feat: new module unicode

add `fixSurrogates` and `unfixSurrogates` functions

Author: 0x04

cli.js

@@ -27,6 +27,8 @@ const usageString = `
     shift
     shiftBits
     toMANS
+    unicode.fixSurrogates
+    unicode.unfixSurrogates
 
   Examples
     $ string-mutilator shift "Hello World!" 6

readme.md

@@ -43,6 +43,9 @@ More or less just for fun.
 <dt><a href="#module_stringMutilator">stringMutilator</a></dt>
 <dd><p>Functions for mutilating strings.</p>
 </dd>
+<dt><a href="#module_stringMutilator/unicode">stringMutilator/unicode</a></dt>
+<dd><p>Functions for handle unicode stuff.</p>
+</dd>
 </dl>
 
 ## Functions
@@ -240,6 +243,39 @@ stringMutilator.compressor.signature('䡥汬漠坯牬搡');
 ## stringMutilator
 Functions for mutilating strings.
 
+<a name="module_stringMutilator/unicode"></a>
+
+## stringMutilator/unicode
+Functions for handle unicode stuff.
+
+
+* [stringMutilator/unicode](#module_stringMutilator/unicode)
+    * [~fixSurrogates(string)](#module_stringMutilator/unicode..fixSurrogates) ⇒ <code>string</code>
+    * [~unfixSurrogates(string)](#module_stringMutilator/unicode..unfixSurrogates) ⇒ <code>string</code>
+
+<a name="module_stringMutilator/unicode..fixSurrogates"></a>
+
+### stringMutilator/unicode~fixSurrogates(string) ⇒ <code>string</code>
+Fix unpaired high/low surrogates by adding a blank high/low surrogate
+(U+D800 or U+DC00) to the required location.
+
+**Kind**: inner method of [<code>stringMutilator/unicode</code>](#module_stringMutilator/unicode)  
+
+| Param | Type | Description |
+| --- | --- | --- |
+| string | <code>string</code> | The input string |
+
+<a name="module_stringMutilator/unicode..unfixSurrogates"></a>
+
+### stringMutilator/unicode~unfixSurrogates(string) ⇒ <code>string</code>
+Remove the by `fixSurrogates` added blank high/low surrogates.
+
+**Kind**: inner method of [<code>stringMutilator/unicode</code>](#module_stringMutilator/unicode)  
+
+| Param | Type | Description |
+| --- | --- | --- |
+| string | <code>string</code> | The input string |
+
 <a name="flipBits"></a>
 
 ## flipBits(string) ⇒ <code>string</code>
@@ -532,6 +568,8 @@ $ string-mutilator --help
       shift
       shiftBits
       toMANS
+      unicode.fixSurrogates
+      unicode.unfixSurrogates
 
     Examples
       $ string-mutilator shift "Hello World!" 6

src/index.js

@@ -16,6 +16,7 @@ import scramble from './scramble';
 import shift from './shift';
 import shiftBits from './shift-bits';
 import toMANS from './to-mans';
+import * as unicode from './unicode';
 
 export default
 {
@@ -31,7 +32,8 @@ export default
   scramble,
   shift,
   shiftBits,
-  toMANS
+  toMANS,
+  unicode
 };
 
 export
@@ -48,5 +50,6 @@ export
   scramble,
   shift,
   shiftBits,
-  toMANS
+  toMANS,
+  unicode
 };

src/unicode.js

@@ -0,0 +1,71 @@
+/**
+ * Functions for handle unicode stuff.
+ * @module stringMutilator/unicode
+ */
+
+/**
+ * Checks if the given `char` is a high surrogate (U+D800 - U+DBFF).
+ * @private
+ * @param {string} char
+ * @returns {boolean}
+ */
+const isHighSurrogate = char => (/^[\uD800-\uDBFF]$/.test(char));
+
+/**
+ * Checks if the given `char` is a low surrogate (U+DC00 - U+DFFF).
+ * @private
+ * @param {string} char
+ * @returns {boolean}
+ */
+const isLowSurrogate = char => (/^[\uDC00-\uDFFF]$/.test(char));
+
+/**
+ * Fix unpaired high/low surrogates by adding a blank high/low surrogate
+ * (U+D800 or U+DC00) to the required location.
+ * @param {string} string The input string
+ * @returns {string}
+ */
+const fixSurrogates = string =>
+{
+  let result = string;
+
+  for (let i = 0; i < result.length; i++)
+  {
+    if (isHighSurrogate(result[i])
+      && (i + 1 === result.length || !isLowSurrogate(result[i + 1])))
+    {
+      result = result.slice(0, i + 1)
+        + String.fromCharCode(0xDC00)
+        + result.slice(i + 1);
+      i++;
+    }
+    else if (isLowSurrogate(result[i])
+      && (i === 0 || !isHighSurrogate(result[i - 1])))
+    {
+      result = result.slice(0, i)
+        + String.fromCharCode(0xD800)
+        + result.slice(i);
+      i++;
+    }
+  }
+
+  return result;
+};
+
+/**
+ * Remove the by `fixSurrogates` added blank high/low surrogates.
+ * @param {string} string The input string
+ * @returns {string}
+ */
+const unfixSurrogates = string =>
+  string.replace(/([\uD800|\uDC00])/g, '');
+
+export default {
+  fixSurrogates,
+  unfixSurrogates
+};
+
+export {
+  fixSurrogates,
+  unfixSurrogates
+};

template-readme.hbs

@@ -112,6 +112,8 @@ $ string-mutilator --help
       shift
       shiftBits
       toMANS
+      unicode.fixSurrogates
+      unicode.unfixSurrogates
 
     Examples
       $ string-mutilator shift "Hello World!" 6

test/unicode.js

@@ -0,0 +1,45 @@
+import test from 'ava';
+import unicode from '../src/unicode.js';
+
+let highSurrogates = Array.from({ length: 1024 }, (v, i) => i + 0xD800);
+let lowSurrogates = Array.from({ length: 1024 }, (v, i) => i + 0xDC00);
+
+let hs = highSurrogates
+  // Exclude U+D800 because we use it as filler
+  .slice(1)
+  .map(code => String.fromCodePoint(code))
+  .join('');
+
+let ls = lowSurrogates
+  // Exclude U+DC00 because we use it as filler
+  .slice(1)
+  .map(code => String.fromCodePoint(code))
+  .join('');
+
+let hsf = unicode.fixSurrogates(hs);
+let hsu = unicode.unfixSurrogates(hsf);
+
+test(
+  `High surrogates fixed length `,
+  t => t.is(hsf.length, hs.length * 2)
+);
+
+test(
+  `High surrogates unfixed length`,
+  t => t.is(hsu.length, hs.length)
+);
+
+let lsf = unicode.fixSurrogates(ls);
+let lsu = unicode.unfixSurrogates(lsf);
+
+test(
+  `Low surrogates fixed length`,
+  t => t.is(lsf.length, ls.length * 2)
+);
+
+test(
+  `Low surrogates unfixed length`,
+  t => t.is(lsu.length, ls.length)
+);
+
+// TODO: Add more detailed tests if necessary