Initial commit: Tree-sitter grammar for XML language

grammar.js

@@ -0,0 +1,62 @@
+module.exports = grammar({
+  name: 'xml',
+
+  rules: {
+    source_file: $ => repeat($._item),
+
+    _item: $ => choice(
+      $.element,
+      $.comment,
+      $.text,
+      $._whitespace
+    ),
+
+    // XML element
+    element: $ => seq(
+      '<',
+      $.tag_name,
+      repeat($.attribute),
+      choice(
+        seq('>', repeat($._item), '</', $.tag_name, '>'),
+        '/>'
+      )
+    ),
+
+    // Tag name
+    tag_name: $ => /[A-Za-z][A-Za-z0-9_-]*/,
+
+    // Attribute
+    attribute: $ => seq(
+      $.attribute_name,
+      '=',
+      $.attribute_value
+    ),
+
+    // Attribute name
+    attribute_name: $ => /[A-Za-z][A-Za-z0-9_-]*/,
+
+    // Attribute value
+    attribute_value: $ => choice(
+      seq('"', $.quoted_value, '"'),
+      seq("'", $.quoted_value, "'")
+    ),
+
+    // Quoted value content
+    quoted_value: $ => /[^"']*/,
+
+    // XML comment
+    comment: $ => token(seq('<!--', /[^-]*(?:-[^-]+)*/, '-->')),
+
+    // Text content
+    text: $ => /[^<]+/,
+
+    // Whitespace
+    _whitespace: $ => /\s+/
+  },
+
+  extras: $ => [
+    /\s/,
+    $.comment
+  ]
+});
+