enh(cpp) simpler struct matching

CHANGES.md

@@ -4,9 +4,20 @@ Parser:
 
 - enh(parser) support multi-class matchers (#3081) [Josh Goebel][]
 - enh(parser) Detect comments based on english like text, rather than keyword list [Josh Goebel][]
+- adds `title.class` sub-scope support (#3078) [Josh Goebel][]
+- adds `title.function` sub-scope support (#3078) [Josh Goebel][]
+- adds `beforeMatch` compiler extension (#3078) [Josh Goebel][]
 
 Grammars:
 
+- enh(java) Simplified class-like matcher (#3078) [Josh Goebel][]
+- enh(cpp) Simplified class-like matcher (#3078) [Josh Goebel][]
+- enh(rust) Simplified class-like matcher (#3078) [Josh Goebel][]
+- enh(actionscript) Simplified class-like matcher (#3078) [Josh Goebel][]
+- enh(arcade) `function.title` => `title.function` (#3078) [Josh Goebel][]
+- enh(autoit) `function.title` => `title.function` (#3078) [Josh Goebel][]
+- enh(c) `function.title` => `title.function` (#3078) [Josh Goebel][]
+- enh(rust) support function invoke and `impl` (#3078) [Josh Goebel][]
 - chore(properties) disable auto-detection #3102 [Josh Goebel][]
 - fix(properties) fix incorrect handling of non-alphanumeric keys #3102 [Egor Rogov][]
 - enh(java) support functions with nested template types (#2641) [Josh Goebel][]

docs/mode-reference.rst

@@ -250,6 +250,29 @@ This callback is triggered the moment an end match is detected. ``matchData`` in
 For an example of usage see ``END_SAME_AS_BEGIN`` in ``modes.js``.
 
 
+beforeMatch
+^^^^^^^^^^^
+
+- **type**: string
+
+Used to qualify a match by the content that immediately precedes it.  This is syntactic sugar that generates a much more complex mode that first matches the entire sequence (using look-ahead), then glosses over the ``beforeMatch`` portion and ``starts`` a new rule to handle the actual match.
+
+Notes:
+
+- Any ``keywords`` specified are applied to the ``beforeMatch`` text as well (as shown in the example below)
+- Do not get this confused with any type of look-behind.  We are always looking forward.
+- This is incompatible with ``starts``.
+
+::
+
+  {
+    beforeMatch: /\b(enum|class|struct|union)\s+/,
+    keywords: "enum class struct union",
+    begin: /\w+/,
+    className: "title.class"
+  }
+
+
 beginKeywords
 ^^^^^^^^^^^^^
 

src/highlight.js

@@ -252,7 +252,8 @@ const HLJS = function(hljs) {
      */
     function emitMultiClass(mode, match) {
       let i = 1;
-      while (match[i]) {
+      // eslint-disable-next-line no-undefined
+      while (match[i] !== undefined) {
         const klass = language.classNameAliases[mode.className[i]] || mode.className[i];
         const text = match[i];
         if (klass) { emitter.addKeyword(text, klass); } else {
@@ -452,7 +453,7 @@ const HLJS = function(hljs) {
         modeBuffer += codeToHighlight.slice(match.index, match.index + 1);
         if (!SAFE_MODE) {
           /** @type {AnnotatedError} */
-          const err = new Error('0 width match regex');
+          const err = new Error(`0 width match regex (${languageName})`);
           err.languageName = languageName;
           err.badRule = lastMatch.rule;
           throw err;

src/languages/actionscript.js

@@ -10,6 +10,10 @@ import * as regex from '../lib/regex.js';
 /** @type LanguageFn */
 export default function(hljs) {
   const IDENT_RE = /[a-zA-Z_$][a-zA-Z0-9_$]*/;
+  const PKG_NAME_RE = regex.concat(
+    IDENT_RE,
+    regex.concat("(\\.", IDENT_RE, ")*")
+  );
   const IDENT_FUNC_RETURN_TYPE_RE = /([*]|[a-zA-Z_$][a-zA-Z0-9_$]*)/;
 
   const AS3_REST_ARG_MODE = {
@@ -92,20 +96,16 @@ export default function(hljs) {
       hljs.C_BLOCK_COMMENT_MODE,
       hljs.C_NUMBER_MODE,
       {
-        className: 'class',
-        beginKeywords: 'package',
-        end: /\{/,
-        contains: [ hljs.TITLE_MODE ]
+        beforeMatch: /\b(package)\s+/,
+        keywords: "package",
+        match: PKG_NAME_RE,
+        className: "title.class"
       },
       {
-        className: 'class',
-        beginKeywords: 'class interface',
-        end: /\{/,
-        excludeEnd: true,
-        contains: [
-          { beginKeywords: 'extends implements' },
-          hljs.TITLE_MODE
-        ]
+        beforeMatch: /\b(class|interface|extends|implements)\s+/,
+        keywords: "class interface extends implements",
+        match: IDENT_RE,
+        className: "title.class"
       },
       {
         className: 'meta',
@@ -114,13 +114,12 @@ export default function(hljs) {
         keywords: { 'meta-keyword': 'import include' }
       },
       {
-        className: 'function',
         beginKeywords: 'function',
         end: /[{;]/,
         excludeEnd: true,
         illegal: /\S/,
         contains: [
-          hljs.TITLE_MODE,
+          hljs.inherit(hljs.TITLE_MODE, { className: "title.function" }),
           {
             className: 'params',
             begin: /\(/,

src/languages/arcade.js

@@ -135,12 +135,12 @@ export default function(hljs) {
         relevance: 0
       },
       {
-        className: 'function',
         beginKeywords: 'function',
         end: /\{/,
         excludeEnd: true,
         contains: [
           hljs.inherit(hljs.TITLE_MODE, {
+            className: "title.function",
             begin: IDENT_RE
           }),
           {

src/languages/autoit.js

@@ -29,7 +29,7 @@ export default function(hljs) {
     "Tidy_On",
     "Tidy_Parameters"
   ]
-  
+
   const LITERAL = 'True False And Null Not Or Default';
 
   const BUILT_IN
@@ -140,12 +140,11 @@ export default function(hljs) {
   };
 
   const FUNCTION = {
-    className: 'function',
     beginKeywords: 'Func',
     end: '$',
     illegal: '\\$|\\[|%',
     contains: [
-      hljs.UNDERSCORE_TITLE_MODE,
+      hljs.inherit(hljs.UNDERSCORE_TITLE_MODE, { className: "title.function" }),
       {
         className: 'params',
         begin: '\\(',

src/languages/c.js

@@ -235,7 +235,6 @@ export default function(hljs) {
   };
 
   const FUNCTION_DECLARATION = {
-    className: 'function',
     begin: '(' + FUNCTION_TYPE_RE + '[\\*&\\s]+)+' + FUNCTION_TITLE,
     returnBegin: true,
     end: /[{;=]/,
@@ -251,7 +250,9 @@ export default function(hljs) {
       {
         begin: FUNCTION_TITLE,
         returnBegin: true,
-        contains: [ TITLE_MODE ],
+        contains: [
+          hljs.inherit(TITLE_MODE, { className: "title.function" })
+        ],
         relevance: 0
       },
       {

src/languages/cpp.js

@@ -21,7 +21,7 @@ export default function(hljs) {
   const DECLTYPE_AUTO_RE = 'decltype\\(auto\\)';
   const NAMESPACE_RE = '[a-zA-Z_]\\w*::';
   const TEMPLATE_ARGUMENT_RE = '<[^<>]+>';
-  const FUNCTION_TYPE_RE = '(' +
+  const FUNCTION_TYPE_RE = '(?!struct)(' +
     DECLTYPE_AUTO_RE + '|' +
     regex.optional(NAMESPACE_RE) +
     '[a-zA-Z_]\\w*' + regex.optional(TEMPLATE_ARGUMENT_RE) +
@@ -233,7 +233,7 @@ export default function(hljs) {
       'atomic_bool atomic_char atomic_schar ' +
       'atomic_uchar atomic_short atomic_ushort atomic_int atomic_uint atomic_long atomic_ulong atomic_llong ' +
       'atomic_ullong new throw return ' +
-      'and and_eq bitand bitor compl not not_eq or or_eq xor xor_eq',
+      'and and_eq bitand bitor compl not not_eq or or_eq xor xor_eq struct',
     built_in: '_Bool _Complex _Imaginary',
     _relevance_hints: COMMON_CPP_HINTS,
     literal: 'true false nullptr NULL'
@@ -403,15 +403,10 @@ export default function(hljs) {
           keywords: CPP_KEYWORDS
         },
         {
-          className: 'class',
-          beginKeywords: 'enum class struct union',
-          end: /[{;:<>=]/,
-          contains: [
-            {
-              beginKeywords: "final class struct"
-            },
-            hljs.TITLE_MODE
-          ]
+          beforeMatch: /\b(enum|class|struct|union)\s+/,
+          keywords: "enum class struct union",
+          match: /\w+/,
+          className: "title.class"
         }
       ])
   };