bump: tree-sitter: v0.26.3

- Updated to tree-sitter v0.26.3
- **Breaking Changes in tree-sitter 0.26.x API**:
  - `ts_language_version` renamed to `ts_language_abi_version`
  - `TSInputEncodingUTF16` split into `TSInputEncodingUTF16LE` and
`TSInputEncodingUTF16BE`
    (now using UTF16LE as default for backward compatibility)
  - Cancellation flag API (`ts_parser_cancellation_flag`,
`ts_parser_set_cancellation_flag`) removed
    - `Parser#cancellation_flag` and `Parser#cancellation_flag=` are now
no-ops for backward compatibility
  - Timeout API (`ts_parser_timeout_micros`,
`ts_parser_set_timeout_micros`) removed
    - `Parser#timeout_micros` and `Parser#timeout_micros=` are now
no-ops for backward compatibility
  - Use `TSParseOptions` with `progress_callback` for
cancellation/timeout functionality in 0.26+
- `TREE_SITTER_LANGUAGE_VERSION` is now 15 (was 14)
- `TREE_SITTER_MIN_COMPATIBLE_LANGUAGE_VERSION` is now 13 (was 6)
- Grammar files (.so) must be built against tree-sitter 0.26+ to work
with this version

Closes #94

Author: pboling

.gitignore

@@ -3,6 +3,7 @@
 *.gemfile
 *.lock
 *.log
+*.o
 *.so
 
 logs
@@ -15,3 +16,9 @@ logs
 /tree-sitter
 /tree-sitter-parsers
 /vendor
+
+# Build artifacts in ext/tree_sitter/
+/ext/tree_sitter/Makefile
+/ext/tree_sitter/extconf.h
+/ext/tree_sitter/tree-sitter-*/
+

.justfile

@@ -94,6 +94,7 @@ setup-bundler:
 setup-ts:
   #!/usr/bin/env bash
   set -euo pipefail
+  version=$(sed -nE 's/^tree-sitter-version[[:space:]]*=[[:space:]]*"([^"]+)".*/\1/p' parsers.toml)
   if [ -d tree-sitter ]; then
     cd tree-sitter
     git reset --hard HEAD
@@ -103,7 +104,7 @@ setup-ts:
     git init
     git remote add origin https://github.com/tree-sitter/tree-sitter
   fi
-  git fetch origin --depth 1 v0.22.6
+  git fetch origin --depth 1 "v${version}"
   git reset --hard FETCH_HEAD
   make
   echo "now you can:"

Gemfile

@@ -5,13 +5,13 @@ source 'https://rubygems.org'
 gemspec
 
 group :development, :test do
-  gem 'bundler', '~> 2.3'
   gem 'debug', '~> 1.9', require: false
   gem 'gemstash', require: false
+  gem 'ostruct'
   gem 'rake', '~> 13.2'
   gem 'rake-compiler', '~> 1.2'
   gem 'rake-compiler-dock', '~> 1.5'
-  gem 'rubocop', '~> 1.39.0', require: false
+  gem 'rubocop'
   gem 'ruby-lsp', '~> 0.17', require: false
   gem 'ruby_memcheck', '~> 1.3'
   gem 'tapioca', '~> 0.11', require: false

News.md

@@ -2,6 +2,22 @@
 
 # [unreleaseed]
 
+## API Changes for tree-sitter 0.26.3 compatibility
+
+- Updated to tree-sitter v0.26.3
+- **Breaking Changes in tree-sitter 0.26.x API**:
+  - `ts_language_version` renamed to `ts_language_abi_version`
+  - `TSInputEncodingUTF16` split into `TSInputEncodingUTF16LE` and `TSInputEncodingUTF16BE`
+    (now using UTF16LE as default for backward compatibility)
+  - Cancellation flag API (`ts_parser_cancellation_flag`, `ts_parser_set_cancellation_flag`) removed
+    - `Parser#cancellation_flag` and `Parser#cancellation_flag=` are now no-ops for backward compatibility
+  - Timeout API (`ts_parser_timeout_micros`, `ts_parser_set_timeout_micros`) removed
+    - `Parser#timeout_micros` and `Parser#timeout_micros=` are now no-ops for backward compatibility
+  - Use `TSParseOptions` with `progress_callback` for cancellation/timeout functionality in 0.26+
+- `TREE_SITTER_LANGUAGE_VERSION` is now 15 (was 14)
+- `TREE_SITTER_MIN_COMPATIBLE_LANGUAGE_VERSION` is now 13 (was 6)
+- Grammar files (.so) must be built against tree-sitter 0.26+ to work with this version
+
 # v2.0.0 (22-01-2025)
 
 - Bump minimum supported Ruby version to 3.1.

ext/tree_sitter/encoding.c

@@ -14,7 +14,8 @@ TSInputEncoding value_to_encoding(VALUE encoding) {
 
   // NOTE: should we emit a warning instead of defaulting to UTF8?
   if (enc == u16) {
-    return TSInputEncodingUTF16;
+    // tree-sitter 0.26+ split UTF16 into UTF16LE and UTF16BE
+    return TSInputEncodingUTF16LE;
   } else {
     return TSInputEncodingUTF8;
   }

ext/tree_sitter/language.c

@@ -3,7 +3,8 @@
 #include <stdint.h>
 #include <stdio.h>
 
-typedef TSLanguage *(tree_sitter_lang)(void);
+// tree-sitter 0.26+ returns const TSLanguage*
+typedef const TSLanguage *(tree_sitter_lang)(void);
 const char *tree_sitter_prefix = "tree_sitter_";
 
 extern VALUE mTreeSitter;
@@ -40,23 +41,29 @@ static VALUE language_load(VALUE self, VALUE name, VALUE path) {
   VALUE path_s = rb_funcall(path, rb_intern("to_s"), 0);
   char *path_cstr = StringValueCStr(path_s);
   void *lib = dlopen(path_cstr, RTLD_NOW);
-  const char *err = dlerror();
-  if (err != NULL) {
+  if (lib == NULL) {
+    const char *err = dlerror();
     VALUE parser_not_found = rb_const_get(mTreeSitter, rb_intern("ParserNotFoundError"));
     rb_raise(parser_not_found,
-             "Could not load shared library `%s'.\nReason: %s", path_cstr, err);
+             "Could not load shared library `%s'.\nReason: %s", path_cstr, err ? err : "unknown error");
   }
 
   char buf[256];
   snprintf(buf, sizeof(buf), "tree_sitter_%s", StringValueCStr(name));
+
+  // Clear any previous error before dlsym (POSIX requirement)
+  dlerror();
+
   tree_sitter_lang *make_ts_language = dlsym(lib, buf);
-  err = dlerror();
-  if (err != NULL) {
+
+  // Only check dlerror if dlsym returned NULL
+  if (make_ts_language == NULL) {
+    const char *err = dlerror();
     dlclose(lib);
     VALUE symbol_not_found = rb_const_get(mTreeSitter, rb_intern("SymbolNotFoundError"));
     rb_raise(symbol_not_found,
-             "Could not load symbol `%s' from library `%s'.\nReason:%s",
-             StringValueCStr(name), path_cstr, err);
+             "Could not load symbol `%s' from library `%s'.\nReason: %s",
+             buf, path_cstr, err ? err : "symbol not found");
   }
 
   TSLanguage *lang = make_ts_language();
@@ -69,7 +76,8 @@ static VALUE language_load(VALUE self, VALUE name, VALUE path) {
              StringValueCStr(name), path_cstr);
   }
 
-  uint32_t version = ts_language_version(lang);
+  // tree-sitter 0.26+ renamed ts_language_version to ts_language_abi_version
+  uint32_t version = ts_language_abi_version(lang);
   if (version < TREE_SITTER_MIN_COMPATIBLE_LANGUAGE_VERSION) {
     VALUE version_error = rb_const_get(mTreeSitter, rb_intern("ParserVersionError"));
     rb_raise(version_error,
@@ -194,7 +202,8 @@ static VALUE language_symbol_type(VALUE self, VALUE symbol) {
  * @see Parser#language=
  */
 static VALUE language_version(VALUE self) {
-  return UINT2NUM(ts_language_version(SELF));
+  // tree-sitter 0.26+ renamed ts_language_version to ts_language_abi_version
+  return UINT2NUM(ts_language_abi_version(SELF));
 }
 
 void init_language(void) {
@@ -220,4 +229,5 @@ void init_language(void) {
   rb_define_method(cLanguage, "symbol_name", language_symbol_name, 1);
   rb_define_method(cLanguage, "symbol_type", language_symbol_type, 1);
   rb_define_method(cLanguage, "version", language_version, 0);
+  rb_define_method(cLanguage, "abi_version", language_version, 0);
 }

ext/tree_sitter/parser.c

@@ -44,9 +44,14 @@ static VALUE parser_allocate(VALUE klass) {
  * Get the parser's current cancellation flag pointer.
  *
  * @return [Integer]
+ *
+ * @note DEPRECATED in tree-sitter 0.26+. This API was removed.
+ *       Use TSParseOptions with progress_callback instead.
  */
 static VALUE parser_get_cancellation_flag(VALUE self) {
-  return SIZET2NUM(*ts_parser_cancellation_flag(SELF));
+  // tree-sitter 0.26+ removed cancellation_flag API
+  // Return the stored value for backward compatibility
+  return SIZET2NUM(unwrap(self)->cancellation_flag);
 }
 
 /**
@@ -58,13 +63,15 @@ static VALUE parser_get_cancellation_flag(VALUE self) {
  *
  * @see parse
  *
- * @note This is not well-tested in the bindings.
+ * @note DEPRECATED in tree-sitter 0.26+. This API was removed.
+ *       Use TSParseOptions with progress_callback instead.
  *
  * @return nil
  */
 static VALUE parser_set_cancellation_flag(VALUE self, VALUE flag) {
+  // tree-sitter 0.26+ removed cancellation_flag API
+  // Store the value for backward compatibility but it won't affect parsing
   unwrap(self)->cancellation_flag = NUM2SIZET(flag);
-  ts_parser_set_cancellation_flag(SELF, &unwrap(self)->cancellation_flag);
   return Qnil;
 }
 
@@ -349,9 +356,15 @@ static VALUE parser_reset(VALUE self) {
  * Get the duration in microseconds that parsing is allowed to take.
  *
  * @return [Integer]
+ *
+ * @note DEPRECATED in tree-sitter 0.26+. This API was removed.
+ *       Use TSParseOptions with progress_callback instead.
  */
 static VALUE parser_get_timeout_micros(VALUE self) {
-  return ULL2NUM(ts_parser_timeout_micros(SELF));
+  // tree-sitter 0.26+ removed timeout_micros API
+  // Return 0 to indicate no timeout (was the default behavior)
+  (void)self;  // suppress unused parameter warning
+  return ULL2NUM(0);
 }
 
 /**
@@ -362,12 +375,18 @@ static VALUE parser_get_timeout_micros(VALUE self) {
  *
  * @see parse
  *
+ * @note DEPRECATED in tree-sitter 0.26+. This API was removed.
+ *       Use TSParseOptions with progress_callback instead.
+ *
  * @param timeout [Integer]
  *
  * @return [nil]
  */
 static VALUE parser_set_timeout_micros(VALUE self, VALUE timeout) {
-  ts_parser_set_timeout_micros(SELF, NUM2ULL(timeout));
+  // tree-sitter 0.26+ removed timeout_micros API
+  // This is a no-op for backward compatibility
+  (void)self;     // suppress unused parameter warning
+  (void)timeout;  // suppress unused parameter warning
   return Qnil;
 }
 

lib/tree_sitter/version.rb

@@ -2,7 +2,7 @@
 
 module TreeSitter
   # The version of the tree-sitter library.
-  TREESITTER_VERSION = '0.24.7'
+  TREESITTER_VERSION = '0.26.3'
   # The current version of the gem.
   VERSION = '2.0.0'
 end

parsers.toml

@@ -1,6 +1,6 @@
 build-dir = "vendor/parsers/build"
 out-dir = "vendor/parsers"
-tree-sitter-version = "0.24.7"
+tree-sitter-version = "0.26.3"
 
 [parsers]
 embedded-template = "0.23.2"