paquettg / php-html-parser

src/PHPHtmlParser/Options.php

Indentation +65 added lines, -65 removed lines

@@ -12,76 +12,76 @@
 class Options
 {
 
-    /**
-     * The default options array
-     *
-     * @param array
-     */
-    protected $defaults = [
-        'whitespaceTextNode' => true,
-        'strict'             => false,
-        'enforceEncoding'    => null,
-        'cleanupInput'       => true,
-        'removeScripts'      => true,
-        'removeStyles'       => true,
-        'preserveLineBreaks' => false,
-    ];
+	/**
+	 * The default options array
+	 *
+	 * @param array
+	 */
+	protected $defaults = [
+		'whitespaceTextNode' => true,
+		'strict'             => false,
+		'enforceEncoding'    => null,
+		'cleanupInput'       => true,
+		'removeScripts'      => true,
+		'removeStyles'       => true,
+		'preserveLineBreaks' => false,
+	];
 
-    /**
-     * The list of all current options set.
-     *
-     * @param array
-     */
-    protected $options = [];
+	/**
+	 * The list of all current options set.
+	 *
+	 * @param array
+	 */
+	protected $options = [];
 
-    /**
-     * Sets the default options in the options array
-     */
-    public function __construct()
-    {
-        $this->options = $this->defaults;
-    }
+	/**
+	 * Sets the default options in the options array
+	 */
+	public function __construct()
+	{
+		$this->options = $this->defaults;
+	}
 
-    /**
-     * A magic get to call the get() method.
-     *
-     * @param string $key
-     * @return mixed
-     * @uses $this->get()
-     */
-    public function __get($key)
-    {
-        return $this->get($key);
-    }
+	/**
+	 * A magic get to call the get() method.
+	 *
+	 * @param string $key
+	 * @return mixed
+	 * @uses $this->get()
+	 */
+	public function __get($key)
+	{
+		return $this->get($key);
+	}
 
-    /**
-     * Sets a new options param to override the current option array.
-     *
-     * @param array $options
-     * @return $this
-     */
-    public function setOptions(array $options)
-    {
-        foreach ($options as $key => $option) {
-            $this->options[$key] = $option;
-        }
+	/**
+	 * Sets a new options param to override the current option array.
+	 *
+	 * @param array $options
+	 * @return $this
+	 */
+	public function setOptions(array $options)
+	{
+		foreach ($options as $key => $option) {
+			$this->options[$key] = $option;
+		}
 
-        return $this;
-    }
+		return $this;
+	}
 
-    /**
-     * Gets the value associated to the key, or null if the key is not
-     * found.
-     *
-     * @param string
-     * @return mixed
-     */
-    public function get($key)
-    {
-        if (isset($this->options[$key])) {
-            return $this->options[$key];
-        }
+	/**
+	 * Gets the value associated to the key, or null if the key is not
+	 * found.
+	 *
+	 * @param string
+	 * @return mixed
+	 */
+	public function get($key)
+	{
+		if (isset($this->options[$key])) {
+			return $this->options[$key];
+		}
 
-        return null;
-    }
+		return null;
+	}
 }

src/PHPHtmlParser/Content.php

Indentation +240 added lines, -240 removed lines

@@ -9,244 +9,244 @@
 class Content
 {
 
-    /**
-     * The content string.
-     *
-     * @var string
-     */
-    protected $content;
-
-    /**
-     * The size of the content.
-     *
-     * @var integer
-     */
-    protected $size;
-
-    /**
-     * The current position we are in the content.
-     *
-     * @var integer
-     */
-    protected $pos;
-
-    /**
-     * The following 4 strings are tags that are important to us.
-     *
-     * @var string
-     */
-    protected $blank = " \t\r\n";
-    protected $equal = ' =/>';
-    protected $slash = " />\r\n\t";
-    protected $attr = ' >';
-
-    /**
-     * Content constructor.
-     *
-     * @param $content
-     */
-    public function __construct($content)
-    {
-        $this->content = $content;
-        $this->size    = strlen($content);
-        $this->pos     = 0;
-    }
-
-    /**
-     * Returns the current position of the content.
-     *
-     * @return int
-     */
-    public function getPosition()
-    {
-        return $this->pos;
-    }
-
-    /**
-     * Gets the current character we are at.
-     *
-     * @param int $char
-     * @return string
-     */
-    public function char($char = null)
-    {
-        $pos = $this->pos;
-        if ( ! is_null($char)) {
-            $pos = $char;
-        }
-
-        if ( ! isset($this->content[$pos])) {
-            return '';
-        }
-
-        return $this->content[$pos];
-    }
-
-    /**
-     * Moves the current position forward.
-     *
-     * @param int $count
-     * @return $this
-     */
-    public function fastForward($count)
-    {
-        $this->pos += $count;
-
-        return $this;
-    }
-
-    /**
-     * Moves the current position backward.
-     *
-     * @param int $count
-     * @return $this
-     */
-    public function rewind($count)
-    {
-        $this->pos -= $count;
-        if ($this->pos < 0) {
-            $this->pos = 0;
-        }
-
-        return $this;
-    }
-
-    /**
-     * Copy the content until we find the given string.
-     *
-     * @param string $string
-     * @param bool $char
-     * @param bool $escape
-     * @return string
-     */
-    public function copyUntil($string, $char = false, $escape = false)
-    {
-        if ($this->pos >= $this->size) {
-            // nothing left
-            return '';
-        }
-
-        if ($escape) {
-            $position = $this->pos;
-            $found    = false;
-            while ( ! $found) {
-                $position = strpos($this->content, $string, $position);
-                if ($position === false) {
-                    // reached the end
-                    $found = true;
-                    continue;
-                }
-
-                if ($this->char($position - 1) == '\\') {
-                    // this character is escaped
-                    ++$position;
-                    continue;
-                }
-
-                $found = true;
-            }
-        } elseif ($char) {
-            $position = strcspn($this->content, $string, $this->pos);
-            $position += $this->pos;
-        } else {
-            $position = strpos($this->content, $string, $this->pos);
-        }
-
-        if ($position === false) {
-            // could not find character, just return the remaining of the content
-            $return    = substr($this->content, $this->pos, $this->size - $this->pos);
-            $this->pos = $this->size;
-
-            return $return;
-        }
-
-        if ($position == $this->pos) {
-            // we are at the right place
-            return '';
-        }
-
-        $return = substr($this->content, $this->pos, $position - $this->pos);
-        // set the new position
-        $this->pos = $position;
-
-        return $return;
-    }
-
-    /**
-     * Copies the content until the string is found and return it
-     * unless the 'unless' is found in the substring.
-     *
-     * @param string $string
-     * @param string $unless
-     * @return string
-     */
-    public function copyUntilUnless($string, $unless)
-    {
-        $lastPos = $this->pos;
-        $this->fastForward(1);
-        $foundString = $this->copyUntil($string, true, true);
-
-        $position = strcspn($foundString, $unless);
-        if ($position == strlen($foundString)) {
-            return $string.$foundString;
-        }
-        // rewind changes and return nothing
-        $this->pos = $lastPos;
-
-        return '';
-    }
-
-    /**
-     * Copies the content until it reaches the token string.,
-     *
-     * @param string $token
-     * @param bool $char
-     * @param bool $escape
-     * @return string
-     * @uses $this->copyUntil()
-     */
-    public function copyByToken($token, $char = false, $escape = false)
-    {
-        $string = $this->$token;
-
-        return $this->copyUntil($string, $char, $escape);
-    }
-
-    /**
-     * Skip a given set of characters.
-     *
-     * @param string $string
-     * @param bool $copy
-     * @return $this|string
-     */
-    public function skip($string, $copy = false)
-    {
-        $len = strspn($this->content, $string, $this->pos);
-
-        // make it chainable if they don't want a copy
-        $return = $this;
-        if ($copy) {
-            $return = substr($this->content, $this->pos, $len);
-        }
-
-        // update the position
-        $this->pos += $len;
-
-        return $return;
-    }
-
-    /**
-     * Skip a given token of pre-defined characters.
-     *
-     * @param string $token
-     * @param bool $copy
-     * @return null|string
-     * @uses $this->skip()
-     */
-    public function skipByToken($token, $copy = false)
-    {
-        $string = $this->$token;
-
-        return $this->skip($string, $copy);
-    }
+	/**
+	 * The content string.
+	 *
+	 * @var string
+	 */
+	protected $content;
+
+	/**
+	 * The size of the content.
+	 *
+	 * @var integer
+	 */
+	protected $size;
+
+	/**
+	 * The current position we are in the content.
+	 *
+	 * @var integer
+	 */
+	protected $pos;
+
+	/**
+	 * The following 4 strings are tags that are important to us.
+	 *
+	 * @var string
+	 */
+	protected $blank = " \t\r\n";
+	protected $equal = ' =/>';
+	protected $slash = " />\r\n\t";
+	protected $attr = ' >';
+
+	/**
+	 * Content constructor.
+	 *
+	 * @param $content
+	 */
+	public function __construct($content)
+	{
+		$this->content = $content;
+		$this->size    = strlen($content);
+		$this->pos     = 0;
+	}
+
+	/**
+	 * Returns the current position of the content.
+	 *
+	 * @return int
+	 */
+	public function getPosition()
+	{
+		return $this->pos;
+	}
+
+	/**
+	 * Gets the current character we are at.
+	 *
+	 * @param int $char
+	 * @return string
+	 */
+	public function char($char = null)
+	{
+		$pos = $this->pos;
+		if ( ! is_null($char)) {
+			$pos = $char;
+		}
+
+		if ( ! isset($this->content[$pos])) {
+			return '';
+		}
+
+		return $this->content[$pos];
+	}
+
+	/**
+	 * Moves the current position forward.
+	 *
+	 * @param int $count
+	 * @return $this
+	 */
+	public function fastForward($count)
+	{
+		$this->pos += $count;
+
+		return $this;
+	}
+
+	/**
+	 * Moves the current position backward.
+	 *
+	 * @param int $count
+	 * @return $this
+	 */
+	public function rewind($count)
+	{
+		$this->pos -= $count;
+		if ($this->pos < 0) {
+			$this->pos = 0;
+		}
+
+		return $this;
+	}
+
+	/**
+	 * Copy the content until we find the given string.
+	 *
+	 * @param string $string
+	 * @param bool $char
+	 * @param bool $escape
+	 * @return string
+	 */
+	public function copyUntil($string, $char = false, $escape = false)
+	{
+		if ($this->pos >= $this->size) {
+			// nothing left
+			return '';
+		}
+
+		if ($escape) {
+			$position = $this->pos;
+			$found    = false;
+			while ( ! $found) {
+				$position = strpos($this->content, $string, $position);
+				if ($position === false) {
+					// reached the end
+					$found = true;
+					continue;
+				}
+
+				if ($this->char($position - 1) == '\\') {
+					// this character is escaped
+					++$position;
+					continue;
+				}
+
+				$found = true;
+			}
+		} elseif ($char) {
+			$position = strcspn($this->content, $string, $this->pos);
+			$position += $this->pos;
+		} else {
+			$position = strpos($this->content, $string, $this->pos);
+		}
+
+		if ($position === false) {
+			// could not find character, just return the remaining of the content
+			$return    = substr($this->content, $this->pos, $this->size - $this->pos);
+			$this->pos = $this->size;
+
+			return $return;
+		}
+
+		if ($position == $this->pos) {
+			// we are at the right place
+			return '';
+		}
+
+		$return = substr($this->content, $this->pos, $position - $this->pos);
+		// set the new position
+		$this->pos = $position;
+
+		return $return;
+	}
+
+	/**
+	 * Copies the content until the string is found and return it
+	 * unless the 'unless' is found in the substring.
+	 *
+	 * @param string $string
+	 * @param string $unless
+	 * @return string
+	 */
+	public function copyUntilUnless($string, $unless)
+	{
+		$lastPos = $this->pos;
+		$this->fastForward(1);
+		$foundString = $this->copyUntil($string, true, true);
+
+		$position = strcspn($foundString, $unless);
+		if ($position == strlen($foundString)) {
+			return $string.$foundString;
+		}
+		// rewind changes and return nothing
+		$this->pos = $lastPos;
+
+		return '';
+	}
+
+	/**
+	 * Copies the content until it reaches the token string.,
+	 *
+	 * @param string $token
+	 * @param bool $char
+	 * @param bool $escape
+	 * @return string
+	 * @uses $this->copyUntil()
+	 */
+	public function copyByToken($token, $char = false, $escape = false)
+	{
+		$string = $this->$token;
+
+		return $this->copyUntil($string, $char, $escape);
+	}
+
+	/**
+	 * Skip a given set of characters.
+	 *
+	 * @param string $string
+	 * @param bool $copy
+	 * @return $this|string
+	 */
+	public function skip($string, $copy = false)
+	{
+		$len = strspn($this->content, $string, $this->pos);
+
+		// make it chainable if they don't want a copy
+		$return = $this;
+		if ($copy) {
+			$return = substr($this->content, $this->pos, $len);
+		}
+
+		// update the position
+		$this->pos += $len;
+
+		return $return;
+	}
+
+	/**
+	 * Skip a given token of pre-defined characters.
+	 *
+	 * @param string $token
+	 * @param bool $copy
+	 * @return null|string
+	 * @uses $this->skip()
+	 */
+	public function skipByToken($token, $copy = false)
+	{
+		$string = $this->$token;
+
+		return $this->skip($string, $copy);
+	}
 }

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 or
-                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() != '>' and
-            $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 or
+				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() != '>' and
+			$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/CurlInterface.php

Indentation +7 added lines, -7 removed lines

@@ -9,11 +9,11 @@
 interface CurlInterface
 {
 
-    /**
-     * This method should return the content of the url in a string
-     *
-     * @param string $url
-     * @return string
-     */
-    public function get($url);
+	/**
+	 * This method should return the content of the url in a string
+	 *
+	 * @param string $url
+	 * @return string
+	 */
+	public function get($url);
 }

src/PHPHtmlParser/Curl.php

Indentation +23 added lines, -23 removed lines

@@ -11,31 +11,31 @@
 class Curl implements CurlInterface
 {
 
-    /**
-     * A simple curl implementation to get the content of the url.
-     *
-     * @param string $url
-     * @return string
-     * @throws CurlException
-     */
-    public function get($url)
-    {
-        $ch = curl_init($url);
+	/**
+	 * A simple curl implementation to get the content of the url.
+	 *
+	 * @param string $url
+	 * @return string
+	 * @throws CurlException
+	 */
+	public function get($url)
+	{
+		$ch = curl_init($url);
 
-        if ( ! ini_get('open_basedir')) {
-            curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
-        }
+		if ( ! ini_get('open_basedir')) {
+			curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
+		}
 
-        curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
-        curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
+		curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
+		curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 5);
 
-        $content = curl_exec($ch);
-        if ($content === false) {
-            // there was a problem
-            $error = curl_error($ch);
-            throw new CurlException('Error retrieving "'.$url.'" ('.$error.')');
-        }
+		$content = curl_exec($ch);
+		if ($content === false) {
+			// there was a problem
+			$error = curl_error($ch);
+			throw new CurlException('Error retrieving "'.$url.'" ('.$error.')');
+		}
 
-        return $content;
-    }
+		return $content;
+	}
 }

src/PHPHtmlParser/Selector.php

Indentation +353 added lines, -353 removed lines

@@ -13,357 +13,357 @@
 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]) AND $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 $options)
-    {
-        // XPath index
-        if ( ! empty($rule['tag']) AND ! empty($rule['key']) AND
-            is_numeric($rule['key'])
-        ) {
-            $count = 0;
-            /** @var AbstractNode $node */
-            foreach ($nodes as $node) {
-                if ($rule['tag'] == '*' OR $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 ($rule['tag'] == '*' AND is_null($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']) AND $rule['tag'] != $child->getTag()->name() AND
-                    $rule['tag'] != '*'
-                ) {
-                    // child failed tag check
-                    $pass = false;
-                }
-
-                // check key
-                if ($pass AND ! is_null($rule['key'])) {
-                    if ($rule['noKey']) {
-                        if ( ! is_null($child->getAttribute($rule['key']))) {
-                            $pass = false;
-                        }
-                    } else {
-                        if ($rule['key'] != 'plaintext' and
-                            is_null($child->getAttribute($rule['key']))
-                        ) {
-                            $pass = false;
-                        }
-                    }
-                }
-
-                // compare values
-                if ($pass and ! is_null($rule['key']) and
-                    ! is_null($rule['value']) and $rule['value'] != '*'
-                ) {
-                    if ($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 and $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]) AND $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 $options)
+	{
+		// XPath index
+		if ( ! empty($rule['tag']) AND ! empty($rule['key']) AND
+			is_numeric($rule['key'])
+		) {
+			$count = 0;
+			/** @var AbstractNode $node */
+			foreach ($nodes as $node) {
+				if ($rule['tag'] == '*' OR $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 ($rule['tag'] == '*' AND is_null($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']) AND $rule['tag'] != $child->getTag()->name() AND
+					$rule['tag'] != '*'
+				) {
+					// child failed tag check
+					$pass = false;
+				}
+
+				// check key
+				if ($pass AND ! is_null($rule['key'])) {
+					if ($rule['noKey']) {
+						if ( ! is_null($child->getAttribute($rule['key']))) {
+							$pass = false;
+						}
+					} else {
+						if ($rule['key'] != 'plaintext' and
+							is_null($child->getAttribute($rule['key']))
+						) {
+							$pass = false;
+						}
+					}
+				}
+
+				// compare values
+				if ($pass and ! is_null($rule['key']) and
+					! is_null($rule['value']) and $rule['value'] != '*'
+				) {
+					if ($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 and $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;
+	}
 }

src/PHPHtmlParser/StaticDom.php

Indentation +88 added lines, -88 removed lines

@@ -11,102 +11,102 @@
 final class StaticDom
 {
 
-    private static $dom = null;
+	private static $dom = null;
 
-    /**
-     * Attempts to call the given method on the most recent created dom
-     * from bellow.
-     *
-     * @param string $method
-     * @param array $arguments
-     * @throws NotLoadedException
-     * @return mixed
-     */
-    public static function __callStatic($method, $arguments)
-    {
-        if (self::$dom instanceof Dom) {
-            return call_user_func_array([self::$dom, $method], $arguments);
-        } else {
-            throw new NotLoadedException('The dom is not loaded. Can not call a dom method.');
-        }
-    }
+	/**
+	 * Attempts to call the given method on the most recent created dom
+	 * from bellow.
+	 *
+	 * @param string $method
+	 * @param array $arguments
+	 * @throws NotLoadedException
+	 * @return mixed
+	 */
+	public static function __callStatic($method, $arguments)
+	{
+		if (self::$dom instanceof Dom) {
+			return call_user_func_array([self::$dom, $method], $arguments);
+		} else {
+			throw new NotLoadedException('The dom is not loaded. Can not call a dom method.');
+		}
+	}
 
-    /**
-     * Call this to mount the static facade. The facade allows you to use
-     * this object as a $className.
-     *
-     * @param string $className
-     * @param Dom $dom
-     * @return bool
-     */
-    public static function mount($className = 'Dom', Dom $dom = null)
-    {
-        if (class_exists($className)) {
-            return false;
-        }
-        class_alias(__CLASS__, $className);
-        if ($dom instanceof Dom) {
-            self::$dom = $dom;
-        }
+	/**
+	 * Call this to mount the static facade. The facade allows you to use
+	 * this object as a $className.
+	 *
+	 * @param string $className
+	 * @param Dom $dom
+	 * @return bool
+	 */
+	public static function mount($className = 'Dom', Dom $dom = null)
+	{
+		if (class_exists($className)) {
+			return false;
+		}
+		class_alias(__CLASS__, $className);
+		if ($dom instanceof Dom) {
+			self::$dom = $dom;
+		}
 
-        return true;
-    }
+		return true;
+	}
 
-    /**
-     * Creates a new dom object and calls load() on the
-     * new object.
-     *
-     * @param string $str
-     * @return $this
-     */
-    public static function load($str)
-    {
-        $dom       = new Dom;
-        self::$dom = $dom;
+	/**
+	 * Creates a new dom object and calls load() on the
+	 * new object.
+	 *
+	 * @param string $str
+	 * @return $this
+	 */
+	public static function load($str)
+	{
+		$dom       = new Dom;
+		self::$dom = $dom;
 
-        return $dom->load($str);
-    }
+		return $dom->load($str);
+	}
 
-    /**
-     * Creates a new dom object and calls loadFromFile() on the
-     * new object.
-     *
-     * @param string $file
-     * @return $this
-     */
-    public static function loadFromFile($file)
-    {
-        $dom       = new Dom;
-        self::$dom = $dom;
+	/**
+	 * Creates a new dom object and calls loadFromFile() on the
+	 * new object.
+	 *
+	 * @param string $file
+	 * @return $this
+	 */
+	public static function loadFromFile($file)
+	{
+		$dom       = new Dom;
+		self::$dom = $dom;
 
-        return $dom->loadFromFile($file);
-    }
+		return $dom->loadFromFile($file);
+	}
 
-    /**
-     * Creates a new dom object and calls loadFromUrl() on the
-     * new object.
-     *
-     * @param string $url
-     * @param CurlInterface $curl
-     * @return $this
-     */
-    public static function loadFromUrl($url, CurlInterface $curl = null)
-    {
-        $dom       = new Dom;
-        self::$dom = $dom;
-        if (is_null($curl)) {
-            // use the default curl interface
-            $curl = new Curl;
-        }
+	/**
+	 * Creates a new dom object and calls loadFromUrl() on the
+	 * new object.
+	 *
+	 * @param string $url
+	 * @param CurlInterface $curl
+	 * @return $this
+	 */
+	public static function loadFromUrl($url, CurlInterface $curl = null)
+	{
+		$dom       = new Dom;
+		self::$dom = $dom;
+		if (is_null($curl)) {
+			// use the default curl interface
+			$curl = new Curl;
+		}
 
-        return $dom->loadFromUrl($url, $curl);
-    }
+		return $dom->loadFromUrl($url, $curl);
+	}
 
-    /**
-     * Sets the $dom variable to null.
-     */
-    public static function unload()
-    {
-        self::$dom = null;
-    }
+	/**
+	 * Sets the $dom variable to null.
+	 */
+	public static function unload()
+	{
+		self::$dom = null;
+	}
 }

src/PHPHtmlParser/Dom/MockNode.php

Indentation +12 added lines, -12 removed lines

@@ -11,19 +11,19 @@
 class MockNode extends AbstractNode
 {
 
-    public function innerHtml()
-    {
-    }
+	public function innerHtml()
+	{
+	}
 
-    public function outerHtml()
-    {
-    }
+	public function outerHtml()
+	{
+	}
 
-    public function text()
-    {
-    }
+	public function text()
+	{
+	}
 
-    protected function clear()
-    {
-    }
+	protected function clear()
+	{
+	}
 }

src/PHPHtmlParser/Dom/TextNode.php

Indentation +82 added lines, -82 removed lines

@@ -9,97 +9,97 @@
 class TextNode extends AbstractNode
 {
 
-    /**
-     * This is a text node.
-     *
-     * @var Tag
-     */
-    protected $tag;
+	/**
+	 * This is a text node.
+	 *
+	 * @var Tag
+	 */
+	protected $tag;
 
-    /**
-     * This is the text in this node.
-     *
-     * @var string
-     */
-    protected $text;
+	/**
+	 * This is the text in this node.
+	 *
+	 * @var string
+	 */
+	protected $text;
 
-    /**
-     * This is the converted version of the text.
-     *
-     * @var string
-     */
-    protected $convertedText = null;
+	/**
+	 * This is the converted version of the text.
+	 *
+	 * @var string
+	 */
+	protected $convertedText = null;
 
-    /**
-     * Sets the text for this node.
-     *
-     * @param string $text
-     */
-    public function __construct($text)
-    {
-        // remove double spaces
-        $text = mb_ereg_replace('\s+', ' ', $text);
+	/**
+	 * Sets the text for this node.
+	 *
+	 * @param string $text
+	 */
+	public function __construct($text)
+	{
+		// remove double spaces
+		$text = mb_ereg_replace('\s+', ' ', $text);
 
-        // restore line breaks
-        $text = str_replace('&#10', "\n", $text);
+		// restore line breaks
+		$text = str_replace('&#10', "\n", $text);
 
-        $this->text = $text;
-        $this->tag  = new Tag('text');
-        parent::__construct();
-    }
+		$this->text = $text;
+		$this->tag  = new Tag('text');
+		parent::__construct();
+	}
 
-    /**
-     * Returns the text of this node.
-     *
-     * @return string
-     */
-    public function text()
-    {
-        // convert charset
-        if ( ! is_null($this->encode)) {
-            if ( ! is_null($this->convertedText)) {
-                // we already know the converted value
-                return $this->convertedText;
-            }
-            $text = $this->encode->convert($this->text);
+	/**
+	 * Returns the text of this node.
+	 *
+	 * @return string
+	 */
+	public function text()
+	{
+		// convert charset
+		if ( ! is_null($this->encode)) {
+			if ( ! is_null($this->convertedText)) {
+				// we already know the converted value
+				return $this->convertedText;
+			}
+			$text = $this->encode->convert($this->text);
 
-            // remember the conversion
-            $this->convertedText = $text;
+			// remember the conversion
+			$this->convertedText = $text;
 
-            return $text;
-        } else {
-            return $this->text;
-        }
-    }
+			return $text;
+		} else {
+			return $this->text;
+		}
+	}
 
-    /**
-     * This node has no html, just return the text.
-     *
-     * @return string
-     * @uses $this->text()
-     */
-    public function innerHtml()
-    {
-        return $this->text();
-    }
+	/**
+	 * This node has no html, just return the text.
+	 *
+	 * @return string
+	 * @uses $this->text()
+	 */
+	public function innerHtml()
+	{
+		return $this->text();
+	}
 
-    /**
-     * This node has no html, just return the text.
-     *
-     * @return string
-     * @uses $this->text()
-     */
-    public function outerHtml()
-    {
-        return $this->text();
-    }
+	/**
+	 * This node has no html, just return the text.
+	 *
+	 * @return string
+	 * @uses $this->text()
+	 */
+	public function outerHtml()
+	{
+		return $this->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->convertedText = null;
-    }
+	/**
+	 * 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->convertedText = null;
+	}
 }