Added double space option

fixes paquettg#80
close paquettg#80

Author: paquettg

README.md

@@ -138,7 +138,7 @@ $dom->load('http://google.com', [
 $dom->load('http://gmail.com'); // will not have whitespaceTextNode set to false.
 ```
 
-At the moment we support 7 options.
+At the moment we support 8 options.
 
 **Strict**
 
@@ -168,6 +168,9 @@ Set this to `false` to skip removing of style tags from the document body. This
 
 Preserves Line Breaks if set to `true`. If set to `false` line breaks are cleaned up as part of the input clean up process. Defaults to `false`.
 
+**removeDoubleSpace**
+
+Set this to `false` if you want to preserver whitespace inside of text nodes. It is set to `true` by default.
 Static Facade
 -------------
 

src/PHPHtmlParser/Dom.php

@@ -565,7 +565,7 @@ protected function parse()
                 trim($str) != ''
             ) {
                 // we found text we care about
-                $textNode = new TextNode($str);
+                $textNode = new TextNode($str, $this->options->removeDoubleSpace);
                 $activeNode->addChild($textNode);
             }
         }

src/PHPHtmlParser/Dom/TextNode.php

@@ -35,10 +35,12 @@ class TextNode extends LeafNode
      *
      * @param string $text
      */
-    public function __construct($text)
+    public function __construct($text, $removeDoubleSpace = true)
     {
-        // remove double spaces
-        $text = mb_ereg_replace('\s+', ' ', $text);
+        if ($removeDoubleSpace) {
+            // remove double spaces
+            $text = mb_ereg_replace('\s+', ' ', $text);
+        }
 
         // restore line breaks
         $text = str_replace('
', "\n", $text);

src/PHPHtmlParser/Options.php

@@ -25,6 +25,7 @@ class Options
         'removeScripts'      => true,
         'removeStyles'       => true,
         'preserveLineBreaks' => false,
+        'removeDoubleSpace'  => true,
     ];
 
     /**

tests/DomTest.php

@@ -377,4 +377,14 @@ public function testFindByIdNotFountEleement()
         $result = $dom->findById(8);
         $this->assertFalse($result);
     }
+
+    public function testWhitespaceInText()
+    {
+        $dom = new Dom();
+        $dom->setOptions(array(
+            'removeDoubleSpace' => false,
+        ));
+        $dom->load('<pre>    Hello world</pre>');
+        $this->assertEquals('<pre>    Hello world</pre>', (string) $dom);
+    }
 }