Use tree-sitter-html from crates.io

CHANGELOG.md

@@ -11,7 +11,7 @@ with YAML.
 Improved language detection when one argument is a named pipe.
 
 Updated to the latest tree-sitter parser for Bash, C, C++, C#, Go,
-Haskell, Java, JavaScript, JSON, Julia, Objective-C, OCaml, PHP,
+Haskell, HTML, Java, JavaScript, JSON, Julia, Objective-C, OCaml, PHP,
 Python, Ruby, Scala and TypeScript.
 
 ### Syntax Highlighting

Cargo.toml

@@ -95,6 +95,7 @@ tree-sitter-go = "0.23.4"
 tree-sitter-bash = "0.23.3"
 tree-sitter-php = "0.23.11"
 tree-sitter-json = "0.24.8"
+tree-sitter-html = "0.23.2"
 
 [dev-dependencies]
 # assert_cmd 2.0.10 requires predicates 3.

build.rs

@@ -41,8 +41,6 @@ impl TreeSitterParser {
                 .flag_if_supported("-Wno-implicit-fallthrough")
                 .flag_if_supported("-Wno-unused-parameter")
                 .flag_if_supported("-Wno-ignored-qualifiers")
-                // Ignore warning from tree-sitter-html.
-                .flag_if_supported("-Wno-sign-compare")
                 .link_lib_modifier("+whole-archive");
 
             for file in cpp_files {
@@ -154,11 +152,6 @@ fn main() {
             src_dir: "vendored_parsers/tree-sitter-hcl-src",
             extra_files: vec!["scanner.cc"],
         },
-        TreeSitterParser {
-            name: "tree-sitter-html",
-            src_dir: "vendored_parsers/tree-sitter-html-src",
-            extra_files: vec!["scanner.cc"],
-        },
         TreeSitterParser {
             name: "tree-sitter-janet-simple",
             src_dir: "vendored_parsers/tree-sitter-janet-simple-src",

src/parse/tree_sitter_parser.rs

@@ -79,7 +79,6 @@ extern "C" {
     fn tree_sitter_hare() -> ts::Language;
     fn tree_sitter_hack() -> ts::Language;
     fn tree_sitter_hcl() -> ts::Language;
-    fn tree_sitter_html() -> ts::Language;
     fn tree_sitter_janet_simple() -> ts::Language;
     fn tree_sitter_kotlin() -> ts::Language;
     fn tree_sitter_latex() -> ts::Language;
@@ -507,7 +506,9 @@ pub(crate) fn from_language(language: guess::Language) -> TreeSitterConfig {
             }
         }
         Html => {
-            let language = unsafe { tree_sitter_html() };
+            let language_fn = tree_sitter_html::LANGUAGE;
+            let language = tree_sitter::Language::new(language_fn);
+
             TreeSitterConfig {
                 language: language.clone(),
                 atom_nodes: vec![
@@ -522,11 +523,8 @@ pub(crate) fn from_language(language: guess::Language) -> TreeSitterConfig {
                 delimiter_tokens: vec![("<", ">"), ("<!", ">"), ("<!--", "-->")]
                     .into_iter()
                     .collect(),
-                highlight_query: ts::Query::new(
-                    &language,
-                    include_str!("../../vendored_parsers/highlights/html.scm"),
-                )
-                .unwrap(),
+                highlight_query: ts::Query::new(&language, tree_sitter_html::HIGHLIGHTS_QUERY)
+                    .unwrap(),
                 sub_languages: vec![
                     TreeSitterSubLanguage {
                         query: ts::Query::new(&language, "(style_element (raw_text) @contents)")