Speed up both encode() and decode() methods.

Author: mdevils

CHANGELOG.md

@@ -1,3 +1,8 @@
+2.1.1
+-----
+
+ * Speed up both `encode()` and `decode()` methods.
+
 2.1.0
 -----
 

README.md

@@ -97,68 +97,68 @@ Common
 
     Initialization / Load speed
 
-      * #1: html-entities x 2,992,640 ops/sec ±1.76% (82 runs sampled)
-        #2: entities x 2,093,859 ops/sec ±1.17% (82 runs sampled)
-        #3: he x 1,815,445 ops/sec ±1.30% (87 runs sampled)
+      * #1: html-entities x 2,941,745 ops/sec ±1.87% (81 runs sampled)
+        #2: entities x 2,061,661 ops/sec ±1.16% (82 runs sampled)
+        #3: he x 1,861,758 ops/sec ±1.15% (86 runs sampled)
 
 HTML5
 
     Encode test
 
-      * #1: html-entities.encode - html5, nonAsciiPrintable x 427,051 ops/sec ±0.25% (96 runs sampled)
-      * #2: html-entities.encode - html5, nonAscii x 427,332 ops/sec ±0.68% (96 runs sampled)
-        #3: entities.encodeNonAsciiHTML x 333,348 ops/sec ±1.08% (93 runs sampled)
-      * #4: html-entities.encode - html5, extensive x 269,630 ops/sec ±0.26% (98 runs sampled)
-        #5: entities.encodeHTML x 126,117 ops/sec ±0.27% (93 runs sampled)
-        #6: he.encode x 114,119 ops/sec ±0.20% (96 runs sampled)
+      * #1: html-entities.encode - html5, nonAscii x 439,350 ops/sec ±0.21% (96 runs sampled)
+      * #2: html-entities.encode - html5, nonAsciiPrintable x 410,462 ops/sec ±0.22% (93 runs sampled)
+        #3: entities.encodeNonAsciiHTML x 332,966 ops/sec ±0.54% (92 runs sampled)
+      * #4: html-entities.encode - html5, extensive x 280,865 ops/sec ±0.22% (95 runs sampled)
+        #5: entities.encodeHTML x 125,338 ops/sec ±0.30% (92 runs sampled)
+        #6: he.encode x 112,572 ops/sec ±0.25% (97 runs sampled)
 
     Decode test
 
-      * #1: html-entities.decode - html5, strict x 347,055 ops/sec ±0.27% (94 runs sampled)
-      * #2: html-entities.decode - html5, attribute x 340,751 ops/sec ±0.22% (97 runs sampled)
-      * #3: html-entities.decode - html5, body x 333,538 ops/sec ±0.28% (94 runs sampled)
-        #4: entities.decodeHTMLStrict x 329,206 ops/sec ±1.64% (92 runs sampled)
-        #5: entities.decodeHTML x 278,862 ops/sec ±0.24% (97 runs sampled)
-        #6: he.decode x 185,834 ops/sec ±0.23% (96 runs sampled)
+      * #1: html-entities.decode - html5, body x 428,051 ops/sec ±0.22% (98 runs sampled)
+      * #2: html-entities.decode - html5, strict x 402,821 ops/sec ±0.22% (91 runs sampled)
+      * #3: html-entities.decode - html5, attribute x 391,007 ops/sec ±0.33% (90 runs sampled)
+        #4: entities.decodeHTMLStrict x 332,909 ops/sec ±0.56% (95 runs sampled)
+        #5: entities.decodeHTML x 274,700 ops/sec ±0.29% (97 runs sampled)
+        #6: he.decode x 184,440 ops/sec ±0.27% (95 runs sampled)
 
 HTML4
 
     Encode test
 
-      * #1: html-entities.encode - html4, nonAscii x 413,667 ops/sec ±0.51% (94 runs sampled)
-      * #2: html-entities.encode - html4, nonAsciiPrintable x 390,540 ops/sec ±0.39% (95 runs sampled)
-      * #3: html-entities.encode - html4, extensive x 199,258 ops/sec ±0.20% (97 runs sampled)
+      * #1: html-entities.encode - html4, nonAscii x 419,600 ops/sec ±0.65% (94 runs sampled)
+      * #2: html-entities.encode - html4, nonAsciiPrintable x 413,954 ops/sec ±0.83% (91 runs sampled)
+      * #3: html-entities.encode - html4, extensive x 216,838 ops/sec ±0.22% (96 runs sampled)
 
     Decode test
 
-      * #1: html-entities.decode - html4, strict x 369,977 ops/sec ±1.13% (93 runs sampled)
-      * #2: html-entities.decode - html4, body x 366,084 ops/sec ±0.30% (94 runs sampled)
-      * #3: html-entities.decode - html4, attribute x 363,317 ops/sec ±0.33% (94 runs sampled)
+      * #1: html-entities.decode - html4, strict x 420,850 ops/sec ±0.23% (92 runs sampled)
+      * #2: html-entities.decode - html4, body x 413,042 ops/sec ±0.49% (94 runs sampled)
+      * #3: html-entities.decode - html4, attribute x 408,538 ops/sec ±2.59% (92 runs sampled)
 
 XML
 
     Encode test
 
-      * #1: html-entities.encode - xml, nonAscii x 478,394 ops/sec ±2.54% (92 runs sampled)
-      * #2: html-entities.encode - xml, nonAsciiPrintable x 459,013 ops/sec ±0.20% (97 runs sampled)
-        #3: entities.encodeXML x 352,570 ops/sec ±1.05% (93 runs sampled)
-      * #4: html-entities.encode - xml, extensive x 269,313 ops/sec ±0.24% (92 runs sampled)
+      * #1: html-entities.encode - xml, nonAscii x 511,788 ops/sec ±0.21% (97 runs sampled)
+      * #2: html-entities.encode - xml, nonAsciiPrintable x 482,136 ops/sec ±0.40% (93 runs sampled)
+        #3: entities.encodeXML x 353,189 ops/sec ±0.57% (95 runs sampled)
+      * #4: html-entities.encode - xml, extensive x 291,091 ops/sec ±0.23% (96 runs sampled)
 
     Decode test
 
-      * #1: html-entities.decode - xml, body x 429,601 ops/sec ±0.20% (96 runs sampled)
-      * #2: html-entities.decode - xml, strict x 428,820 ops/sec ±0.22% (96 runs sampled)
-        #3: entities.decodeXML x 423,011 ops/sec ±0.28% (94 runs sampled)
-      * #4: html-entities.decode - xml, attribute x 419,337 ops/sec ±0.66% (94 runs sampled)
+      * #1: html-entities.decode - xml, body x 543,327 ops/sec ±0.25% (89 runs sampled)
+      * #2: html-entities.decode - xml, attribute x 533,470 ops/sec ±0.22% (94 runs sampled)
+      * #3: html-entities.decode - xml, strict x 528,014 ops/sec ±2.27% (95 runs sampled)
+        #4: entities.decodeXML x 421,154 ops/sec ±0.32% (96 runs sampled)
 
 Escaping
 
     Escape test
 
-        #1: he.escape x 1,126,149 ops/sec ±0.23% (98 runs sampled)
-      * #2: html-entities.encode - xml, specialChars x 1,077,095 ops/sec ±1.09% (94 runs sampled)
-        #3: entities.escapeUTF8 x 724,973 ops/sec ±0.25% (98 runs sampled)
-        #4: entities.escape x 316,363 ops/sec ±0.20% (97 runs sampled)
+      * #1: html-entities.encode - xml, specialChars x 1,583,074 ops/sec ±0.24% (95 runs sampled)
+        #2: he.escape x 1,131,879 ops/sec ±1.65% (94 runs sampled)
+        #3: entities.escapeUTF8 x 736,205 ops/sec ±0.28% (94 runs sampled)
+        #4: entities.escape x 314,225 ops/sec ±0.24% (93 runs sampled)
 ```
 
 License

src/index.ts

@@ -46,17 +46,42 @@ export function encode(
     if (!text) {
         return '';
     }
+
+    const encodeRegExp = encodeRegExps[mode];
+    encodeRegExp.lastIndex = 0;
+
+    let match = encodeRegExp.exec(text);
+
+    if (!match) {
+        return text;
+    }
+
     const references = allNamedReferences[level].characters;
     const isHex = numeric === 'hexadecimal';
 
-    return text.replace(encodeRegExps[mode], function (input) {
+    let lastIndex = 0;
+    let result = '';
+
+    do {
+        if (lastIndex !== match.index) {
+            result += text.substring(lastIndex, match.index);
+        }
+        const input = match[0];
         const entity = references[input];
         if (entity) {
-            return entity;
+            result += entity;
+        } else {
+            const code = input.length > 1 ? getCodePoint(input, 0)! : input.charCodeAt(0);
+            result += (isHex ? '&#x' + code.toString(16) : '&#' + code) + ';';
         }
-        const code = input.length > 1 ? getCodePoint(input, 0)! : input.charCodeAt(0);
-        return (isHex ? '&#x' + code.toString(16) : '&#' + code) + ';';
-    });
+        lastIndex = match.index + input.length;
+    } while ((match = encodeRegExp.exec(text)));
+
+    if (lastIndex !== text.length) {
+        result += text.substring(lastIndex, text.length);
+    }
+
+    return result;
 }
 
 const defaultDecodeOptions: DecodeOptions = {
@@ -100,24 +125,48 @@ export function decode(
     if (!text) {
         return '';
     }
+    const decodeRegExp = decodeRegExps[level][scope];
+
+    let match = decodeRegExp.exec(text);
+
+    if (!match) {
+        return text;
+    }
+
     const references = allNamedReferences[level].entities;
     const isAttribute = scope === 'attribute';
 
-    return text.replace(decodeRegExps[level][scope], function (entity) {
-        if (isAttribute && entity[entity.length - 1] === '=') {
-            return entity;
+    let lastIndex = 0;
+    let result = '';
+
+    do {
+        const entity = match[0];
+        if (lastIndex !== match.index) {
+            result += text.substring(lastIndex, match.index);
         }
-        if (entity[1] != '#') {
-            return references[entity] || entity;
+        if (isAttribute && entity[entity.length - 1] === '=') {
+            result += entity;
+        } else if (entity[1] != '#') {
+            result += references[entity] || entity;
+        } else {
+            const secondChar = entity[2];
+            const code =
+                secondChar == 'x' || secondChar == 'X' ? parseInt(entity.substr(3), 16) : parseInt(entity.substr(2));
+
+            result +=
+                code >= 0x10ffff
+                    ? outOfBoundsChar
+                    : code > 65535
+                    ? fromCodePoint(code)
+                    : fromCharCode(numericUnicodeMap[code] || code);
         }
-        const secondChar = entity[2];
-        const code =
-            secondChar == 'x' || secondChar == 'X' ? parseInt(entity.substr(3), 16) : parseInt(entity.substr(2));
-
-        return code >= 0x10ffff
-            ? outOfBoundsChar
-            : code > 65535
-            ? fromCodePoint(code)
-            : fromCharCode(numericUnicodeMap[code] || code);
-    });
+
+        lastIndex = match.index + entity.length;
+    } while ((match = decodeRegExp.exec(text)));
+
+    if (lastIndex !== text.length) {
+        result += text.substring(lastIndex, text.length);
+    }
+
+    return result;
 }

test/index.test.ts

@@ -19,6 +19,9 @@ describe('encode()', () => {
             expect(encode('a\n<>"\'&©∆℞😂\0\x01', {mode: 'specialChars'})).to.equal(
                 'a\n&lt;&gt;&quot;&apos;&amp;©∆℞😂\0\x01'
             );
+            expect(encode('a\n<>"\'&©∆℞😂\0\x01END', {mode: 'specialChars'})).to.equal(
+                'a\n&lt;&gt;&quot;&apos;&amp;©∆℞😂\0\x01END'
+            );
             expect(encode('a\n<>"\'&©∆℞😂\0\x01', {mode: 'nonAscii'})).to.equal(
                 'a\n&lt;&gt;&quot;&apos;&amp;&copy;&#8710;&rx;&#128514;\0\x01'
             );