tree-sitter-beancount 2.3.2 → 2.4.0

package/grammar.js

@@ -40,32 +40,34 @@ module.exports = grammar({
 
         _nl: _ => choice('\n', '\r'),
         _eol: $ => choice('\n', '\r', $._eof),
-        _any: $ => /.*/,
+        _any: $ => /[^\r\n]*/,
 
 
-        /*
-         * Org Header Sections
-         */
+        /*--------------------------------------------------------------------------------*/
+        /* SECTION AND HEADLINE RULES (Org-mode/Markdown support) */
+        /*--------------------------------------------------------------------------------*/
         section: $ => seq(
             field('headline', $.headline),
             repeat(choice(
                 $._declarations,
-                $._nl
+                $._nl,
+                field('subsection', $.section)
             )),
-            repeat(field('subsection', $.section)),
             $._sectionend
         ),
-        _org_stars: $ => seq($._stars, /\*+/),
+        _org_stars: $ => seq($._stars, /(\*|#)+/),
         headline: $ => seq(
             $._org_stars,
-            /[ \t]+/, // so it's not part of (item)
+            token(/[ \t]+/), // tokenized for performance
             optional(field('item', $.item)),
             $._nl,
         ),
-        item: $ => $._any,
+        item: $ => token(/[^\r\n]+/),
 
 
-        /* Types for terminal symbols */
+        /*--------------------------------------------------------------------------------*/
+        /* TERMINAL SYMBOLS */
+        /*--------------------------------------------------------------------------------*/
         _indent: $ => token(/[ \r\t]+/),
         _eol: $ => token(/\n/),
         atat: $ => token('@@'),
@@ -77,10 +79,9 @@ module.exports = grammar({
         flag: $ => token(/[!&?%PSTCURM*#]/),
         _none: $ => token('NULL'),
         bool: $ => token(/TRUE|FALSE/),
-        date: $ => token(/([12]\d{3}-(0[1-9]|1[0-2])-(0[1-9]|[12]\d|3[01]))/),
-        //An account name is a colon-separated list of capitalized words which begin with a letter, and whose first word must be one of five account types:
-        //Each component of the account names begin with a capital letter or a number and are followed by letters, numbers or dash (-) characters.
-        //All other characters are disallowed.
+        date: $ => token(/([12]\d{3}[-\/](0[1-9]|1[0-2])[-\/](0[1-9]|[12]\d|3[01]))/),
+        // Account names: Assets|Liabilities|Equity|Income|Expenses followed by colon-separated components
+        // Components can contain Unicode letters/numbers including CJK characters
         account: $ =>
             token(
                 seq(
@@ -88,31 +89,21 @@ module.exports = grammar({
                     repeat1(
                         seq(
                             ":",
-                            /[\p{Lu}\p{N}][\p{L}\p{N}\-]*/,
+                            /[\p{Lu}\p{N}\u3040-\u309f\u30a0-\u30ff\u4e00-\u9fff\uac00-\ud7a3][\p{L}\p{N}\u3040-\u309f\u30a0-\u30ff\u4e00-\u9fff\uac00-\ud7a3\-]*/,
                         ),
                     ),
                 ),
             ),
-        currency: $ => token(/[A-Z][A-Z0-9\'\._\-]{0,22}[A-Z0-9]/),
-        string: $ => token(/"[^"]*"/),
+        currency: $ => token(/[A-Z]([A-Z0-9\'\._\-]{0,22}[A-Z0-9])?/),
+        string: $ => token(/"([^"]|\\")*"/),
         number: $ => token(/([0-9]+|[0-9][0-9,]+[0-9])(\.[0-9]*)?/),
         tag: $ => token(/#[A-Za-z0-9\-_/.]+/),
         link: $ => token(/\^[A-Za-z0-9\-_/.]+/),
 
-        /* Operator precedence.
-         * This is pulled straight out of the textbook example:
-         * https://www.gnu.org/software/bison/manual/html_node/Infix-Calc.html#Infix-Calc
-         */
-        // %left MINUS PLUS
-        // %left ASTERISK SLASH
-        // %precedence NEGATIVE /* negation--unary minus */
-
-        // Start symbol: file
-        /* We have some number of expected shift/reduce conflicts at 'eol'. */
-        //%expect 7
-
         /*--------------------------------------------------------------------------------*/
-        /* Grammar Rules */
+        /* ARITHMETIC EXPRESSIONS */
+        /*--------------------------------------------------------------------------------*/
+        // Operator precedence: PLUS/MINUS (left, 1), MULTIPLY/DIVIDE (left, 2), UNARY (3)
 
         txn: $ => choice(
             "txn",
@@ -211,7 +202,7 @@ module.exports = grammar({
                             $.posting,
                             seq(
                                 $._indent,
-                                $.comment,
+                                field("comment", $.comment),
                                 $._eol
                             )
                         )
@@ -241,46 +232,20 @@ module.exports = grammar({
         //),
 
         posting: $ =>
-            choice(
-                seq(
-                    $._indent,
-                    field("optflag", optional($.optflag)),
-                    field("account", $.account),
-                    field("amount", optional($.incomplete_amount)),
-                    field("cost_spec", optional($.cost_spec)),
-                    field("comment", optional($.comment)),
-                    $._eol
-                ),
-                seq(
-                    $._indent,
-                    field("optflag", optional($.optflag)),
-                    field("account", $.account),
-                    field("amount", optional($.incomplete_amount)),
-                    field("cost_spec", optional($.cost_spec)),
-                    $.at,
-                    field("price_annotation", optional($.price_annotation)),
-                    field("comment", optional($.comment)),
-                    $._eol
-                ),
-                seq(
-                    $._indent,
-                    field("optflag", optional($.optflag)),
-                    field("account", $.account),
-                    field("amount", optional($.incomplete_amount)),
-                    field("cost_spec", optional($.cost_spec)),
-                    $.atat,
-                    field("price_annotation", optional($.price_annotation)),
-                    field("comment", optional($.comment)),
-                    $._eol
+            seq(
+                $._indent,
+                field("optflag", optional($.optflag)),
+                field("account", $.account),
+                field("amount", optional($.incomplete_amount)),
+                field("cost_spec", optional($.cost_spec)),
+                optional(
+                    seq(
+                        choice($.at, $.atat),
+                        field("price_annotation", optional($.price_annotation))
+                    )
                 ),
-                seq(
-                    $._indent,
-                    field("optflag", optional($.optflag)),
-                    field("account", $.account),
-                    field("amount", optional($.incomplete_amount)),
-                    field("comment", optional($.comment)),
-                    $._eol
-                )
+                field("comment", optional($.comment)),
+                $._eol
             ),
 
         key: $ => token(/[a-z][a-zA-Z0-9\-_]+/),
@@ -326,7 +291,12 @@ module.exports = grammar({
                     ),
                     seq(
                         $._key_value_line
-                    )
+                    ),
+                    seq(
+                        $._indent,
+                        field("comment", $.comment),
+                        $._eol
+                    ),
                 )
             ),
 
@@ -522,6 +492,7 @@ module.exports = grammar({
                 "price",
                 field("currency", $.currency),
                 field("amount", $.amount),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -532,6 +503,7 @@ module.exports = grammar({
                 "event",
                 field("type", $.string),
                 field("desc", $.string),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -542,6 +514,7 @@ module.exports = grammar({
                 "query",
                 field("name", $.string),
                 field("query", $.string),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -552,6 +525,7 @@ module.exports = grammar({
                 "note",
                 field("account", $.account),
                 field("note", $.string),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -565,6 +539,7 @@ module.exports = grammar({
                 field("account", $.account),
                 field("filename", $.filename),
                 field("tags_links", optional($.tags_links)),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -590,6 +565,7 @@ module.exports = grammar({
                         $.custom_value
                     ),
                 )),
+                field("comment", optional($.comment)),
                 $._eol,
                 optional($._key_value_list)
             ),
@@ -655,10 +631,11 @@ module.exports = grammar({
             $._skipped_lines,
         ),
 
-        /* End Grammar Rules */
+        /*--------------------------------------------------------------------------------*/
+        /* UTILITY AND LEXICAL RULES */
         /*--------------------------------------------------------------------------------*/
 
-        comment: $ => seq(';', /.*/),
+        comment: $ => token(seq(';', /[^\r\n]*/)),
 
         // NOTE: includes reserved identifiers
         identifier: $ => /[a-z]+/,
@@ -667,12 +644,12 @@ module.exports = grammar({
             choice(
                 seq(
                     $.flag,
-                    /.*/,
+                    /[^\r\n]*/,
                     $._eol
                 ),
                 seq(
                     ":",
-                    /.*/,
+                    /[^\r\n]*/,
                     $._eol
                 ),
                 $._eol,
@@ -682,58 +659,5 @@ module.exports = grammar({
                 ),
             ),
 
-        _ASCII: $ => /[\x00-\x7f]/,
-        _UTF_8_1: $ => /[\x80-\xbf]/,
-        _UTF_8_2: $ => seq(
-            /[\xc2-\xdf]/,
-            $._UTF_8_1,
-        ),
-        _UTF_8_3: $ => choice(
-            seq(
-                /\xe0[\xa0-\xbf]/,
-                $._UTF_8_1,
-            ),
-            seq(
-                /[\xe1-\xec]/,
-                $._UTF_8_1,
-                $._UTF_8_1,
-            ),
-            seq(
-                /\xed[\x80-\x9f]/,
-                $._UTF_8_1,
-            ),
-            seq(
-                /[\xee-\xef]/,
-                $._UTF_8_1,
-                $._UTF_8_1,
-            ),
-        ),
-        _UTF_8_4: $ => choice(
-            seq(
-                /\xf0[\x90-\xbf]/,
-                $._UTF_8_1,
-                $._UTF_8_1,
-            ),
-            seq(
-                /[\xf1-\xf3]/,
-                $._UTF_8_1,
-                $._UTF_8_1,
-                $._UTF_8_1,
-            ),
-            seq(
-                /\xf4[\x80-\x8f]/,
-                $._UTF_8_1,
-                $._UTF_8_1,
-            ),
-        ),
-        _UTF_8_ONLY: $ => choice(
-            $._UTF_8_2,
-            $._UTF_8_3,
-            $._UTF_8_4,
-        ),
-        _UTF_8: $ => choice(
-            $._ASCII,
-            $._UTF_8_ONLY,
-        ),
     }
 })

package/package.json

@@ -1,11 +1,14 @@
 {
   "name": "tree-sitter-beancount",
-  "version": "2.3.2",
+  "version": "2.4.0",
   "description": "a tree-sitter parser for the beancount syntax",
   "main": "bindings/node",
+  "types": "bindings/node",
   "scripts": {
     "build": "tree-sitter generate",
-    "test": "tree-sitter test"
+    "test": "tree-sitter test",
+    "install": "node-gyp-build",
+    "prebuildify": "prebuildify --napi --strip"
   },
   "repository": {
     "type": "git",
@@ -15,6 +18,14 @@
     "tree-sitter",
     "beancount"
   ],
+  "files": [
+    "grammar.js",
+    "binding.gyp",
+    "prebuilds/**",
+    "bindings/node/*",
+    "queries/*",
+    "src/**"
+  ],
   "author": "Brian Ryall",
   "license": "MIT",
   "bugs": {
@@ -22,9 +33,28 @@
   },
   "homepage": "https://github.com/bryall/tree-sitter-beancount#readme",
   "dependencies": {
-    "nan": "^2.15.0"
+    "node-addon-api": "^7.1.0",
+    "node-gyp-build": "^4.8.0"
+  },
+  "peerDependencies": {
+    "tree-sitter": "^0.25.0"
+  },
+  "peerDependenciesMeta": {
+    "tree_sitter": {
+      "optional": true
+    }
   },
   "devDependencies": {
-    "tree-sitter-cli": "^0.20.0"
-  }
+    "tree-sitter-cli": "^0.25.0",
+    "prebuildify": "^6.0.0"
+  },
+  "engines": {
+    "node": ">=22.0.0"
+  },
+  "tree-sitter": [
+    {
+      "scope": "source.beancount",
+      "injection-regex": "^beancount$"
+    }
+  ]
 }