@algosail/tree-sitter 0.1.1 → 0.1.3

package/grammar.js

@@ -1,92 +1,179 @@
 /// <reference types="tree-sitter-cli/dsl" />
 // @ts-check
 
+const UPPERNAME = /[A-Z][a-zA-Z0-9_]*/
+const LOWERNAME = /[a-z][a-zA-Z0-9_]*/
+
 module.exports = grammar({
   name: 'sail',
 
   extras: ($) => [/\s+/],
 
   conflicts: ($) => [
-    // '(' can start a comment OR a signature — tree-sitter resolves by
-    // looking for '--' inside.
     [$.comment, $.signature],
     [$.comment, $.sig_quotation],
-    // After word_def's signature, '(' is ambiguous: body comment vs next
-    // top-level item. Declare as GLR self-conflict so both are explored.
-    [$.word_def],
-    // tag_group contains tag_def children; both can follow an uppercase name
-    [$.tag_def, $.tag_group],
-    // module_alias and type_name are both /[A-Z][a-zA-Z0-9_]*/
-    // [$.type_name, $.module_alias],
-    // type_variable and identifier are both lowercase words
-    [$.type_variable, $.identifier],
+
+    [$.module_def, $.module_ref],
+    [$.group_def, $.group_ref],
+    [$.tag_def, $.tag_ref],
+    [$.map_def, $.map_ref],
+    [$.field_def, $.field_ref],
+
+    [$.group],
+    [$.tag],
+    [$.map],
+    [$.field],
+    [$.word],
+
+    [$.group_def],
+    [$.tag, $.group],
+    [$.word, $._expr],
   ],
 
   rules: {
     source_file: ($) => repeat($._top_level),
+    _top_level: ($) => choice($.comment, $.import, $.group, $.map, $.word),
+
+    // ~Module
+    module_def: ($) => /~[A-Z][a-zA-Z0-9_]*/,
+    // ~Module
+    module_ref: ($) => /~[A-Z][a-zA-Z0-9_]*/,
+    // &Group
+    group_def: ($) => /&[A-Z][a-zA-Z0-9_]*/,
+    // &Group
+    group_ref: ($) => /&[A-Z][a-zA-Z0-9_]*/,
+    // #Tag
+    tag_def: ($) => /#[A-Z][a-zA-Z0-9_]*/,
+    // #Tag
+    tag_ref: ($) => /#[A-Z][a-zA-Z0-9_]*/,
+    // _Tag
+    tag_pattern: ($) => /\_[A-Z][a-zA-Z0-9_]*/,
+    // _
+    default_pattern: ($) => token('_'),
+    // $Map
+    map_def: ($) => /\$[A-Z][a-zA-Z0-9_]*/,
+    // $Map
+    map_ref: ($) => /\$[A-Z][a-zA-Z0-9_]*/,
+    // .field
+    field_def: ($) => /\.[a-z][a-zA-Z0-9_]*/,
+    // $Map.field
+    field_ref: ($) => /\$[A-Z][a-zA-Z0-9_]*\.[a-z][a-zA-Z0-9_]*/,
+    // @word
+    word_def: ($) => /@[a-z][a-zA-Z0-9_]*/,
+    // /word
+    word_ref: ($) => /\/[a-z][a-zA-Z0-9_]*/,
+
+    // ~Module&Group
+    module_group_ref: ($) => /~[A-Z][a-zA-Z0-9_]*&[A-Z][a-zA-Z0-9_]*/,
+    // ~Module#Tag
+    module_tag_ref: ($) => /~[A-Z][a-zA-Z0-9_]*#[A-Z][a-zA-Z0-9_]*/,
+    // ~Module$Map
+    module_map_ref: ($) => /~[A-Z][a-zA-Z0-9_]*\$[A-Z][a-zA-Z0-9_]*/,
+    // ~Module$Map.field
+    module_field_ref: ($) => /~[A-Z][a-zA-Z0-9_]*\$[A-Z][a-zA-Z0-9_]*\.[a-z][a-zA-Z0-9_]*/,
+    // ~Module/word
+    module_word_ref: ($) => /~[A-Z][a-zA-Z0-9_]*\/[a-z][a-zA-Z0-9_]*/,
 
-    _top_level: ($) =>
-      choice($.comment, $.module_def, $.import_decl, $.tag_group, $.map_def, $.word_def),
-
-    // Comment / doc block:  ( any text )
-    // Intentionally simple — no nested-paren support inside comment text.
-    comment: ($) => seq('(', optional($.comment_content), ')'),
-    comment_content: ($) => /[^()]+/,
-
-    // Module definition:  !ModuleName
-    module_def: ($) => field('name', $.module_name),
-    module_name: ($) => /![A-Z][a-zA-Z0-9_]*/,
-
-    // Import:  |path/or/npm:pkg  Alias
-    import_decl: ($) => seq(field('path', $.import_path), field('alias', $.module_alias)),
-
-    import_path: ($) => /\|[^\s]+/,
-
-    module_alias: ($) => /~[A-Z][a-zA-Z0-9_]*/,
-
-    // Tag group:  &Name typeParam*  (>TagCase typeParam*)*
-    tag_group: ($) =>
-      seq(field('name', $.tag_group_name), repeat($.type_variable), repeat($.tag_def)),
-
-    tag_group_name: ($) => /&[A-Z][a-zA-Z0-9_]*/,
+    // Uppercase type: Int, Str, Maybe, List, etc.
+    type: ($) => /[A-Z][a-zA-Z0-9_]*/,
+    // Lowercase type variable: a, b, elem, etc.
+    type_var: ($) => /[a-z][a-zA-Z0-9_]*/,
+    // Lowercase type variable: a, b, elem, etc.
+    spread: ($) => /\.\.[a-z][a-zA-Z0-9_]*/,
 
-    tag_def: ($) => seq(field('name', $.tag_name), repeat($.type_variable)),
+    // +Effect
+    effect_add: ($) => /\+[A-Z][a-zA-Z0-9_]*/,
+    // -Effect
+    effect_remove: ($) => /\-[A-Z][a-zA-Z0-9_]*/,
 
-    tag_name: ($) => />[A-Z][a-zA-Z0-9_]*/,
+    // :name
+    slot_push: ($) => /:[a-z][a-zA-Z0-9_]*/,
 
-    // Map definition:  %Name  (:field Type)*
-    map_def: ($) => seq(field('name', $.map_name), repeat($.map_field)),
+    // ;name
+    slot_pop: ($) => /;[a-z][a-zA-Z0-9_]*/,
 
-    map_name: ($) => /%[A-Z][a-zA-Z0-9_]*/,
+    // 'raw string literal'
+    raw_string: ($) => /\'[^\']*\'/,
 
-    map_field: ($) => seq(field('key', $.map_field_name), field('type', $.type_name)),
+    // Catch-all: any non-whitespace sequence that doesn't match a more specific
+    // rule. prec(-1) gives it the lowest priority so every other token wins
+    // when there is a tie. Structural characters ( ) [ ] are excluded because
+    // they are needed by the parser to delimit blocks and comments.
+    raw_value: ($) => token(prec(-1, /[^\s\[\]()']+/)),
 
-    map_field_name: ($) => /:[a-z][a-zA-Z0-9_]*/,
+    // Comment / doc block:  ( any text )
+    // comment_content is recursive: it can contain plain text and/or nested
+    // parenthesised groups, so  ( use f(x) here )  parses correctly.
+    comment: ($) => seq('(', optional($.comment_content), ')'),
+    comment_content: ($) => repeat1(choice(/[^()]+/, $.comment)),
+
+    // Import:  +path/or/+pkg  Alias
+    import: ($) => seq(field('path', $.path), field('module', $.module_def)),
+    path: ($) => /\+[^\s]+/,
+    // seq('+', field('url', alias(/[^\s]+/, $.url))),
+
+    // Tag group:  &Name typeParam*  (#TagCase typeParam*)*
+    group: ($) =>
+      seq(
+        field('def', $.group_def),
+        repeat($.type_var),
+        optional(field('doc', $.comment)),
+        repeat($.tag),
+      ),
+    tag: ($) =>
+      seq(
+        field('def', $.tag_def),
+        optional(field('type_param', $.type_var)),
+        optional(field('doc', $.comment)),
+      ),
+    group_type: ($) =>
+      seq(
+        field('group', choice($.group_ref, $.module_group_ref)),
+        optional(field('params', $._generic)),
+      ),
+    _generic: ($) => seq('{', repeat($._generic_content), '}'),
+    _generic_content: ($) => choice($.type, $.group_type, $.map_ref, $.module_map_ref),
+
+    // Map definition:  %Name  (.field Type)*
+    map: ($) => seq(field('def', $.map_def), optional(field('doc', $.comment)), repeat($.field)),
+    field: ($) =>
+      seq(
+        field('key', $.field_def),
+        field('type', $._field_types),
+        optional(field('doc', $.comment)),
+      ),
+    _field_types: ($) => choice($.type, $.group_type, $.map_ref, $.module_map_ref),
 
     // Word definition:  @name ( sig ) expr*
     // Signature is required per the spec ("Word definition must have a signature").
     // prec.right makes the body's repeat greedy: prefer consuming '(' as a body
     // comment rather than ending the word_def early.
-    word_def: ($) =>
-      prec.right(seq(field('name', $.word_name), field('sig', $.signature), repeat($._expr))),
-
-    word_name: ($) => /@[a-z][a-zA-Z0-9_]*/,
+    word: ($) =>
+      prec.right(
+        seq(
+          field('name_def', $.word_def),
+          field('sig', $.signature),
+          optional(field('doc', $.comment)),
+          repeat($._expr),
+        ),
+      ),
 
     // Signature:  ( inputs -- outputs +effects )
     // The required '--' token is what makes it unambiguous vs a comment.
     signature: ($) => seq('(', repeat($._sig_item), $.sig_arrow, repeat($._sig_item), ')'),
-
     sig_arrow: ($) => token('--'),
-
     _sig_item: ($) =>
       choice(
         $.effect_add,
         $.effect_remove,
         $.spread,
-        $.type_name,
-        $.type_variable,
+        $.type,
+        $.type_var,
         $.sig_list,
         $.sig_quotation,
+        $.group_type,
+        $.map_ref,
+        $.module_map_ref,
       ),
 
     // [ Type Type ... ] — list / tuple type in a signature
@@ -95,37 +182,24 @@ module.exports = grammar({
     // ( a b -- c d ) — higher-order function type nested inside a signature
     sig_quotation: ($) => seq('(', repeat($._sig_item), $.sig_arrow, repeat($._sig_item), ')'),
 
-    // +IO, +FAIL, etc.
-    effect_add: ($) => /\+[A-Z][a-zA-Z0-9_]*/,
-
-    // -IO, -FAIL, etc.  (uppercase after dash avoids matching negative numbers)
-    effect_remove: ($) => /-[A-Z][a-zA-Z0-9_]*/,
-
-    // ..a, ..row — spread / row-variable in a signature
-    spread: ($) => /\.\.[a-zA-Z][a-zA-Z0-9_]*/,
-
-    // Uppercase type: Int, Str, Maybe, List, etc.
-    type_name: ($) => /[A-Z][a-zA-Z0-9_]*/,
-
-    // Lowercase type variable: a, b, elem, etc.
-    type_variable: ($) => /[a-z][a-zA-Z0-9_]*/,
-
     // Expressions inside word bodies
     _expr: ($) =>
       choice(
         $.comment, // doc / inline comment block
         $.quotation,
         $.builtin_word,
-        $.word_call,
-        $.module_call,
-        $.map_access,
-        $.tag_constructor,
+        $.word_ref,
+        $.module_word_ref,
+        $.tag_ref,
+        $.module_tag_ref,
         $.tag_pattern,
+        $.default_pattern,
+        $.field_ref,
+        $.module_field_ref,
         $.slot_push,
         $.slot_pop,
         $.raw_string,
-        $.number,
-        $.identifier,
+        $.raw_value,
       ),
 
     // [ expr* ] — quotation (anonymous code block or list literal)
@@ -159,35 +233,5 @@ module.exports = grammar({
           'ERROR',
         ),
       ),
-
-    // /wordName — call a locally defined word
-    word_call: ($) => /\/[a-z][a-zA-Z0-9_]*/,
-
-    // ~Module/word  or  ~Module — module-qualified word call
-    module_call: ($) => /~[A-Z][a-zA-Z0-9_]*(\/[a-zA-Z][a-zA-Z0-9_]*)?/,
-
-    // *Map/field — map field accessor / lens
-    map_access: ($) => /\*[A-Z][a-zA-Z0-9_]*\/[a-z][a-zA-Z0-9_]*/,
-
-    // #TagName — construct a tagged union value
-    tag_constructor: ($) => /#[A-Z][a-zA-Z0-9_]*/,
-
-    // _TagName — match/destructure a tag in MATCH
-    tag_pattern: ($) => /_[A-Z][a-zA-Z0-9_]*/,
-
-    // .name — pop the top of the stack into a named local slot
-    slot_push: ($) => /\.[a-z][a-zA-Z0-9_]*/,
-
-    // ,name — push a named local slot back onto the stack
-    slot_pop: ($) => /,[a-z][a-zA-Z0-9_]*/,
-
-    // 'raw string literal'
-    raw_string: ($) => /\'[^\']*\'/,
-
-    // Numeric literal: integer or decimal
-    number: ($) => /[0-9]+(\.[0-9]+)?/,
-
-    // Generic bare identifier (raw data tokens, unrecognised lowercase words)
-    identifier: ($) => /[a-zA-Z_][a-zA-Z0-9_]*/,
   },
 })

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@algosail/tree-sitter",
-  "version": "0.1.1",
+  "version": "0.1.3",
   "description": "Tree-sitter grammar for the Sail language",
   "main": "bindings/node",
   "types": "bindings/node",
@@ -8,7 +8,8 @@
     "build:wasm": "tree-sitter build --wasm",
     "build:native": "node-gyp rebuild",
     "generate": "tree-sitter generate",
-    "install": "echo 'Skipping native build. Run npm run build:native manually if needed.'"
+    "install": "echo 'Skipping native build. Run npm run build:native manually if needed.'",
+    "playground": "tree-sitter playground"
   },
   "devDependencies": {
     "node-addon-api": "^8.6.0",