tree-sitter-sed 0.5.0 → 0.7.0

package/CMakeLists.txt

@@ -2,7 +2,7 @@ cmake_minimum_required(VERSION 3.13)
 
 project(
   tree-sitter-sed
-  VERSION 0.5.0
+  VERSION 0.6.0
   DESCRIPTION "Tree-sitter grammars for POSIX sed."
   HOMEPAGE_URL "https://github.com/konomanoasa/tree-sitter-sed"
   LANGUAGES C

package/Makefile

@@ -1,6 +1,6 @@
 LANGUAGE_NAME := tree-sitter-sed
 HOMEPAGE_URL := https://github.com/konomanoasa/tree-sitter-sed
-VERSION := 0.5.0
+VERSION := 0.6.0
 DESCRIPTION := Tree-sitter grammars for POSIX sed.
 
 PREFIX ?= /usr/local

package/README.md

@@ -21,6 +21,14 @@ This repository contains the following two grammars.
 | `sed`     | BRE    | `tree_sitter_sed()`     |
 | `sed_ere` | ERE    | `tree_sitter_sed_ere()` |
 
+## Development
+
+```sh
+npm install
+npm run parse -- script.sed
+npm run parse:ere -- script.sed
+```
+
 ## Specifications
 
 - [POSIX.1-2024 `sed`](https://pubs.opengroup.org/onlinepubs/9799919799.2024edition/utilities/sed.html)

package/common/grammar.js

@@ -1276,19 +1276,19 @@ function issueDefinitions(mode) {
           },
           {
             reason: "bre_subexpression_left_anchor",
-            outcome: "implementation_option_syntax",
+            outcome: "invalid_syntax",
             rule: ($) => $._regex_bre_subexpression_caret,
           },
           {
             reason: "bre_subexpression_right_anchor",
-            outcome: "implementation_option_syntax",
+            outcome: "invalid_syntax",
             rule: ($) => $._regex_bre_subexpression_dollar,
           },
         ]
       : []),
     {
       reason: "omitted_file_separator",
-      outcome: "implementation_option_syntax",
+      outcome: "invalid_syntax",
       rule: missing("omitted_file_separator"),
     },
     {

package/common/regex.js

@@ -64,24 +64,21 @@ function intervalExpression($, openingName, closingName) {
 function compoundBracketExpression(
   $,
   openingExternal,
-  openingName,
+  marker,
   contentField,
   content,
   closingExternal,
-  closingName,
 ) {
-  const closing = namedExternal($, closingExternal, closingName);
+  const closing = seq(alias(closingExternal, marker), "]");
   return seq(
-    field("opening", namedExternal($, openingExternal, openingName)),
+    alias(openingExternal, "["),
+    marker,
     choice(
       seq(
         field(contentField, content),
         choice(
-          field("closing", closing),
-          seq(
-            issueField($, "malformed_bracket_term"),
-            optional(field("closing", closing)),
-          ),
+          closing,
+          seq(issueField($, "malformed_bracket_term"), optional(closing)),
           issueField($, "incomplete_bracket_term"),
         ),
       ),
@@ -90,7 +87,7 @@ function compoundBracketExpression(
           issueField($, "malformed_bracket_term"),
           issueField($, "incomplete_bracket_term"),
         ),
-        optional(field("closing", closing)),
+        optional(closing),
       ),
     ),
   );
@@ -255,7 +252,7 @@ function bracketRules() {
       compoundBracketExpression(
         $,
         $._regex_open_dot,
-        "open_dot",
+        ".",
         "element",
         choice(
           namedExternal($, $._regex_coll_elem_single, "coll_elem_single"),
@@ -263,32 +260,29 @@ function bracketRules() {
           namedExternal($, $._regex_meta_char, "meta_char"),
         ),
         $._regex_dot_close,
-        "dot_close",
       ),
 
     equivalence_class: ($) =>
       compoundBracketExpression(
         $,
         $._regex_open_equal,
-        "open_equal",
+        "=",
         "element",
         choice(
           namedExternal($, $._regex_coll_elem_single, "coll_elem_single"),
           namedExternal($, $._regex_coll_elem_multi, "coll_elem_multi"),
         ),
         $._regex_equal_close,
-        "equal_close",
       ),
 
     character_class: ($) =>
       compoundBracketExpression(
         $,
         $._regex_open_colon,
-        "open_colon",
+        ":",
         "name",
         namedExternal($, $._regex_class_name, "class_name"),
         $._regex_colon_close,
-        "colon_close",
       ),
   };
 }

package/common/scanner.h

@@ -1234,7 +1234,6 @@ static bool scan_regex_bracket_opener(
   if (!valid_symbols[candidate]) {
     return false;
   }
-  consume(lexer);
   begin_compound_bracket_element(state);
   finish_first_bracket_element(state);
   state->regex_bracket_term_state = term_state;
@@ -1386,11 +1385,10 @@ static bool scan_regex_bracket_term_close(
     return false;
   }
 
-  advance(lexer);
+  consume(lexer);
   if (lexer->lookahead != ']') {
     return false;
   }
-  consume(lexer);
   state->regex_bracket_term_state = REGEX_BRACKET_TERM_NONE;
   finish_compound_bracket_element(state);
   return emit_symbol(valid_symbols, candidate, symbol);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "tree-sitter-sed",
-  "version": "0.5.0",
+  "version": "0.7.0",
   "description": "Tree-sitter grammars for POSIX sed.",
   "license": "MIT",
   "repository": "git+https://github.com/konomanoasa/tree-sitter-sed.git",
@@ -27,13 +27,15 @@
     "check": "biome check . && node scripts/check-c.js && node scripts/check-generated.js && npm test",
     "format": "biome check --write . && node scripts/check-c.js --write",
     "generate": "npm run tree-sitter -- generate-all",
+    "parse": "npm run tree-sitter -- parse --scope source.sed",
+    "parse:ere": "npm run tree-sitter -- parse --scope source.sed.ere",
     "test": "npm run tree-sitter -- test-corpus --rebuild && node --test",
     "test:fuzz": "npm run tree-sitter -- fuzz-all --iterations 100 --edits 5",
     "tree-sitter": "node scripts/tree-sitter.js"
   },
   "devDependencies": {
     "@biomejs/biome": "2.5.11",
-    "tree-sitter-cli": "0.26.13"
+    "tree-sitter-cli": "0.27.0"
   },
   "allowScripts": {
     "tree-sitter-cli": true

package/queries/highlights.scm

@@ -65,11 +65,39 @@
   (back_open_parenthesis)
   (close_bracket_token)
   (closing_brace_token)
-  (colon_close)
   (open_bracket)
-  (open_colon)
 ] @punctuation.bracket
 
+(character_class
+  "[" @punctuation.bracket
+  .
+  ":" @punctuation.delimiter)
+
+(character_class
+  ":" @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
+(collating_symbol
+  "[" @punctuation.bracket
+  .
+  "." @punctuation.delimiter)
+
+(collating_symbol
+  "." @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
+(equivalence_class
+  "[" @punctuation.bracket
+  .
+  "=" @punctuation.delimiter)
+
+(equivalence_class
+  "=" @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
 [
   (nonmatching_list_operator)
   (range_operator)
@@ -85,11 +113,7 @@
   (coll_elem_multi)
   (coll_elem_single)
   (collating_element_token)
-  (dot_close)
-  (equal_close)
   (meta_char)
-  (open_dot)
-  (open_equal)
   (period_token)
 ] @character.special
 

package/sed_ere/queries/highlights.scm

@@ -63,13 +63,41 @@
   (close_bracket_token)
   (close_parenthesis_token)
   (closing_brace_token)
-  (colon_close)
   (open_brace)
   (open_bracket)
-  (open_colon)
   (open_parenthesis)
 ] @punctuation.bracket
 
+(character_class
+  "[" @punctuation.bracket
+  .
+  ":" @punctuation.delimiter)
+
+(character_class
+  ":" @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
+(collating_symbol
+  "[" @punctuation.bracket
+  .
+  "." @punctuation.delimiter)
+
+(collating_symbol
+  "." @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
+(equivalence_class
+  "[" @punctuation.bracket
+  .
+  "=" @punctuation.delimiter)
+
+(equivalence_class
+  "=" @punctuation.delimiter
+  .
+  "]" @punctuation.bracket)
+
 [
   (nonmatching_list_operator)
   (range_operator)
@@ -85,11 +113,7 @@
   (coll_elem_multi)
   (coll_elem_single)
   (collating_element_token)
-  (dot_close)
-  (equal_close)
   (meta_char)
-  (open_dot)
-  (open_equal)
   (period_token)
 ] @character.special