Use tree-sitter-xml from crates.io

Wilfred · Dec 20, 2024 · b3606fc · b3606fc
1 parent cef0b56
commit b3606fc
Show file tree

Hide file tree

Showing 50 changed files with 19 additions and 23,125 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -12,7 +12,7 @@ Improved language detection when one argument is a named pipe.
 
 Updated to the latest tree-sitter parser for Bash, C, C++, C#, CSS,
 Go, Haskell, HTML, Java, JavaScript, JSON, Julia, Lua, Objective-C,
-OCaml, PHP, Python, Ruby, Scala, TOML and TypeScript.
+OCaml, PHP, Python, Ruby, Scala, TOML, TypeScript and XML.
 
 ### Syntax Highlighting
 

diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -99,6 +99,7 @@ tree-sitter-html = "0.23.2"
 tree-sitter-css = "0.23.1"
 tree-sitter-toml-ng = "0.7.0"
 tree-sitter-lua = "0.2.0"
+tree-sitter-xml = "0.7.0"
 
 [dev-dependencies]
 # assert_cmd 2.0.10 requires predicates 3.

diff --git a/build.rs b/build.rs
@@ -247,11 +247,6 @@ fn main() {
             src_dir: "vendored_parsers/tree-sitter-vhdl-src",
             extra_files: vec![],
         },
-        TreeSitterParser {
-            name: "tree-sitter-xml",
-            src_dir: "vendored_parsers/tree-sitter-xml-src/tree-sitter-xml/src",
-            extra_files: vec!["scanner.c"],
-        },
         TreeSitterParser {
             name: "tree-sitter-yaml",
             src_dir: "vendored_parsers/tree-sitter-yaml-src",

diff --git a/manual/src/languages_supported.md b/manual/src/languages_supported.md
@@ -70,6 +70,6 @@ with `difft --list-languages`.
 | Newick   | [delehef/tree-sitter-newick](https://github.com/delehef/tree-sitter-newick)                       |
 | SCSS     | [serenadeai/tree-sitter-scss](https://github.com/serenadeai/tree-sitter-scss)                     |
 | TOML     | [tree-sitter-grammars/tree-sitter-toml](https://github.com/tree-sitter-grammars/tree-sitter-toml) |
-| XML      | [ObserverOfTime/tree-sitter-xml](https://github.com/ObserverOfTime/tree-sitter-xml)               |
+| XML      | [tree-sitter-grammars/tree-sitter-xml](https://github.com/tree-sitter-grammars/tree-sitter-xml)   |
 | YAML     | [ikatyang/tree-sitter-yaml](https://github.com/ikatyang/tree-sitter-yaml)                         |
 
diff --git a/src/parse/tree_sitter_parser.rs b/src/parse/tree_sitter_parser.rs
@@ -97,7 +97,6 @@ extern "C" {
     fn tree_sitter_sql() -> ts::Language;
     fn tree_sitter_swift() -> ts::Language;
     fn tree_sitter_vhdl() -> ts::Language;
-    fn tree_sitter_xml() -> ts::Language;
     fn tree_sitter_yaml() -> ts::Language;
     fn tree_sitter_zig() -> ts::Language;
 }
@@ -1080,19 +1079,18 @@ pub(crate) fn from_language(language: guess::Language) -> TreeSitterConfig {
             }
         }
         Xml => {
-            let language = unsafe { tree_sitter_xml() };
+            let language_fn = tree_sitter_xml::LANGUAGE_XML;
+            let language = tree_sitter::Language::new(language_fn);
+
             TreeSitterConfig {
                 language: language.clone(),
                 // XMLDecl is the <?xml ...?> header, but the parser
                 // just treats it as a sequence of tokens rather than
                 // e.g. string subexpressions, so flatten.
                 atom_nodes: vec!["AttValue", "XMLDecl"].into_iter().collect(),
                 delimiter_tokens: (vec![("<", ">")]),
-                highlight_query: ts::Query::new(
-                    &language,
-                    include_str!("../../vendored_parsers/highlights/xml.scm"),
-                )
-                .unwrap(),
+                highlight_query: ts::Query::new(&language, tree_sitter_xml::XML_HIGHLIGHT_QUERY)
+                    .unwrap(),
                 sub_languages: vec![],
             }
         }

diff --git a/vendored_parsers/highlights/xml.scm b/vendored_parsers/highlights/xml.scm
diff --git a/vendored_parsers/tree-sitter-xml-src/common b/vendored_parsers/tree-sitter-xml-src/common
diff --git a/vendored_parsers/tree-sitter-xml-src/tree-sitter-xml b/vendored_parsers/tree-sitter-xml-src/tree-sitter-xml
diff --git a/vendored_parsers/tree-sitter-xml/.editorconfig b/vendored_parsers/tree-sitter-xml/.editorconfig
diff --git a/vendored_parsers/tree-sitter-xml/.gitattributes b/vendored_parsers/tree-sitter-xml/.gitattributes
diff --git a/vendored_parsers/tree-sitter-xml/.github/FUNDING.yml b/vendored_parsers/tree-sitter-xml/.github/FUNDING.yml
diff --git a/vendored_parsers/tree-sitter-xml/.github/workflows/ci.yml b/vendored_parsers/tree-sitter-xml/.github/workflows/ci.yml
diff --git a/vendored_parsers/tree-sitter-xml/.gitignore b/vendored_parsers/tree-sitter-xml/.gitignore
diff --git a/vendored_parsers/tree-sitter-xml/Cargo.toml b/vendored_parsers/tree-sitter-xml/Cargo.toml
diff --git a/vendored_parsers/tree-sitter-xml/LICENSE b/vendored_parsers/tree-sitter-xml/LICENSE
diff --git a/vendored_parsers/tree-sitter-xml/README.md b/vendored_parsers/tree-sitter-xml/README.md
diff --git a/vendored_parsers/tree-sitter-xml/binding.gyp b/vendored_parsers/tree-sitter-xml/binding.gyp
diff --git a/vendored_parsers/tree-sitter-xml/bindings/node/binding.cc b/vendored_parsers/tree-sitter-xml/bindings/node/binding.cc
diff --git a/vendored_parsers/tree-sitter-xml/bindings/node/dtd.js b/vendored_parsers/tree-sitter-xml/bindings/node/dtd.js
diff --git a/vendored_parsers/tree-sitter-xml/bindings/node/index.js b/vendored_parsers/tree-sitter-xml/bindings/node/index.js
diff --git a/vendored_parsers/tree-sitter-xml/bindings/node/xml.js b/vendored_parsers/tree-sitter-xml/bindings/node/xml.js
diff --git a/vendored_parsers/tree-sitter-xml/bindings/rust/build.rs b/vendored_parsers/tree-sitter-xml/bindings/rust/build.rs
diff --git a/vendored_parsers/tree-sitter-xml/bindings/rust/lib.rs b/vendored_parsers/tree-sitter-xml/bindings/rust/lib.rs