Run the tests unconditionally and skip if the shaper is not available.
This fixes distcheck (https://github.com/harfbuzz/harfbuzz/pull/3504)
and shows SKIP for these tests instead of ignoring them.
For builds from release tarballs, the tests fail in the DirectWrite and
Uniscribe tests when these platform shapers are enabled, since the data files
were not found in the source tree, when building with Meson at least.
Fix this by dist'ing the platform shaper test data files.
we should not cache visited langsys cause 2 different Record<Langsys>
could have different Tag while pointing to the same Langsys, a langsys
is redundant in Record<Langsys> A does not mean it's redundant in Record
B. Same thing for visited_script.
Also adding the number of features in the LangSys's feature list to the
visited langsys count so it's more accurate.
Plus some improvement in langsys compare()
Add a variation of hb_ot_layout_get_baseline that
synthesizes missing baselines, using heuristics in part
taken from the CSS Inline Layout Module, Level 3.
Includes some new tests for synthesized baselines.
The base2.ttf is a subset of Noto Sans Bengali that
includes just the Bengali Ka.
New API: hb_ot_layout_get_baseline_with_fallback
[Boring Expansion] >64k loca & hmtx tables
This does two things:
The num-glyphs reported by the face now is the maximum reported by the maxp and that deduced from the length of the loca table; I think this is the right thing to do anyway; According to OpenType such loca tables are invalid.
The interpretation hmtx tables that have excessive bytes at the end, again, invalid according to OpenType, has changed. Previously we were interpreting those excessive bytes as extra lsb values. Now we interpret them as extra advance values, the last of which is repeated for all missing glyphs. Again, these are tables that are invalid according to OpenType, and the advances are for glyph indices beyond maxp table's num-glyphs.
The combined effect is that the font can have shapes and advances for gid's beyond the maxp limit of 64k. In fact, maxp table becomes optional.
As CI failure, apparently the my local freetype and CI one have different
result so let's switch the case with a simpler one just to test quadratic command
is emitted correctly.
This is unnecessary overhead. Up to rasterizers to handle this. Plus,
this throws off point-numbers in uses that rely on it.
Disabled one test that broke with this.