Initial commit

.gitignore

@@ -0,0 +1,11 @@
+Cargo.toml
+Cargo.lock
+node_modules
+build
+package-lock.json
+pnpm-lock.yaml
+/bindings/
+/src/
+/target/
+.build/
+binding.gyp

README.md

@@ -0,0 +1,3 @@
+# tree-sitter-moshell
+
+Moshell grammar subset for [tree-sitter](https://github.com/tree-sitter/tree-sitter).

corpus/commands.txt

@@ -0,0 +1,112 @@
+===============================
+Command
+===============================
+
+whoami
+
+---
+
+(source_file
+  (command (command_name (identifier))))
+
+===============================
+Command without trailing new line
+===============================
+
+echo hello
+---
+
+(source_file (command (command_name (identifier)) (word)))
+
+===============================
+Variable expansion
+===============================
+
+cat $FOO
+
+---
+
+(source_file
+  (command
+    (command_name (identifier))
+    (expansion (variable (identifier)))))
+
+===============================
+Redirection
+===============================
+
+cat file 2> /dev/null
+echo message >&2
+python <<< 'print("hello")'
+
+---
+
+(source_file
+  (command
+    (command_name
+      (identifier))
+    (word)
+    (word)
+    (redirection
+      (word)))
+  (command
+    (command_name
+      (identifier))
+    (word)
+    (redirection
+      (word)))
+  (command
+    (command_name
+      (identifier))
+    (redirection
+      (raw_string))))
+
+===============================
+Pipeline
+===============================
+
+cat path/to/file.txt | wc -l
+
+---
+
+(source_file
+  (pipeline
+    (command
+      (command_name
+        (identifier))
+      (word))
+    (command
+      (command_name
+        (identifier))
+      (word))))
+
+===============================
+String interpolation
+===============================
+
+echo "The result is $result"
+
+---
+
+(source_file
+  (command
+  (command_name
+    (identifier))
+  (template_string
+    (expansion
+      (variable
+        (identifier))))))
+
+===============================
+Negated command
+===============================
+
+! grep foo
+
+---
+
+(source_file
+  (command
+    (command_name
+      (identifier))
+    (word)))

corpus/control_flow.txt

@@ -0,0 +1,138 @@
+===============================
+If expression
+===============================
+
+if grep found file {
+  echo present
+} else {
+  echo absent
+}
+
+---
+
+(source_file
+  (if
+    (command
+      (command_name
+        (identifier))
+      (word)
+      (word))
+    (block
+      (command
+        (command_name
+          (identifier))
+        (word)))
+    (block
+      (command
+        (command_name
+          (identifier))
+        (word)))))
+
+===============================
+While statement
+===============================
+
+val i = 0
+while $i < 10 {
+  i += 1
+}
+
+---
+
+(source_file
+  (variable_declaration
+    (identifier)
+    (primary
+      (number)))
+  (while
+    (infix_expression
+      (expansion
+        (variable
+          (identifier)))
+      (primary
+        (number)))
+    (block
+      (assignation
+        (identifier)
+        (primary
+          (number))))))
+
+===============================
+Function declaration
+===============================
+
+fun square(n: Int) -> Int = $n * $n
+
+---
+
+(source_file
+  (function_declaration
+    (identifier)
+    (parameter_list
+      (function_parameter
+        (identifier)
+        (type_identifier
+          (identifier))))
+    (type_identifier
+      (identifier))
+    (infix_expression
+      (expansion
+        (variable
+          (identifier)))
+      (expansion
+        (variable
+          (identifier))))))
+
+===============================
+Implicit return
+===============================
+
+fun fibonacci(n: Int) -> Int =
+  if $n <= 1 {
+    1
+  } else {
+    fibonacci($n - 1) + fibonacci($n - 2)
+  }
+
+---
+
+(source_file
+  (function_declaration
+    (identifier)
+    (parameter_list
+      (function_parameter
+        (identifier)
+        (type_identifier
+          (identifier))))
+    (type_identifier
+      (identifier))
+    (if
+      (infix_expression
+        (expansion
+          (variable
+            (identifier)))
+        (primary
+          (number)))
+      (block
+        (primary
+          (number)))
+      (block
+        (infix_expression
+          (call_expression
+            (identifier)
+            (argument_list
+              (infix_expression
+                (expansion
+                  (variable
+                    (identifier)))
+                (primary
+                  (number)))))
+          (call_expression
+            (identifier)
+            (argument_list
+              (infix_expression
+                (expansion
+                  (variable
+                    (identifier)))
+                (primary
+                  (number))))))))))

grammar.js

@@ -0,0 +1,194 @@
+const commaSep = (rule) => seq(rule, repeat(seq(',', rule)));
+
+module.exports = grammar({
+  name: 'moshell',
+  extras: $ => [
+    /\s/,
+    /\\\r?\n/,
+    $.line_comment,
+    $.block_comment,
+  ],
+  rules: {
+    source_file: $ => optional($._declarations),
+
+    _declarations: $ => seq(
+      repeat(seq(
+        $._declaration,
+        $._terminator
+      )),
+      $._declaration,
+      optional($._terminator)
+    ),
+    _declaration: $ => choice(
+      $.function_declaration,
+      $.use_declaration,
+      $.variable_declaration,
+      $._statement,
+      $.return,
+      $.while,
+      'break',
+      'continue',
+    ),
+    function_declaration: $ => seq(
+      'fun',
+      field('name', $.identifier),
+      field('parameters', $.parameter_list),
+      optional(seq(
+        '->',
+        field('return_type', $.type_identifier)
+      )),
+      '=',
+      $._expression
+    ),
+    parameter_list: $ => seq(
+      '(',
+      optional(commaSep($.function_parameter)),
+      ')'
+    ),
+    function_parameter: $ => seq(
+      $.identifier,
+      ':',
+      $.type_identifier
+    ),
+    use_declaration: $ => seq(
+      'use',
+      $._use_clause,
+    ),
+    _use_clause: $ => seq(
+      $.identifier,
+      repeat(seq('::', $.identifier)),
+      optional(seq('as', $.identifier)),
+    ),
+    variable_declaration: $ => seq(
+      choice('var', 'val'),
+      $.identifier,
+      optional(seq(':', $.type_identifier)),
+      optional(seq('=', $._expression))
+    ),
+    command: $ => seq(
+      $.command_name,
+      repeat($._argument),
+      repeat($.redirection),
+    ),
+
+    _statement: $ => choice(
+      $.command,
+      $.pipeline,
+      seq('!', $.command),
+      $._expression,
+    ),
+    while: $ => seq(
+      'while',
+      $._statement,
+      $.block
+    ),
+    return: $ => seq(
+      'return',
+      optional($._expression)
+    ),
+
+    _expression: $ => choice(
+      $.block,
+      $.if,
+      $.call_expression,
+      $.assignation,
+      $.infix_expression,
+      $.prefix_expression,
+      $.expansion,
+      $.primary
+    ),
+    block: $ => seq(
+      '{',
+      optional($._declarations),
+      '}'
+    ),
+    if: $ => seq(
+      'if',
+      field('condition', $._statement),
+      field('then', $.block),
+      optional(seq(
+        'else',
+        field('else', $.block)
+      ))
+    ),
+    call_expression: $ => seq(
+      field('name', $.identifier),
+      field('arguments', $.argument_list),
+    ),
+    argument_list: $ => seq(
+      '(',
+      optional($._expression),
+      repeat(seq(',', $._expression)),
+      ')'
+    ),
+    assignation: $ => prec.right(seq(
+      choice($.identifier, $._expression),
+      choice('=', '+=', '-=', '*=', '/=', '%='),
+      $._expression
+    )),
+    cast_expression: $ => prec.left(seq(
+      field('value', $._expression),
+      'as',
+      field('type', $.type_identifier)
+    )),
+    infix_expression: $ => prec.left(seq(
+      $._expression,
+      choice('+', '-', '*', '/', '%', '==', '!=', '<', '>', '<=', '>=', '&&', '||', 'in'),
+      $._expression
+    )),
+    prefix_expression: $ => prec.right(seq(
+      choice('!', '-'),
+      $._expression
+    )),
+
+    expansion: $ => choice(
+      $.variable,
+      $.command_substitution
+    ),
+    pipeline: $ => prec.left(seq(
+      $._statement,
+      '|',
+      $._statement
+    )),
+    redirection: $ => seq(
+      field('descriptor', optional($.number)),
+      choice('<', '>', '>>', '<<<', '&>', '&>>', '<&', '>&'),
+      field('destination', $._argument)
+    ),
+    command_substitution: $ => seq('$(', $._declaration, ')'),
+
+    command_name: $ => $.identifier,
+    _argument: $ => choice(
+      $.raw_string,
+      $.template_string,
+      $.expansion,
+      $.word
+    ),
+
+    primary: $ => choice(
+      $.number,
+      $.boolean,
+      $.raw_string,
+      $.template_string,
+    ),
+    line_comment: $ => token(seq('//', /.*/)),
+    block_comment: $ => token(seq('/*', /[^*]*\*+([^/*][^*]*\*+)*/, '/')),
+    identifier: $ => /[a-zA-Z_][a-zA-Z0-9_]*/,
+    word: $ => /[^#'"\\\s$<>{}&;]+/,
+    number: $ => /\d+/,
+    boolean: $ => choice('true', 'false'),
+    template_string: $ => seq(
+      '"',
+      repeat(choice($._string_content, $.expansion)),
+      '"'
+    ),
+    raw_string: $ => /'[^']*'/,
+    _string_content: $ => token(prec(-1, /([^"`$\\]|\\(.|\r?\n))+/)),
+    _terminator: $ => choice(';', '\n'),
+    type_identifier: $ => $.identifier,
+    variable: $ => seq(
+      '$',
+      $.identifier
+    ),
+  }
+});

package.json

@@ -0,0 +1,32 @@
+{
+  "name": "tree-sitter-moshell",
+  "version": "0.1.0",
+  "description": "C grammar for node-tree-sitter",
+  "main": "bindings/node",
+  "keywords": [
+    "parser",
+    "lexer"
+  ],
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/moshell-lang/tree-sitter-moshell.git"
+  },
+  "scripts": {
+    "test": "tree-sitter test"
+  },
+  "license": "GPL-3.0-only",
+  "dependencies": {
+    "nan": "^2.17.0"
+  },
+  "devDependencies": {
+    "tree-sitter-cli": "^0.20.8"
+  },
+  "tree-sitter": [
+    {
+      "scope": "source.moshell",
+      "file-types": [
+        "msh"
+      ]
+    }
+  ]
+}

queries/highlights.scm

@@ -0,0 +1,73 @@
+; Literals
+
+(raw_string) @string
+(template_string) @string
+(number) @number
+(command_name) @function.macro
+(expansion) @embedded
+
+; Functions
+
+(function_declaration
+  name: (identifier) @function)
+(function_parameter (identifier) @variable.parameter)
+(call_expression
+  name: (identifier) @function)
+
+(line_comment) @comment
+(block_comment) @comment
+
+; Punctuation and operators
+
+"(" @punctuation.bracket
+")" @punctuation.bracket
+"{" @punctuation.bracket
+"}" @punctuation.bracket
+":" @punctuation.delimiter
+"::" @punctuation.delimiter
+"," @punctuation.delimiter
+";" @punctuation.delimiter
+
+[
+  "+"
+  "-"
+  "*"
+  "/"
+  "%"
+  "="
+  "+="
+  "-="
+  "*="
+  "/="
+  "%="
+  "<"
+  "<="
+  ">"
+  ">="
+  "&&"
+  "||"
+  "!"
+  "|"
+] @operator
+
+; Identifiers
+
+(type_identifier) @type
+[
+    "as"
+    "break"
+    "continue"
+    "else"
+    "fun"
+    "if"
+    "in"
+    "return"
+    "use"
+    "val"
+    "var"
+    "while"
+] @keyword
+[
+    "true"
+    "false"
+] @constant.builtin