From 2a5e10cfd2e9eef9be3e6f152fe567c9a6ef9e68 Mon Sep 17 00:00:00 2001
From: Lars Moelleken <lars@moelleken.org>
Date: Sun, 23 Oct 2016 05:54:42 +0200
Subject: [PATCH] [+]: optimize performance

---
 src/voku/helper/HtmlDomParser.php | 68 ++++++++++---------------------
 src/voku/helper/SimpleHtmlDom.php |  7 ----
 tests/HtmlDomParserTest.php       | 10 +++++
 3 files changed, 31 insertions(+), 54 deletions(-)

diff --git a/src/voku/helper/HtmlDomParser.php b/src/voku/helper/HtmlDomParser.php
index b5ec981..0e6cf42 100644
--- a/src/voku/helper/HtmlDomParser.php
+++ b/src/voku/helper/HtmlDomParser.php
@@ -48,10 +48,10 @@ class HtmlDomParser
   protected static $domLinkReplaceHelper = array(
       'orig' => array('[', ']', '{', '}',),
       'tmp'  => array(
-          '!!!!HTML_DOM__SQUARE_BRACKET_LEFT!!!!',
-          '!!!!HTML_DOM__SQUARE_BRACKET_RIGHT!!!!',
-          '!!!!HTML_DOM__BRACKET_LEFT!!!!',
-          '!!!!HTML_DOM__BRACKET_RIGHT!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_LEFT!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__SQUARE_BRACKET_RIGHT!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__BRACKET_LEFT!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__BRACKET_RIGHT!!!!',
       ),
   );
 
@@ -61,10 +61,10 @@ class HtmlDomParser
   protected static $domReplaceHelper = array(
       'orig' => array('&', '|', '+', '%'),
       'tmp'  => array(
-          '!!!!HTML_DOM__AMP!!!!',
-          '!!!!HTML_DOM__PIPE!!!!',
-          '!!!!HTML_DOM__PLUS!!!!',
-          '!!!!HTML_DOM__PERCENT!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__AMP!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__PIPE!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__PLUS!!!!',
+          '!!!!SIMPLE_HTML_DOM__VOKU__PERCENT!!!!',
       ),
   );
 
@@ -93,13 +93,6 @@ class HtmlDomParser
    */
   protected $isDOMDocumentCreatedWithoutHtmlWrapper = false;
 
-  /**
-   * An random md5-hash, generated via "random_bytes()".
-   *
-   * @var string
-   */
-  protected $randomHash;
-
   /**
    * Constructor
    *
@@ -107,11 +100,8 @@ class HtmlDomParser
    */
   public function __construct($element = null)
   {
-    $this->randomHash = md5(Bootup::get_random_bytes(16));
     $this->document = new \DOMDocument('1.0', $this->getEncoding());
 
-    $this->addRandBytesToDomReplaceHelpers();
-
     // DOMDocument settings
     $this->document->preserveWhiteSpace = true;
     $this->document->formatOutput = true;
@@ -135,22 +125,6 @@ public function __construct($element = null)
     }
   }
 
-  /**
-   * Add rand-bytes to the "Dom-Replace-Helper"-variables.
-   */
-  protected function addRandBytesToDomReplaceHelpers()
-  {
-    /** @noinspection AlterInForeachInspection */
-    foreach (self::$domLinkReplaceHelper['tmp'] as &$linkHelper) {
-      $linkHelper .= $this->randomHash;
-    }
-
-    /** @noinspection AlterInForeachInspection */
-    foreach (self::$domReplaceHelper['tmp'] as &$domHelper) {
-      $domHelper .= $this->randomHash;
-    }
-  }
-
   /**
    * @param $name
    * @param $arguments
@@ -300,19 +274,19 @@ public static function replaceToPreserveHtmlEntities($html)
    */
   public static function putReplacedBackToPreserveHtmlEntities($html)
   {
-    return str_replace(
-        array_merge(
-            self::$domLinkReplaceHelper['tmp'],
-            self::$domReplaceHelper['tmp'],
-            array('&#13;')
-        ),
-        array_merge(
-            self::$domLinkReplaceHelper['orig'],
-            self::$domReplaceHelper['orig'],
-            array('')
-        ),
-        $html
-    );
+    static $DOM_REPLACE__HELPER_CACHE = null;
+    if ($DOM_REPLACE__HELPER_CACHE === null) {
+      $DOM_REPLACE__HELPER_CACHE['tmp'] = array_merge(
+          self::$domLinkReplaceHelper['tmp'],
+          self::$domReplaceHelper['tmp']
+      );
+      $DOM_REPLACE__HELPER_CACHE['orig'] = array_merge(
+          self::$domLinkReplaceHelper['orig'],
+          self::$domReplaceHelper['orig']
+      );
+    }
+
+    return str_replace($DOM_REPLACE__HELPER_CACHE['tmp'], $DOM_REPLACE__HELPER_CACHE['orig'], $html);
   }
 
   /**
diff --git a/src/voku/helper/SimpleHtmlDom.php b/src/voku/helper/SimpleHtmlDom.php
index 0737ddb..a22e9f4 100644
--- a/src/voku/helper/SimpleHtmlDom.php
+++ b/src/voku/helper/SimpleHtmlDom.php
@@ -494,11 +494,8 @@ protected function replaceChild($string)
     }
 
     if (!empty($newDocument)) {
-
       $newDocument = $this->cleanHtmlWrapper($newDocument);
-
       $newNode = $this->node->ownerDocument->importNode($newDocument->getDocument()->documentElement, true);
-
       $this->node->appendChild($newNode);
     }
 
@@ -522,10 +519,6 @@ protected function replaceNode($string)
 
     $newDocument = new HtmlDomParser($string);
 
-    // DEBUG
-    //echo $this->normalizeStringForComparision($newDocument->outertext) . "\n";
-    //echo $this->normalizeStringForComparision($string) . "\n\n";
-
     if ($this->normalizeStringForComparision($newDocument->outertext) != $this->normalizeStringForComparision($string)) {
       throw new RuntimeException('Not valid HTML fragment');
     }
diff --git a/tests/HtmlDomParserTest.php b/tests/HtmlDomParserTest.php
index 2da08cb..bd344d0 100644
--- a/tests/HtmlDomParserTest.php
+++ b/tests/HtmlDomParserTest.php
@@ -924,6 +924,16 @@ public function testGetElementsByClass()
     );
   }
 
+  public function testUtf8AndBrokenHtmlEncoding()
+  {
+    $dom = new HtmlDomParser();
+    $dom->load('hi</b><p>سلام<div>の家庭に、9 ☆<><');
+    self::assertSame(
+        '<p>hi</p><p>سلام</p><div>の家庭に、9 ☆</div>',
+        $dom->innerHtml
+    );
+  }
+
   public function testEnforceEncoding()
   {
     $dom = new HtmlDomParser();