Ticket #30966: 30966.14.patch

src/wp-admin/js/word-count.js

@@ -14 +14 @@
                 shortcodes = this.settings.l10n.shortcodes;
 
                 if ( shortcodes && shortcodes.length ) {
-                        this.settings.shortcodesRegExp = new RegExp( '\\[\\/?(?:' + shortcodes.join( '|' ) + ')[^\\]]*?\\]', 'gi' );
+                        this.settings.shortcodesRegExp = new RegExp( '\\[\\/?(?:' + shortcodes.join( '|' ) + ')[^\\]]*?\\]', 'g' );
                 }
         }
 
         WordCounter.prototype.settings = {
                 HTMLRegExp: /<\/?[a-z][^>]*?>/gi,
+                HTMLcommentRegExp: /<!--[\s\S]*?-->/g,
                 spaceRegExp: /&nbsp;|&#160;/gi,
-                connectorRegExp: /--|\u2014/gi,
+                HTMLEntityRegExp: /&\S+?;/g,
+                connectorRegExp: /--|\u2014/g,
                 removeRegExp: new RegExp( [
                         '[',
                                 // Basic Latin (extract)
@@ -60 +62 @@
                 astralRegExp: /[\uD800-\uDBFF][\uDC00-\uDFFF]/g,
                 wordsRegExp: /\S\s+/g,
                 charactersRegExp: /\S/g,
-                allRegExp: /[^\f\n\r\t\v\u00ad\u2028\u2029]/g,
+                allRegExp: /[^\f\n\r\t\v\u00AD\u2028\u2029]/g,
                 l10n: window.wordCountL10n || {}
         };
 
@@ -73 +75 @@
                         text = text + '\n';
 
                         text = text.replace( this.settings.HTMLRegExp, '\n' );
+                        text = text.replace( this.settings.HTMLcommentRegExp, '' );
 
                         if ( this.settings.shortcodesRegExp ) {
                                 text = text.replace( this.settings.shortcodesRegExp, '\n' );
@@ -81 +84 @@
                         text = text.replace( this.settings.spaceRegExp, ' ' );
 
                         if ( type === 'words' ) {
+                                text = text.replace( this.settings.HTMLEntityRegExp, '' );
                                 text = text.replace( this.settings.connectorRegExp, ' ' );
                                 text = text.replace( this.settings.removeRegExp, '' );
                         } else {
+                                text = text.replace( this.settings.HTMLEntityRegExp, 'a' );
                                 text = text.replace( this.settings.astralRegExp, 'a' );
                         }
 

tests/qunit/wp-admin/js/word-count.js

@@ -58 +58 @@
                                 words: 1,
                                 characters: 1,
                                 all: 1
+                        },
+                        {
+                                message: 'HTML comment.',
+                                string: 'one<!-- comment -->two three',
+                                words: 2,
+                                characters: 11,
+                                all: 12
+                        },
+                        {
+                                message: 'HTML entity.',
+                                string: '&gt; test',
+                                words: 1,
+                                characters: 5,
+                                all: 6
                         }
                 ], function( test ) {
                         _.each( [ 'words', 'characters', 'all' ], function( type ) {