pref: rework unesc for a 63+% performance boost (#239)

* In profiling postcss I found that a significant amount of time was being
spent in [`unesc`](https://github.com/postcss/postcss-selector-parser/commits/master/src/util/unesc.js), this was due to the expensive regex checks that were
being performed on the fly for every selector in the codebase which looked to be performing quite poorly inside of modern node and v8.

![image](https://user-images.githubusercontent.com/883126/114136698-fdd98a80-98bf-11eb-8068-ace4f6f2274d.png)

----

As an experiment and based on some prior experience with this class of slowdown I migrated the implementation to one that performs a scan through the string instead of running a regex replace. By testing this on my local application I instantly saw the work from this function go from > 900 ms to ~100ms.

![image](https://user-images.githubusercontent.com/883126/114136734-0c27a680-98c0-11eb-82ab-f0c9529fd32d.png)

This implementation passes all of the existing test cases and aims to mirror the prior implementation's implementation details :)

-----

Based on my application I am seeing the major wins come from purgecss dropping my total application build by multiple seconds! 🔥

* Expand unesc handling to correctly handle spec edgecase for lone
surrogates and out of bound codepoint values.

Author: samccone

src/__tests__/classes.js

@@ -264,3 +264,4 @@ test('class selector with escaping (36)', '.not-pseudo\\:\\:focus', (t, tree) =>
     t.deepEqual(tree.nodes[0].nodes[0].type, 'class');
     t.deepEqual(tree.nodes[0].nodes[0].raws.value, 'not-pseudo\\:\\:focus');
 });
+

src/__tests__/util/unesc.js

@@ -0,0 +1,62 @@
+import {test} from '../util/helpers';
+
+test('id selector', '#foo', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'foo');
+});
+
+test('escaped special char', '#w\\+', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'w+');
+});
+
+test('tailing escape', '#foo\\', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'foo\\');
+});
+
+test('double escape', '#wow\\\\k', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'wow\\k');
+});
+
+test('leading numeric', '.\\31 23', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, '123');
+});
+
+test('emoji', '.\\🐐', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, '🐐');
+});
+
+// https://www.w3.org/International/questions/qa-escapes#cssescapes
+test('hex escape', '.\\E9motion', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'émotion');
+});
+
+test('hex escape with space', '.\\E9 dition', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'édition');
+});
+
+test('hex escape with hex number', '.\\0000E9dition', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'édition');
+});
+
+test('class selector with escaping', '.\\1D306', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, '𝌆');
+});
+
+test('class selector with escaping with more chars', '.\\1D306k', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, '𝌆k');
+});
+
+test('class selector with escaping with more chars with whitespace', '.wow\\1D306 k', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, 'wow𝌆k');
+});
+
+test('handles 0 value hex', '\\0', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, String.fromCodePoint(0xFFFD));
+});
+
+test('handles lone surrogate value hex', '\\DBFF', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, String.fromCodePoint(0xFFFD));
+});
+
+test('handles out of bound values', '\\110000', (t, tree) => {
+    t.deepEqual(tree.nodes[0].nodes[0].value, String.fromCodePoint(0xFFFD));
+});

src/util/unesc.js

@@ -1,19 +1,82 @@
-const whitespace = '[\\x20\\t\\r\\n\\f]';
-const unescapeRegExp = new RegExp('\\\\([\\da-f]{1,6}' + whitespace + '?|(' + whitespace + ')|.)', 'ig');
+// Many thanks for this post which made this migration much easier.
+// https://mathiasbynens.be/notes/css-escapes
+
+/**
+ * 
+ * @param {string} str 
+ * @returns {[string, number]|undefined}
+ */
+function gobbleHex (str) {
+    const lower = str.toLowerCase();
+    let hex = '';
+    let spaceTerminated = false;
+    for (let i = 0; i < 6 && lower[i] !== undefined; i++) {
+        const code =  lower.charCodeAt(i);
+        // check to see if we are dealing with a valid hex char [a-f|0-9]
+        const valid = (code >= 97 && code <= 102) || (code >= 48 && code <= 57);
+        // https://drafts.csswg.org/css-syntax/#consume-escaped-code-point
+        spaceTerminated = code === 32;
+        if (!valid) {
+            break;
+        }
+        hex += lower[i];
+    }
+
+    if (hex.length === 0) {
+        return undefined;
+    }
+    const codePoint = parseInt(hex, 16);
+
+    const isSurrogate = codePoint >= 0xD800 && codePoint <= 0xDFFF;
+    // Add special case for
+    // "If this number is zero, or is for a surrogate, or is greater than the maximum allowed code point"
+    // https://drafts.csswg.org/css-syntax/#maximum-allowed-code-point
+    if (isSurrogate || codePoint === 0x0000 || codePoint > 0x10FFFF) {
+        return ['\uFFFD', hex.length + (spaceTerminated ? 1 : 0)];
+    }
+
+    return [
+        String.fromCodePoint(codePoint),
+        hex.length + (spaceTerminated ? 1 : 0),
+    ];
+}
+
+const CONTAINS_ESCAPE = /\\/;
 
 export default function unesc (str) {
-    return str.replace(unescapeRegExp, (_, escaped, escapedWhitespace) => {
-        const high = '0x' + escaped - 0x10000;
-
-        // NaN means non-codepoint
-        // Workaround erroneous numeric interpretation of +"0x"
-        // eslint-disable-next-line no-self-compare
-        return high !== high || escapedWhitespace
-            ? escaped
-            : high < 0
-                ? // BMP codepoint
-                String.fromCharCode(high + 0x10000)
-                : // Supplemental Plane codepoint (surrogate pair)
-                String.fromCharCode((high >> 10) | 0xd800, (high & 0x3ff) | 0xdc00);
-    });
+    let needToProcess = CONTAINS_ESCAPE.test(str);
+    if (!needToProcess) {
+        return str;
+    }
+    let ret = "";
+
+    for (let i = 0; i < str.length; i++) {
+        if ((str[i] === "\\")) {
+            const gobbled = gobbleHex(str.slice(i + 1, i + 7));
+            if (gobbled !== undefined) {
+                ret += gobbled[0];
+                i += gobbled[1];
+                continue;
+            }
+
+            // Retain a pair of \\ if double escaped `\\\\`
+            // https://github.com/postcss/postcss-selector-parser/commit/268c9a7656fb53f543dc620aa5b73a30ec3ff20e
+            if (str[i + 1] === "\\") {
+                ret += "\\";
+                i++;
+                continue;
+            }
+
+            // if \\ is at the end of the string retain it
+            // https://github.com/postcss/postcss-selector-parser/commit/01a6b346e3612ce1ab20219acc26abdc259ccefb
+            if (str.length === i + 1) {
+                ret += str[i];
+            }
+            continue;
+        }
+      
+        ret += str[i];
+    }
+
+    return ret;
 }