paquettg / php-html-parser

src/PHPHtmlParser/Dom.php

Indentation +628 added lines, -628 removed lines

@@ -16,632 +16,632 @@
 class Dom
 {
 
-    /**
-     * The charset we would like the output to be in.
-     *
-     * @var string
-     */
-    protected $defaultCharset = 'UTF-8';
-
-    /**
-     * Contains the root node of this dom tree.
-     *
-     * @var HtmlNode
-     */
-    public $root;
-
-    /**
-     * The raw version of the document string.
-     *
-     * @var string
-     */
-    protected $raw;
-
-    /**
-     * The document string.
-     *
-     * @var Content
-     */
-    protected $content = null;
-
-    /**
-     * The original file size of the document.
-     *
-     * @var int
-     */
-    protected $rawSize;
-
-    /**
-     * The size of the document after it is cleaned.
-     *
-     * @var int
-     */
-    protected $size;
-
-    /**
-     * A global options array to be used by all load calls.
-     *
-     * @var array
-     */
-    protected $globalOptions = [];
-
-    /**
-     * A persistent option object to be used for all options in the
-     * parsing of the file.
-     *
-     * @var Options
-     */
-    protected $options;
-
-    /**
-     * A list of tags which will always be self closing
-     *
-     * @var array
-     */
-    protected $selfClosing = [
-        'img',
-        'br',
-        'input',
-        'meta',
-        'link',
-        'hr',
-        'base',
-        'embed',
-        'spacer',
-    ];
-
-    /**
-     * Returns the inner html of the root node.
-     *
-     * @return string
-     */
-    public function __toString()
-    {
-        return $this->root->innerHtml();
-    }
-
-    /**
-     * A simple wrapper around the root node.
-     *
-     * @param string $name
-     * @return mixed
-     */
-    public function __get($name)
-    {
-        return $this->root->$name;
-    }
-
-    /**
-     * Attempts to load the dom from any resource, string, file, or URL.
-     *
-     * @param string $str
-     * @param array $options
-     * @return $this
-     */
-    public function load($str, $options = [])
-    {
-        // check if it's a file
-        if (strpos($str, "\n") === false && is_file($str)) {
-            return $this->loadFromFile($str, $options);
-        }
-        // check if it's a url
-        if (preg_match("/^https?:\/\//i", $str)) {
-            return $this->loadFromUrl($str, $options);
-        }
-
-        return $this->loadStr($str, $options);
-    }
-
-    /**
-     * Loads the dom from a document file/url
-     *
-     * @param string $file
-     * @param array $options
-     * @return $this
-     */
-    public function loadFromFile($file, $options = [])
-    {
-        return $this->loadStr(file_get_contents($file), $options);
-    }
-
-    /**
-     * Use a curl interface implementation to attempt to load
-     * the content from a url.
-     *
-     * @param string $url
-     * @param array $options
-     * @param CurlInterface $curl
-     * @return $this
-     */
-    public function loadFromUrl($url, $options = [], CurlInterface $curl = null)
-    {
-        if (is_null($curl)) {
-            // use the default curl interface
-            $curl = new Curl;
-        }
-        $content = $curl->get($url);
-
-        return $this->loadStr($content, $options);
-    }
-
-    /**
-     * Parsers the html of the given string. Used for load(), loadFromFile(),
-     * and loadFromUrl().
-     *
-     * @param string $str
-     * @param array $option
-     * @return $this
-     */
-    public function loadStr($str, $option)
-    {
-        $this->options = new Options;
-        $this->options->setOptions($this->globalOptions)
-                      ->setOptions($option);
-
-        $this->rawSize = strlen($str);
-        $this->raw     = $str;
-
-        $html = $this->clean($str);
-
-        $this->size    = strlen($str);
-        $this->content = new Content($html);
-
-        $this->parse();
-        $this->detectCharset();
-
-        return $this;
-    }
-
-    /**
-     * Sets a global options array to be used by all load calls.
-     *
-     * @param array $options
-     * @return $this
-     */
-    public function setOptions(array $options)
-    {
-        $this->globalOptions = $options;
-
-        return $this;
-    }
-
-    /**
-     * Find elements by css selector on the root node.
-     *
-     * @param string $selector
-     * @param int $nth
-     * @return array
-     */
-    public function find($selector, $nth = null)
-    {
-        $this->isLoaded();
-
-        return $this->root->find($selector, $nth);
-    }
-
-    /**
-     * Adds the tag (or tags in an array) to the list of tags that will always
-     * be self closing.
-     *
-     * @param string|array $tag
-     * @return $this
-     */
-    public function addSelfClosingTag($tag)
-    {
-        if ( ! is_array($tag)) {
-            $tag = [$tag];
-        }
-        foreach ($tag as $value) {
-            $this->selfClosing[] = $value;
-        }
-
-        return $this;
-    }
-
-    /**
-     * Removes the tag (or tags in an array) from the list of tags that will
-     * always be self closing.
-     *
-     * @param string|array $tag
-     * @return $this
-     */
-    public function removeSelfClosingTag($tag)
-    {
-        if ( ! is_array($tag)) {
-            $tag = [$tag];
-        }
-        $this->selfClosing = array_diff($this->selfClosing, $tag);
-
-        return $this;
-    }
-
-    /**
-     * Sets the list of self closing tags to empty.
-     *
-     * @return $this
-     */
-    public function clearSelfClosingTags()
-    {
-        $this->selfClosing = [];
-
-        return $this;
-    }
-
-    /**
-     * Simple wrapper function that returns the first child.
-     *
-     * @return \PHPHtmlParser\Dom\AbstractNode
-     */
-    public function firstChild()
-    {
-        $this->isLoaded();
-
-        return $this->root->firstChild();
-    }
-
-    /**
-     * Simple wrapper function that returns the last child.
-     *
-     * @return \PHPHtmlParser\Dom\AbstractNode
-     */
-    public function lastChild()
-    {
-        $this->isLoaded();
-
-        return $this->root->lastChild();
-    }
-
-    /**
-     * Simple wrapper function that returns an element by the
-     * id.
-     *
-     * @param string $id
-     * @return \PHPHtmlParser\Dom\AbstractNode
-     */
-    public function getElementById($id)
-    {
-        $this->isLoaded();
-
-        return $this->find('#'.$id, 0);
-    }
-
-    /**
-     * Simple wrapper function that returns all elements by
-     * tag name.
-     *
-     * @param string $name
-     * @return array
-     */
-    public function getElementsByTag($name)
-    {
-        $this->isLoaded();
-
-        return $this->find($name);
-    }
-
-    /**
-     * Simple wrapper function that returns all elements by
-     * class name.
-     *
-     * @param string $class
-     * @return array
-     */
-    public function getElementsByClass($class)
-    {
-        $this->isLoaded();
-
-        return $this->find('.'.$class);
-    }
-
-    /**
-     * Checks if the load methods have been called.
-     *
-     * @throws NotLoadedException
-     */
-    protected function isLoaded()
-    {
-        if (is_null($this->content)) {
-            throw new NotLoadedException('Content is not loaded!');
-        }
-    }
-
-    /**
-     * Cleans the html of any none-html information.
-     *
-     * @param string $str
-     * @return string
-     */
-    protected function clean($str)
-    {
-        if ($this->options->get('cleanupInput') != true) {
-            // skip entire cleanup step
-            return $str;
-        }
-
-        // clean out the \n\r
-        $replace = ' ';
-        if ($this->options->get('preserveLineBreaks')) {
-            $replace = '&#10';
-        }
-        $str = str_replace(["\r\n", "\r", "\n"], $replace, $str);
-
-        // strip the doctype
-        $str = mb_eregi_replace("<!doctype(.*?)>", '', $str);
-
-        // strip out comments
-        $str = mb_eregi_replace("<!--(.*?)-->", '', $str);
-
-        // strip out cdata
-        $str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str);
-
-        // strip out <script> tags
-        if ($this->options->get('removeScripts') == true) {
-            $str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str);
-            $str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str);
-        }
-
-        // strip out <style> tags
-        if ($this->options->get('removeStyles') == true) {
-            $str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str);
-            $str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str);
-        }
-
-        // strip out preformatted tags
-        $str = mb_eregi_replace("<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>", '', $str);
-
-        // strip out server side scripts
-        $str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str);
-
-        // strip smarty scripts
-        $str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str);
-
-        return $str;
-    }
-
-    /**
-     * Attempts to parse the html in content.
-     */
-    protected function parse()
-    {
-        // add the root node
-        $this->root = new HtmlNode('root');
-        $activeNode = $this->root;
-        while ( ! is_null($activeNode)) {
-            $str = $this->content->copyUntil('<');
-            if ($str == '') {
-                $info = $this->parseTag();
-                if ( ! $info['status']) {
-                    // we are done here
-                    $activeNode = null;
-                    continue;
-                }
-
-                // check if it was a closing tag
-                if ($info['closing']) {
-                    $originalNode = $activeNode;
-                    while ($activeNode->getTag()->name() != $info['tag']) {
-                        $activeNode = $activeNode->getParent();
-                        if (is_null($activeNode)) {
-                            // we could not find opening tag
-                            $activeNode = $originalNode;
-                            break;
-                        }
-                    }
-                    if ( ! is_null($activeNode)) {
-                        $activeNode = $activeNode->getParent();
-                    }
-                    continue;
-                }
-
-                if ( ! isset($info['node'])) {
-                    continue;
-                }
-
-                /** @var AbstractNode $node */
-                $node = $info['node'];
-                $activeNode->addChild($node);
-
-                // check if node is self closing
-                if ( ! $node->getTag()->isSelfClosing()) {
-                    $activeNode = $node;
-                }
-            } else if ($this->options->whitespaceTextNode ||
-                trim($str) != ''
-            ) {
-                // we found text we care about
-                $textNode = new TextNode($str);
-                $activeNode->addChild($textNode);
-            }
-        }
-    }
-
-    /**
-     * Attempt to parse a tag out of the content.
-     *
-     * @return array
-     * @throws StrictException
-     */
-    protected function parseTag()
-    {
-        $return = [
-            'status'  => false,
-            'closing' => false,
-            'node'    => null,
-        ];
-        if ($this->content->char() != '<') {
-            // we are not at the beginning of a tag
-            return $return;
-        }
-
-        // check if this is a closing tag
-        if ($this->content->fastForward(1)->char() == '/') {
-            // end tag
-            $tag = $this->content->fastForward(1)
-                                 ->copyByToken('slash', true);
-            // move to end of tag
-            $this->content->copyUntil('>');
-            $this->content->fastForward(1);
-
-            // check if this closing tag counts
-            $tag = strtolower($tag);
-            if (in_array($tag, $this->selfClosing)) {
-                $return['status'] = true;
-
-                return $return;
-            } else {
-                $return['status']  = true;
-                $return['closing'] = true;
-                $return['tag']     = strtolower($tag);
-            }
-
-            return $return;
-        }
-
-        $tag  = strtolower($this->content->copyByToken('slash', true));
-        $node = new HtmlNode($tag);
-
-        // attributes
-        while ($this->content->char() != '>' &&
-            $this->content->char() != '/') {
-            $space = $this->content->skipByToken('blank', true);
-            if (empty($space)) {
-                $this->content->fastForward(1);
-                continue;
-            }
-
-            $name = $this->content->copyByToken('equal', true);
-            if ($name == '/') {
-                break;
-            }
-
-            if (empty($name)) {
-                $this->content->fastForward(1);
-                continue;
-            }
-
-            $this->content->skipByToken('blank');
-            if ($this->content->char() == '=') {
-                $attr = [];
-                $this->content->fastForward(1)
-                              ->skipByToken('blank');
-                switch ($this->content->char()) {
-                    case '"':
-                        $attr['doubleQuote'] = true;
-                        $this->content->fastForward(1);
-                        $string = $this->content->copyUntil('"', true, true);
-                        do {
-                            $moreString = $this->content->copyUntilUnless('"', '=>');
-                            $string .= $moreString;
-                        } while ( ! empty($moreString));
-                        $attr['value'] = $string;
-                        $this->content->fastForward(1);
-                        $node->getTag()->$name = $attr;
-                        break;
-                    case "'":
-                        $attr['doubleQuote'] = false;
-                        $this->content->fastForward(1);
-                        $string = $this->content->copyUntil("'", true, true);
-                        do {
-                            $moreString = $this->content->copyUntilUnless("'", '=>');
-                            $string .= $moreString;
-                        } while ( ! empty($moreString));
-                        $attr['value'] = $string;
-                        $this->content->fastForward(1);
-                        $node->getTag()->$name = $attr;
-                        break;
-                    default:
-                        $attr['doubleQuote']   = true;
-                        $attr['value']         = $this->content->copyByToken('attr', true);
-                        $node->getTag()->$name = $attr;
-                        break;
-                }
-            } else {
-                // no value attribute
-                if ($this->options->strict) {
-                    // can't have this in strict html
-                    $character = $this->content->getPosition();
-                    throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)");
-                }
-                $node->getTag()->$name = [
-                    'value'       => null,
-                    'doubleQuote' => true,
-                ];
-                if ($this->content->char() != '>') {
-                    $this->content->rewind(1);
-                }
-            }
-        }
-
-        $this->content->skipByToken('blank');
-        if ($this->content->char() == '/') {
-            // self closing tag
-            $node->getTag()->selfClosing();
-            $this->content->fastForward(1);
-        } elseif (in_array($tag, $this->selfClosing)) {
-
-            // Should be a self closing tag, check if we are strict
-            if ($this->options->strict) {
-                $character = $this->content->getPosition();
-                throw new StrictException("Tag '$tag' is not self closing! (character #$character)");
-            }
-
-            // We force self closing on this tag.
-            $node->getTag()->selfClosing();
-        }
-
-        $this->content->fastForward(1);
-
-        $return['status'] = true;
-        $return['node']   = $node;
-
-        return $return;
-    }
-
-    /**
-     * Attempts to detect the charset that the html was sent in.
-     *
-     * @return bool
-     */
-    protected function detectCharset()
-    {
-        // set the default
-        $encode = new Encode;
-        $encode->from($this->defaultCharset);
-        $encode->to($this->defaultCharset);
-
-        if ( ! is_null($this->options->enforceEncoding)) {
-            //  they want to enforce the given encoding
-            $encode->from($this->options->enforceEncoding);
-            $encode->to($this->options->enforceEncoding);
-
-            return false;
-        }
-
-        $meta = $this->root->find('meta[http-equiv=Content-Type]', 0);
-        if (is_null($meta)) {
-            // could not find meta tag
-            $this->root->propagateEncoding($encode);
-
-            return false;
-        }
-        $content = $meta->content;
-        if (empty($content)) {
-            // could not find content
-            $this->root->propagateEncoding($encode);
-
-            return false;
-        }
-        $matches = [];
-        if (preg_match('/charset=(.+)/', $content, $matches)) {
-            $encode->from(trim($matches[1]));
-            $this->root->propagateEncoding($encode);
-
-            return true;
-        }
-
-        // no charset found
-        $this->root->propagateEncoding($encode);
-
-        return false;
-    }
+	/**
+	 * The charset we would like the output to be in.
+	 *
+	 * @var string
+	 */
+	protected $defaultCharset = 'UTF-8';
+
+	/**
+	 * Contains the root node of this dom tree.
+	 *
+	 * @var HtmlNode
+	 */
+	public $root;
+
+	/**
+	 * The raw version of the document string.
+	 *
+	 * @var string
+	 */
+	protected $raw;
+
+	/**
+	 * The document string.
+	 *
+	 * @var Content
+	 */
+	protected $content = null;
+
+	/**
+	 * The original file size of the document.
+	 *
+	 * @var int
+	 */
+	protected $rawSize;
+
+	/**
+	 * The size of the document after it is cleaned.
+	 *
+	 * @var int
+	 */
+	protected $size;
+
+	/**
+	 * A global options array to be used by all load calls.
+	 *
+	 * @var array
+	 */
+	protected $globalOptions = [];
+
+	/**
+	 * A persistent option object to be used for all options in the
+	 * parsing of the file.
+	 *
+	 * @var Options
+	 */
+	protected $options;
+
+	/**
+	 * A list of tags which will always be self closing
+	 *
+	 * @var array
+	 */
+	protected $selfClosing = [
+		'img',
+		'br',
+		'input',
+		'meta',
+		'link',
+		'hr',
+		'base',
+		'embed',
+		'spacer',
+	];
+
+	/**
+	 * Returns the inner html of the root node.
+	 *
+	 * @return string
+	 */
+	public function __toString()
+	{
+		return $this->root->innerHtml();
+	}
+
+	/**
+	 * A simple wrapper around the root node.
+	 *
+	 * @param string $name
+	 * @return mixed
+	 */
+	public function __get($name)
+	{
+		return $this->root->$name;
+	}
+
+	/**
+	 * Attempts to load the dom from any resource, string, file, or URL.
+	 *
+	 * @param string $str
+	 * @param array $options
+	 * @return $this
+	 */
+	public function load($str, $options = [])
+	{
+		// check if it's a file
+		if (strpos($str, "\n") === false && is_file($str)) {
+			return $this->loadFromFile($str, $options);
+		}
+		// check if it's a url
+		if (preg_match("/^https?:\/\//i", $str)) {
+			return $this->loadFromUrl($str, $options);
+		}
+
+		return $this->loadStr($str, $options);
+	}
+
+	/**
+	 * Loads the dom from a document file/url
+	 *
+	 * @param string $file
+	 * @param array $options
+	 * @return $this
+	 */
+	public function loadFromFile($file, $options = [])
+	{
+		return $this->loadStr(file_get_contents($file), $options);
+	}
+
+	/**
+	 * Use a curl interface implementation to attempt to load
+	 * the content from a url.
+	 *
+	 * @param string $url
+	 * @param array $options
+	 * @param CurlInterface $curl
+	 * @return $this
+	 */
+	public function loadFromUrl($url, $options = [], CurlInterface $curl = null)
+	{
+		if (is_null($curl)) {
+			// use the default curl interface
+			$curl = new Curl;
+		}
+		$content = $curl->get($url);
+
+		return $this->loadStr($content, $options);
+	}
+
+	/**
+	 * Parsers the html of the given string. Used for load(), loadFromFile(),
+	 * and loadFromUrl().
+	 *
+	 * @param string $str
+	 * @param array $option
+	 * @return $this
+	 */
+	public function loadStr($str, $option)
+	{
+		$this->options = new Options;
+		$this->options->setOptions($this->globalOptions)
+					  ->setOptions($option);
+
+		$this->rawSize = strlen($str);
+		$this->raw     = $str;
+
+		$html = $this->clean($str);
+
+		$this->size    = strlen($str);
+		$this->content = new Content($html);
+
+		$this->parse();
+		$this->detectCharset();
+
+		return $this;
+	}
+
+	/**
+	 * Sets a global options array to be used by all load calls.
+	 *
+	 * @param array $options
+	 * @return $this
+	 */
+	public function setOptions(array $options)
+	{
+		$this->globalOptions = $options;
+
+		return $this;
+	}
+
+	/**
+	 * Find elements by css selector on the root node.
+	 *
+	 * @param string $selector
+	 * @param int $nth
+	 * @return array
+	 */
+	public function find($selector, $nth = null)
+	{
+		$this->isLoaded();
+
+		return $this->root->find($selector, $nth);
+	}
+
+	/**
+	 * Adds the tag (or tags in an array) to the list of tags that will always
+	 * be self closing.
+	 *
+	 * @param string|array $tag
+	 * @return $this
+	 */
+	public function addSelfClosingTag($tag)
+	{
+		if ( ! is_array($tag)) {
+			$tag = [$tag];
+		}
+		foreach ($tag as $value) {
+			$this->selfClosing[] = $value;
+		}
+
+		return $this;
+	}
+
+	/**
+	 * Removes the tag (or tags in an array) from the list of tags that will
+	 * always be self closing.
+	 *
+	 * @param string|array $tag
+	 * @return $this
+	 */
+	public function removeSelfClosingTag($tag)
+	{
+		if ( ! is_array($tag)) {
+			$tag = [$tag];
+		}
+		$this->selfClosing = array_diff($this->selfClosing, $tag);
+
+		return $this;
+	}
+
+	/**
+	 * Sets the list of self closing tags to empty.
+	 *
+	 * @return $this
+	 */
+	public function clearSelfClosingTags()
+	{
+		$this->selfClosing = [];
+
+		return $this;
+	}
+
+	/**
+	 * Simple wrapper function that returns the first child.
+	 *
+	 * @return \PHPHtmlParser\Dom\AbstractNode
+	 */
+	public function firstChild()
+	{
+		$this->isLoaded();
+
+		return $this->root->firstChild();
+	}
+
+	/**
+	 * Simple wrapper function that returns the last child.
+	 *
+	 * @return \PHPHtmlParser\Dom\AbstractNode
+	 */
+	public function lastChild()
+	{
+		$this->isLoaded();
+
+		return $this->root->lastChild();
+	}
+
+	/**
+	 * Simple wrapper function that returns an element by the
+	 * id.
+	 *
+	 * @param string $id
+	 * @return \PHPHtmlParser\Dom\AbstractNode
+	 */
+	public function getElementById($id)
+	{
+		$this->isLoaded();
+
+		return $this->find('#'.$id, 0);
+	}
+
+	/**
+	 * Simple wrapper function that returns all elements by
+	 * tag name.
+	 *
+	 * @param string $name
+	 * @return array
+	 */
+	public function getElementsByTag($name)
+	{
+		$this->isLoaded();
+
+		return $this->find($name);
+	}
+
+	/**
+	 * Simple wrapper function that returns all elements by
+	 * class name.
+	 *
+	 * @param string $class
+	 * @return array
+	 */
+	public function getElementsByClass($class)
+	{
+		$this->isLoaded();
+
+		return $this->find('.'.$class);
+	}
+
+	/**
+	 * Checks if the load methods have been called.
+	 *
+	 * @throws NotLoadedException
+	 */
+	protected function isLoaded()
+	{
+		if (is_null($this->content)) {
+			throw new NotLoadedException('Content is not loaded!');
+		}
+	}
+
+	/**
+	 * Cleans the html of any none-html information.
+	 *
+	 * @param string $str
+	 * @return string
+	 */
+	protected function clean($str)
+	{
+		if ($this->options->get('cleanupInput') != true) {
+			// skip entire cleanup step
+			return $str;
+		}
+
+		// clean out the \n\r
+		$replace = ' ';
+		if ($this->options->get('preserveLineBreaks')) {
+			$replace = '&#10';
+		}
+		$str = str_replace(["\r\n", "\r", "\n"], $replace, $str);
+
+		// strip the doctype
+		$str = mb_eregi_replace("<!doctype(.*?)>", '', $str);
+
+		// strip out comments
+		$str = mb_eregi_replace("<!--(.*?)-->", '', $str);
+
+		// strip out cdata
+		$str = mb_eregi_replace("<!\[CDATA\[(.*?)\]\]>", '', $str);
+
+		// strip out <script> tags
+		if ($this->options->get('removeScripts') == true) {
+			$str = mb_eregi_replace("<\s*script[^>]*[^/]>(.*?)<\s*/\s*script\s*>", '', $str);
+			$str = mb_eregi_replace("<\s*script\s*>(.*?)<\s*/\s*script\s*>", '', $str);
+		}
+
+		// strip out <style> tags
+		if ($this->options->get('removeStyles') == true) {
+			$str = mb_eregi_replace("<\s*style[^>]*[^/]>(.*?)<\s*/\s*style\s*>", '', $str);
+			$str = mb_eregi_replace("<\s*style\s*>(.*?)<\s*/\s*style\s*>", '', $str);
+		}
+
+		// strip out preformatted tags
+		$str = mb_eregi_replace("<\s*(?:code)[^>]*>(.*?)<\s*/\s*(?:code)\s*>", '', $str);
+
+		// strip out server side scripts
+		$str = mb_eregi_replace("(<\?)(.*?)(\?>)", '', $str);
+
+		// strip smarty scripts
+		$str = mb_eregi_replace("(\{\w)(.*?)(\})", '', $str);
+
+		return $str;
+	}
+
+	/**
+	 * Attempts to parse the html in content.
+	 */
+	protected function parse()
+	{
+		// add the root node
+		$this->root = new HtmlNode('root');
+		$activeNode = $this->root;
+		while ( ! is_null($activeNode)) {
+			$str = $this->content->copyUntil('<');
+			if ($str == '') {
+				$info = $this->parseTag();
+				if ( ! $info['status']) {
+					// we are done here
+					$activeNode = null;
+					continue;
+				}
+
+				// check if it was a closing tag
+				if ($info['closing']) {
+					$originalNode = $activeNode;
+					while ($activeNode->getTag()->name() != $info['tag']) {
+						$activeNode = $activeNode->getParent();
+						if (is_null($activeNode)) {
+							// we could not find opening tag
+							$activeNode = $originalNode;
+							break;
+						}
+					}
+					if ( ! is_null($activeNode)) {
+						$activeNode = $activeNode->getParent();
+					}
+					continue;
+				}
+
+				if ( ! isset($info['node'])) {
+					continue;
+				}
+
+				/** @var AbstractNode $node */
+				$node = $info['node'];
+				$activeNode->addChild($node);
+
+				// check if node is self closing
+				if ( ! $node->getTag()->isSelfClosing()) {
+					$activeNode = $node;
+				}
+			} else if ($this->options->whitespaceTextNode ||
+				trim($str) != ''
+			) {
+				// we found text we care about
+				$textNode = new TextNode($str);
+				$activeNode->addChild($textNode);
+			}
+		}
+	}
+
+	/**
+	 * Attempt to parse a tag out of the content.
+	 *
+	 * @return array
+	 * @throws StrictException
+	 */
+	protected function parseTag()
+	{
+		$return = [
+			'status'  => false,
+			'closing' => false,
+			'node'    => null,
+		];
+		if ($this->content->char() != '<') {
+			// we are not at the beginning of a tag
+			return $return;
+		}
+
+		// check if this is a closing tag
+		if ($this->content->fastForward(1)->char() == '/') {
+			// end tag
+			$tag = $this->content->fastForward(1)
+								 ->copyByToken('slash', true);
+			// move to end of tag
+			$this->content->copyUntil('>');
+			$this->content->fastForward(1);
+
+			// check if this closing tag counts
+			$tag = strtolower($tag);
+			if (in_array($tag, $this->selfClosing)) {
+				$return['status'] = true;
+
+				return $return;
+			} else {
+				$return['status']  = true;
+				$return['closing'] = true;
+				$return['tag']     = strtolower($tag);
+			}
+
+			return $return;
+		}
+
+		$tag  = strtolower($this->content->copyByToken('slash', true));
+		$node = new HtmlNode($tag);
+
+		// attributes
+		while ($this->content->char() != '>' &&
+			$this->content->char() != '/') {
+			$space = $this->content->skipByToken('blank', true);
+			if (empty($space)) {
+				$this->content->fastForward(1);
+				continue;
+			}
+
+			$name = $this->content->copyByToken('equal', true);
+			if ($name == '/') {
+				break;
+			}
+
+			if (empty($name)) {
+				$this->content->fastForward(1);
+				continue;
+			}
+
+			$this->content->skipByToken('blank');
+			if ($this->content->char() == '=') {
+				$attr = [];
+				$this->content->fastForward(1)
+							  ->skipByToken('blank');
+				switch ($this->content->char()) {
+					case '"':
+						$attr['doubleQuote'] = true;
+						$this->content->fastForward(1);
+						$string = $this->content->copyUntil('"', true, true);
+						do {
+							$moreString = $this->content->copyUntilUnless('"', '=>');
+							$string .= $moreString;
+						} while ( ! empty($moreString));
+						$attr['value'] = $string;
+						$this->content->fastForward(1);
+						$node->getTag()->$name = $attr;
+						break;
+					case "'":
+						$attr['doubleQuote'] = false;
+						$this->content->fastForward(1);
+						$string = $this->content->copyUntil("'", true, true);
+						do {
+							$moreString = $this->content->copyUntilUnless("'", '=>');
+							$string .= $moreString;
+						} while ( ! empty($moreString));
+						$attr['value'] = $string;
+						$this->content->fastForward(1);
+						$node->getTag()->$name = $attr;
+						break;
+					default:
+						$attr['doubleQuote']   = true;
+						$attr['value']         = $this->content->copyByToken('attr', true);
+						$node->getTag()->$name = $attr;
+						break;
+				}
+			} else {
+				// no value attribute
+				if ($this->options->strict) {
+					// can't have this in strict html
+					$character = $this->content->getPosition();
+					throw new StrictException("Tag '$tag' has an attribute '$name' with out a value! (character #$character)");
+				}
+				$node->getTag()->$name = [
+					'value'       => null,
+					'doubleQuote' => true,
+				];
+				if ($this->content->char() != '>') {
+					$this->content->rewind(1);
+				}
+			}
+		}
+
+		$this->content->skipByToken('blank');
+		if ($this->content->char() == '/') {
+			// self closing tag
+			$node->getTag()->selfClosing();
+			$this->content->fastForward(1);
+		} elseif (in_array($tag, $this->selfClosing)) {
+
+			// Should be a self closing tag, check if we are strict
+			if ($this->options->strict) {
+				$character = $this->content->getPosition();
+				throw new StrictException("Tag '$tag' is not self closing! (character #$character)");
+			}
+
+			// We force self closing on this tag.
+			$node->getTag()->selfClosing();
+		}
+
+		$this->content->fastForward(1);
+
+		$return['status'] = true;
+		$return['node']   = $node;
+
+		return $return;
+	}
+
+	/**
+	 * Attempts to detect the charset that the html was sent in.
+	 *
+	 * @return bool
+	 */
+	protected function detectCharset()
+	{
+		// set the default
+		$encode = new Encode;
+		$encode->from($this->defaultCharset);
+		$encode->to($this->defaultCharset);
+
+		if ( ! is_null($this->options->enforceEncoding)) {
+			//  they want to enforce the given encoding
+			$encode->from($this->options->enforceEncoding);
+			$encode->to($this->options->enforceEncoding);
+
+			return false;
+		}
+
+		$meta = $this->root->find('meta[http-equiv=Content-Type]', 0);
+		if (is_null($meta)) {
+			// could not find meta tag
+			$this->root->propagateEncoding($encode);
+
+			return false;
+		}
+		$content = $meta->content;
+		if (empty($content)) {
+			// could not find content
+			$this->root->propagateEncoding($encode);
+
+			return false;
+		}
+		$matches = [];
+		if (preg_match('/charset=(.+)/', $content, $matches)) {
+			$encode->from(trim($matches[1]));
+			$this->root->propagateEncoding($encode);
+
+			return true;
+		}
+
+		// no charset found
+		$this->root->propagateEncoding($encode);
+
+		return false;
+	}
 }

src/PHPHtmlParser/Dom/HtmlNode.php

Indentation +185 added lines, -185 removed lines

@@ -12,189 +12,189 @@
 class HtmlNode extends ArrayNode
 {
 
-    /**
-     * Remembers what the innerHtml was if it was scanned previously.
-     */
-    protected $innerHtml = null;
-
-    /**
-     * Remembers what the outerHtml was if it was scanned previously.
-     *
-     * @var string
-     */
-    protected $outerHtml = null;
-
-    /**
-     * Remembers what the text was if it was scanned previously.
-     *
-     * @var string
-     */
-    protected $text = null;
-
-    /**
-     * Remembers what the text was when we looked into all our
-     * children nodes.
-     *
-     * @var string
-     */
-    protected $textWithChildren = null;
-
-    /**
-     * Sets up the tag of this node.
-     *
-     * @param $tag
-     */
-    public function __construct($tag)
-    {
-        if ( ! $tag instanceof Tag) {
-            $tag = new Tag($tag);
-        }
-        $this->tag = $tag;
-        parent::__construct();
-    }
-
-    /**
-     * Gets the inner html of this node.
-     *
-     * @return string
-     * @throws UnknownChildTypeException
-     */
-    public function innerHtml()
-    {
-        if ( ! $this->hasChildren()) {
-            // no children
-            return '';
-        }
-
-        if ( ! is_null($this->innerHtml)) {
-            // we already know the result.
-            return $this->innerHtml;
-        }
-
-        $child  = $this->firstChild();
-        $string = '';
-
-        // continue to loop until we are out of children
-        while ( ! is_null($child)) {
-            if ($child instanceof TextNode) {
-                $string .= $child->text();
-            } elseif ($child instanceof HtmlNode) {
-                $string .= $child->outerHtml();
-            } else {
-                throw new UnknownChildTypeException('Unknown child type "'.get_class($child).'" found in node');
-            }
-
-            try {
-                $child = $this->nextChild($child->id());
-            } catch (ChildNotFoundException $e) {
-                // no more children
-                $child = null;
-            }
-        }
-
-        // remember the results
-        $this->innerHtml = $string;
-
-        return $string;
-    }
-
-    /**
-     * Gets the html of this node, including it's own
-     * tag.
-     *
-     * @return string
-     */
-    public function outerHtml()
-    {
-        // special handling for root
-        if ($this->tag->name() == 'root') {
-            return $this->innerHtml();
-        }
-
-        if ( ! is_null($this->outerHtml)) {
-            // we already know the results.
-            return $this->outerHtml;
-        }
-
-        $return = $this->tag->makeOpeningTag();
-        if ($this->tag->isSelfClosing()) {
-            // ignore any children... there should not be any though
-            return $return;
-        }
-
-        // get the inner html
-        $return .= $this->innerHtml();
-
-        // add closing tag
-        $return .= $this->tag->makeClosingTag();
-
-        // remember the results
-        $this->outerHtml = $return;
-
-        return $return;
-    }
-
-    /**
-     * Gets the text of this node (if there is any text). Or get all the text
-     * in this node, including children.
-     *
-     * @param bool $lookInChildren
-     * @return string
-     */
-    public function text($lookInChildren = false)
-    {
-        if ($lookInChildren) {
-            if ( ! is_null($this->textWithChildren)) {
-                // we already know the results.
-                return $this->textWithChildren;
-            }
-        } elseif ( ! is_null($this->text)) {
-            // we already know the results.
-            return $this->text;
-        }
-
-        // find out if this node has any text children
-        $text = '';
-        foreach ($this->children as $child) {
-            /** @var AbstractNode $node */
-            $node = $child['node'];
-            if ($node instanceof TextNode) {
-                $text .= $child['node']->text;
-            } elseif ($lookInChildren &&
-                $node instanceof HtmlNode
-            ) {
-                $text .= $node->text($lookInChildren);
-            }
-        }
-
-        // remember our result
-        if ($lookInChildren) {
-            $this->textWithChildren = $text;
-        } else {
-            $this->text = $text;
-        }
-
-        return $text;
-    }
-
-    /**
-     * Call this when something in the node tree has changed. Like a child has been added
-     * or a parent has been changed.
-     */
-    protected function clear()
-    {
-        $this->innerHtml = null;
-        $this->outerHtml = null;
-        $this->text      = null;
-    }
-
-    /**
-     * Returns all children of this html node.
-     *
-     * @return array
-     */
-    protected function getIteratorArray()
-    {
-        return $this->getChildren();
-    }
+	/**
+	 * Remembers what the innerHtml was if it was scanned previously.
+	 */
+	protected $innerHtml = null;
+
+	/**
+	 * Remembers what the outerHtml was if it was scanned previously.
+	 *
+	 * @var string
+	 */
+	protected $outerHtml = null;
+
+	/**
+	 * Remembers what the text was if it was scanned previously.
+	 *
+	 * @var string
+	 */
+	protected $text = null;
+
+	/**
+	 * Remembers what the text was when we looked into all our
+	 * children nodes.
+	 *
+	 * @var string
+	 */
+	protected $textWithChildren = null;
+
+	/**
+	 * Sets up the tag of this node.
+	 *
+	 * @param $tag
+	 */
+	public function __construct($tag)
+	{
+		if ( ! $tag instanceof Tag) {
+			$tag = new Tag($tag);
+		}
+		$this->tag = $tag;
+		parent::__construct();
+	}
+
+	/**
+	 * Gets the inner html of this node.
+	 *
+	 * @return string
+	 * @throws UnknownChildTypeException
+	 */
+	public function innerHtml()
+	{
+		if ( ! $this->hasChildren()) {
+			// no children
+			return '';
+		}
+
+		if ( ! is_null($this->innerHtml)) {
+			// we already know the result.
+			return $this->innerHtml;
+		}
+
+		$child  = $this->firstChild();
+		$string = '';
+
+		// continue to loop until we are out of children
+		while ( ! is_null($child)) {
+			if ($child instanceof TextNode) {
+				$string .= $child->text();
+			} elseif ($child instanceof HtmlNode) {
+				$string .= $child->outerHtml();
+			} else {
+				throw new UnknownChildTypeException('Unknown child type "'.get_class($child).'" found in node');
+			}
+
+			try {
+				$child = $this->nextChild($child->id());
+			} catch (ChildNotFoundException $e) {
+				// no more children
+				$child = null;
+			}
+		}
+
+		// remember the results
+		$this->innerHtml = $string;
+
+		return $string;
+	}
+
+	/**
+	 * Gets the html of this node, including it's own
+	 * tag.
+	 *
+	 * @return string
+	 */
+	public function outerHtml()
+	{
+		// special handling for root
+		if ($this->tag->name() == 'root') {
+			return $this->innerHtml();
+		}
+
+		if ( ! is_null($this->outerHtml)) {
+			// we already know the results.
+			return $this->outerHtml;
+		}
+
+		$return = $this->tag->makeOpeningTag();
+		if ($this->tag->isSelfClosing()) {
+			// ignore any children... there should not be any though
+			return $return;
+		}
+
+		// get the inner html
+		$return .= $this->innerHtml();
+
+		// add closing tag
+		$return .= $this->tag->makeClosingTag();
+
+		// remember the results
+		$this->outerHtml = $return;
+
+		return $return;
+	}
+
+	/**
+	 * Gets the text of this node (if there is any text). Or get all the text
+	 * in this node, including children.
+	 *
+	 * @param bool $lookInChildren
+	 * @return string
+	 */
+	public function text($lookInChildren = false)
+	{
+		if ($lookInChildren) {
+			if ( ! is_null($this->textWithChildren)) {
+				// we already know the results.
+				return $this->textWithChildren;
+			}
+		} elseif ( ! is_null($this->text)) {
+			// we already know the results.
+			return $this->text;
+		}
+
+		// find out if this node has any text children
+		$text = '';
+		foreach ($this->children as $child) {
+			/** @var AbstractNode $node */
+			$node = $child['node'];
+			if ($node instanceof TextNode) {
+				$text .= $child['node']->text;
+			} elseif ($lookInChildren &&
+				$node instanceof HtmlNode
+			) {
+				$text .= $node->text($lookInChildren);
+			}
+		}
+
+		// remember our result
+		if ($lookInChildren) {
+			$this->textWithChildren = $text;
+		} else {
+			$this->text = $text;
+		}
+
+		return $text;
+	}
+
+	/**
+	 * Call this when something in the node tree has changed. Like a child has been added
+	 * or a parent has been changed.
+	 */
+	protected function clear()
+	{
+		$this->innerHtml = null;
+		$this->outerHtml = null;
+		$this->text      = null;
+	}
+
+	/**
+	 * Returns all children of this html node.
+	 *
+	 * @return array
+	 */
+	protected function getIteratorArray()
+	{
+		return $this->getChildren();
+	}
 }

src/PHPHtmlParser/Dom/AbstractNode.php

Indentation +683 added lines, -683 removed lines

@@ -17,687 +17,687 @@
 abstract class AbstractNode
 {
 
-    /**
-     * Contains the tag name/type
-     *
-     * @var \PHPHtmlParser\Dom\Tag
-     */
-    protected $tag;
-
-    /**
-     * Contains a list of attributes on this tag.
-     *
-     * @var array
-     */
-    protected $attr = [];
-
-    /**
-     * An array of all the children.
-     *
-     * @var array
-     */
-    protected $children = [];
-
-    /**
-     * Contains the parent Node.
-     *
-     * @var AbstractNode
-     */
-    protected $parent = null;
-
-    /**
-     * The unique id of the class. Given by PHP.
-     *
-     * @var string
-     */
-    protected $id;
-
-    /**
-     * The encoding class used to encode strings.
-     *
-     * @var mixed
-     */
-    protected $encode;
-
-    /**
-     * Creates a unique spl hash for this node.
-     */
-    public function __construct()
-    {
-        $this->id = spl_object_hash($this);
-    }
-
-    /**
-     * Magic get method for attributes and certain methods.
-     *
-     * @param string $key
-     * @return mixed
-     */
-    public function __get($key)
-    {
-        // check attribute first
-        if ( ! is_null($this->getAttribute($key))) {
-            return $this->getAttribute($key);
-        }
-        switch (strtolower($key)) {
-            case 'outerhtml':
-                return $this->outerHtml();
-            case 'innerhtml':
-                return $this->innerHtml();
-            case 'text':
-                return $this->text();
-        }
-
-        return null;
-    }
-
-    /**
-     * Attempts to clear out any object references.
-     */
-    public function __destruct()
-    {
-        $this->tag      = null;
-        $this->attr     = [];
-        $this->parent   = null;
-        $this->children = [];
-    }
-
-    /**
-     * Simply calls the outer text method.
-     *
-     * @return string
-     */
-    public function __toString()
-    {
-        return $this->outerHtml();
-    }
-
-    /**
-     * Returns the id of this object.
-     */
-    public function id()
-    {
-        return $this->id;
-    }
-
-    /**
-     * Returns the parent of node.
-     *
-     * @return AbstractNode
-     */
-    public function getParent()
-    {
-        return $this->parent;
-    }
-
-    /**
-     * Sets the parent node.
-     *
-     * @param AbstractNode $parent
-     * @return $this
-     * @throws CircularException
-     */
-    public function setParent(AbstractNode $parent)
-    {
-        // check integrity
-        if ($this->isDescendant($parent->id())) {
-            throw new CircularException('Can not add descendant "'.$parent->id().'" as my parent.');
-        }
-
-        // remove from old parent
-        if ( ! is_null($this->parent)) {
-            if ($this->parent->id() == $parent->id()) {
-                // already the parent
-                return $this;
-            }
-
-            $this->parent->removeChild($this->id);
-        }
-
-        $this->parent = $parent;
-
-        // assign child to parent
-        $this->parent->addChild($this);
-
-        //clear any cache
-        $this->clear();
-
-        return $this;
-    }
-
-    /**
-     * Sets the encoding class to this node and propagates it
-     * to all its children.
-     *
-     * @param Encode $encode
-     */
-    public function propagateEncoding(Encode $encode)
-    {
-        $this->encode = $encode;
-        $this->tag->setEncoding($encode);
-        // check children
-        foreach ($this->children as $id => $child) {
-            /** @var AbstractNode $node */
-            $node = $child['node'];
-            $node->propagateEncoding($encode);
-        }
-    }
-
-    /**
-     * Checks if this node has children.
-     *
-     * @return bool
-     */
-    public function hasChildren()
-    {
-        return ! empty($this->children);
-    }
-
-    /**
-     * Returns the child by id.
-     *
-     * @param int $id
-     * @return AbstractNode
-     * @throws ChildNotFoundException
-     */
-    public function getChild($id)
-    {
-        if ( ! isset($this->children[$id])) {
-            throw new ChildNotFoundException("Child '$id' not found in this node.");
-        }
-
-        return $this->children[$id]['node'];
-    }
-
-    /**
-     * Returns a new array of child nodes
-     *
-     * @return array
-     */
-    public function getChildren()
-    {
-        $nodes = [];
-        try {
-            $child = $this->firstChild();
-            do {
-                $nodes[] = $child;
-                $child   = $this->nextChild($child->id());
-            } while ( ! is_null($child));
-        } catch (ChildNotFoundException $e) {
-            // we are done looking for children
-        }
-
-        return $nodes;
-    }
-
-    /**
-     * Counts children
-     *
-     * @return int
-     */
-    public function countChildren()
-    {
-        return count($this->children);
-    }
-
-    /**
-     * Adds a child node to this node and returns the id of the child for this
-     * parent.
-     *
-     * @param AbstractNode $child
-     * @return bool
-     * @throws CircularException
-     */
-    public function addChild(AbstractNode $child)
-    {
-        $key = null;
-
-        // check integrity
-        if ($this->isAncestor($child->id())) {
-            throw new CircularException('Can not add child. It is my ancestor.');
-        }
-
-        // check if child is itself
-        if ($child->id() == $this->id) {
-            throw new CircularException('Can not set itself as a child.');
-        }
-
-        if ($this->hasChildren()) {
-            if (isset($this->children[$child->id()])) {
-                // we already have this child
-                return false;
-            }
-            $sibling                      = $this->lastChild();
-            $key                          = $sibling->id();
-            $this->children[$key]['next'] = $child->id();
-        }
-
-        // add the child
-        $this->children[$child->id()] = [
-            'node' => $child,
-            'next' => null,
-            'prev' => $key,
-        ];
-
-        // tell child I am the new parent
-        $child->setParent($this);
-
-        //clear any cache
-        $this->clear();
-
-        return true;
-    }
-
-    /**
-     * Removes the child by id.
-     *
-     * @param int $id
-     * @return $this
-     */
-    public function removeChild($id)
-    {
-        if ( ! isset($this->children[$id])) {
-            return $this;
-        }
-
-        // handle moving next and previous assignments.
-        $next = $this->children[$id]['next'];
-        $prev = $this->children[$id]['prev'];
-        if ( ! is_null($next)) {
-            $this->children[$next]['prev'] = $prev;
-        }
-        if ( ! is_null($prev)) {
-            $this->children[$prev]['next'] = $next;
-        }
-
-        // remove the child
-        unset($this->children[$id]);
-
-        //clear any cache
-        $this->clear();
-
-        return $this;
-    }
-
-    /**
-     * Attempts to get the next child.
-     *
-     * @param int $id
-     * @return AbstractNode
-     * @uses $this->getChild()
-     */
-    public function nextChild($id)
-    {
-        $child = $this->getChild($id);
-        $next  = $this->children[$child->id()]['next'];
-
-        return $this->getChild($next);
-    }
-
-    /**
-     * Attempts to get the previous child.
-     *
-     * @param int $id
-     * @return AbstractNode
-     * @uses $this->getChild()
-     */
-    public function previousChild($id)
-    {
-        $child = $this->getchild($id);
-        $next  = $this->children[$child->id()]['prev'];
-
-        return $this->getChild($next);
-    }
-
-    /**
-     * Checks if the given node id is a child of the
-     * current node.
-     *
-     * @param int $id
-     * @return bool
-     */
-    public function isChild($id)
-    {
-        foreach ($this->children as $childId => $child) {
-            if ($id == $childId) {
-                return true;
-            }
-        }
-
-        return false;
-    }
-
-    /**
-     * Checks if the given node id is a descendant of the
-     * current node.
-     *
-     * @param int $id
-     * @return bool
-     */
-    public function isDescendant($id)
-    {
-        if ($this->isChild($id)) {
-            return true;
-        }
-
-        foreach ($this->children as $childId => $child) {
-            /** @var AbstractNode $node */
-            $node = $child['node'];
-            if ($node->hasChildren() &&
-                $node->isDescendant($id)
-            ) {
-                return true;
-            }
-        }
-
-        return false;
-    }
-
-    /**
-     * Checks if the given node id is an ancestor of
-     * the current node.
-     *
-     * @param int $id
-     * @return bool
-     */
-    public function isAncestor($id)
-    {
-        if ( ! is_null($this->getAncestor($id))) {
-            return true;
-        }
-
-        return false;
-    }
-
-    /**
-     * Attempts to get an ancestor node by the given id.
-     *
-     * @param int $id
-     * @return null|AbstractNode
-     */
-    public function getAncestor($id)
-    {
-        if ( ! is_null($this->parent)) {
-            if ($this->parent->id() == $id) {
-                return $this->parent;
-            }
-
-            return $this->parent->getAncestor($id);
-        }
-
-        return null;
-    }
-
-    /**
-     * Shortcut to return the first child.
-     *
-     * @return AbstractNode
-     * @uses $this->getChild()
-     */
-    public function firstChild()
-    {
-        reset($this->children);
-        $key = key($this->children);
-
-        return $this->getChild($key);
-    }
-
-    /**
-     * Attempts to get the last child.
-     *
-     * @return AbstractNode
-     */
-    public function lastChild()
-    {
-        end($this->children);
-        $key = key($this->children);
-
-        return $this->getChild($key);
-    }
-
-    /**
-     * Attempts to get the next sibling.
-     *
-     * @return AbstractNode
-     * @throws ParentNotFoundException
-     */
-    public function nextSibling()
-    {
-        if (is_null($this->parent)) {
-            throw new ParentNotFoundException('Parent is not set for this node.');
-        }
-
-        return $this->parent->nextChild($this->id);
-    }
-
-    /**
-     * Attempts to get the previous sibling
-     *
-     * @return AbstractNode
-     * @throws ParentNotFoundException
-     */
-    public function previousSibling()
-    {
-        if (is_null($this->parent)) {
-            throw new ParentNotFoundException('Parent is not set for this node.');
-        }
-
-        return $this->parent->previousChild($this->id);
-    }
-
-    /**
-     * Gets the tag object of this node.
-     *
-     * @return Tag
-     */
-    public function getTag()
-    {
-        return $this->tag;
-    }
-
-    /**
-     * A wrapper method that simply calls the getAttribute method
-     * on the tag of this node.
-     *
-     * @return array
-     */
-    public function getAttributes()
-    {
-        $attributes = $this->tag->getAttributes();
-        foreach ($attributes as $name => $info) {
-            $attributes[$name] = $info['value'];
-        }
-
-        return $attributes;
-    }
-
-    /**
-     * A wrapper method that simply calls the getAttribute method
-     * on the tag of this node.
-     *
-     * @param string $key
-     * @return mixed
-     */
-    public function getAttribute($key)
-    {
-        $attribute = $this->tag->getAttribute($key);
-        if ( ! is_null($attribute)) {
-            $attribute = $attribute['value'];
-        }
-
-        return $attribute;
-    }
-
-    /**
-     * A wrapper method that simply calls the setAttribute method
-     * on the tag of this node.
-     *
-     * @param string $key
-     * @param string $value
-     * @return $this
-     */
-    public function setAttribute($key, $value)
-    {
-        $this->tag->setAttribute($key, $value);
-
-        return $this;
-    }
-
-    /**
-     * Function to locate a specific ancestor tag in the path to the root.
-     *
-     * @param  string $tag
-     * @return AbstractNode
-     * @throws ParentNotFoundException
-     */
-    public function ancestorByTag($tag)
-    {
-        // Start by including ourselves in the comparison.
-        $node = $this;
-
-        while ( ! is_null($node)) {
-            if ($node->tag->name() == $tag) {
-                return $node;
-            }
-
-            $node = $node->getParent();
-        }
-
-        throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag');
-    }
-
-    /**
-     * Find elements by css selector
-     *
-     * @param string $selector
-     * @param int $nth
-     * @return array|AbstractNode
-     */
-    public function find($selector, $nth = null)
-    {
-        $selector = new Selector($selector);
-        $nodes    = $selector->find($this);
-
-        if ( ! is_null($nth)) {
-            // return nth-element or array
-            if (isset($nodes[$nth])) {
-                return $nodes[$nth];
-            }
-
-            return null;
-        }
-
-        return $nodes;
-    }
-
-    /**
-     * Function to try a few tricks to determine the displayed size of an img on the page.
-     * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types.
-     *
-     * Future enhancement:
-     * Look in the tag to see if there is a class or id specified that has a height or width attribute to it.
-     *
-     * Far future enhancement
-     * Look at all the parent tags of this image to see if they specify a class or id that has an img selector that specifies a height or width
-     * Note that in this case, the class or id will have the img sub-selector for it to apply to the image.
-     *
-     * ridiculously far future development
-     * If the class or id is specified in a SEPARATE css file that's not on the page, go get it and do what we were just doing for the ones on the page.
-     *
-     * @author John Schlick
-     * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out.
-     */
-    public function get_display_size()
-    {
-        $width  = -1;
-        $height = -1;
-
-        if ($this->tag->name() != 'img') {
-            return false;
-        }
-
-        // See if there is a height or width attribute in the tag itself.
-        if ( ! is_null($this->tag->getAttribute('width'))) {
-            $width = $this->tag->getAttribute('width');
-        }
-
-        if ( ! is_null($this->tag->getAttribute('height'))) {
-            $height = $this->tag->getAttribute('height');
-        }
-
-        // Now look for an inline style.
-        if ( ! is_null($this->tag->getAttribute('style'))) {
-            // Thanks to user 'gnarf' from stackoverflow for this regular expression.
-            $attributes = [];
-            preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches,
-                PREG_SET_ORDER);
-            foreach ($matches as $match) {
-                $attributes[$match[1]] = $match[2];
-            }
-
-            $width = $this->getLength($attributes, $width, 'width');
-            $height = $this->getLength($attributes, $width, 'height');
-        }
-
-        $result = [
-            'height' => $height,
-            'width'  => $width,
-        ];
-
-        return $result;
-    }
-
-    /**
-     * If there is a length in the style attributes use it.
-     *
-     * @param array $attributes
-     * @param int $length
-     * @param string $key
-     * @return int
-     */
-    protected function getLength(array $attributes, $length, $key)
-    {
-        if (isset($attributes[$key]) && $length == -1) {
-            // check that the last two characters are px (pixels)
-            if (strtolower(substr($attributes[$key], -2)) == 'px') {
-                $proposed_length = substr($attributes[$key], 0, -2);
-                // Now make sure that it's an integer and not something stupid.
-                if (filter_var($proposed_length, FILTER_VALIDATE_INT)) {
-                    $length = $proposed_length;
-                }
-            }
-        }
-
-        return $length;
-    }
-
-    /**
-     * Gets the inner html of this node.
-     *
-     * @return string
-     */
-    abstract public function innerHtml();
-
-    /**
-     * Gets the html of this node, including it's own
-     * tag.
-     *
-     * @return string
-     */
-    abstract public function outerHtml();
-
-    /**
-     * Gets the text of this node (if there is any text).
-     *
-     * @return string
-     */
-    abstract public function text();
-
-    /**
-     * Call this when something in the node tree has changed. Like a child has been added
-     * or a parent has been changed.
-     *
-     * @return void
-     */
-    abstract protected function clear();
+	/**
+	 * Contains the tag name/type
+	 *
+	 * @var \PHPHtmlParser\Dom\Tag
+	 */
+	protected $tag;
+
+	/**
+	 * Contains a list of attributes on this tag.
+	 *
+	 * @var array
+	 */
+	protected $attr = [];
+
+	/**
+	 * An array of all the children.
+	 *
+	 * @var array
+	 */
+	protected $children = [];
+
+	/**
+	 * Contains the parent Node.
+	 *
+	 * @var AbstractNode
+	 */
+	protected $parent = null;
+
+	/**
+	 * The unique id of the class. Given by PHP.
+	 *
+	 * @var string
+	 */
+	protected $id;
+
+	/**
+	 * The encoding class used to encode strings.
+	 *
+	 * @var mixed
+	 */
+	protected $encode;
+
+	/**
+	 * Creates a unique spl hash for this node.
+	 */
+	public function __construct()
+	{
+		$this->id = spl_object_hash($this);
+	}
+
+	/**
+	 * Magic get method for attributes and certain methods.
+	 *
+	 * @param string $key
+	 * @return mixed
+	 */
+	public function __get($key)
+	{
+		// check attribute first
+		if ( ! is_null($this->getAttribute($key))) {
+			return $this->getAttribute($key);
+		}
+		switch (strtolower($key)) {
+			case 'outerhtml':
+				return $this->outerHtml();
+			case 'innerhtml':
+				return $this->innerHtml();
+			case 'text':
+				return $this->text();
+		}
+
+		return null;
+	}
+
+	/**
+	 * Attempts to clear out any object references.
+	 */
+	public function __destruct()
+	{
+		$this->tag      = null;
+		$this->attr     = [];
+		$this->parent   = null;
+		$this->children = [];
+	}
+
+	/**
+	 * Simply calls the outer text method.
+	 *
+	 * @return string
+	 */
+	public function __toString()
+	{
+		return $this->outerHtml();
+	}
+
+	/**
+	 * Returns the id of this object.
+	 */
+	public function id()
+	{
+		return $this->id;
+	}
+
+	/**
+	 * Returns the parent of node.
+	 *
+	 * @return AbstractNode
+	 */
+	public function getParent()
+	{
+		return $this->parent;
+	}
+
+	/**
+	 * Sets the parent node.
+	 *
+	 * @param AbstractNode $parent
+	 * @return $this
+	 * @throws CircularException
+	 */
+	public function setParent(AbstractNode $parent)
+	{
+		// check integrity
+		if ($this->isDescendant($parent->id())) {
+			throw new CircularException('Can not add descendant "'.$parent->id().'" as my parent.');
+		}
+
+		// remove from old parent
+		if ( ! is_null($this->parent)) {
+			if ($this->parent->id() == $parent->id()) {
+				// already the parent
+				return $this;
+			}
+
+			$this->parent->removeChild($this->id);
+		}
+
+		$this->parent = $parent;
+
+		// assign child to parent
+		$this->parent->addChild($this);
+
+		//clear any cache
+		$this->clear();
+
+		return $this;
+	}
+
+	/**
+	 * Sets the encoding class to this node and propagates it
+	 * to all its children.
+	 *
+	 * @param Encode $encode
+	 */
+	public function propagateEncoding(Encode $encode)
+	{
+		$this->encode = $encode;
+		$this->tag->setEncoding($encode);
+		// check children
+		foreach ($this->children as $id => $child) {
+			/** @var AbstractNode $node */
+			$node = $child['node'];
+			$node->propagateEncoding($encode);
+		}
+	}
+
+	/**
+	 * Checks if this node has children.
+	 *
+	 * @return bool
+	 */
+	public function hasChildren()
+	{
+		return ! empty($this->children);
+	}
+
+	/**
+	 * Returns the child by id.
+	 *
+	 * @param int $id
+	 * @return AbstractNode
+	 * @throws ChildNotFoundException
+	 */
+	public function getChild($id)
+	{
+		if ( ! isset($this->children[$id])) {
+			throw new ChildNotFoundException("Child '$id' not found in this node.");
+		}
+
+		return $this->children[$id]['node'];
+	}
+
+	/**
+	 * Returns a new array of child nodes
+	 *
+	 * @return array
+	 */
+	public function getChildren()
+	{
+		$nodes = [];
+		try {
+			$child = $this->firstChild();
+			do {
+				$nodes[] = $child;
+				$child   = $this->nextChild($child->id());
+			} while ( ! is_null($child));
+		} catch (ChildNotFoundException $e) {
+			// we are done looking for children
+		}
+
+		return $nodes;
+	}
+
+	/**
+	 * Counts children
+	 *
+	 * @return int
+	 */
+	public function countChildren()
+	{
+		return count($this->children);
+	}
+
+	/**
+	 * Adds a child node to this node and returns the id of the child for this
+	 * parent.
+	 *
+	 * @param AbstractNode $child
+	 * @return bool
+	 * @throws CircularException
+	 */
+	public function addChild(AbstractNode $child)
+	{
+		$key = null;
+
+		// check integrity
+		if ($this->isAncestor($child->id())) {
+			throw new CircularException('Can not add child. It is my ancestor.');
+		}
+
+		// check if child is itself
+		if ($child->id() == $this->id) {
+			throw new CircularException('Can not set itself as a child.');
+		}
+
+		if ($this->hasChildren()) {
+			if (isset($this->children[$child->id()])) {
+				// we already have this child
+				return false;
+			}
+			$sibling                      = $this->lastChild();
+			$key                          = $sibling->id();
+			$this->children[$key]['next'] = $child->id();
+		}
+
+		// add the child
+		$this->children[$child->id()] = [
+			'node' => $child,
+			'next' => null,
+			'prev' => $key,
+		];
+
+		// tell child I am the new parent
+		$child->setParent($this);
+
+		//clear any cache
+		$this->clear();
+
+		return true;
+	}
+
+	/**
+	 * Removes the child by id.
+	 *
+	 * @param int $id
+	 * @return $this
+	 */
+	public function removeChild($id)
+	{
+		if ( ! isset($this->children[$id])) {
+			return $this;
+		}
+
+		// handle moving next and previous assignments.
+		$next = $this->children[$id]['next'];
+		$prev = $this->children[$id]['prev'];
+		if ( ! is_null($next)) {
+			$this->children[$next]['prev'] = $prev;
+		}
+		if ( ! is_null($prev)) {
+			$this->children[$prev]['next'] = $next;
+		}
+
+		// remove the child
+		unset($this->children[$id]);
+
+		//clear any cache
+		$this->clear();
+
+		return $this;
+	}
+
+	/**
+	 * Attempts to get the next child.
+	 *
+	 * @param int $id
+	 * @return AbstractNode
+	 * @uses $this->getChild()
+	 */
+	public function nextChild($id)
+	{
+		$child = $this->getChild($id);
+		$next  = $this->children[$child->id()]['next'];
+
+		return $this->getChild($next);
+	}
+
+	/**
+	 * Attempts to get the previous child.
+	 *
+	 * @param int $id
+	 * @return AbstractNode
+	 * @uses $this->getChild()
+	 */
+	public function previousChild($id)
+	{
+		$child = $this->getchild($id);
+		$next  = $this->children[$child->id()]['prev'];
+
+		return $this->getChild($next);
+	}
+
+	/**
+	 * Checks if the given node id is a child of the
+	 * current node.
+	 *
+	 * @param int $id
+	 * @return bool
+	 */
+	public function isChild($id)
+	{
+		foreach ($this->children as $childId => $child) {
+			if ($id == $childId) {
+				return true;
+			}
+		}
+
+		return false;
+	}
+
+	/**
+	 * Checks if the given node id is a descendant of the
+	 * current node.
+	 *
+	 * @param int $id
+	 * @return bool
+	 */
+	public function isDescendant($id)
+	{
+		if ($this->isChild($id)) {
+			return true;
+		}
+
+		foreach ($this->children as $childId => $child) {
+			/** @var AbstractNode $node */
+			$node = $child['node'];
+			if ($node->hasChildren() &&
+				$node->isDescendant($id)
+			) {
+				return true;
+			}
+		}
+
+		return false;
+	}
+
+	/**
+	 * Checks if the given node id is an ancestor of
+	 * the current node.
+	 *
+	 * @param int $id
+	 * @return bool
+	 */
+	public function isAncestor($id)
+	{
+		if ( ! is_null($this->getAncestor($id))) {
+			return true;
+		}
+
+		return false;
+	}
+
+	/**
+	 * Attempts to get an ancestor node by the given id.
+	 *
+	 * @param int $id
+	 * @return null|AbstractNode
+	 */
+	public function getAncestor($id)
+	{
+		if ( ! is_null($this->parent)) {
+			if ($this->parent->id() == $id) {
+				return $this->parent;
+			}
+
+			return $this->parent->getAncestor($id);
+		}
+
+		return null;
+	}
+
+	/**
+	 * Shortcut to return the first child.
+	 *
+	 * @return AbstractNode
+	 * @uses $this->getChild()
+	 */
+	public function firstChild()
+	{
+		reset($this->children);
+		$key = key($this->children);
+
+		return $this->getChild($key);
+	}
+
+	/**
+	 * Attempts to get the last child.
+	 *
+	 * @return AbstractNode
+	 */
+	public function lastChild()
+	{
+		end($this->children);
+		$key = key($this->children);
+
+		return $this->getChild($key);
+	}
+
+	/**
+	 * Attempts to get the next sibling.
+	 *
+	 * @return AbstractNode
+	 * @throws ParentNotFoundException
+	 */
+	public function nextSibling()
+	{
+		if (is_null($this->parent)) {
+			throw new ParentNotFoundException('Parent is not set for this node.');
+		}
+
+		return $this->parent->nextChild($this->id);
+	}
+
+	/**
+	 * Attempts to get the previous sibling
+	 *
+	 * @return AbstractNode
+	 * @throws ParentNotFoundException
+	 */
+	public function previousSibling()
+	{
+		if (is_null($this->parent)) {
+			throw new ParentNotFoundException('Parent is not set for this node.');
+		}
+
+		return $this->parent->previousChild($this->id);
+	}
+
+	/**
+	 * Gets the tag object of this node.
+	 *
+	 * @return Tag
+	 */
+	public function getTag()
+	{
+		return $this->tag;
+	}
+
+	/**
+	 * A wrapper method that simply calls the getAttribute method
+	 * on the tag of this node.
+	 *
+	 * @return array
+	 */
+	public function getAttributes()
+	{
+		$attributes = $this->tag->getAttributes();
+		foreach ($attributes as $name => $info) {
+			$attributes[$name] = $info['value'];
+		}
+
+		return $attributes;
+	}
+
+	/**
+	 * A wrapper method that simply calls the getAttribute method
+	 * on the tag of this node.
+	 *
+	 * @param string $key
+	 * @return mixed
+	 */
+	public function getAttribute($key)
+	{
+		$attribute = $this->tag->getAttribute($key);
+		if ( ! is_null($attribute)) {
+			$attribute = $attribute['value'];
+		}
+
+		return $attribute;
+	}
+
+	/**
+	 * A wrapper method that simply calls the setAttribute method
+	 * on the tag of this node.
+	 *
+	 * @param string $key
+	 * @param string $value
+	 * @return $this
+	 */
+	public function setAttribute($key, $value)
+	{
+		$this->tag->setAttribute($key, $value);
+
+		return $this;
+	}
+
+	/**
+	 * Function to locate a specific ancestor tag in the path to the root.
+	 *
+	 * @param  string $tag
+	 * @return AbstractNode
+	 * @throws ParentNotFoundException
+	 */
+	public function ancestorByTag($tag)
+	{
+		// Start by including ourselves in the comparison.
+		$node = $this;
+
+		while ( ! is_null($node)) {
+			if ($node->tag->name() == $tag) {
+				return $node;
+			}
+
+			$node = $node->getParent();
+		}
+
+		throw new ParentNotFoundException('Could not find an ancestor with "'.$tag.'" tag');
+	}
+
+	/**
+	 * Find elements by css selector
+	 *
+	 * @param string $selector
+	 * @param int $nth
+	 * @return array|AbstractNode
+	 */
+	public function find($selector, $nth = null)
+	{
+		$selector = new Selector($selector);
+		$nodes    = $selector->find($this);
+
+		if ( ! is_null($nth)) {
+			// return nth-element or array
+			if (isset($nodes[$nth])) {
+				return $nodes[$nth];
+			}
+
+			return null;
+		}
+
+		return $nodes;
+	}
+
+	/**
+	 * Function to try a few tricks to determine the displayed size of an img on the page.
+	 * NOTE: This will ONLY work on an IMG tag. Returns FALSE on all other tag types.
+	 *
+	 * Future enhancement:
+	 * Look in the tag to see if there is a class or id specified that has a height or width attribute to it.
+	 *
+	 * Far future enhancement
+	 * Look at all the parent tags of this image to see if they specify a class or id that has an img selector that specifies a height or width
+	 * Note that in this case, the class or id will have the img sub-selector for it to apply to the image.
+	 *
+	 * ridiculously far future development
+	 * If the class or id is specified in a SEPARATE css file that's not on the page, go get it and do what we were just doing for the ones on the page.
+	 *
+	 * @author John Schlick
+	 * @return array an array containing the 'height' and 'width' of the image on the page or -1 if we can't figure it out.
+	 */
+	public function get_display_size()
+	{
+		$width  = -1;
+		$height = -1;
+
+		if ($this->tag->name() != 'img') {
+			return false;
+		}
+
+		// See if there is a height or width attribute in the tag itself.
+		if ( ! is_null($this->tag->getAttribute('width'))) {
+			$width = $this->tag->getAttribute('width');
+		}
+
+		if ( ! is_null($this->tag->getAttribute('height'))) {
+			$height = $this->tag->getAttribute('height');
+		}
+
+		// Now look for an inline style.
+		if ( ! is_null($this->tag->getAttribute('style'))) {
+			// Thanks to user 'gnarf' from stackoverflow for this regular expression.
+			$attributes = [];
+			preg_match_all("/([\w-]+)\s*:\s*([^;]+)\s*;?/", $this->tag->getAttribute('style'), $matches,
+				PREG_SET_ORDER);
+			foreach ($matches as $match) {
+				$attributes[$match[1]] = $match[2];
+			}
+
+			$width = $this->getLength($attributes, $width, 'width');
+			$height = $this->getLength($attributes, $width, 'height');
+		}
+
+		$result = [
+			'height' => $height,
+			'width'  => $width,
+		];
+
+		return $result;
+	}
+
+	/**
+	 * If there is a length in the style attributes use it.
+	 *
+	 * @param array $attributes
+	 * @param int $length
+	 * @param string $key
+	 * @return int
+	 */
+	protected function getLength(array $attributes, $length, $key)
+	{
+		if (isset($attributes[$key]) && $length == -1) {
+			// check that the last two characters are px (pixels)
+			if (strtolower(substr($attributes[$key], -2)) == 'px') {
+				$proposed_length = substr($attributes[$key], 0, -2);
+				// Now make sure that it's an integer and not something stupid.
+				if (filter_var($proposed_length, FILTER_VALIDATE_INT)) {
+					$length = $proposed_length;
+				}
+			}
+		}
+
+		return $length;
+	}
+
+	/**
+	 * Gets the inner html of this node.
+	 *
+	 * @return string
+	 */
+	abstract public function innerHtml();
+
+	/**
+	 * Gets the html of this node, including it's own
+	 * tag.
+	 *
+	 * @return string
+	 */
+	abstract public function outerHtml();
+
+	/**
+	 * Gets the text of this node (if there is any text).
+	 *
+	 * @return string
+	 */
+	abstract public function text();
+
+	/**
+	 * Call this when something in the node tree has changed. Like a child has been added
+	 * or a parent has been changed.
+	 *
+	 * @return void
+	 */
+	abstract protected function clear();
 }

src/PHPHtmlParser/Dom/Tag.php

Indentation +233 added lines, -233 removed lines

@@ -12,237 +12,237 @@
 class Tag
 {
 
-    /**
-     * The name of the tag.
-     *
-     * @var string
-     */
-    protected $name;
-
-    /**
-     * The attributes of the tag.
-     *
-     * @var array
-     */
-    protected $attr = [];
-
-    /**
-     * Is this tag self closing.
-     *
-     * @var bool
-     */
-    protected $selfClosing = false;
-
-    /**
-     * Tag noise
-     */
-    protected $noise = '';
-
-    /**
-     * The encoding class to... encode the tags
-     *
-     * @var mixed
-     */
-    protected $encode = null;
-
-    /**
-     * Sets up the tag with a name.
-     *
-     * @param $name
-     */
-    public function __construct($name)
-    {
-        $this->name = $name;
-    }
-
-    /**
-     * Magic method to get any of the attributes.
-     *
-     * @param string $key
-     * @return mixed
-     */
-    public function __get($key)
-    {
-        return $this->getAttribute($key);
-    }
-
-    /**
-     * Magic method to set any attribute.
-     *
-     * @param string $key
-     * @param mixed $value
-     */
-    public function __set($key, $value)
-    {
-        $this->setAttribute($key, $value);
-    }
-
-    /**
-     * Returns the name of this tag.
-     *
-     * @return string
-     */
-    public function name()
-    {
-        return $this->name;
-    }
-
-    /**
-     * Sets the tag to be self closing.
-     *
-     * @return $this
-     */
-    public function selfClosing()
-    {
-        $this->selfClosing = true;
-
-        return $this;
-    }
-
-    /**
-     * Checks if the tag is self closing.
-     *
-     * @return bool
-     */
-    public function isSelfClosing()
-    {
-        return $this->selfClosing;
-    }
-
-    /**
-     * Sets the encoding type to be used.
-     *
-     * @param Encode $encode
-     */
-    public function setEncoding(Encode $encode)
-    {
-        $this->encode = $encode;
-    }
-
-    /**
-     * Sets the noise for this tag (if any)
-     *
-     * @param $noise
-     * @return $this
-     */
-    public function noise($noise)
-    {
-        $this->noise = $noise;
-
-        return $this;
-    }
-
-    /**
-     * Set an attribute for this tag.
-     *
-     * @param string $key
-     * @param mixed $value
-     * @return $this
-     */
-    public function setAttribute($key, $value)
-    {
-        $key = strtolower($key);
-        if ( ! is_array($value)) {
-            $value = [
-                'value'       => $value,
-                'doubleQuote' => true,
-            ];
-        }
-        $this->attr[$key] = $value;
-
-        return $this;
-    }
-
-    /**
-     * Sets the attributes for this tag
-     *
-     * @param array $attr
-     * @return $this
-     */
-    public function setAttributes(array $attr)
-    {
-        foreach ($attr as $key => $value) {
-            $this->setAttribute($key, $value);
-        }
-
-        return $this;
-    }
-
-    /**
-     * Returns all attributes of this tag.
-     *
-     * @return array
-     */
-    public function getAttributes()
-    {
-        $return = [];
-        foreach ($this->attr as $attr => $info) {
-            $return[$attr] = $this->getAttribute($attr);
-        }
-
-        return $return;
-    }
-
-    /**
-     * Returns an attribute by the key
-     *
-     * @param string $key
-     * @return mixed
-     */
-    public function getAttribute($key)
-    {
-        if ( ! isset($this->attr[$key])) {
-            return null;
-        }
-        $value = $this->attr[$key]['value'];
-        if (is_string($value) && ! is_null($this->encode)) {
-            // convert charset
-            $this->attr[$key]['value'] = $this->encode->convert($value);
-        }
-
-        return $this->attr[$key];
-    }
-
-    /**
-     * Generates the opening tag for this object.
-     *
-     * @return string
-     */
-    public function makeOpeningTag()
-    {
-        $return = '<'.$this->name;
-
-        // add the attributes
-        foreach ($this->attr as $key => $info) {
-            $info = $this->getAttribute($key);
-            $val  = $info['value'];
-            if (is_null($val)) {
-                $return .= ' '.$key;
-            } elseif ($info['doubleQuote']) {
-                $return .= ' '.$key.'="'.$val.'"';
-            } else {
-                $return .= ' '.$key.'=\''.$val.'\'';
-            }
-        }
-
-        if ($this->selfClosing) {
-            return $return.' />';
-        } else {
-            return $return.'>';
-        }
-    }
-
-    /**
-     * Generates the closing tag for this object.
-     *
-     * @return string
-     */
-    public function makeClosingTag()
-    {
-        if ($this->selfClosing) {
-            return '';
-        }
-
-        return '</'.$this->name.'>';
-    }
+	/**
+	 * The name of the tag.
+	 *
+	 * @var string
+	 */
+	protected $name;
+
+	/**
+	 * The attributes of the tag.
+	 *
+	 * @var array
+	 */
+	protected $attr = [];
+
+	/**
+	 * Is this tag self closing.
+	 *
+	 * @var bool
+	 */
+	protected $selfClosing = false;
+
+	/**
+	 * Tag noise
+	 */
+	protected $noise = '';
+
+	/**
+	 * The encoding class to... encode the tags
+	 *
+	 * @var mixed
+	 */
+	protected $encode = null;
+
+	/**
+	 * Sets up the tag with a name.
+	 *
+	 * @param $name
+	 */
+	public function __construct($name)
+	{
+		$this->name = $name;
+	}
+
+	/**
+	 * Magic method to get any of the attributes.
+	 *
+	 * @param string $key
+	 * @return mixed
+	 */
+	public function __get($key)
+	{
+		return $this->getAttribute($key);
+	}
+
+	/**
+	 * Magic method to set any attribute.
+	 *
+	 * @param string $key
+	 * @param mixed $value
+	 */
+	public function __set($key, $value)
+	{
+		$this->setAttribute($key, $value);
+	}
+
+	/**
+	 * Returns the name of this tag.
+	 *
+	 * @return string
+	 */
+	public function name()
+	{
+		return $this->name;
+	}
+
+	/**
+	 * Sets the tag to be self closing.
+	 *
+	 * @return $this
+	 */
+	public function selfClosing()
+	{
+		$this->selfClosing = true;
+
+		return $this;
+	}
+
+	/**
+	 * Checks if the tag is self closing.
+	 *
+	 * @return bool
+	 */
+	public function isSelfClosing()
+	{
+		return $this->selfClosing;
+	}
+
+	/**
+	 * Sets the encoding type to be used.
+	 *
+	 * @param Encode $encode
+	 */
+	public function setEncoding(Encode $encode)
+	{
+		$this->encode = $encode;
+	}
+
+	/**
+	 * Sets the noise for this tag (if any)
+	 *
+	 * @param $noise
+	 * @return $this
+	 */
+	public function noise($noise)
+	{
+		$this->noise = $noise;
+
+		return $this;
+	}
+
+	/**
+	 * Set an attribute for this tag.
+	 *
+	 * @param string $key
+	 * @param mixed $value
+	 * @return $this
+	 */
+	public function setAttribute($key, $value)
+	{
+		$key = strtolower($key);
+		if ( ! is_array($value)) {
+			$value = [
+				'value'       => $value,
+				'doubleQuote' => true,
+			];
+		}
+		$this->attr[$key] = $value;
+
+		return $this;
+	}
+
+	/**
+	 * Sets the attributes for this tag
+	 *
+	 * @param array $attr
+	 * @return $this
+	 */
+	public function setAttributes(array $attr)
+	{
+		foreach ($attr as $key => $value) {
+			$this->setAttribute($key, $value);
+		}
+
+		return $this;
+	}
+
+	/**
+	 * Returns all attributes of this tag.
+	 *
+	 * @return array
+	 */
+	public function getAttributes()
+	{
+		$return = [];
+		foreach ($this->attr as $attr => $info) {
+			$return[$attr] = $this->getAttribute($attr);
+		}
+
+		return $return;
+	}
+
+	/**
+	 * Returns an attribute by the key
+	 *
+	 * @param string $key
+	 * @return mixed
+	 */
+	public function getAttribute($key)
+	{
+		if ( ! isset($this->attr[$key])) {
+			return null;
+		}
+		$value = $this->attr[$key]['value'];
+		if (is_string($value) && ! is_null($this->encode)) {
+			// convert charset
+			$this->attr[$key]['value'] = $this->encode->convert($value);
+		}
+
+		return $this->attr[$key];
+	}
+
+	/**
+	 * Generates the opening tag for this object.
+	 *
+	 * @return string
+	 */
+	public function makeOpeningTag()
+	{
+		$return = '<'.$this->name;
+
+		// add the attributes
+		foreach ($this->attr as $key => $info) {
+			$info = $this->getAttribute($key);
+			$val  = $info['value'];
+			if (is_null($val)) {
+				$return .= ' '.$key;
+			} elseif ($info['doubleQuote']) {
+				$return .= ' '.$key.'="'.$val.'"';
+			} else {
+				$return .= ' '.$key.'=\''.$val.'\'';
+			}
+		}
+
+		if ($this->selfClosing) {
+			return $return.' />';
+		} else {
+			return $return.'>';
+		}
+	}
+
+	/**
+	 * Generates the closing tag for this object.
+	 *
+	 * @return string
+	 */
+	public function makeClosingTag()
+	{
+		if ($this->selfClosing) {
+			return '';
+		}
+
+		return '</'.$this->name.'>';
+	}
 }

src/PHPHtmlParser/Selector.php

Indentation +355 added lines, -355 removed lines

@@ -13,359 +13,359 @@
 class Selector
 {
 
-    /**
-     * Pattern of CSS selectors, modified from 'mootools'
-     *
-     * @var string
-     */
-    protected $pattern = "/([\w-:\*>]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is";
-
-    protected $selectors = [];
-
-    /**
-     * Constructs with the selector string
-     *
-     * @param string $selector
-     */
-    public function __construct($selector)
-    {
-        $this->parseSelectorString($selector);
-    }
-
-    /**
-     * Returns the selectors that where found in __construct
-     *
-     * @return array
-     */
-    public function getSelectors()
-    {
-        return $this->selectors;
-    }
-
-    /**
-     * Attempts to find the selectors starting from the given
-     * node object.
-     *
-     * @param AbstractNode $node
-     * @return array|Collection
-     */
-    public function find(AbstractNode $node)
-    {
-        $results = new Collection;
-        foreach ($this->selectors as $selector) {
-            $nodes = [$node];
-            if (count($selector) == 0) {
-                continue;
-            }
-
-            $options = [];
-            foreach ($selector as $rule) {
-                if ($rule['alterNext']) {
-                    $options[] = $this->alterNext($rule);
-                    continue;
-                }
-                $nodes = $this->seek($nodes, $rule, $options);
-                // clear the options
-                $options = [];
-            }
-
-            // this is the final set of nodes
-            foreach ($nodes as $result) {
-                $results[] = $result;
-            }
-        }
-
-        return $results;
-    }
-
-    /**
-     * Parses the selector string
-     *
-     * @param string $selector
-     */
-    protected function parseSelectorString($selector)
-    {
-        $matches = [];
-        preg_match_all($this->pattern, trim($selector).' ', $matches, PREG_SET_ORDER);
-
-        // skip tbody
-        $result = [];
-        foreach ($matches as $match) {
-            // default values
-            $tag       = strtolower(trim($match[1]));
-            $operator  = '=';
-            $key       = null;
-            $value     = null;
-            $noKey     = false;
-            $alterNext = false;
-
-            // check for elements that alter the behavior of the next element
-            if ($tag == '>') {
-                $alterNext = true;
-            }
-
-            // check for id selector
-            if ( ! empty($match[2])) {
-                $key   = 'id';
-                $value = $match[2];
-            }
-
-            // check for class selector
-            if ( ! empty($match[3])) {
-                $key   = 'class';
-                $value = $match[3];
-            }
-
-            // and final attribute selector
-            if ( ! empty($match[4])) {
-                $key = strtolower($match[4]);
-            }
-            if ( ! empty($match[5])) {
-                $operator = $match[5];
-            }
-            if ( ! empty($match[6])) {
-                $value = $match[6];
-            }
-
-            // check for elements that do not have a specified attribute
-            if (isset($key[0]) && $key[0] == '!') {
-                $key   = substr($key, 1);
-                $noKey = true;
-            }
-
-            $result[] = [
-                'tag'       => $tag,
-                'key'       => $key,
-                'value'     => $value,
-                'operator'  => $operator,
-                'noKey'     => $noKey,
-                'alterNext' => $alterNext,
-            ];
-            if (trim($match[7]) == ',') {
-                $this->selectors[] = $result;
-                $result            = [];
-            }
-        }
-
-        // save last results
-        if (count($result) > 0) {
-            $this->selectors[] = $result;
-        }
-    }
-
-    /**
-     * Attempts to find all children that match the rule
-     * given.
-     *
-     * @param array $nodes
-     * @param array $rule
-     * @param array $options
-     * @return array
-     * @recursive
-     */
-    protected function seek(array $nodes, array $rule, array $options)
-    {
-        // XPath index
-        if (count($rule['tag']) > 0 &&
-            count($rule['key']) > 0 &&
-            is_numeric($rule['key'])
-        ) {
-            $count = 0;
-            /** @var AbstractNode $node */
-            foreach ($nodes as $node) {
-                if ($rule['tag'] == '*' ||
-                    $rule['tag'] == $node->getTag()->name()) {
-                    ++$count;
-                    if ($count == $rule['key']) {
-                        // found the node we wanted
-                        return [$node];
-                    }
-                }
-            }
-
-            return [];
-        }
-
-        $options = $this->flattenOptions($options);
-
-        $return = [];
-        /** @var AbstractNode $node */
-        foreach ($nodes as $node) {
-            // check if we are a leaf
-            if ( ! $node->hasChildren()) {
-                continue;
-            }
-
-            $children = [];
-            $child    = $node->firstChild();
-            while ( ! is_null($child)) {
-                // wild card, grab all
-                if ($rule['tag'] == '*' && is_null($rule['key'])) {
-                    $return[] = $child;
-                    try {
-                        $child = $node->nextChild($child->id());
-                    } catch (ChildNotFoundException $e) {
-                        // no more children
-                        $child = null;
-                    }
-                    continue;
-                }
-
-                $pass = true;
-                // check tag
-                if ( ! empty($rule['tag']) && $rule['tag'] != $child->getTag()->name() &&
-                    $rule['tag'] != '*'
-                ) {
-                    // child failed tag check
-                    $pass = false;
-                }
-
-                // check key
-                if ($pass && ! is_null($rule['key'])) {
-                    if ($rule['noKey']) {
-                        if ( ! is_null($child->getAttribute($rule['key']))) {
-                            $pass = false;
-                        }
-                    } else {
-                        if ($rule['key'] != 'plaintext' &&
-                            is_null($child->getAttribute($rule['key']))
-                        ) {
-                            $pass = false;
-                        }
-                    }
-                }
-
-                // compare values
-                if ($pass && ! is_null($rule['key']) &&
-                    ! is_null($rule['value']) && $rule['value'] != '*'
-                ) {
-                    if ($rule['key'] == 'plaintext') {
-                        // plaintext search
-                        $nodeValue = $child->text();
-                    } else {
-                        // normal search
-                        $nodeValue = $child->getAttribute($rule['key']);
-                    }
-
-                    $check = $this->match($rule['operator'], $rule['value'], $nodeValue);
-
-                    // handle multiple classes
-                    if ( ! $check && $rule['key'] == 'class') {
-                        $childClasses = explode(' ', $child->getAttribute('class'));
-                        foreach ($childClasses as $class) {
-                            if ( ! empty($class)) {
-                                $check = $this->match($rule['operator'], $rule['value'], $class);
-                            }
-                            if ($check) {
-                                break;
-                            }
-                        }
-                    }
-
-                    if ( ! $check) {
-                        $pass = false;
-                    }
-                }
-
-                if ($pass) {
-                    // it passed all checks
-                    $return[] = $child;
-                } else {
-                    // this child failed to be matched
-                    if ($child->hasChildren()) {
-                        // we still want to check its children
-                        $children[] = $child;
-                    }
-                }
-
-                try {
-                    // get next child
-                    $child = $node->nextChild($child->id());
-                } catch (ChildNotFoundException $e) {
-                    // no more children
-                    $child = null;
-                }
-            }
-
-            if (( ! isset($options['checkGrandChildren']) ||
-                    $options['checkGrandChildren'])
-                && count($children) > 0
-            ) {
-                // we have children that failed but are not leaves.
-                $matches = $this->seek($children, $rule, $options);
-                foreach ($matches as $match) {
-                    $return[] = $match;
-                }
-            }
-        }
-
-        return $return;
-    }
-
-    /**
-     * Attempts to match the given arguments with the given operator.
-     *
-     * @param string $operator
-     * @param string $pattern
-     * @param string $value
-     * @return bool
-     */
-    protected function match($operator, $pattern, $value)
-    {
-        $value   = strtolower($value);
-        $pattern = strtolower($pattern);
-        switch ($operator) {
-            case '=':
-                return $value === $pattern;
-            case '!=':
-                return $value !== $pattern;
-            case '^=':
-                return preg_match('/^'.preg_quote($pattern, '/').'/', $value);
-            case '$=':
-                return preg_match('/'.preg_quote($pattern, '/').'$/', $value);
-            case '*=':
-                if ($pattern[0] == '/') {
-                    return preg_match($pattern, $value);
-                }
-
-                return preg_match("/".$pattern."/i", $value);
-        }
-
-        return false;
-    }
-
-    /**
-     * Attempts to figure out what the alteration will be for
-     * the next element.
-     *
-     * @param array $rule
-     * @return array
-     */
-    protected function alterNext($rule)
-    {
-        $options = [];
-        if ($rule['tag'] == '>') {
-            $options['checkGrandChildren'] = false;
-        }
-
-        return $options;
-    }
-
-    /**
-     * Flattens the option array.
-     *
-     * @param array $optionsArray
-     * @return array
-     */
-    protected function flattenOptions(array $optionsArray)
-    {
-        $options = [];
-        foreach ($optionsArray as $optionArray) {
-            foreach ($optionArray as $key => $option) {
-                $options[$key] = $option;
-            }
-        }
-
-        return $options;
-    }
+	/**
+	 * Pattern of CSS selectors, modified from 'mootools'
+	 *
+	 * @var string
+	 */
+	protected $pattern = "/([\w-:\*>]*)(?:\#([\w-]+)|\.([\w-]+))?(?:\[@?(!?[\w-:]+)(?:([!*^$]?=)[\"']?(.*?)[\"']?)?\])?([\/, ]+)/is";
+
+	protected $selectors = [];
+
+	/**
+	 * Constructs with the selector string
+	 *
+	 * @param string $selector
+	 */
+	public function __construct($selector)
+	{
+		$this->parseSelectorString($selector);
+	}
+
+	/**
+	 * Returns the selectors that where found in __construct
+	 *
+	 * @return array
+	 */
+	public function getSelectors()
+	{
+		return $this->selectors;
+	}
+
+	/**
+	 * Attempts to find the selectors starting from the given
+	 * node object.
+	 *
+	 * @param AbstractNode $node
+	 * @return array|Collection
+	 */
+	public function find(AbstractNode $node)
+	{
+		$results = new Collection;
+		foreach ($this->selectors as $selector) {
+			$nodes = [$node];
+			if (count($selector) == 0) {
+				continue;
+			}
+
+			$options = [];
+			foreach ($selector as $rule) {
+				if ($rule['alterNext']) {
+					$options[] = $this->alterNext($rule);
+					continue;
+				}
+				$nodes = $this->seek($nodes, $rule, $options);
+				// clear the options
+				$options = [];
+			}
+
+			// this is the final set of nodes
+			foreach ($nodes as $result) {
+				$results[] = $result;
+			}
+		}
+
+		return $results;
+	}
+
+	/**
+	 * Parses the selector string
+	 *
+	 * @param string $selector
+	 */
+	protected function parseSelectorString($selector)
+	{
+		$matches = [];
+		preg_match_all($this->pattern, trim($selector).' ', $matches, PREG_SET_ORDER);
+
+		// skip tbody
+		$result = [];
+		foreach ($matches as $match) {
+			// default values
+			$tag       = strtolower(trim($match[1]));
+			$operator  = '=';
+			$key       = null;
+			$value     = null;
+			$noKey     = false;
+			$alterNext = false;
+
+			// check for elements that alter the behavior of the next element
+			if ($tag == '>') {
+				$alterNext = true;
+			}
+
+			// check for id selector
+			if ( ! empty($match[2])) {
+				$key   = 'id';
+				$value = $match[2];
+			}
+
+			// check for class selector
+			if ( ! empty($match[3])) {
+				$key   = 'class';
+				$value = $match[3];
+			}
+
+			// and final attribute selector
+			if ( ! empty($match[4])) {
+				$key = strtolower($match[4]);
+			}
+			if ( ! empty($match[5])) {
+				$operator = $match[5];
+			}
+			if ( ! empty($match[6])) {
+				$value = $match[6];
+			}
+
+			// check for elements that do not have a specified attribute
+			if (isset($key[0]) && $key[0] == '!') {
+				$key   = substr($key, 1);
+				$noKey = true;
+			}
+
+			$result[] = [
+				'tag'       => $tag,
+				'key'       => $key,
+				'value'     => $value,
+				'operator'  => $operator,
+				'noKey'     => $noKey,
+				'alterNext' => $alterNext,
+			];
+			if (trim($match[7]) == ',') {
+				$this->selectors[] = $result;
+				$result            = [];
+			}
+		}
+
+		// save last results
+		if (count($result) > 0) {
+			$this->selectors[] = $result;
+		}
+	}
+
+	/**
+	 * Attempts to find all children that match the rule
+	 * given.
+	 *
+	 * @param array $nodes
+	 * @param array $rule
+	 * @param array $options
+	 * @return array
+	 * @recursive
+	 */
+	protected function seek(array $nodes, array $rule, array $options)
+	{
+		// XPath index
+		if (count($rule['tag']) > 0 &&
+			count($rule['key']) > 0 &&
+			is_numeric($rule['key'])
+		) {
+			$count = 0;
+			/** @var AbstractNode $node */
+			foreach ($nodes as $node) {
+				if ($rule['tag'] == '*' ||
+					$rule['tag'] == $node->getTag()->name()) {
+					++$count;
+					if ($count == $rule['key']) {
+						// found the node we wanted
+						return [$node];
+					}
+				}
+			}
+
+			return [];
+		}
+
+		$options = $this->flattenOptions($options);
+
+		$return = [];
+		/** @var AbstractNode $node */
+		foreach ($nodes as $node) {
+			// check if we are a leaf
+			if ( ! $node->hasChildren()) {
+				continue;
+			}
+
+			$children = [];
+			$child    = $node->firstChild();
+			while ( ! is_null($child)) {
+				// wild card, grab all
+				if ($rule['tag'] == '*' && is_null($rule['key'])) {
+					$return[] = $child;
+					try {
+						$child = $node->nextChild($child->id());
+					} catch (ChildNotFoundException $e) {
+						// no more children
+						$child = null;
+					}
+					continue;
+				}
+
+				$pass = true;
+				// check tag
+				if ( ! empty($rule['tag']) && $rule['tag'] != $child->getTag()->name() &&
+					$rule['tag'] != '*'
+				) {
+					// child failed tag check
+					$pass = false;
+				}
+
+				// check key
+				if ($pass && ! is_null($rule['key'])) {
+					if ($rule['noKey']) {
+						if ( ! is_null($child->getAttribute($rule['key']))) {
+							$pass = false;
+						}
+					} else {
+						if ($rule['key'] != 'plaintext' &&
+							is_null($child->getAttribute($rule['key']))
+						) {
+							$pass = false;
+						}
+					}
+				}
+
+				// compare values
+				if ($pass && ! is_null($rule['key']) &&
+					! is_null($rule['value']) && $rule['value'] != '*'
+				) {
+					if ($rule['key'] == 'plaintext') {
+						// plaintext search
+						$nodeValue = $child->text();
+					} else {
+						// normal search
+						$nodeValue = $child->getAttribute($rule['key']);
+					}
+
+					$check = $this->match($rule['operator'], $rule['value'], $nodeValue);
+
+					// handle multiple classes
+					if ( ! $check && $rule['key'] == 'class') {
+						$childClasses = explode(' ', $child->getAttribute('class'));
+						foreach ($childClasses as $class) {
+							if ( ! empty($class)) {
+								$check = $this->match($rule['operator'], $rule['value'], $class);
+							}
+							if ($check) {
+								break;
+							}
+						}
+					}
+
+					if ( ! $check) {
+						$pass = false;
+					}
+				}
+
+				if ($pass) {
+					// it passed all checks
+					$return[] = $child;
+				} else {
+					// this child failed to be matched
+					if ($child->hasChildren()) {
+						// we still want to check its children
+						$children[] = $child;
+					}
+				}
+
+				try {
+					// get next child
+					$child = $node->nextChild($child->id());
+				} catch (ChildNotFoundException $e) {
+					// no more children
+					$child = null;
+				}
+			}
+
+			if (( ! isset($options['checkGrandChildren']) ||
+					$options['checkGrandChildren'])
+				&& count($children) > 0
+			) {
+				// we have children that failed but are not leaves.
+				$matches = $this->seek($children, $rule, $options);
+				foreach ($matches as $match) {
+					$return[] = $match;
+				}
+			}
+		}
+
+		return $return;
+	}
+
+	/**
+	 * Attempts to match the given arguments with the given operator.
+	 *
+	 * @param string $operator
+	 * @param string $pattern
+	 * @param string $value
+	 * @return bool
+	 */
+	protected function match($operator, $pattern, $value)
+	{
+		$value   = strtolower($value);
+		$pattern = strtolower($pattern);
+		switch ($operator) {
+			case '=':
+				return $value === $pattern;
+			case '!=':
+				return $value !== $pattern;
+			case '^=':
+				return preg_match('/^'.preg_quote($pattern, '/').'/', $value);
+			case '$=':
+				return preg_match('/'.preg_quote($pattern, '/').'$/', $value);
+			case '*=':
+				if ($pattern[0] == '/') {
+					return preg_match($pattern, $value);
+				}
+
+				return preg_match("/".$pattern."/i", $value);
+		}
+
+		return false;
+	}
+
+	/**
+	 * Attempts to figure out what the alteration will be for
+	 * the next element.
+	 *
+	 * @param array $rule
+	 * @return array
+	 */
+	protected function alterNext($rule)
+	{
+		$options = [];
+		if ($rule['tag'] == '>') {
+			$options['checkGrandChildren'] = false;
+		}
+
+		return $options;
+	}
+
+	/**
+	 * Flattens the option array.
+	 *
+	 * @param array $optionsArray
+	 * @return array
+	 */
+	protected function flattenOptions(array $optionsArray)
+	{
+		$options = [];
+		foreach ($optionsArray as $optionArray) {
+			foreach ($optionArray as $key => $option) {
+				$options[$key] = $option;
+			}
+		}
+
+		return $options;
+	}
 }