Glyph insertion is an expensive operation and we like to have it limited
based on buffer's input size which is handled by buffer's max_ops.
clusterfuzz-testcase-minimized-harfbuzz_fuzzer-5754958982021120:
Before the change: 0.67s user 0.00s system 99% cpu 0.674 total
After the change: 0.02s user 0.00s system 98% cpu 0.024 total
Which takes much longer on valgrind and tsan bots.
* [aat] Correct array indexing when looking up actions in KerxSubTableFormat4.
- For action_type 0 and 1, there are 2 values per action record; for action_type 2, there are 4. So we need to account for these factors when indexing into the ankrData array.
Fixes https://github.com/harfbuzz/harfbuzz/issues/2531.
Very low use, only two distinct font files, Apple Chancery.ttf and Hoefler Text.ttc
have it so it really doesn't worth the size addition and so, but one may argue that
whole ligature caret is low use but guess we better to encourage GDEF one anyway.
Fixing these complains raised by newer versions of clang,
./hb-ot-layout-common.hh:1720:53: error: loop variable 'gid_klass_pair' is always a copy because the range of type 'hb_map_iter_t<hb_sorted_array_t<OT::HBGlyphID>, (lambda at ./hb-ot-layout-common.hh:1672:29), hb_function_sortedness_t::RETAINS_SORTING, nullptr>' does not return a reference [-Werror,-Wrange-loop-analysis]
for (const hb_pair_t<hb_codepoint_t, unsigned>& gid_klass_pair : + it)
And
./hb-ot-color-colr-table.hh:177:44: error: loop variable '_' is always a copy because the range of type 'hb_map_iter_t<hb_filter_iter_t<hb_map_iter_t<hb_range_iter_t<unsigned int, unsigned int>, (lambda at ./hb-ot-color-colr-table.hh:209:31), hb_function_sortedness_t::RETAINS_SORTING, nullptr>, (anonymous struct at ./hb-algs.hh:331:1) &, (anonymous struct at ./hb-algs.hh:51:1) &, nullptr>, (anonymous struct at ./hb-algs.hh:338:1) &, hb_function_sortedness_t::RETAINS_SORTING, nullptr>' does not return a reference [-Werror,-Wrange-loop-analysis]
for (const hb_item_type<BaseIterator>& _ : + base_it.iter ())
Use the WINAPI_PARTITION macro to filter desktop/app flavors.
We use a negated desktop check because the default (for mingw-w64 at
least) is to allow all API by combining desktop + app partitions.
This causes build failures if we were to filter using
WINAPI_PARTITION(WINAPI_FAMILY_APP) because it would always be true, but
those API also require Windows 8 or later, while we only require Vista
Fixes warnings like
../src/hb-blob.cc:572:47: warning: 'WINAPI_FAMILY_PC_APP' is not defined, evaluates to 0 [-Wundef]
#if defined(WINAPI_FAMILY) && (WINAPI_FAMILY==WINAPI_FAMILY_PC_APP || WINAPI_FAMILY==WINAPI_FAMILY_PHONE_APP)
^
../src/hb-blob.cc:572:86: warning: 'WINAPI_FAMILY_PHONE_APP' is not defined, evaluates to 0 [-Wundef]
#if defined(WINAPI_FAMILY) && (WINAPI_FAMILY==WINAPI_FAMILY_PC_APP || WINAPI_FAMILY==WINAPI_FAMILY_PHONE_APP)
How this works? gen-harfbuzzcc.py operates at its own source path (see its 7th line)
and that is reliable when used both on meson and autotools.
Just like 19ecabed, weirdly this didn't come up sooner, guess it has something
to do with timestamps. Fortunately whole harfbuzz.cc just doesn't matter for
packagers but we can tag a release only for this if needed.
The way it used to work was a bit nonidiomatic but the replacment
is idiomatic way of iterating used elsewhere.
The new code just doesn't check nullability of "characters", which isn't
what we do anywhere else.
All the other similar iterating API are like this and don't have such
comment, written at the time I wasn't familiar enough with the way such
API are shaped.
First time we do this in a way that if target object doesn't have the matching
function we basically "ignore". Risky but I feel like is the right decision
for this case.
I'm going to put back the template varargs and use those, which would make
the dispatcher be just that: "dispatcher", and wouldn't need to carry the
call context. That would be a refreshing change I think.
This was done in #770 but no indication of anyone is using it,
let's remove it from our meson port and we can just don't care about
it in autotools port after the migration to meson.
Instead of passing dependencies as required we used one giant shared
dependency list containing all dependencies for every library/executable.
While this kinda works, the specified deps are also used for generating
the pkg-config files and this leads to lots of Requires.private and Libs.private
entries which aren't really needed.
This removes the "deps" array and replaces it with a few smaller ones and
makes sure the public libraries only get passed the dependencies actually
needed.
Fixes#2441
ptem, used for AAT's tracking/`trak` table is equivalent to opsz of variable fonts.
For variable AAT fonts, such as SFNS, ideally variable axis of the hb_font_t
should be set and equivalent to ptem, https://crbug.com/1005969#c37
Searches variation axes of a hb_font_t object for a specific axis first,
if not set, then tries to get default style values from different
tables of the font.
So one can run a category of interested tests like
meson test -Cbuild --suite aots --suite src --print-errorlogs
Intead issuing particular tests which also is possible like
meson test -Cbuild test-shape --print-errorlogs
This way it won't ruin incremental builds.
We need a better way to declare source altering tasks
https://github.com/mesonbuild/meson/issues/7156
yet this is good enough despite being not idiomatic.
It is however not that smooth yet as the change may is detected on the
next meson run. One of course can issue ./gen-ragel-artifacts.py
manually for better experience before running meson.
Using a C linker was limited to non-Windows as 20a840c7, let's
revisit this while transition to meson.
Packagers easily override that via the option and use a C++ linker
if needed.
As https://circleci.com/gh/harfbuzz/harfbuzz/140481
Traceback (most recent call last):
File "./gen-hb-version.py", line 14, in <module>
output_file.write (input_file.read ()
File "/usr/lib/python3.4/encodings/ascii.py", line 26, in decode
return codecs.ascii_decode(input, self.errors)[0]
UnicodeDecodeError: 'ascii' codec can't decode byte 0xc2 in position 16: ordinal not in range(128)
In file included from hb-ot-face.cc:34:
In file included from ./hb-ot-kern-table.hh:30:
In file included from ./hb-aat-layout-kerx-table.hh:31:
In file included from ./hb-kern.hh:32:
In file included from ./hb-ot-layout-gpos-table.hh:32:
./hb-ot-layout-gsubgpos.hh:1878:63: error: loop variable '_' binds to a temporary value produced by a range of type 'decltype((hb_forward<hb_filter_iter_factory_t<hb_map_t &, const (anonymous struct at ./hb-algs.hh:331:1) &>>(rhs)(hb_forward<hb_zip_iter_t<hb_iota_iter_t<unsigned int, unsigned int>, hb_array_t<const OT::OffsetTo<OT::RuleSet, OT::IntType<unsigned short, 2>, true>>>>(lhs))))' (aka 'hb_filter_iter_t<hb_zip_iter_t<hb_iota_iter_t<unsigned int, unsigned int>, hb_array_t<const OT::OffsetTo<OT::RuleSet, OT::IntType<unsigned short, 2>, true>>>, hb_map_t &, const (anonymous struct at ./hb-algs.hh:331:1) &>') [-Werror,-Wrange-loop-bind-reference]
for (const hb_pair_t<unsigned, const OffsetTo<RuleSet>&>& _ : + hb_enumerate (ruleSet)
^
./hb-ot-layout-gsubgpos.hh:1878:10: note: use non-reference type 'hb_pair_t<unsigned int, const OffsetTo<OT::RuleSet> &>'
for (const hb_pair_t<unsigned, const OffsetTo<RuleSet>&>& _ : + hb_enumerate (ruleSet)
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
AAT::Lookup has no other way to detect whether it is returned from
a real and sanitized font data or from a null pool, this checks if
the table has been recognized valid by sanitizer by checking
table's major version which is zero if returned from a null pool and
non-zero if is from a sanitized font data, it is expected the other
calls of the table (unlikely to have more calls however) also do a
similar version check before calling the lookups used on the table.
In case a user passed -Dintrospection=enabled the build would just ignore
it by default because gobject defaults to disabled and the introspection build
gets skipped.
Instead, if introspection is explicitly enabled but gobject is for some reason
missing error out.
Fixes#2404
* [aat] Fix implementation of AAT kerning for Geeza Pro.
Despite what the comment in the code used to say, it appears that Geeza Pro
does rely on accumulating kerning values from successive subtables. With
this change, the results now match Core Text rendering (and avoid the clear
visual breakage reported in #2358).
Testcase: U+0644,U+064E,U+0645,U+064E,U+0651,U+0627
Fixes#2358.
* [aat] Update test expectations, add new testcase.
If the size of opened file is zero, try opening resource fork by
appending "/..namedfork/rsrc" to the file name. This is guarded with
__APPLE__ ifdef and uses _PATH_RSRCFORKSPEC macro from sys/paths.h.
Defining HB_NO_RESOURCE_FORK will disable this fallback.
Fixes https://github.com/harfbuzz/harfbuzz/issues/2383
This adds a seperate library like with autotools.
This also fixes the ico feature option which was just set to required:false
when disabled instead of really disabling it.
Disabling is still broken with msvc because it then tries to find the library
another way, but that's broken for all other deps as well so I left it as is.
For tests only test-unicode.c is using icu specific functions so split it out
into its own category which depends on harfbuzz-icu.
Fixes#2338
Pass the same config to gobject-introspection as with cmake/autotools.
This makes sure the c-include and package name is included in the gir
and also fixes the build because of the missing HB_AAT_H* defines.
Fixes#2336
Providing lengthier explanation for how the destroy function works on the hb_font_t but that destroying the FT_Face (and not destroying it too early) remains the client's responsibility.
Fixes https://crbug.com/oss-fuzz/21560
revert () does not clean up useless object_t. Adjust the order of
subsetting substitutions and conditions to avoid dangling object_t.
This may be implemented by selectors under either LETTER_CASE or LOWER_CASE feature types
in AAT, so we need to check for the presence of either one.
Fixes#2307.
If an AAT feature type is not exposed in the 'feat' table, we assume it is not intended
to be user-controllable and so we should not map any OT feature tag requests to it.
Fixes#2285.
* Search src/ build directory for objects in check-static-inits.sh
* Find .def files in src/ build directory in src/check-symbols.sh
* Pass builddir also in autotools also, we may just remove libs passing after autotools removal
* Move harfbuzz_subset_def target so can be referenced as a check-static-inits.sh dependency
hb_shape_plan_key_t::equal expects hb_ot_shape_plan_key_t be initialized by
hb_ot_layout_table_find_feature_variations calls but it won't get initialized
when HB_NO_VAR build config is used.
Related to https://github.com/harfbuzz/harfbuzz/issues/2280
MSVC doens't like its NullPool,
test-bimap.cc.obj : error LNK2019: unresolved external symbol "unsigned __int64 const * const _hb_NullPool" (?_hb_NullPool@@3QB_KB) referenced in function
test-unicode.c:960: undefined reference to `hb_icu_get_unicode_funcs'
test-unicode.c:961: undefined reference to `hb_icu_get_unicode_funcs'
For now add the icu sources to libharfbuzz also for the amalgam
build, later we need to have a separate harfbuzz-icu module and
link against that, and/or generate harfbuzz.cc.
...and supersede the configuration option uniscribe with gdi, as Uniscribe is
tightly tied to GDI. This means that enabling GDI would also mean enabling
Uniscribe.
We don't actually need the .def files (vs_module_defs) entry when we are
building DLLs with Visual Studio as long as we have HB_DLL_EXPORT defined.
Plus, to maintain compatibility with the CMake builds, for Visual Studio builds
we do not prefix the libraries with 'lib', nor have a '-0' suffix for the DLL
file name.
This was, we also avoid Visual Studio compiler warnings C4828 as that sign is
not favored when /utf-8 is enabled, which is the norm nowadays as Visual Studio
2015 or later is required to build harfbuzz nowadays.
HarfBuzz headers are under src/ not the root directory, without using
incsrc no headers will be found by the dependent project. I think
incbase is superfluous, it should be replaced by incsrc or dropped.
Fix symbol export with MSVC when features are disabled,
such as GLib. We need to generate the list of exported
symbols on the fly to make sure we only export symbols
that are actually available.
Needs some minor modifications to the gen-def.py script:
- accept header list also via command line args; we can't
pass things to a configure_file() command via the environment
in Meson.
- strip any leading 'src/' from library filename. This might
be there because in Meson the script might be called from
the top-level directory and not the current source directory.
Remove .def files again which had been checked in for earlier
versions of the Meson port.