Remove redundant name table records

[m4rc1e]

I've needed a way to remove redundant name table records for many years so I thought I'd just write it. I'm using Behdad's visitor pattern to collect all the name records which are being used by the fvar, STAT, CPAL etc.

You can get redundant name records when you replace a font's STAT table using the builder's buildSTAT function or when hot fixing with other scripts etc.

My plan is to make removeUnusedNameRecords a method of table__n_a_m_e. I just wanted to open this pr to get feedback before I progress too far. No worries if you don't want this pr either.

So far I've covered every table in ttLib/tables/otData, STAT, fvar and CPAL. Please let me know if I'm missing any.

[behdad]

Looks good!

[behdad]

Maybe remove this from the visitor object, move it somewhere else.

[m4rc1e]

Thanks for looking. I've moved and converted the function so it's a static method of the name table. Imo it feels like belongs here. Happy to drop the static method and just have it as a plain func instead. I've moved it away from the Snippets dir so it can be imported into other libs and tools.

Happy to add tests if you're ok with the above. Still not too sure if I've found all the places in a font which can have a nameID though.

[behdad]

Thanks. We should see if we can use this, or at least compare notes with, the subsetter. That one looks very incomplete to me.

https://github.com/fonttools/fonttools/blob/main/Lib/fontTools/subset/__init__.py#L2917

[behdad]

[Use NameRecordVisitor in subsetter](/fonttools/fonttools/pull/3185/commits/edf8891fba54015c5d37a6397b7b95f14dfe5662)

Slick! I just wonder if we need to add a table order dependency to process name last.

[m4rc1e]

I just wonder if we need to add a table order dependency to process name last.

I'm not too familiar with all of the subsetting code. However, if I skim, I see self._prune_pre_subset and _subset_glyphs methods call _sort_tables. Is it sufficient to just move the "name" table to last pos in _sort_tables? if not, I'll keep digging.

[behdad]

Thanks. Checking the subset code, I think we need to move the name-table subsetting from prune_pre_subset to prune_post_subset. That might be enough.

There's also some "name" table code in the "CPAL" subsetting method, which should be removed.

I don't remember why "name" table is in _sort_tables. May be possible to remove it now, maybe not.

[behdad]

There's also some "name" table code in the "CPAL" subsetting method, which should be removed.

Can you please do this as well?

Thanks. I let @anthrotype review the rest.

[m4rc1e]

There's also some "name" table code in the "CPAL" subsetting method, which should be removed.

Can you please do this as well?

Think I got it all. Thanks for your help Behdad!

[anthrotype]

LGTM, with comments

[anthrotype]

the snippet doesn't do much besides loading a TTFont, calling remoteUnusedNames method, and saving back; do we actually need it?

[m4rc1e]

Thanks Cosimo. I've implemented your feedback and removed the snippet. The comment may be too long for your liking. Lmk and I'll rephrase.