When passing a parsed feature file that has variable anchors to
addOpenTypeFeatures(), builder would modify the anchors in place and
discard the variations, which break any subsequent use of the feature
file.
I encountered this building a font that has variable cursing anchors
with ufo2ft. The cursFeatureWriter would write the variable anchors, but
then when kernFeatreWriter compiles the file to get GSUB closure, the
variation would be dropped from the anchors, and later when when the
feature data is compiled into the font, the anchors would be compiled
without variations.
technically we are tweaking the original example from the spec but it keeps the spirit, so that the product of glyph classes produces the same representation in the font as if the sequences were manually enumerated (while keeping the declaration order)
If a FeatureVariationRecord with the same ConditionTable exists re-use
it and append FeatureTableSubstitutionRecord’s.
Without this, in the following feature code only the first lookup will
be applied since there will be two FeatureVariationRecord with the same
ConditionTable, so the first will be matched and the other will be
skipped:
conditionset test {
wght 600 1000;
wdth 150 200;
} test;
variation ccmp test {
sub e by a;
} ccmp;
variation rlig test {
sub b by c;
} rlig;
With this change only one FeatureVariationRecord will be created with
two FeatureTableSubstitutionRecord’s.
Allow variable scaler in ligature caret position and build
CaretValueFormat3 with DeviceTable. Does not support non-variable device
table, but can be added if someone really really wants it.
If a variable scalar does not vary (i.e. all values are the same), we
can simplify it and use a simple value instead. Arguably users shouldn’t
be using variable scalar in this case, but it helps when the feature
code is auto generated so each feaLib users doesn’t have to do the
check done here themselves.
This is a GlyphsApp extension, (partially) documented [here][1], but it is a
useful one as it allows concise glyph insertion using decomposition
without having to maintain the list of glyphs in two places. E.g.
```fea
@upper = [A-Z];
lookup insert {
sub @upper by @upper connector;
} insert;
feature ccmp {
sub @upper' lookup insert x;
} ccmp;
```
As apposed to:
```fea
@upper = [A-Z];
lookup insert {
sub A by A connector;
sub B by B connector;
sub C by C connector;
sub D by D connector;
sub E by E connector;
sub F by F connector;
sub G by G connector;
sub H by H connector;
sub I by I connector;
sub J by J connector;
sub K by K connector;
sub L by L connector;
sub M by M connector;
sub N by N connector;
sub O by O connector;
sub P by P connector;
sub R by R connector;
sub S by S connector;
sub T by T connector;
sub U by U connector;
sub V by V connector;
sub W by W connector;
sub X by X connector;
sub Y by Y connector;
sub Z by Z connector;
} insert;
feature ccmp {
sub @upper' lookup insert x;
} ccmp;
```
1. http://handbook.glyphsapp.com/en/layout/multiple-substitution-with-classes/
Fixes https://github.com/fonttools/fonttools/issues/2949
- If there are no marked glyphs in an "ignore" statement, issue a
warning. The spec disallows this but makeotf allows it. It is most
likely a typo, so a warning is warranted.
- Mark the first glyph not all the glyphs, this matches makeotf.
- In asFea() always mark the input glyph.
- Udpate the tests.
In what I presume was error, this test case included a space on the line
immediately following the <ttFont> element; this space is not present in
any other ttx files.
For reasons that are unclear to me, and likely not worth investigating,
the presence (or absence) of this space does not influence the passing
or failure of this test case; however it *does* cause a failure in my
feature compiler, which is reusing this test suite, and so I would like
to fix it here, and simplify my own life somewhat.
This patch ensures that feaLib always produces a name table with entries
sorted in the order proscribed by the spec: platform id, encoding id,
language id, name id.
This breaks some tests, and so I have manually updated the test data to
match the new outputs.
* add test that fails for #2293
* fixing #2293: rewrite of contextual positioning logic, ensure len(suffix) > 1 yields the correct result; checking more edge cases and raising errors inspired by makeotf
* test error cases
* only check when we actually have a value
* catch one more case that makeotf errors on and we didn't
* Replaced all from ...py23 import * with explicit name imports, or removed completely when possible.
* Replaced tounicode() with tostr()
* Changed all BytesIO ans StringIO imports to from io import ..., replaced all UnicodeIO with StringIO.
* Replaced all unichr() with chr()
* Misc minor tweaks and fixes
