tree-sitter-lang/grammar.js

module.exports = grammar({
  name: 'lang',

  word: $ => $.identifier,

  extras: $ => [
    /\s/,
  ],

  rules: {

    // --- sources

    source_file: $ => repeat1($._statement),

    _statement: $ => choice(
      $.import,
      $.function_definition,
      $.type_definition,
      $.extra,
      $.empty_lines,
    ),

    import: $ => seq(
      choice('::', 'import'),
      field('name', choice($.simple_name_identifier, $.placeholder)),
      optional(seq('=', field('module', $.simple_name_identifier))),
      optional(seq(':', repeat1(choice(
        $.simple_name_identifier,
        seq('(',$.operator, ')'),
        $.simple_type_identifier,
        $.typeclass_identifier,
      )))),
      ';',
    ),

    constraint: $ => seq('?', $._expression),

    // --- definitions

    function_definition: $ => seq(
      optional($.definition_info),
      repeat($.annotation_info),
      repeat(seq($.constraint, ';')),
      optional('.'), // for methods
      choice(field('name', $.simple_name_identifier), seq('(', field('name', $.operator), ')')),
      optional($._optional_or_result),
      repeat(seq(
        optional($.annotation_identifier),
        optional($._reference),
        $.argument_name_identifier,
        optional($._optional_or_result),
      )),
      optional(seq(
        ':',
        repeat1(seq(
          optional($.annotation_identifier),
          optional($._reference),
          $.type,
        )),
      )),
      choice(seq('=', choice(prec(2, choice($.block, $.array)), seq($._super_expression, ';'))), ';'),
    ),

    // datatype or typeclass definition
    type_definition: $ => seq(
      optional($.definition_info),
      repeat($.annotation_info), // for datatypes only
      optional('^'),
      field('name', choice($.simple_type_identifier, $.typeclass_identifier)),
      optional(seq('[', repeat($.typeclass_identifier), ']')), // parametric typeclasses ??
      repeat($.argument_type_identifier), // for datatypes only
      optional(seq('=', $.variant_type)), // for datatypes only
      ';',
    ),

    // --- flow control

    case: $ => seq(
      choice(':=', '=:'),
      $._expression,
      optional(seq(choice('??', 'if'), field('condition', $._expression))),
      optional(field('expression', seq($._do, $._expression))),
    ),

    match: $ => seq(
      $._expression,
      repeat1($.case),
    ),

    condition: $ => seq(
      choice('??', 'if'),
      $._expression,
      $._do,
      $._expression,
      repeat(seq(
        choice('!!', 'elif'),
        $._expression,
        $._do,
        $._expression,
      )),
      optional(seq(choice('!!=>', 'else'), $._expression)),
    ),

    loop: $ => seq(
      choice('@', 'for'),
      optional(choice(
        $._expression,
        seq($._expression, ':', $._expression),
      )),
      $._do,
      $._expression,
    ),

    // --- operators

    comma_expression: $ => prec.left(seq($._super_expression, ',', $._super_expression)),

    operator_expression: $ => prec.left(choice(
      prec(4, seq($._expression, field('name', $.operator), $._expression)),
      prec(3, seq($._expression, field('name', $.operator_tail1), $._expression)),
      prec(2, seq($._expression, field('name', $.operator_tail2), $._expression)),
      prec(1, seq($._expression, field('name', $.operator_tail3), $._expression)),
    )),

    // --- continers

    block: $ => seq('{', repeat(choice(seq($._super_expression, ';'), $.extra, $.empty_lines)), '}'),

    array: $ => seq('[[', repeat1($._scoped_expression), ']]'),

    // --- modifiers

    return: $ => seq(choice('return', 'bring'), $._expression),

    name_definition: $ => seq(
      choice(choice('%', 'let'), choice('$', 'var')),
      choice($.simple_name_identifier, $.placeholder),
    ),

    array_access: $ => seq($._scoped_expression, '[', $._super_expression, ']'),

    tuple_access: $ => seq($._scoped_expression, '.', $.number_literal),

    loop_control: $ => choice('break', 'continue'),

    reference_expression: $ => prec(-1, seq($._reference, $._scoped_expression)),

    suffix_expression: $ => seq($._scoped_expression, $._optional_or_result),

    // --- other

    name_expression: $ => seq(
      choice(
        seq($.type, '.', field('name', $.simple_name_identifier)),
        seq($._scoped_expression, '.', field('name', $.simple_name_identifier)),
        field('name', $._name_identifier),
        seq('(', field('name', $.operator), ')'),
      ),
      repeat(seq(optional($.annotation_identifier), $._scoped_expression)),
    ),

    constructor: $ => seq(
      field('type', $.type),
      repeat1(seq(optional($.annotation_identifier), $._scoped_expression)),
    ),

    lambda: $ => seq(
      choice('\\', 'lambda'),
      repeat($.argument_name_identifier),
      $._do,
      $._expression,
    ),

    // --- expression

    _super_expression: $ => choice(
      $.match,
      $.condition,
      $.loop,
      $.comma_expression,
      $._expression,
    ),

    _expression: $ => choice(
      $.operator_expression,
      $.return,
      $.name_expression,
      $.constructor,
      $.lambda,
      $._not_name_scoped_expression,
    ),

    _scoped_expression: $ => choice(
      $._name_identifier,
      $._not_name_scoped_expression,
    ),

    _not_name_scoped_expression: $ => choice(
      $.block,
      $.array,
      $.placeholder,
      $.name_definition,
      $.array_access,
      $.tuple_access,
      $.loop_control,
      $.reference_expression,
      $.suffix_expression,
      $._literal,
      seq('(', $._super_expression, ')'),
    ),

    // --- type

    variant_type: $ => seq(
      optional('|'),
      $.tuple_type,
      repeat(seq('|', $.tuple_type)),
    ),

    tuple_type: $ => seq(
      optional('&'),
      $._annotated_type,
      repeat(seq('&', $._annotated_type)),
    ),

    _annotated_type: $ => seq(optional($.annotation_identifier), $.type),

    type: $ => seq(
      optional('^'),
      field('name', $._type_identifier),
      optional($._optional_or_result),
      optional(seq('[', repeat1($.type), ']'))
    ),

    // --- comments

    definition_info: $ => repeat1(seq(': ', $.info)),
    annotation_info: $ => seq($.annotation_identifier, ' ', $.info),

    info: $ => /[^\n]*/,

    extra: $ => choice($._exec_comment, $._line_comment, $._block_comment),
    empty_lines: $ => prec.left(repeat1('\n')),

    _exec_comment: $ => token(seq('#!', /[^\n]*/)),
    _line_comment: $ => token(seq('//', /[^\n]*/)),
    _block_comment: $ => token(seq('/*', /([^*]|(\*[^/]))*/, '*/')),

    // --- tokens

    _do: $ => choice('=>', 'do'),
    _optional_or_result: $ => choice('?', '!'),

    _reference: $ => choice(choice('<-', '<>', '--', '->',
                                   '<-|<>', '<-|--', '<>|->', '--|->', '<>|--', '<-|->',
                                   '<-|<>|->', '<-|--|->', '<-|<>|--', '<>|--|->',
                                   '<-|<>|--|->'),
                            choice('in', 'ref', 'const', 'out',
                                   'in|ref', 'in|const', 'ref|out', 'const|out', 'ref|const', 'in|out',
                                   'in|ref|out', 'in|const|out', 'in|ref|const', 'ref|const|out',
                                   'in|ref|const|out')),

    _name_identifier: $ => choice($.argument_name_identifier, $.simple_name_identifier),
    _type_identifier: $ => choice($.argument_type_identifier, $.simple_type_identifier),

    placeholder: $ => '_',
    simple_name_identifier: $ => token(seq(repeat(seq(/[a-z_][a-z0-9_]*/, '.')), /[a-z_][a-z0-9_]*/)),
    simple_type_identifier: $ => /([a-z_][a-z0-9_]*\.)*[A-Z][a-zA-Z0-9]*/,
    typeclass_identifier: $ => /([a-z_][a-z0-9_]*\.)*#[A-Z][a-zA-Z0-9]*/,
    argument_name_identifier: $ => /'[a-z_][a-z0-9_]*/,
    argument_type_identifier: $ => /'[A-Z][a-zA-Z0-9]*/,
    annotation_identifier: $ => /@[a-z_][a-z0-9_]*/,

    operator: $ => /([+\-*/%^!?|&,<>=]+)|\.+/,
    operator_tail1: $ => /[+\-*/%^!?|&,<>=]+\./,
    operator_tail2: $ => /[+\-*/%^!?|&,<>=]+\.\./,
    operator_tail3: $ => /[+\-*/%^!?|&,<>=]+\.\.\./,

    float_number_literal: $ => /[0-9]+\.[0-9]+/,
    number_literal: $ => /[0-9]+/,
    string_literal: $ => seq('\"', /([^\\\"]|(\\.))*/, '\"'),
    char_literal: $ => seq('\'\'', /[^\\\']|(\\.)/, '\'\''),
    bool_literal: $ => choice('true', 'false'),
    unit_literal: $ => '()',
    null_literal: $ => 'null',

    _literal: $ => choice(
      $.float_number_literal,
      $.number_literal,
      $.string_literal,
      $.char_literal,
      $.bool_literal,
      $.unit_literal,
      $.null_literal,
    ),

    identifier: $ => /([@'#]?[a-zA-Z0-9_]+)|([+\-*/%^!?|&,<>=]+\.?\.?\.?)|\.+/,
  }
});