gfsubsets 2025.11.4

Codepoint definitions for the Google Fonts subsetter

gfsubsets

This is the collection of nam files (codepoint subsets) that are used to subset fonts before serving on the Google Fonts CSS API.

The subset files

The subset definitions are contained in two forms; the "editable" definitions are found in the subsets-input directory, and the machine readable defitions can be found in Lib/gfsubsets/data.

The machine readable files consist of text files with one Unicode codepoint (a hexadecimal number beginning with 0x) on each line; everything after the codepoint is a comment. Blank lines and comment lines beginning # are ignored.

The editable definitions are similar, but for convenience allow the following "directives":

• Two codepoints separated by .. denote an (inclusive) range; all codepoints within the range are included.
• The symbol - before a codepoint or range excludes it from the output set.
• The directive @include(filename) includes the codepoints in filename.
• The directive @block(XYZ) includes the codepoints in the Unicode block named XYZ.
• The directive @script(XYZ) includes all codepoints with either the Unicode Script property equal to XYZ or the ScriptExtensions property equal to XYZ

These editable definitions are turned into their machine-readable equivalents using the script scripts/preprocess_namfile.py.

Python interface

This repository also includes the Python module gfsubsets which provides an interface to these subset definitions. It exports the following functions:

• CodepointsInFont(filename): Lists the Unicode codepoints supported by the font
• ListSubsets(): Returns the name of all defined subsets.
• SubsetsForCodepoint(cp): Returns the names of all subsets including the codepoint.
• SubsetForCodepoint(cp): Returns the name of the "most relevant" subset including the codepoint.
• CodepointsInSubset(subset): Returns a set of codepoints included in the subset.
• SubsetsInFont(filename, min_pct, ext_min_pct): Returns the name of subsets "well" supported by a font.