Fix regex look-ahead for end matchers

CHANGES.md

@@ -7,6 +7,7 @@ New styles:
   none.
 
 Improvements:
+- fix(parser): Look-ahead regex now work for end matches also (#2237) [Josh Goebel][]
 - fix(parser): Better errors when a language is missing (#2236) [Josh Goebel][]
 
 [Josh Goebel]: https://github.com/yyyc514

docs/language-guide.rst

@@ -265,13 +265,14 @@ The goal of Highlight.js is to support whatever regex features Javascript itself
 
 Things we support now that we did not always:
 
-* look-ahead matching for `begin` (#2135)
-* look-ahead matching for `illegal` (#2135)
-* back-references within your regex (#1897)
+* look-ahead regex matching for `begin` (#2135)
+* look-ahead regex matching for `end` (#2237)
+* look-ahead regex matching for `illegal` (#2135)
+* back-references within your regex matches (#1897)
+* look-behind matching (when JS supports it) for `begin` (#2135)
 
 Things we currently know are still issues:
 
-* look-ahead matching for `end` matchers
 * look-behind matching (when JS supports it) for `end` matchers
 
 

src/highlight.js

@@ -614,7 +614,8 @@ https://highlightjs.org/
 
     function doEndMatch(match) {
       var lexeme = match[0];
-      var end_mode = endOfMode(top, lexeme);
+      var matchPlusRemainder = value.substr(match.index);
+      var end_mode = endOfMode(top, matchPlusRemainder);
       if (!end_mode) { return; }
 
       var origin = top;

test/parser/index.js

@@ -3,4 +3,5 @@
 describe('hljs', function() {
   require('./reuse-endsWithParent');
   require('./should-not-destroyData');
+  require('./look-ahead-end-matchers');
 });

test/parser/look-ahead-end-matchers.js

@@ -0,0 +1,31 @@
+const hljs = require('../../build');
+
+describe("parser specifics", function () {
+
+  // CONTEXT: https://github.com/highlightjs/highlight.js/pull/2219
+  describe("a grammar with look-ahead end matchers", () => {
+    it("should match successfully", () => {
+      hljs.registerLanguage('test-language', (hljs) => {
+
+        // broken regex from old Fortran ruleset
+        const PATTERN = {
+          className: "pattern",
+          begin: '[A-Z]{3}',
+          // followed by at least one space
+          end: '\\d{3}(?=\\s+)'
+        }
+
+        return {
+          contains: [PATTERN]
+        };
+      });
+
+      hljs.highlight('test-language', 'ABC123 is the secret. XYZ123. End of string: ABC123').value
+        .should.equal(
+        // one full match at beginning, other match begins with XYZ but then never terminates,
+        // so the end of the parsing finally closes the span tag
+        '<span class="hljs-pattern">ABC123</span> is the secret. <span class="hljs-pattern">XYZ123. End of string: ABC123</span>'
+      )
+    })
+  })
+})