added utf16 functionality (compressToUTF16 / decompressFromUTF16)

Tobias Seipke · Tobias Seipke · commit b08fc07df528 · 2016-03-23T10:49:56.000+01:00
diff --git a/README.md b/README.md
@@ -24,6 +24,9 @@ composer require nullpunkt/lz-string-php
 ```
 
 ## Changelog
+### 2016-03-23 
+- v1.2.0 Added utf16 functionality
+
 ### 2016-02-28 
 - v1.1.0 Completely rewritten LZString component to match the output of js-lz-string version 1.4.4
 - PHPUnit tests for continuous testing / comparision of lz-string js
diff --git a/boilerplate/README.md b/boilerplate/README.md
@@ -0,0 +1,9 @@
+lz-string-php boilerplate
+=========================
+
+HTML Boilerplate to test the php implementiation of lz-string manually 
+
+## Usage
+```cmd
+php -S localhost:8000
+```
diff --git a/boilerplate/index.php b/boilerplate/index.php
@@ -15,7 +15,6 @@
         }
     </style>
     <link rel="stylesheet" href="bower_components/bootstrap/dist/css/bootstrap-theme.min.css">
-
 </head>
 <body>
 
@@ -30,52 +29,86 @@
                 </div>
                 <button type="submit" class="btn btn-primary" data-ng-click="vm.encode()">Encode!</button>
                 &nbsp;&nbsp;&nbsp;&nbsp;
-                <input id="display-compressed" type="checkbox" class="form-control" data-ng-model="vm.displayCompressed">
-                <label for="display-compressed">Display Compressed Bytes</label>
+                <span data-ng-if="vm.activeTab == 'compress64'">
+                    <input id="display-compressed" type="checkbox" class="form-control" data-ng-model="vm.displayCompressed">
+                    <label for="display-compressed">Display Compressed Bytes</label>
+                </span>
             </form>
         </div>
         <br>
         <br>
-        <hr>
         <br>
-        <div class="col-md-12">
-            <table class="table table-condensed">
-                <thead>
-                <tr>
-                    <th>Source</th>
-                    <th data-ng-show="vm.displayCompressed">Compressed</th>
-                    <th data-ng-show="vm.displayCompressed">Compressed [Bytes]</th>
-                    <th>Compressed64</th>
-                    <th>Decompressed</th>
-                    <th>Decompressed64</th>
-                </tr>
-                </thead>
-                <tbody>
-                <tr data-ng-repeat="row in vm.results | orderBy:'$index':true">
-                    <td data-ng-bind="row.input"></td>
-                    <td data-ng-show="vm.displayCompressed" ng-class="{warning: row.compressed!=row.php.compressed}">
-                        <div data-ng-bind="row.compressed"></div>
-                        <div data-ng-bind="row.php.compressed"></div>
-                    </td>
-                    <td data-ng-show="vm.displayCompressed" ng-class="{warning: row.compressedBytesString!=row.php.compressedBytesString}">
-                        <div data-ng-bind="row.compressedBytesString"></div>
-                        <div data-ng-bind="row.php.compressedBytesString"></div>
-                    </td>
-                    <td ng-class="{warning: row.compressed64!=row.php.compressed64}">
-                        <div data-ng-bind="row.compressed64"></div>
-                        <div data-ng-bind="row.php.compressed64"></div>
-                    </td>
-                    <td ng-class="{warning: row.decompressed!=row.php.decompressed}">
-                        <div data-ng-bind="row.decompressed"></div>
-                        <div data-ng-bind="row.php.decompressed"></div>
-                    </td>
-                    <td ng-class="{warning: row.decompressed64!=row.php.decompressed64}">
-                        <div data-ng-bind="row.decompressed64"></div>
-                        <div data-ng-bind="row.php.decompressed64"></div>
-                    </td>
-                </tr>
-                </tbody>
-            </table>
+        <ul class="nav nav-tabs">
+            <li data-ng-class="{active: (vm.activeTab == 'compress64')}"><a href data-ng-click="vm.activeTab = 'compress64'">Compressed64</a></li>
+            <li data-ng-class="{active: (vm.activeTab == 'utf16')}"><a href data-ng-click="vm.activeTab = 'utf16'">UTF-16</a></li>
+        </ul>
+        <br>
+        <div data-ng-if="vm.activeTab == 'compress64'">
+            <div class="col-md-12">
+                <table class="table table-condensed">
+                    <thead>
+                    <tr>
+                        <th>Source</th>
+                        <th data-ng-show="vm.displayCompressed">Compressed</th>
+                        <th data-ng-show="vm.displayCompressed">Compressed [Bytes]</th>
+                        <th>Compressed64</th>
+                        <th>Decompressed</th>
+                        <th>Decompressed64</th>
+                    </tr>
+                    </thead>
+                    <tbody>
+                    <tr data-ng-repeat="row in vm.results | orderBy:'$index':true">
+                        <td data-ng-bind="row.input"></td>
+                        <td data-ng-show="vm.displayCompressed" ng-class="{warning: row.compressed!=row.php.compressed}">
+                            <div data-ng-bind="row.compressed"></div>
+                            <div data-ng-bind="row.php.compressed"></div>
+                        </td>
+                        <td data-ng-show="vm.displayCompressed" ng-class="{warning: row.compressedBytesString!=row.php.compressedBytesString}">
+                            <div data-ng-bind="row.compressedBytesString"></div>
+                            <div data-ng-bind="row.php.compressedBytesString"></div>
+                        </td>
+                        <td ng-class="{warning: row.compressed64!=row.php.compressed64}">
+                            <div data-ng-bind="row.compressed64"></div>
+                            <div data-ng-bind="row.php.compressed64"></div>
+                        </td>
+                        <td ng-class="{warning: row.decompressed!=row.php.decompressed}">
+                            <div data-ng-bind="row.decompressed"></div>
+                            <div data-ng-bind="row.php.decompressed"></div>
+                        </td>
+                        <td ng-class="{warning: row.decompressed64!=row.php.decompressed64}">
+                            <div data-ng-bind="row.decompressed64"></div>
+                            <div data-ng-bind="row.php.decompressed64"></div>
+                        </td>
+                    </tr>
+                    </tbody>
+                </table>
+            </div>
+        </div>
+        <div data-ng-if="vm.activeTab == 'utf16'">
+            <div class="col-md-12">
+                <table class="table table-condensed">
+                    <thead>
+                    <tr>
+                        <th>Source</th>
+                        <th>Compressed [Bytes]</th>
+                        <th>Decompressed</th>
+                    </tr>
+                    </thead>
+                    <tbody>
+                    <tr data-ng-repeat="row in vm.results16 | orderBy:'$index':true">
+                        <td data-ng-bind="row.input"></td>
+                        <td ng-class="{warning: row.compressedBytesString!=row.php.compressedBytesString}">
+                            <div data-ng-bind="row.compressedBytesString"></div>
+                            <div data-ng-bind="row.php.compressedBytesString"></div>
+                        </td>
+                        <td ng-class="{warning: row.decompressed!=row.php.decompressed}">
+                            <div data-ng-bind="row.decompressed"></div>
+                            <div data-ng-bind="row.php.decompressed"></div>
+                        </td>
+                    </tr>
+                    </tbody>
+                </table>
+            </div>
         </div>
     </div>
 </div>
diff --git a/boilerplate/main.js b/boilerplate/main.js
@@ -5,21 +5,29 @@ angular.module('lzapp').controller('LZStringCtrl', LZStringCtrl);
 LZStringCtrl.$inject = ['$http'];
 
 function LZStringCtrl($http) {
-    var vm = this;
+    var vm = this,
+        // def = 'متن شگفت انگیز در اینجا'
+        def = ' «sauvegardes», «'
+        ;
 
-    vm.source = '';
+    vm.source = def;
     vm.displayCompressed = false;
     vm.results = [];
-
+    vm.results16 = [];
+    vm.activeTab = 'utf16';
 
     vm.encode = encode;
     vm.encodeLong = encodeLong;
 
     vm.longResult = null;
 
     function encode() {
-        vm.results.push(generate(vm.source));
-        vm.source = '';
+        if(vm.activeTab == 'utf16') {
+            vm.results16.push(generate16(vm.source));
+        } else {
+            vm.results.push(generate(vm.source));
+        }
+        vm.source = def;
     }
 
     function encodeLong() {
@@ -65,6 +73,30 @@ function LZStringCtrl($http) {
         return result;
     }
 
+    function generate16(str) {
+        var com = LZString.compressToUTF16(str), compressedBytes = bytes(com);
+        var result = {
+            input: str,
+            compressed: com,
+            compressedBytes: compressedBytes,
+            compressedBytesString: bytesToHexStr(compressedBytes),
+            decompressed: LZString.decompressFromUTF16(com),
+            php: {
+
+            }
+        };
+
+        $http.post('service16.php', {str: str}).then(function(res) {
+            result.php = res.data;
+            if(angular.isDefined(result.php.compressedBytes)) {
+                // result.php.compressedBytesString = result.php.compressedBytes;
+                result.php.compressedBytesString = bytesToHexStr(result.php.compressedBytes)
+            }
+        });
+
+        return result;
+    }
+
     function bytes(str) {
         var bytes = [];
         for (var i = 0; i < str.length; ++i) {
diff --git a/boilerplate/service16.php b/boilerplate/service16.php
@@ -0,0 +1,38 @@
+<?php
+require_once '../vendor/autoload.php';
+
+use \LZCompressor\LZString as LZ;
+use \LZCompressor\LZUtil16 as Util;
+
+$log = new \Monolog\Logger('name');
+\Monolog\ErrorHandler::register($log);
+
+$log->pushHandler(new \Monolog\Handler\StreamHandler(getcwd().'/log/service.log'));
+$request = json_decode(file_get_contents('php://input'), true);
+
+
+
+$log->debug('Request: '.json_encode($request));
+
+$compressed = LZ::compressToUTF16($request['str']);
+
+$length = Util::utf16_strlen($compressed);
+$compressedBytes = [];
+for($i=0; $i<$length; $i++) {
+    $val = Util::charCodeAt($compressed, $i);
+    $log->debug('val: |'.$val.'|');
+    $compressedBytes[] = $val & 255;
+    $compressedBytes[] = ($val>>8) & 255;
+}
+
+
+$result = [
+    'compressedBytes' => $compressedBytes,
+    'decompressed' => LZ::decompressFromUTF16($compressed)
+];
+
+$log->debug('Result: '.json_encode($result));
+
+header('Content-type: text/application');
+echo json_encode($result);
+//echo json_encode(['hi']);
diff --git a/src/LZCompressor/LZString.php b/src/LZCompressor/LZString.php
@@ -25,6 +25,18 @@ public static function decompressFromBase64($input)
         });
     }
 
+    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 $uncompressed
      * @return string
@@ -273,7 +285,7 @@ private static function _decompress($compressed, $resetValue, $getNextValue)
             }
 
             $result .= $entry;
-            $dictionary->addEntry($w . $entry{0});
+            $dictionary->addEntry($w . LZUtil::utf8_charAt($entry, 0));
             $w = $entry;
 
             $enlargeIn--;
diff --git a/src/LZCompressor/LZUtil.php b/src/LZCompressor/LZUtil.php
@@ -107,6 +107,4 @@ public static function utf8_charAt($str, $num)
     public static function utf8_strlen($str) {
         return mb_strlen($str, 'UTF-8');
     }
-
-
 }
diff --git a/src/LZCompressor/LZUtil16.php b/src/LZCompressor/LZUtil16.php
@@ -0,0 +1,90 @@
+<?php
+/**
+ * Created by PhpStorm.
+ * User: sics
+ * Date: 27.02.2016
+ * Time: 15:54
+ */
+
+namespace LZCompressor;
+
+
+class LZUtil16
+{
+
+    /**
+     * @return string
+     */
+    public static function fromCharCode()
+    {
+        return array_reduce(func_get_args(), function ($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');
+    }
+
+}
diff --git a/test/TestLZString.php b/test/TestLZString.php
diff --git a/test/TestLZStringUTF16.php b/test/TestLZStringUTF16.php
diff --git a/test/TestLZStringUTF8.php b/test/TestLZStringUTF8.php

Original file line number	Diff line number	Diff line change
`@@ -107,6 +107,4 @@ public static function utf8_charAt($str, $num)`
`107`	`107`	`public static function utf8_strlen($str) {`
`108`	`108`	`return mb_strlen($str, 'UTF-8');`
`109`	`109`	`}`
`110`		`-`
`111`		`-`
`112`	`110`	`}`