Predicates/directives are documented to end in either `!` or `?`.
However, `query.c` allows them to be any valid identifier, and also
allows `?` or `!` characters anywhere inside an identifier.
This commit removes `?` and `!` from the list of valid identifier
characters, and asserts that predicates/directives only *end* in `?` or
`!`, respectively.
This commit is breaking because you can no longer do something like
`(#eq? @capture foo!bar)` (`foo!bar` must now be quoted).
* Fix pkgconfig
Init CMAKE_INSTALL_INCLUDEDIR before pc file generation.
Install pc file to CMAKE_INSTALL_LIBDIR/pkgconfig -
it accompanies the architecture-dependent library.
* Include GNUInstallDirs early
The CMake module initializes variables which are used for
exported information (CMake and pkgconfig).
* Change pc file install destination
This reverts commit dc4e232e6e.
Reason: The sha in the generated output (which most distro builds of
tree-sitter, including `cargo install`, strip) produces too many
conflicts when verifying via CI that parsers are regenerated on every
grammar change.
It is very common practice to ignore
these lock files for libraries, since they do not apply to applications
that use the libraries. The lock files are especially not useful in
tree-sitter grammar repos, since tree-sitter grammars should not have
dependencies. The lock files are just a source of merge conflicts and
spurious CI failures.
This makes `parser.c` follow the https://generated.at/ convention for generated files. This potentially allows any compatible IDE to discourage editing it directly.
* Rename corpus test functions to allow easy filtering by language
* Use usize for seed argument
* Avoid retaining useless stack versions when reductions merge
We found this problem when debugging an infinite loop that happened
during error recovery when using the Zig grammar. The large number of
unnecessary paused stack versions were preventing the correct recovery
strategy from being tried.
* Fix leaked lookahead token when reduction results in a merged stack
* Enable running PHP tests in CI
* Fix possible infinite loop during error recovery at EOF
* Account for external scanner state changes when detecting changed ranges in subtrees
Problem: Linking different parsers into one executable fails due to duplicate symbols.
Solution: Mark `TSCharacterRange` as `static` when generating parsers.
fixes#4209
Building tree-sitter on Windows within a Cygwin or MSYS2 environment
for MinGW-w64 targets should work. As there's no configure-like step,
the tools have to be specified with:
make CC=x86_64-w64-mingw32-gcc \
AR=x86_64-w64-mingw32-ar \
STRIP=x86_64-w64-mingw32-strip
Useful reference:
[How to build and use DLLs on Windows](https://nullprogram.com/blog/2021/05/31/)
This commit doesn't tag public functions with dllexport or
dllimport. This results in exporting non-static functions visible
between translation units, and generating slightly less efficient code
for calling exported functions from the DLL.
This commit doesn't include support to build libtree-sitter with MSVC or
clang-cl, but generates a libtree-sitter.lib file for MSVC/clang-cl
consumers.
Instead of having users declare the extern function themselves, they can
pass in the language to `Language.create` in the zig bindings. If they
really want, they can always opt into the `extern fn tree_sitter_LANG()
*const ts.Language` approach.
**Problem:** When resetting the parser during subtree balancing, an
error is thrown:
```
parser.c:2198: ts_parser_parse: Assertion `self->finished_tree.ptr' failed.
```
**Solution:** Reset `canceled_balancing` to false in
`ts_parser_reset()`.
