Initial commit: Tree-sitter grammar for Hack Assembly

grammar.js

@@ -0,0 +1,97 @@
+module.exports = grammar({
+  name: 'hack_assembly',
+
+  rules: {
+    source_file: $ => repeat($._item),
+
+    _item: $ => choice(
+      $.a_instruction,
+      $.c_instruction,
+      $.label_declaration,
+      $.comment,
+      $._whitespace
+    ),
+
+    // A-instruction: @value or @symbol
+    a_instruction: $ => seq(
+      '@',
+      choice(
+        $.constant,
+        $.symbol
+      )
+    ),
+
+    // C-instruction: [dest=]comp[;jump]
+    c_instruction: $ => seq(
+      optional(seq($.dest, '=')),
+      $.comp,
+      optional(seq(';', $.jump))
+    ),
+
+    // Label declaration: (SYMBOL)
+    label_declaration: $ => seq(
+      '(',
+      $.symbol,
+      ')'
+    ),
+
+    // Constants: -32768 to 32767 (16-bit signed)
+    constant: $ => /-?\d+/,
+
+    // Symbols: letters, digits, _, ., $, : (not starting with digit)
+    symbol: $ => choice(
+      $.predefined_symbol,
+      $.user_symbol
+    ),
+
+    // Predefined symbols
+    predefined_symbol: $ => choice(
+      // Virtual registers R0-R15
+      /R(0|1[0-5]|[2-9])/,
+      // Special symbols
+      'SP', 'LCL', 'ARG', 'THIS', 'THAT',
+      // I/O symbols
+      'SCREEN', 'KBD'
+    ),
+
+    // User-defined symbols
+    user_symbol: $ => /[A-Za-z_.$:][A-Za-z0-9_.$:]*/,
+
+    // Destination field
+    dest: $ => choice(
+      'M', 'D', 'MD', 'A', 'AM', 'AD', 'AMD'
+    ),
+
+    // Computation field
+    comp: $ => choice(
+      // Zero and one
+      '0', '1', '-1',
+      // D register operations
+      'D', '!D', '-D', 'D+1', 'D-1',
+      // A register operations
+      'A', '!A', '-A', 'A+1', 'A-1',
+      // M register operations
+      'M', '!M', '-M', 'M+1', 'M-1',
+      // Two operand operations with D and A
+      'D+A', 'D-A', 'A-D', 'D&A', 'D|A',
+      // Two operand operations with D and M
+      'D+M', 'D-M', 'M-D', 'D&M', 'D|M'
+    ),
+
+    // Jump field
+    jump: $ => choice(
+      'JGT', 'JEQ', 'JGE', 'JLT', 'JNE', 'JLE', 'JMP'
+    ),
+
+    // Comments
+    comment: $ => token(seq('//', /.*/)),
+
+    // Whitespace
+    _whitespace: $ => /\s+/
+  },
+
+  extras: $ => [
+    /\s/,
+    $.comment
+  ]
+});