LZCompress now including UTF16

LZCompressor/LZContext.php

@@ -6,12 +6,12 @@ class LZContext
     /**
      * @var array
      */
-    public $dictionary = [];
+    public $dictionary = array();
 
     /**
      * @var array
      */
-    public $dictionaryToCreate = [];
+    public $dictionaryToCreate = array();
 
     /**
      * @var string

LZCompressor/LZString.php

@@ -25,6 +25,18 @@ class LZString
         });
     }
 
+    public static function compressToUTF16($input) {
+        return self::_compress($input, 15, function($a) {
+            return LZUtil16::fromCharCode($a+32);
+        }) . LZUtil16::utf16_chr(32);
+    }
+
+    public static function decompressFromUTF16($input) {
+        return self::_decompress($input, 16384, function($feed, $index) {
+            return LZUtil16::charCodeAt($feed, $index)-32;
+        });
+    }
+
     /**
      * @param string $uncompressed
      * @return string
@@ -273,7 +285,7 @@ class LZString
             }
 
             $result .= $entry;
-            $dictionary->addEntry($w . $entry{0});
+            $dictionary->addEntry($w . LZUtil::utf8_charAt($entry, 0));
             $w = $entry;
 
             $enlargeIn--;

LZCompressor/LZUtil.php

@@ -16,7 +16,7 @@ class LZUtil
     */
     public static $keyStrBase64 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=";
     public static $keyStrUriSafe = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+-$";
-    private static $baseReverseDic = [];
+    private static $baseReverseDic = array();
 
     /**
      * @param string $alphabet
@@ -26,7 +26,7 @@ class LZUtil
     public static function getBaseValue($alphabet, $character)
     {
         if(!array_key_exists($alphabet, self::$baseReverseDic)) {
-            self::$baseReverseDic[$alphabet] = [];
+            self::$baseReverseDic[$alphabet] = array();
             for($i=0; $i<strlen($alphabet); $i++) {
                 self::$baseReverseDic[$alphabet][$alphabet{$i}] = $i;
             }
@@ -107,6 +107,4 @@ class LZUtil
     public static function utf8_strlen($str) {
         return mb_strlen($str, 'UTF-8');
     }
-
-
 }

LZCompressor/LZUtil16.php

@@ -0,0 +1,90 @@
+<?php
+/**
+ * Created by PhpStorm.
+ * User: sics
+ * Date: 27.02.2016
+ * Time: 15:54
+ */
+
+namespace LZCompressor;
+
+
+class LZUtil16
+{
+
+    /**
+     * @return string
+     */
+    public static function fromCharCode()
+    {
+        return array_reduce(func_get_args(), function ($a, $b) {
+            $a .= self::utf16_chr($b);
+            return $a;
+        });
+    }
+
+    /**
+     * Phps chr() equivalent for UTF-16 encoding
+     *
+     * @param int|string $u
+     * @return string
+     */
+    public static function utf16_chr($u)
+    {
+        return mb_convert_encoding('&#' . intval($u) . ';', 'UTF-16', 'HTML-ENTITIES');
+    }
+
+    /**
+     * @param string $str
+     * @param int $num
+     *
+     * @return bool|integer
+     */
+    public static function charCodeAt($str, $num=0)
+    {
+        return self::utf16_ord(self::utf16_charAt($str, $num));
+    }
+
+    /**
+     * @source http://blog.sarabande.jp/post/35970262740
+     * @param string $ch
+     * @return bool|integer
+     */
+    function utf16_ord($ch) {
+        $length = strlen($ch);
+        if (2 === $length) {
+            return hexdec(bin2hex($ch));
+        } else if (4 === $length) {
+            $w1 = $ch[0].$ch[1];
+            $w2 = $ch[2].$ch[3];
+            if ($w1 < "\xD8\x00" || "\xDF\xFF" < $w1 || $w2 < "\xDC\x00" || "\xDF\xFF" < $w2) {
+                return false;
+            }
+            $w1 = (hexdec(bin2hex($w1)) & 0x3ff) << 10;
+            $w2 =  hexdec(bin2hex($w2)) & 0x3ff;
+            return $w1 + $w2 + 0x10000;
+        }
+        return false;
+    }
+
+    /**
+     * @param string $str
+     * @param integer $num
+     *
+     * @return string
+     */
+    public static function utf16_charAt($str, $num)
+    {
+        return mb_substr($str, $num, 1, 'UTF-16');
+    }
+
+    /**
+     * @param string $str
+     * @return integer
+     */
+    public static function utf16_strlen($str)
+    {
+        return mb_strlen($str, 'UTF-16');
+    }
+
+}

lib/file-control-xhr.php

@@ -10,6 +10,7 @@ include(dirname(__FILE__)."/../LZCompressor/LZData.php");
 include(dirname(__FILE__)."/../LZCompressor/LZReverseDictionary.php");
 include(dirname(__FILE__)."/../LZCompressor/LZString.php");
 include(dirname(__FILE__)."/../LZCompressor/LZUtil.php");
+include(dirname(__FILE__)."/../LZCompressor/LZUtil16.php");
 use LZCompressor\LZString as LZString;
 
 // ===============================

lib/file-control.php

@@ -10,6 +10,7 @@ include(dirname(__FILE__)."/../LZCompressor/LZData.php");
 include(dirname(__FILE__)."/../LZCompressor/LZReverseDictionary.php");
 include(dirname(__FILE__)."/../LZCompressor/LZString.php");
 include(dirname(__FILE__)."/../LZCompressor/LZUtil.php");
+include(dirname(__FILE__)."/../LZCompressor/LZUtil16.php");
 use LZCompressor\LZString as LZString;
 ?>
 <?php if ($_SESSION['githubDiff']) { ?>