Merge pull request paquettg#90 from andreyshade/develop

Add additional features

Author: paquettg

src/PHPHtmlParser/Dom.php

@@ -135,6 +135,7 @@ public function __get($name)
      */
     public function load($str, $options = [])
     {
+        AbstractNode::resetCount();
         // check if it's a file
         if (strpos($str, "\n") === false && is_file($str)) {
             return $this->loadFromFile($str, $options);
@@ -234,6 +235,20 @@ public function find($selector, $nth = null)
         return $this->root->find($selector, $nth);
     }
 
+    /**
+     * Find element by Id on the root node
+     *
+     * @param int $id Element Id
+     * @return mixed
+     *
+     */
+    public function findById($id)
+    {
+        $this->isLoaded();
+
+        return $this->root->findById($id);
+    }
+
     /**
      * Adds the tag (or tags in an array) to the list of tags that will always
      * be self closing.
@@ -353,6 +368,42 @@ public function lastChild()
         return $this->root->lastChild();
     }
 
+    /**
+     * Simple wrapper function that returns count of child elements
+     *
+     * @return int
+     */
+    public function countChildren()
+    {
+        $this->isLoaded();
+
+        return $this->root->countChildren();
+    }
+
+    /**
+     * Get array of children
+     *
+     * @return array
+     */
+    public function getChildren()
+    {
+        $this->isLoaded();
+
+        return $this->root->getChildren();
+    }
+
+    /**
+     * Check if node have children nodes
+     *
+     * @return bool
+     */
+    public function hasChildren()
+    {
+        $this->isLoaded();
+
+        return $this->root->hasChildren();
+    }
+
     /**
      * Simple wrapper function that returns an element by the
      * id.
@@ -453,7 +504,9 @@ protected function clean($str)
         }
 
         // strip out server side scripts
-        $str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str);
+        if ($this->options->get('serverSideScriptis') == true){
+            $str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str);
+        }
 
         // strip smarty scripts
         $str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str);

src/PHPHtmlParser/Dom/AbstractNode.php

@@ -5,6 +5,7 @@
 use PHPHtmlParser\Exceptions\ParentNotFoundException;
 use PHPHtmlParser\Selector;
 use stringEncode\Encode;
+use PHPHtmlParser\Finder;
 
 /**
  * Dom node object.
@@ -17,7 +18,7 @@
  */
 abstract class AbstractNode
 {
-
+    private static $count = 0;
     /**
      * Contains the tag name/type
      *
@@ -54,11 +55,12 @@ abstract class AbstractNode
     protected $encode;
 
     /**
-     * Creates a unique spl hash for this node.
+     * Creates a unique id for this node.
      */
     public function __construct()
     {
-        $this->id = spl_object_hash($this);
+        $this->id = self::$count;
+        self::$count++;
     }
 
     /**
@@ -110,6 +112,14 @@ public function __toString()
         return $this->outerHtml();
     }
 
+    /**
+     * Reset node counter
+     */
+    public static function resetCount()
+    {
+        self::$count = 0;
+    }
+
     /**
      * Returns the id of this object.
      */
@@ -220,6 +230,15 @@ public function getAncestor($id)
         return null;
     }
 
+    public function hasNextSibling()
+    {
+        if (is_null($this->parent) || (!$this->parent->hasChildren())) {
+            return false;
+        }
+
+        return $this->parent->hasNextChild($this->id());
+    }
+
     /**
      * Attempts to get the next sibling.
      *
@@ -398,6 +417,20 @@ public function find($selector, $nth = null)
         return $nodes;
     }
 
+    /**
+     * Find node by id
+     *
+     * @param $id
+     * @return bool|AbstractNode
+     */
+    public function findById($id)
+    {
+        $finder= new Finder($id);
+
+        return $finder->find($this);
+    }
+
+
     /**
      * Gets the inner html of this node.
      *
@@ -427,4 +460,14 @@ abstract public function text();
      * @return void
      */
     abstract protected function clear();
+
+    /**
+     * Check is node type textNode
+     *
+     * @return boolean
+     */
+    public function isTextNode() {
+
+        return false;
+    }
 }

src/PHPHtmlParser/Dom/InnerNode.php

@@ -175,6 +175,18 @@ public function removeChild($id)
         return $this;
     }
 
+    /**
+     * Check if has next Child
+     *
+     * @param $id childId
+     * @return mixed
+     */
+    public function hasNextChild($id)
+    {
+        $child= $this->getChild($id);
+        return $this->children[$child->id()]['next'];
+    }
+
     /**
      * Attempts to get the next child.
      *

src/PHPHtmlParser/Dom/Tag.php

@@ -173,6 +173,51 @@ public function setAttribute($key, $value)
         return $this;
     }
 
+    /**
+     * Set inline style attribute value.
+     *
+     * @param $attr_key
+     * @param $attr_value
+     */
+    public function setStyleAttributeValue($attr_key, $attr_value)
+    {
+
+        $style_array = $this->getStyleAttributeArray();
+        $style_array[$attr_key] = $attr_value;
+
+        $style_string = '';
+        foreach ($style_array as $key => $value) {
+            $style_string .= $key . ':' . $value . ';';
+        }
+
+        $this->setAttribute('style', $style_string);
+    }
+
+    /**
+     * Get style attribute in array
+     *
+     * @return array|null
+     */
+    public function getStyleAttributeArray()
+    {
+        $value = $this->getAttribute('style')['value'];
+
+        if ($value === null) {
+            return null;
+        }
+
+        $value = explode(';', substr(trim($value), 0, -1));
+        $result = [];
+        foreach ($value as $attr) {
+            $attr = explode(':', $attr);
+            $result[$attr[0]] = $attr[1];
+        }
+
+        return $result;
+    }
+
+
+
     /**
      * Removes an attribute from this tag.
      *

src/PHPHtmlParser/Dom/TextNode.php

@@ -120,4 +120,15 @@ protected function clear()
     {
         $this->convertedText = null;
     }
+
+    public function isTextNode()
+    {
+        return true;
+    }
+
+    public function setText($text)
+    {
+        $this->text = $text;
+    }
+
 }

src/PHPHtmlParser/Finder.php

@@ -0,0 +1,54 @@
+<?php
+
+namespace PHPHtmlParser;
+
+use PHPHtmlParser\Dom\AbstractNode;
+
+class Finder
+{
+    private $id;
+
+    /**
+     * Finder constructor.
+     * @param $id
+     */
+    public function __construct($id)
+    {
+        $this->id = $id;
+    }
+
+    /**
+     *
+     * Find node in tree by id
+     *
+     * @param AbstractNode $node
+     * @return bool|AbstractNode
+     */
+    public function find(AbstractNode $node)
+    {
+        if (!$node->id()) {
+            return $this->find($node->firstChild());
+        }
+
+        if ($node->id() == $this->id) {
+            return $node;
+        }
+
+        if ($node->hasNextSibling()) {
+            $nextSibling = $node->nextSibling();
+            if ($nextSibling->id() == $this->id) {
+                return $nextSibling;
+            }
+            if ($nextSibling->id() > $this->id) {
+                return $this->find($node->firstChild());
+            }
+            if ($nextSibling->id() < $this->id) {
+                return $this->find($nextSibling);
+            }
+        } else if (!$node->isTextNode()) {
+            return $this->find($node->firstChild());
+        }
+
+        return false;
+    }
+}

tests/DomTest.php

@@ -329,4 +329,52 @@ public function testDeleteNode()
         unset($a);
         $this->assertEquals('<div class="all"><p>Hey bro, <br /> :)</p></div>', (string) $dom);
     }
+
+    public function testCountChildren()
+    {
+        $dom = new Dom;
+        $dom->load('<strong>hello</strong><code class="language-php">$foo = "bar";</code>');
+        $this->assertEquals(2, $dom->countChildren());
+    }
+
+    public function testGetChildrenArray()
+    {
+        $dom = new Dom;
+        $dom->load('<strong>hello</strong><code class="language-php">$foo = "bar";</code>');
+        $this->assertInternalType('array', $dom->getChildren());
+    }
+
+    public function testHasChildren()
+    {
+        $dom = new Dom;
+        $dom->load('<strong>hello</strong><code class="language-php">$foo = "bar";</code>');
+        $this->assertTrue($dom->hasChildren());
+    }
+
+    public function testFindByIdVar1()
+    {
+        $dom = new Dom;
+        $dom->load('<div class="all"><p>Hey bro, <a href="google.com">click here</a><br /> :)</p></div>');
+        /** @var Dom\AbstractNode $result */
+        $result = $dom->findById(4);
+        $this->assertEquals(4, $result->id());
+    }
+
+    public function testFindByIdVar2()
+    {
+        $dom = new Dom;
+        $dom->load('<div class="all"><p>Hey bro, <a href="google.com">click here</a><br /> :)</p></div>');
+        /** @var Dom\AbstractNode $result */
+        $result = $dom->findById(5);
+        $this->assertEquals(5, $result->id());
+    }
+
+    public function testFindByIdNotFountEleement()
+    {
+        $dom = new Dom;
+        $dom->load('<div class="all"><p>Hey bro, <a href="google.com">click here</a><br /> :)</p></div>');
+        /** @var Dom\AbstractNode $result */
+        $result = $dom->findById(8);
+        $this->assertFalse($result);
+    }
 }