👽️ Support tree-sitter 0.26.x C library API

.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

@@ -1,8 +1,3 @@
-GEM_NAME := 'tree_sitter'
-LIB := 'lib'
-LIB_FILE := LIB / GEM_NAME + '.rb'
-VERSION_FILE := LIB / GEM_NAME / 'version.rb'
-VERSION := shell("ruby -r ./" + VERSION_FILE  + " -e 'puts TreeSitter::VERSION'")
 TS_PARSER_VERSION := '4.8'
 
 default: check
@@ -22,7 +17,9 @@ compile *args:
 
 [group('tree-sitter')]
 dl-parsers platform:
-  curl -o tree-sitter-parsers.zip -L https://github.com/Faveod/tree-sitter-parsers/releases/download/v{{TS_PARSER_VERSION}}/tree-sitter-parsers-{{TS_PARSER_VERSION}}-{{platform}}.zip
+  curl \
+    -o tree-sitter-parsers.zip \
+    -L https://github.com/Faveod/tree-sitter-parsers/releases/download/v{{TS_PARSER_VERSION}}/tree-sitter-parsers-{{TS_PARSER_VERSION}}-{{platform}}.zip
   unzip tree-sitter-parsers.zip
 
 [group('doc')]

.rubocop.yml

@@ -76,6 +76,11 @@ Style/BlockDelimiters:
 Style/DoubleNegation:
   Enabled: false
 
+Style/ArgumentsForwarding:
+  Exclude:
+    - lib/tree_stand.rb
+    - lib/tree_stand/*
+
 Style/FormatString:
   Enabled: false
 

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

@@ -1,6 +1,22 @@
 # News
 
-# [unreleaseed]
+# [unreleased]
+
+## 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)
 

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/extconf.rb

@@ -121,25 +121,27 @@ def env_var_on?(var)
     $warnflags.gsub!(/#{r}/, '')
   end
 
-  cflags.concat %W[
-    -fms-extensions
-    -fdeclspec
-    -fsanitize=#{sanitizers}
-    -fsanitize-blacklist=../../../../.asanignore
-    -fsanitize-recover=#{sanitizers}
-    -fno-sanitize-recover=all
-    -fno-sanitize=null
-    -fno-sanitize=alignment
-    -fno-omit-frame-pointer
-  ]
-
-  ldflags.concat %W[
-    -fsanitize=#{sanitizers}
-    -dynamic-libasan
-  ]
+  cflags.push(*
+    %W[
+      -fms-extensions
+      -fdeclspec
+      -fsanitize=#{sanitizers}
+      -fsanitize-blacklist=../../../../.asanignore
+      -fsanitize-recover=#{sanitizers}
+      -fno-sanitize-recover=all
+      -fno-sanitize=null
+      -fno-sanitize=alignment
+      -fno-omit-frame-pointer
+    ])
+
+  ldflags.push(*
+    %W[
+      -fsanitize=#{sanitizers}
+      -dynamic-libasan
+    ])
 end
 
-cflags.concat %w[-std=c99 -fPIC -Wall]
+cflags.push(*%w[-std=c99 -fPIC -Wall])
 
 append_cflags(cflags)
 append_ldflags(ldflags)

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;
@@ -40,26 +41,32 @@ 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();
+  const TSLanguage *lang = make_ts_language();
   if (lang == NULL) {
     dlclose(lib);
     VALUE language_load_error = rb_const_get(mTreeSitter, rb_intern("LanguageLoadError"));
@@ -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;
 }
 
@@ -345,32 +352,6 @@ static VALUE parser_reset(VALUE self) {
   return Qnil;
 }
 
-/**
- * Get the duration in microseconds that parsing is allowed to take.
- *
- * @return [Integer]
- */
-static VALUE parser_get_timeout_micros(VALUE self) {
-  return ULL2NUM(ts_parser_timeout_micros(SELF));
-}
-
-/**
- * Set the maximum duration in microseconds that parsing should be allowed to
- * take before halting.
- *
- * If parsing takes longer than this, it will halt early, returning +nil+.
- *
- * @see parse
- *
- * @param timeout [Integer]
- *
- * @return [nil]
- */
-static VALUE parser_set_timeout_micros(VALUE self, VALUE timeout) {
-  ts_parser_set_timeout_micros(SELF, NUM2ULL(timeout));
-  return Qnil;
-}
-
 void init_parser(void) {
   cParser = rb_define_class_under(mTreeSitter, "Parser", rb_cObject);
 
@@ -393,6 +374,4 @@ void init_parser(void) {
                    parser_parse_string_encoding, 3);
   rb_define_method(cParser, "print_dot_graphs", parser_print_dot_graphs, 1);
   rb_define_method(cParser, "reset", parser_reset, 0);
-  rb_define_method(cParser, "timeout_micros", parser_get_timeout_micros, 0);
-  rb_define_method(cParser, "timeout_micros=", parser_set_timeout_micros, 1);
 }

ext/tree_sitter/repo.rb

@@ -39,7 +39,7 @@ def exe?(name)
     end
 
     def extract?
-      !exe.filter { |k, v| %i[tar zip].include?(k) && v }.empty?
+      exe.any? { |k, v| %i[tar zip].include?(k) && v }
     end
 
     def download
@@ -50,7 +50,7 @@ def download
       %w[git curl wget].each do |cmd|
         res =
           if find_executable(cmd)
-            send("sources_from_#{cmd}")
+            send("sources_from_#{cmd}?")
           else
             false
           end
@@ -76,7 +76,7 @@ def sh(cmd)
       MSG
     end
 
-    def sources_from_curl
+    def sources_from_curl?
       return false if !exe?(:curl) || !extract?
 
       if exe?(:tar)
@@ -90,7 +90,7 @@ def sources_from_curl
       true
     end
 
-    def sources_from_git
+    def sources_from_git?
       return false if !exe?(:git)
 
       sh <<~SHELL.chomp
@@ -104,7 +104,7 @@ def sources_from_git
       true
     end
 
-    def sources_from_wget
+    def sources_from_wget?
       return false if !exe?(:wget) || !extract?
 
       if exe?(:tar)

lib/tree_sitter/query.rb

@@ -45,11 +45,7 @@ def process(source) # rubocop:disable Metrics/AbcSize,Metrics/CyclomaticComplexi
       pattern_count.times do |i| # rubocop:disable Metrics/BlockLength
         predicate_steps = predicates_for_pattern(i)
         byte_offset = start_byte_for_pattern(i)
-        row =
-          source.chars.map.with_index
-            .take_while { |_, i| i < byte_offset } # rubocop:disable Lint/ShadowingOuterLocalVariable
-            .filter { |c, _| c == "\n" }
-            .size
+        row = source[0...byte_offset].count("\n")
         text_predicates = []
         property_predicates = []
         property_settings = []

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"

test/tree_sitter/language_test.rb

@@ -75,7 +75,6 @@ def mul(a, b)
   end
 
   it 'must be of correct version' do
-    assert ruby.version <= TreeSitter::LANGUAGE_VERSION \
-           && ruby.version >= TreeSitter::MIN_COMPATIBLE_LANGUAGE_VERSION
+    assert ruby.version.between?(TreeSitter::MIN_COMPATIBLE_LANGUAGE_VERSION, TreeSitter::LANGUAGE_VERSION)
   end
 end

test/tree_sitter/parser_test.rb

@@ -124,13 +124,6 @@ def mul(a, b)
   end
 end
 
-describe 'timeout_micros' do
-  it 'must get/set timeout_micros' do
-    parser.timeout_micros = 1
-    assert_equal 1, parser.timeout_micros
-  end
-end
-
 # TODO: included_ranges for parsing partial documents.
 
 # TODO: parsing with non-nil tree.