' table...
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 256 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 257 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 258 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 259 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 260 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table
WARNING: name id 261 missing from name table

madig Feb 9, 2021
Author

Or did you mean to say that #2705 needs to land before this works but you tried it on a simpler font and it worked?

garretrieger Feb 9, 2021
Maintainer

Yes you'll need #2705 merged for extension lookup subsetting to work, currently on master they are dropped.

madig Feb 9, 2021
Author

Ah! Great, now I get it :D tried it on LiberationSans-Regular.ttf v2.1.0 and it works as expected. Thanks!

behdad · 2021-02-09T19:47:49Z

behdad
Feb 9, 2021
Maintainer

@madig for you, pretty much anything you open is probably a bug in our code. So feel free to use Issues instead of Discussion. We don't bite. :)

1 reply

madig Feb 9, 2021
Author

Will do after I know what's going on in the log above 🙀

behdad · 2021-02-09T19:49:53Z

behdad
Feb 9, 2021
Maintainer

So what others said. If you don't specify the mark characters, they won't be included.

What's the alternative? Include all marks that interact with specified characters? That results in a very arbitrary selection. Though I see how that would be useful for production workflows. Feel free to open an issue to discuss the semantics (and we should first add to fonttools subsetter). There has been similar requests in the past by Adam Twardoch and others...

6 replies

garretrieger Feb 9, 2021
Maintainer

We have an issue discussing whether we should do that or not in harfbuzz here: #2283

madig Feb 9, 2021
Author

Another thought is that maybe deep glyph exploration by the subsetter becomes less pressing if you can more easily input glyphs you want to keep. Currently, either --text or --text-file seem to be a hard requirement, you can't just give --gids, or even better, actual glyph names (and files).

behdad Feb 9, 2021
Maintainer

Currently, either --text or --text-file seem to be a hard requirement, you can't just give --gids, or even better, actual glyph names (and files).

That sure isn't how we want it to work indeed. Fonttools subsetter does the way you suggest and that was as designed. What if you give it an empty --text string and --gids?

madig Feb 10, 2021
Author

I was expecting an "either --text or --gids" behavior or maybe union behavior? Anyway, all I really want is #2455

behdad Feb 10, 2021
Maintainer

I was expecting an "either --text or --gids" behavior or maybe union behavior?

Union behavior is how I designed the fonttools subsetter to work. I like to see the same in hb-subset as well.

That is to say, if there's zero characters in the text, we should not produce any GSUB/GPOS/GDEF tables; that is, layout subsetting is automatically dropped because there's no text expected to ever be shaped using the subset font. Internally we don't distinguish between text-only and glyph-only modes. Those should just work as expected given a unified implementation.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

hb-subset discards mark/mkmk feature? #2841

{{title}}

Replies: 4 comments 18 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

hb-subset discards mark/mkmk feature? #2841

madig Feb 2, 2021

Replies: 4 comments · 18 replies

behdad Feb 5, 2021 Maintainer

garretrieger Feb 5, 2021 Maintainer

qxliu76 Feb 9, 2021 Collaborator

madig Feb 9, 2021 Author

madig Feb 9, 2021 Author

garretrieger Feb 9, 2021 Maintainer

madig Feb 9, 2021 Author

behdad Feb 9, 2021 Maintainer

madig Feb 9, 2021 Author

behdad Feb 9, 2021 Maintainer

garretrieger Feb 9, 2021 Maintainer

madig Feb 9, 2021 Author

behdad Feb 9, 2021 Maintainer

madig Feb 10, 2021 Author

behdad Feb 10, 2021 Maintainer

madig
Feb 2, 2021

Replies: 4 comments 18 replies

behdad
Feb 5, 2021
Maintainer

garretrieger
Feb 5, 2021
Maintainer

qxliu76 Feb 9, 2021
Collaborator

madig Feb 9, 2021
Author

madig Feb 9, 2021
Author

garretrieger Feb 9, 2021
Maintainer

madig Feb 9, 2021
Author

behdad
Feb 9, 2021
Maintainer

madig Feb 9, 2021
Author

behdad
Feb 9, 2021
Maintainer

garretrieger Feb 9, 2021
Maintainer

madig Feb 9, 2021
Author

behdad Feb 9, 2021
Maintainer

madig Feb 10, 2021
Author

behdad Feb 10, 2021
Maintainer