|
59c45f6d
|
2022-11-22T12:54:50
|
|
Use hb_memcpy instead of memcpy consistently
|
|
|
a0bde1e1
|
2022-11-16T21:27:12
|
|
[open-type] Remove (Sorted)ArrayOf.sub_array()
|
|
83769b9c
|
2022-10-21T22:37:32
|
|
[subset] add comment for why we retain empty lookups.
|
|
|
70e2de2b
|
2022-10-21T22:33:17
|
|
[subset] Always output Lookup's even if they are empty.
The rest of layout subsetting depends on lookup indices being consistent with those computed during planning. So if an empty lookup is discarded during the subset phase it will invalidate all subsequent lookup indices. Generally we shouldn't end up with an empty lookup as we pre-prune them during the planning phase, but it can happen in rare cases such as when a subtable is considered degenerate (eg. #3853)
|
|
c813f842
|
2022-10-20T19:45:23
|
|
Make build work for arm-none-eabi
|
|
b706c6f7
|
2022-09-12T13:58:55
|
|
[instance] update FeatureList with variations
|
|
|
8f3a7017
|
2022-09-09T15:00:37
|
|
[instance] support FeatureVariations table
|
|
|
f4813e3b
|
2022-09-09T10:34:09
|
|
[instance] update collect_lookups/prune_features()/closure_features() with variations
Some features will be substituted with variations, so we do not collect
lookups from the original feature tables.
|
|
|
64e2f2fc
|
2022-09-09T09:36:19
|
|
[instance] store active featurevariation record/condition idxes in the plan
|
|
|
9ab6605f
|
2022-08-02T09:54:29
|
|
[instance] update GDEF table
|
|
|
b72995ff
|
2022-08-01T14:48:01
|
|
[instance] GDEF table: collect both varidxes and deltas
|
|
|
e03043bd
|
2022-08-01T11:37:47
|
|
[subset] don't let VariationStore subset() produce inner_maps
Make it a subset-planning object so VariationStore can subset to
specified inner_maps.
Also add a layout_variation_idx_delta_map in subset_plan
|
|
|
7fde6ab0
|
2022-08-05T13:33:22
|
|
fuzzer fix: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=49790
|
|
|
c67c1b74
|
2022-08-04T11:42:22
|
|
VarStore: simplify again
|
|
|
d831e935
|
2022-08-04T10:15:21
|
|
bug fix in VarData get_delta ()
when LONG_WORDS flag is set, item row is not computed correctly
lcursor should be interpreted as INT32
|
|
|
3da50c50
|
2022-07-29T11:40:33
|
|
Remove unused const
|
|
|
3f7a74ff
|
2022-07-19T21:50:13
|
|
[repacker] WIP extension promotion implementation.
|
|
|
4cb83967
|
2022-07-23T10:59:42
|
|
[subset/ClassDefFormat2] Fix timeout
Fixes https://oss-fuzz.com/testcase-detail/5417800474165248
|
|
|
f94a3ba1
|
2022-07-23T10:36:08
|
|
[varStore] Better protect against HB_NO_VAR builds
|
|
|
5a9c7930
|
2022-07-22T21:33:15
|
|
Add HB_NO_BEYOND_64K
|
|
|
7cdde6a2
|
2022-07-21T17:17:59
|
|
[ClassDef] Write a loop as range for
|
|
|
00dfbbce
|
2022-07-21T11:39:32
|
|
[Coverage] Rename and templatize intersected_coverage_glyphs
|
|
|
8d923363
|
2022-07-20T12:08:18
|
|
[layout] Reduce number of closure rounds
|
|
|
f3151b65
|
2022-07-17T16:19:28
|
|
[ArrayOf family] Use memory barrier before accessing array
Without it, the compiler was reordering and batching the read
of array length and array[0] if the 0'th member was accessed
constantly and function was inlined. This felt safe to the
compiler because HB_VAR_ARRAY is 1, but could be unsafe actually.
The memory barrier disallows that.
This was found by afl/honggfuzz address sanitizers.
Fixes https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=49187
|
|
|
1327d8e3
|
2022-07-15T16:03:38
|
|
[layout-common] Reshuffle code so Feature is defined before it's used
|
|
|
c1e280ea
|
2022-07-13T22:43:38
|
|
[reorg] Move Coverage, RangeRecord into new namespace layout.
|
|
|
df7eebf4
|
2022-07-11T14:14:55
|
|
[>64k:layout] Fix layout of RangeRecord
https://github.com/be-fonts/boring-expansion-spec/issues/30
|
|
|
9ef9fc01
|
2022-07-11T13:29:23
|
|
[>64k:layout] Templatize GSUBGPOSFormat1
|
|
|
5fd0a3f0
|
2022-07-11T13:10:04
|
|
[>64k:layout] Templatize GSUBGPOSFormat1
|
|
|
ca5c8a64
|
2022-07-06T13:02:49
|
|
[>64k:layout:Coverage] Implement format 3/4
Implements rest of https://github.com/be-fonts/boring-expansion-spec/issues/30
|
|
|
25de6fb4
|
2022-07-06T12:52:03
|
|
[>64k:layout:ClassDef] Implement format 3/4
This implements part of https://github.com/be-fonts/boring-expansion-spec/issues/30
|
|
|
9286526f
|
2022-07-06T12:26:58
|
|
[>64k:layout] Templatize Coverage & ClassDef
Have not added new formats yet.
|
|
|
e8cce9d1
|
2022-07-05T14:05:43
|
|
[>64k:layout] Add SmallTypes & MediumTypes
|
|
|
8a971d01
|
2022-07-06T12:37:43
|
|
[layout] Use is_source_of instead of is_iterator
|
|
|
0dc0da05
|
2022-07-06T12:33:37
|
|
[iter] Add hb_is_sorted_iterator(_of)
|
|
|
587969af
|
2022-06-30T21:37:42
|
|
[subset] Drop scripts that are not in the layout_scripts list.
|
|
|
b96622d1
|
2022-06-05T02:45:41
|
|
[layout] Use a cache for main input ClassDef of (Chain)ContextLookupFormat2
This commit adds a per-lookup caching infrastructure to GSUB/GPOS, and
uses it to cache input ClassDef.get_class value for (Chain)ContextLookupFormat2.
For fonts heavy on use of heave class-based2 context matching, this shows
a good speedup. For NotoNastaliqUrdu for example, I observe 17% speedup.
Unfortunately not many other lookups can use a cache like this :(.
https://github.com/harfbuzz/harfbuzz/pull/3636
|
|
|
e6409d39
|
2022-06-05T06:57:37
|
|
Revert "[layout] Use a cache for main input ClassDef of (Chain)ContextLookups"
This reverts commit 57d1c08739d0acd94b96da2f9d5dd6c0ff3b3722.
Err. This was an accident.
|
|
|
57d1c087
|
2022-06-05T02:45:41
|
|
[layout] Use a cache for main input ClassDef of (Chain)ContextLookups
|
|
|
f13a7954
|
2022-06-03T01:17:20
|
|
[subset] Convert another use of hashmap to unique_ptr
|
|
|
997d9cc4
|
2022-06-02T18:04:12
|
|
[map] Make unique_ptr hashable
|
|
|
5af5a565
|
2022-05-30T08:32:50
|
|
[VarStore] Implement writing 32bit var-store
Untested.
Finishes fixing https://github.com/harfbuzz/harfbuzz/issues/2965
|
|
|
75112098
|
2022-05-23T12:42:15
|
|
[VarStore] Implement reading 32bit var-store
Untested.
Part of https://github.com/harfbuzz/harfbuzz/issues/2965
Serializing is incomplete.
|
|
|
e5d6da79
|
2022-05-23T12:25:39
|
|
[varStore] rename shortCount to wordCount
|
|
|
371e14d9
|
2022-05-28T13:40:30
|
|
Combine uses of map has() then get() with has(.., &..)
|
|
|
6010feee
|
2022-05-24T09:00:44
|
|
[varStore] Rename variable as per review
https://github.com/harfbuzz/harfbuzz/pull/3605
|
|
|
d9acc045
|
2022-05-21T14:16:21
|
|
[VarStore] Sprinkle cache_t type around
It's available so no need to use void*.
|
|
|
da38312e
|
2022-05-20T17:32:56
|
|
[VarStore] Pepper cache with likely()
|
|
|
cf8f00e3
|
2022-05-20T17:14:10
|
|
[VarStore] Don't use NAN
Is faster.
With this, I'm seeing 25 to 28% speedup in glyph_h_advances benchmark
of benchmark-font for var/hb tests.
|
|
|
880f50f7
|
2022-05-20T16:50:00
|
|
Refactor varstore cache
|
|
|
f2a2fb91
|
2022-05-20T16:06:05
|
|
[GPOS] Cache VarStore region scalars
|
|
|
5fbc70c5
|
2022-05-20T14:07:27
|
|
[VarStore] Add cache API
|
|
|
3ab2c793
|
2022-05-19T17:23:36
|
|
[subset-perf] Signficiantly speed up ClassDef*::subset.
Eliminates the usage of a glyph -> klass hash map and replaces it with a vector storing the mapping. This allows us to use the vector directly as the iterator driving the serialize. Approximately 1% speedup for Noto Nastaliq.
|
|
|
0b1c2ff9
|
2022-05-18T23:32:03
|
|
[subset-perf] Remove extra map lookup in ClassDef subset methods.
|
|
|
13ace77f
|
2022-05-18T22:38:43
|
|
[subset-perf] Use glyph_map instead of set in ClassDefFormat.
|
|
|
adae2f22
|
2022-05-18T21:42:28
|
|
[subset-perf] Cache a glyph map for gsub.
This allows us in some cases to avoid using glyph_set_gsub as a filter.
|
|
|
cedf7396
|
2022-05-18T16:52:35
|
|
Add some commented-out code
|
|
|
48dfbd54
|
2022-05-18T21:03:56
|
|
[subset] minor cleanup.
|
|
|
482c6e5d
|
2022-05-18T19:58:55
|
|
[subset-perf] Speed up Coverage::serialize by caching iterator.
|
|
|
14b18725
|
2022-05-18T15:14:32
|
|
In Coverage::iter_t, assume iterators are from same Coverage object
No need to support otherwise.
|
|
|
27141735
|
2022-05-18T15:12:49
|
|
[subset] Add Coverage::__end__ implementation
|
|
|
c476f58a
|
2022-05-18T14:20:23
|
|
[subset] Write CoverageFormat2::intersects_coverage() as bsearch()
|
|
|
63c66951
|
2022-05-18T13:53:52
|
|
[ot-layout] Cosmetic
The implementation of HBUINT16 operator == is slower than just
comparing to ints.
|
|
|
777debd7
|
2022-05-18T13:46:06
|
|
[subset] Rewrite CoverageFormat2::intersects as dagger
|
|
|
cf5001fa
|
2022-05-18T13:38:29
|
|
[subset] Optimize CoverageFormat2::intersected_coverage_glyphs
|
|
|
6f37c207
|
2022-05-18T13:25:42
|
|
[subset] Minor rewrite in CoverageFormat2::serialize()
|
|
|
175319cd
|
2022-05-11T13:47:17
|
|
[gsubgpos] Clean up OT::ClassDefFormat2::intersected_class_glyphs 0 case
|
|
|
137af361
|
2022-05-11T13:39:30
|
|
[gsubgpos] Simplify OT::ClassDefFormat2::intersected_class_glyphs()
|
|
|
3261e05b
|
2022-05-11T13:16:31
|
|
[subset] Optimize ClassDef1::intersected_class_glyphs() for class0
|
|
|
a792e16e
|
2022-03-24T13:08:51
|
|
[ot-layout] Change max nesting level of lookups from 6 to 64
|
|
|
7cb002cb
|
2022-03-14T11:40:35
|
|
[subset] bug fix in prune_langsys
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()
|
|
|
de2cef42
|
2022-01-18T09:21:31
|
|
[bimap] Remove init/fini
|
|
e2e30506
|
2022-01-16T07:00:53
|
|
Fix various typos
Found via `codespell -q 3 -S ./perf/texts -L actualy,ba,beng,fo,gir,inout,nd,ot,pres,ro,te,teh,timne`
|
|
|
067f90a8
|
2021-12-14T16:24:38
|
|
[subset] Fix for fuzzer timeout.
Fixes https://oss-fuzz.com/testcase-detail/5549945449480192
In prune_langsys: move LangSys visited check up before any work is done for a LangSys. In this particular case the compare() method is responsible for the majority of the time spent and wasn't being guarded with a visisted check.
|
|
|
706014f6
|
2021-12-01T20:20:12
|
|
[subset] (Chain)ContextSubst glyph_closure fix
- When pos_glyphs is empty, use current full glyphs set as input for
subsequent recursive closure process
- Also increase max_lookup_visit_count to 35000 cause a real font file hit
previous limit 20000 and some lookups are dropped unexpectedly
|
|
|
394f7729
|
2021-11-19T11:49:23
|
|
[map] Allow storing classes in the hashmap
Fixes https://github.com/harfbuzz/harfbuzz/issues/3293
The trick was to change the type of the invalid key/value to be non-class.
|
|
|
a75b96f7
|
2021-11-16T14:34:07
|
|
[sanitize] In lookup sanitize don't try to access subtable 0 if it isn't there.
|
|
|
540f19b6
|
2021-10-29T17:11:53
|
|
[subset] fix bug in (Chain)ContextFormat2
Only keep rulesets for glyphs class numbers that survived in coverage
|
|
|
6d555ce8
|
2021-11-02T00:18:22
|
|
[meta] Use std::forward instead of hb_forward()
|
|
|
ca7b9dae
|
2021-09-20T14:42:51
|
|
[subset] subset MATH table
|
|
|
c2cc566c
|
2021-09-22T14:15:55
|
|
[subset] Fix subset_offset_array adding unused space to serializer.
ArrayOf.serialize_append allocates space for the new item, but ArrayOf.pop() does not recover the allocated space. So in the case where the revert path was entered the extra space added by serialize_append gets left in the serialization buffer. This moves the snapshot to before ArrayOf.serialize_append is called so that revert cleans up the buffer extend.
|
|
|
14a2df7b
|
2021-09-19T23:06:09
|
|
Add stub HBUINT15
|
|
|
c852b868
|
2021-09-19T16:30:12
|
|
Rename HBGlyphID to HBGlyphID16
|
|
|
b63ac571
|
2021-07-29T18:23:41
|
|
[subset] bail if collection region indices is in error.
|
|
|
30579f5a
|
2021-07-29T11:19:13
|
|
[set] Document & use open-ended del_range()
|
|
|
2d54c346
|
2021-07-28T18:20:53
|
|
[subset/VarStore] Fix OOM in fuzzer test
The test in question is the one added in c68a00b92eb62fdc79f9a2bd1fc7a6acc24a3602.
Culprit is that it's allocating lots of memory because of region_indices that
are out-of-range anyway. So, try to filter those out first.
|
|
|
ff1fe25a
|
2021-07-28T18:02:52
|
|
[VarStore] Rename internal get_scalars() to get_region_scalars()
|
|
|
f0a1892f
|
2021-07-28T17:36:22
|
|
[serialize] Remove unnecessary pointer indirection
|
|
|
1c511ed6
|
2021-07-28T17:33:26
|
|
[subset] Simplify VarStore serialization
|
|
|
ad4c6924
|
2021-07-28T16:50:34
|
|
[subset] Simplify VarStore serialize
|
|
|
25dee609
|
2021-07-28T13:58:24
|
|
[sanitize] Revert some from c68a00b92eb62fdc79f9a2bd1fc7a6acc24a3602
That overflow check was already happening in arrayZ.sanitize().
|
|
|
c68a00b9
|
2021-07-27T13:25:22
|
|
[subset] Fix possible overflows in VarRegionList serialize
Fixes https://oss-fuzz.com/testcase-detail/5362189182566400
|
|
|
03cd9c5c
|
2021-07-22T11:27:33
|
|
Align lfind() API to match bfind()
|
|
|
6269866b
|
2021-07-16T14:06:55
|
|
[subset] Keep bimap consistent if only one of the internal maps is in error.
|
|
|
48edbef9
|
2021-06-29T13:14:26
|
|
[subset] bail if region_map is in error.
Fixes fuzzer case https://oss-fuzz.com/testcase-detail/5499162606043136.
|
|
|
0da9158f
|
2021-06-29T14:23:37
|
|
[subset] Remove OffsetTo::serialize().
Convert remaining uses of it to serialize_serialize() which correctly uses the object packer.
|
|
|
be50779c
|
2021-04-30T12:59:17
|
|
[subset] correct ClassDef format determination.
It was possible for glyph_min to be a glyph mapped to class zero which threw off the number of glyph calculation.
|
|
|
84dbfacf
|
2021-04-15T16:52:11
|
|
[subset] fix collect_variation_idx in PairPosFormat2
|
|
|
596f4258
|
2021-04-02T12:41:09
|
|
[subset] copy Lookup::markFilteringSet when subseting.
|
|
|
092094f7
|
2021-04-01T15:47:21
|
|
Use as_array() and range loops in a few places
|