paquettg / php-html-parser

src/PHPHtmlParser/Selector.php

Doc Comments +1 added lines, -1 removed lines

@@ -47,7 +47,7 @@
      * node object.
      *
      * @param AbstractNode $node
-     * @return array|Collection
+     * @return Collection
      */
     public function find(AbstractNode $node)
     {

Indentation +357 added lines, -357 removed lines

@@ -15,361 +15,361 @@
 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 InnerNode $node */
-        foreach ($nodes as $node) {
-            // check if we are a leaf
-            if ($node instanceof LeafNode ||
-                 ! $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 instanceof InnerNode &&
-                        $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 InnerNode $node */
+		foreach ($nodes as $node) {
+			// check if we are a leaf
+			if ($node instanceof LeafNode ||
+				 ! $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 instanceof InnerNode &&
+						$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

Doc Comments +3 added lines, -3 removed lines

@@ -57,7 +57,7 @@ 
      * new object.
      *
      * @param string $str
-     * @return $this
+     * @return Dom
      */
     public static function load($str)
     {
@@ -72,7 +72,7 @@ 
      * new object.
      *
      * @param string $file
-     * @return $this
+     * @return Dom
      */
     public static function loadFromFile($file)
     {
@@ -88,7 +88,7 @@ 
      *
      * @param string $url
      * @param CurlInterface $curl
-     * @return $this
+     * @return Dom
      */
     public static function loadFromUrl($url, CurlInterface $curl = null)
     {

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/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/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/Dom/Collection.php

Indentation +139 added lines, -139 removed lines

@@ -15,154 +15,154 @@
 class Collection implements IteratorAggregate, ArrayAccess, Countable
 {
 
-    /**
-     * The collection of Nodes.
-     *
-     * @param array
-     */
-    protected $collection = [];
+	/**
+	 * The collection of Nodes.
+	 *
+	 * @param array
+	 */
+	protected $collection = [];
 
-    /**
-     * Attempts to call the method on the first node in
-     * the collection.
-     *
-     * @param string $method
-     * @param array $arguments
-     * @return mixed;
-     * @throws EmptyCollectionException
-     */
-    public function __call($method, $arguments)
-    {
-        $node = reset($this->collection);
-        if ($node instanceof AbstractNode) {
-            return call_user_func_array([$node, $method], $arguments);
-        } else {
-            throw new EmptyCollectionException('The collection does not contain any Nodes.');
-        }
-    }
+	/**
+	 * Attempts to call the method on the first node in
+	 * the collection.
+	 *
+	 * @param string $method
+	 * @param array $arguments
+	 * @return mixed;
+	 * @throws EmptyCollectionException
+	 */
+	public function __call($method, $arguments)
+	{
+		$node = reset($this->collection);
+		if ($node instanceof AbstractNode) {
+			return call_user_func_array([$node, $method], $arguments);
+		} else {
+			throw new EmptyCollectionException('The collection does not contain any Nodes.');
+		}
+	}
 
-    /**
-     * Attempts to apply the magic get to the first node
-     * in the collection.
-     *
-     * @param mixed $key
-     * @return mixed
-     * @throws EmptyCollectionException
-     */
-    public function __get($key)
-    {
-        $node = reset($this->collection);
-        if ($node instanceof AbstractNode) {
-            return $node->$key;
-        } else {
-            throw new EmptyCollectionException('The collection does not contain any Nodes.');
-        }
-    }
+	/**
+	 * Attempts to apply the magic get to the first node
+	 * in the collection.
+	 *
+	 * @param mixed $key
+	 * @return mixed
+	 * @throws EmptyCollectionException
+	 */
+	public function __get($key)
+	{
+		$node = reset($this->collection);
+		if ($node instanceof AbstractNode) {
+			return $node->$key;
+		} else {
+			throw new EmptyCollectionException('The collection does not contain any Nodes.');
+		}
+	}
 
-    /**
-     * Applies the magic string method to the first node in
-     * the collection.
-     *
-     * @return string
-     * @throws EmptyCollectionException
-     */
-    public function __toString()
-    {
-        $node = reset($this->collection);
-        if ($node instanceof AbstractNode) {
-            return (string)$node;
-        } else {
-            throw new EmptyCollectionException('The collection does not contain any Nodes.');
-        }
-    }
+	/**
+	 * Applies the magic string method to the first node in
+	 * the collection.
+	 *
+	 * @return string
+	 * @throws EmptyCollectionException
+	 */
+	public function __toString()
+	{
+		$node = reset($this->collection);
+		if ($node instanceof AbstractNode) {
+			return (string)$node;
+		} else {
+			throw new EmptyCollectionException('The collection does not contain any Nodes.');
+		}
+	}
 
-    /**
-     * Returns the count of the collection.
-     *
-     * @return int
-     */
-    public function count()
-    {
-        return count($this->collection);
-    }
+	/**
+	 * Returns the count of the collection.
+	 *
+	 * @return int
+	 */
+	public function count()
+	{
+		return count($this->collection);
+	}
 
-    /**
-     * Returns an iterator for the collection.
-     *
-     * @return ArrayIterator
-     */
-    public function getIterator()
-    {
-        return new ArrayIterator($this->collection);
-    }
+	/**
+	 * Returns an iterator for the collection.
+	 *
+	 * @return ArrayIterator
+	 */
+	public function getIterator()
+	{
+		return new ArrayIterator($this->collection);
+	}
 
-    /**
-     * Set an attribute by the given offset
-     *
-     * @param mixed $offset
-     * @param mixed $value
-     */
-    public function offsetSet($offset, $value)
-    {
-        if (is_null($offset)) {
-            $this->collection[] = $value;
-        } else {
-            $this->collection[$offset] = $value;
-        }
-    }
+	/**
+	 * Set an attribute by the given offset
+	 *
+	 * @param mixed $offset
+	 * @param mixed $value
+	 */
+	public function offsetSet($offset, $value)
+	{
+		if (is_null($offset)) {
+			$this->collection[] = $value;
+		} else {
+			$this->collection[$offset] = $value;
+		}
+	}
 
-    /**
-     * Checks if an offset exists.
-     *
-     * @param mixed $offset
-     * @return bool
-     */
-    public function offsetExists($offset)
-    {
-        return isset($this->collection[$offset]);
-    }
+	/**
+	 * Checks if an offset exists.
+	 *
+	 * @param mixed $offset
+	 * @return bool
+	 */
+	public function offsetExists($offset)
+	{
+		return isset($this->collection[$offset]);
+	}
 
-    /**
-     * Unset a collection Node.
-     *
-     * @param mixed $offset
-     */
-    public function offsetUnset($offset)
-    {
-        unset($this->collection[$offset]);
-    }
+	/**
+	 * Unset a collection Node.
+	 *
+	 * @param mixed $offset
+	 */
+	public function offsetUnset($offset)
+	{
+		unset($this->collection[$offset]);
+	}
 
-    /**
-     * Gets a node at the given offset, or null
-     *
-     * @param mixed $offset
-     * @return mixed
-     */
-    public function offsetGet($offset)
-    {
-        return isset($this->collection[$offset]) ? $this->collection[$offset] : null;
-    }
+	/**
+	 * Gets a node at the given offset, or null
+	 *
+	 * @param mixed $offset
+	 * @return mixed
+	 */
+	public function offsetGet($offset)
+	{
+		return isset($this->collection[$offset]) ? $this->collection[$offset] : null;
+	}
 
-    /**
-     * Returns this collection as an array.
-     *
-     * @return array
-     */
-    public function toArray()
-    {
-        return $this->collection;
-    }
+	/**
+	 * Returns this collection as an array.
+	 *
+	 * @return array
+	 */
+	public function toArray()
+	{
+		return $this->collection;
+	}
 
-    /**
-     * Similar to jQuery "each" method. Calls the callback with each
-     * Node in this collection.
-     *
-     * @param callback $callback
-     */
-    public function each($callback)
-    {
-        foreach ($this->collection as $key => $value) {
-            $callback($value, $key);
-        }
-    }
+	/**
+	 * Similar to jQuery "each" method. Calls the callback with each
+	 * Node in this collection.
+	 *
+	 * @param callback $callback
+	 */
+	public function each($callback)
+	{
+		foreach ($this->collection as $key => $value) {
+			$callback($value, $key);
+		}
+	}
 }

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/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.'>';
+	}
 }