enh(parser) keyword relevance has a limit

CHANGES.md

@@ -66,6 +66,7 @@ Dev Improvements:
 
 Parser:
 
+- keywords now have a maximum # of times they provide relevance (#3129) [Josh Goebel][]
 - enh(api) add `unregisterLanguage` method (#3009) [Antoine du Hamel][]
 - enh: Make alias registration case insensitive (#3026) [David Ostrovsky][]
 - fix(parser) `highlightAll()` now works if the library is lazy loaded [Josh Goebel][]

src/highlight.js

@@ -19,6 +19,7 @@ import * as logger from "./lib/logger.js";
 const escape = utils.escapeHTML;
 const inherit = utils.inherit;
 const NO_MATCH = Symbol("nomatch");
+const MAX_KEYWORD_HITS = 7;
 
 /**
  * @param {any} hljs - object that is extended (legacy)
@@ -162,15 +163,16 @@ const HLJS = function(hljs) {
    * @returns {HighlightResult} - result of the highlight operation
   */
   function _highlight(languageName, codeToHighlight, ignoreIllegals, continuation) {
+    const keywordHits = Object.create(null);
+
     /**
      * Return keyword data if a match is a keyword
      * @param {CompiledMode} mode - current mode
-     * @param {RegExpMatchArray} match - regexp match data
+     * @param {string} matchText - the textual match
      * @returns {KeywordData | false}
      */
-    function keywordData(mode, match) {
-      const matchText = language.case_insensitive ? match[0].toLowerCase() : match[0];
-      return Object.prototype.hasOwnProperty.call(mode.keywords, matchText) && mode.keywords[matchText];
+    function keywordData(mode, matchText) {
+      return mode.keywords[matchText];
     }
 
     function processKeywords() {
@@ -186,13 +188,15 @@ const HLJS = function(hljs) {
 
       while (match) {
         buf += modeBuffer.substring(lastIndex, match.index);
-        const data = keywordData(top, match);
+        const word = language.case_insensitive ? match[0].toLowerCase() : match[0];
+        const data = keywordData(top, word);
         if (data) {
           const [kind, keywordRelevance] = data;
           emitter.addText(buf);
           buf = "";
 
-          relevance += keywordRelevance;
+          keywordHits[word] = (keywordHits[word] || 0) + 1;
+          if (keywordHits[word] <= MAX_KEYWORD_HITS) relevance += keywordRelevance;
           if (kind.startsWith("_")) {
             // _ implied for relevance only, do not highlight
             // by applying a class name

src/lib/compile_keywords.js

@@ -23,7 +23,7 @@ const DEFAULT_KEYWORD_CLASSNAME = "keyword";
  */
 export function compileKeywords(rawKeywords, caseInsensitive, className = DEFAULT_KEYWORD_CLASSNAME) {
   /** @type KeywordDict */
-  const compiledKeywords = {};
+  const compiledKeywords = Object.create(null);
 
   // input can be a string of keywords, an array of keywords, or a object with
   // named keys representing className (which can then point to a string or array)

test/parser/index.js

@@ -8,4 +8,5 @@ describe('hljs', function() {
   require('./reuse-endsWithParent');
   require('./should-not-destroyData');
   require('./compiler-extensions');
+  require('./max_keyword_hits');
 });

test/parser/max_keyword_hits.js

@@ -0,0 +1,19 @@
+const hljs = require('../../build');
+
+describe("max keyword hits", function() {
+  it("should count a keyword 7 times for relevance, no more", () => {
+    hljs.registerLanguage('test-language', (hljs) => {
+      return {
+        keywords: "bob suzy|2"
+      };
+    });
+
+    let result = hljs.highlight('bob bob bob bob bob bob bob bob bob bob bob bob bob', { language: 'test-language' });
+    result.relevance.should.equal(7);
+
+    result = hljs.highlight('suzy suzy suzy suzy suzy suzy suzy suzy suzy suzy suzy suzy suzy', { language: 'test-language' });
+    result.relevance.should.equal(14);
+
+    hljs.unregisterLanguage("test-language");
+  });
+});