Add XML support

Fixes #10

CHANGELOG.md

@@ -1,5 +1,9 @@
 ## 0.52 (unreleased)
 
+### Parsing
+
+Added support for XML.
+
 ### Display
 
 Improved syntax highlighting of constructors (i.e. type names when

build.rs

@@ -335,6 +335,11 @@ fn main() {
             src_dir: "vendored_parsers/tree-sitter-typescript-src/typescript/src",
             extra_files: vec!["scanner.c"],
         },
+        TreeSitterParser {
+            name: "tree-sitter-xml",
+            src_dir: "vendored_parsers/tree-sitter-xml-src",
+            extra_files: vec!["scanner.c"],
+        },
         TreeSitterParser {
             name: "tree-sitter-yaml",
             src_dir: "vendored_parsers/tree-sitter-yaml-src",

src/parse/guess_language.rs

@@ -69,6 +69,7 @@ pub enum Language {
     Toml,
     TypeScript,
     TypeScriptTsx,
+    Xml,
     Yaml,
     Zig,
 }
@@ -153,6 +154,7 @@ pub fn language_name(language: Language) -> &'static str {
         Toml => "TOML",
         TypeScript => "TypeScript",
         TypeScriptTsx => "TypeScript TSX",
+        Xml => "XML",
         Yaml => "YAML",
         Zig => "Zig",
     }
@@ -340,6 +342,19 @@ pub fn language_globs(language: Language) -> Vec<glob::Pattern> {
         ],
         TypeScript => &["*.ts"],
         TypeScriptTsx => &["*.tsx"],
+        Xml => &[
+            "*.ant",
+            "*.csproj",
+            "*.plist",
+            "*.resx",
+            "*.svg",
+            "*.xml",
+            "App.config",
+            "nuget.config",
+            ".classpath",
+            ".cproject",
+            ".project",
+        ],
         Yaml => &["*.yaml", "*.yml"],
         Zig => &["*.zig"],
     };
@@ -438,6 +453,7 @@ fn from_emacs_mode_header(src: &str) -> Option<Language> {
             "java" => Some(Java),
             "js" | "js2" => Some(JavaScript),
             "lisp" => Some(CommonLisp),
+            "nxml" => Some(Xml),
             "perl" => Some(Perl),
             "python" => Some(Python),
             "racket" => Some(Racket),

src/parse/tree_sitter_parser.rs

@@ -112,6 +112,7 @@ extern "C" {
     fn tree_sitter_toml() -> ts::Language;
     fn tree_sitter_tsx() -> ts::Language;
     fn tree_sitter_typescript() -> ts::Language;
+    fn tree_sitter_xml() -> ts::Language;
     fn tree_sitter_yaml() -> ts::Language;
     fn tree_sitter_zig() -> ts::Language;
 }
@@ -997,6 +998,23 @@ pub fn from_language(language: guess::Language) -> TreeSitterConfig {
                 sub_languages: vec![],
             }
         }
+        Xml => {
+            let language = unsafe { tree_sitter_xml() };
+            TreeSitterConfig {
+                language,
+                // 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(),
+                sub_languages: vec![],
+            }
+        }
         Yaml => {
             let language = unsafe { tree_sitter_yaml() };
             TreeSitterConfig {

vendored_parsers/highlights/xml.scm

@@ -0,0 +1 @@
+../tree-sitter-xml/tree-sitter-xml/queries/highlights.scm

vendored_parsers/tree-sitter-xml-src

@@ -0,0 +1 @@
+tree-sitter-xml/tree-sitter-xml/src