/lib/Molinos/Base/XML.php

<?php
/**
 * ??????????? ????? ??? ???????????? XML ????.
 *
 * @package Molinos_CMS
 * @subpackage base
 * @author Justin Forest <justin.forest@gmail.com>
 * @copyright 2006-2009 molinos.ru
 * @license http://www.gnu.org/copyleft/gpl.html GPL
 */

class Molinos_Base_XML
{
	/**
	 * ???????????? ??? ????? ?????? ? simple().
	 *
	 * @var string
	 */
	protected static $xml_errors;

	/**
	 * Renders an HTML element.
	 *
	 * Returns the HTML representation of an element described by
	 * input parameters which are: element name, an array of attributes,
	 * and the content.	Except for the first parameter, all is optional.
	 *
	 * @param string $em ??? ???????????? ????????.
	 * @param array $attrs ????????.
	 * @param string $content ??????????.
	 * @return string ??????? XML ???.
	 */
	public static function em()
	{
		if (func_num_args() == 0 or func_num_args() > 3)
			throw new InvalidArgumentException('Molinos_Base_XML::em() expects 1 to 3 arguments.');
		else {
			$args = func_get_args();
			$name = array_shift($args);

			if (empty($name))
				throw new InvalidArgumentException('Attempting to create an XML element without a name.');

			$parts = array();
			$content = null;

			if (is_array($tmp = array_shift($args)))
				$parts = $tmp;
			else
				$content = $tmp;

			if (!empty($args))
				$content = array_shift($args);
		}

		if (array_key_exists('#cdata', $parts)) {
			$content = self::cdata($parts['#cdata']);
			unset($parts['#cdata']);
		} elseif (array_key_exists('#text', $parts)) {
			$content = $parts['#text'];
			unset($parts['#text']);
		}

		$name = str_replace('_', '-', $name);
		$output = '<'. $name;

		if (('td' == $name or 'th' == $name) and empty($content))
			$content = '&nbsp;';

		if (empty($parts))
			$parts = array();

		$fixmap = array(
			'a' => 'href',
			'form' => 'action',
			);

		$output .= self::attrs($parts);

		if (null === $content and !in_array($name, array('a', 'script', 'div', 'textarea', 'span', 'base'))) {
			$output .= '/>';
		} else {
			$output .= '>'. $content .'</'. $name .'>';
		}

		return $output;
	}

	/**
	 * ?????????????? ?????????.
	 *
	 * ???????????? ????? ????????? (????? ???????), ??????? ?? ? ???????
	 * ???????? ? ??????? ???????? ????????. ???????? true ???????? ?? "yes",
	 * ?????? ???????? ?? ?????????.
	 *
	 * @param array $attrs ???????? ??? ??????????????.
	 * @return string ?????? ? ?????????? XML ????????.
	 */
	public static function attrs(array $attrs)
	{
		$result = '';

		ksort($attrs);
		foreach ($attrs as $k => $v) {
			if (!ctype_alpha(substr($k, 0, 1)))
				continue;

			if ('class' == $k) {
				if (!is_array($v))
					$v = preg_split('/\s+/', $v, -1, PREG_SPLIT_NO_EMPTY);
				$v = join(' ', array_unique($v));
			}

			if (empty($v) or is_array($v))
				continue;

			if (strlen($v)) {
				if (true === $v)
					$v = 'yes';
				else
					$v = Molinos_Base_XML::plain($v, false);
				$result .= ' '.self::attrname($k).'=\''. $v .'\'';
			} elseif ($k == 'value') {
				$result .= " value=''";
			}
		}

		return $result;
	}

	/**
	 * ??????? ???????????? HTML ??????.
	 *
	 * ?????????? ?????????????????? ????????? <li>, ?????? ?? ??????? ????????
	 * ???? ???????? ?? ???????, ?????????????? ? ??????? htmlspecialchars().
	 * ????? ??????? ????????????. ??? ?????? (<ul> ??? <ol>) ?? ???????????.
	 *
	 * @param array $elements ???????? ??????. ????? ????????????.
	 * @return string ?????????????????? ????????? <li>.
	 */
	public static function simpleList(array $elements)
	{
		$result = '';

		foreach ($elements as $em)
			$result .= self::em('li', htmlspecialchars($em));

		return $result;
	}

	/**
	 * ???????????? ?????? ????????.
	 *
	 * ????? ??????? ????????? ? ??????? value ????????, ???????? — ? ?????,
	 * ??? ???? (??? ?????????????).
	 *
	 * @param array $elements ????????????? ?????? ????????.
	 * @param string $emName ??? ????????? ?? ??????????.
	 * @param string $wrapperName ??? ????????, ? ??????? ??????????????
	 * ?????????, ???? ?????? ?? ????.
	 * @return string XML ???.
	 */
	public static function simpleOptions(array $elements, $emName = 'option', $wrapperName = null)
	{
		$output = '';

		foreach ($elements as $k => $v)
			$output .= Molinos_Base_XML::em($emName, array(
				'value' => $k,
				), $v);

		if (null !== $wrapperName and !empty($output))
			return Molinos_Base_XML::em($wrapperName, $output);

		return $output;
	}

	/**
	 * ????????????? ?????? ? CDATA, ???? ?????.
	 *
	 * @param string $data ?????????????? ??????.
	 * @return string ???????? ??????, ???? ?? ???????? ????????????, ??? ???
	 * ??, ?????????? ? CDATA, ???? ???????? ???? ?? ????????: "<>&".
	 */
	public static function cdata($data)
	{
		if (empty($data))
			return null;

		if (is_array($data) or is_object($data))
			return null;

		if (strlen($data) != strcspn($data, '<>&'))
			return '<![CDATA[' . $data . ']]>';
		else
			return $data;
	}

	/**
	 * ?????????????? ????????????.
	 *
	 * @return string ?????????????????? ????????? <link> ? <script>.
	 *
	 * @deprecated ?????? ?? ????????????, ??????????.
	 * @todo ???????.
	 */
	public static function formatExtras(array $extras)
	{
		$output = '';

		foreach ($extras as $item) {
			switch ($item[0]) {
			case 'style':
				$output .= Molinos_Base_XML::em('link', array(
					'rel' => 'stylesheet',
					'type' => 'text/css',
					'href' => Molinos_Core_Utils::webpath($item[1]),
					));
				break;
			case 'script':
				$output .= Molinos_Base_XML::em('script', array(
					'type' => 'text/javascript',
					'src' => Molinos_Core_Utils::webpath($item[1]),
					));
				break;
			}
		}

		return $output;
	}

	/**
	 * ?????????? ??????????? HTML, ??????????? ?????? ?? ?????.
	 *
	 * @param string $text ??????????? ?????.
	 * @param bool $strip true, ???? ????? ??????? ????.
	 * @return string ?????????????? ?????.
	 */
	public static function plain($text, $strip = true)
	{
		if (is_object($text) or is_array($text))
			throw new InvalidArgumentException('Molinos_Base_XML::plain() expects a string.');
		if ($strip)
			$text = strip_tags($text);
		return str_replace(array('&amp;quot;'), array('&quot;'), htmlspecialchars($text, ENT_QUOTES));
	}

	/**
	 * ????????? ??????.
	 *
	 * @param string $href ????? ??????.
	 * @param string $text ????? ??????.
	 * @param array $options ?????????????? ????????.
	 * @return string ????????????????? ??????.
	 *
	 * @deprecated ??????? ???????????? {@link em()}.
	 * @todo ???????.
	 */
	public static function link($href, $text, array $options = array())
	{
		$options['href'] = $href;
		return Molinos_Base_XML::em('a', $options, $text);
	}

	/**
	 * ????????????? ????????? ??????????? ? ???????.
	 *
	 * @param string $em ??? ????????-???????.
	 * @param string $content ????????????? ??????????.
	 * @param array $options ?????????????? ????????.
	 * @return string ?????????? $content, ?????????? ? ??????? $em. ????
	 * ? $content ?????, ?????????? NULL.
	 */
	public static function wrap($em, $content, $options = null)
	{
		return empty($content)
			? null
			: Molinos_Base_XML::em($em, $options, $content);
	}

	/**
	 * ????????? ????? ????????.
	 *
	 * @param string $string ??? ????????.
	 * @return string ??? ????????, ?? ???????? ??????? ??? ???????, ?????
	 * ????????? ????, ???? ? ???????, ? ????? ? ???????? ???????? ????????.
	 */
	public static function attrname($string)
	{
		return preg_replace('/[^a-z0-9-]/i', '', str_replace(array('_', '.'), array('-', '-'), $string));
	}

	/**
	 * ?????????? ???????? ??????, ????????? ??? ????????????? ? ?????????.
	 *
	 * ??????? ??? ???? ? ??????? ? ????? ????? 0x20, ??? ????????? ?????????
	 * ? ????? ??????????.
	 *
	 * @param string $html ???????? ?????.
	 * @param integer $length ?????, ? ????????.
	 * @return string ????????? ???????? ?????? ?????? ?????.
	 */
	public static function snippet($html, $length = 200)
	{
		$snippet = mb_strimwidth(strip_tags(html_entity_decode($html)), 0, $length, '…', 'utf-8');
		$snippet = preg_replace('/[\x01-\x20]+/', ' ', $snippet);
		return trim($snippet);
	}

	/**
	 * ?????????? SimpleXMLElement.
	 *
	 * ????? ???????? ?????????? ????????? ???????? ??????, ??????? ??????????,
	 * ????????? ? ???, ????? ????????? ??????????.
	 *
	 * @param string $xml ???, ??????? ????? ??????????.
	 * @return SimpleXMLElement ????????? ???????.
	 */
	public static function simple($xml)
	{
		$xml = self::decode_entities($xml);

		self::$xml_errors = '';
		set_error_handler(array(__CLASS__, 'simple_error'), -1);

		try {
			$em = new SimpleXMLElement($xml);
			restore_error_handler();
		} catch (Exception $e) {
			restore_error_handler();
			$logger = Molinos_Core_Logger::getInstance();
			$logger->error(self::$xml_errors . "The full XML document follows.\n" . $xml);
			throw new RuntimeException(t('Could not process the XML document, see the log file for details.'));
		}

		return $em;
	}

	public static function simple_error($errno, $errstr, $errfile, $errline)
	{
		self::$xml_errors .= trim($errstr) . "\n";
	}

	/**
	 * ?????? HTML ????????? ?? ?????????? ?????????????.
	 *
	 * ?????? ???????? ???????? HTML ? ??????? SimpleXML (issue 2155).
	 *
	 * @param string $xml ???????? ?????.
	 * @return string ????? ? ???????? ?????? ????? ?????????.
	 */
	public static function decode_entities($xml)
	{
		return str_replace(array_keys(self::$entity_map), array_values(self::$entity_map), $xml);
	}

	/**
	 * ?????????? ????? ?????????.
	 *
	 * ????? ?? ????????????.  ??? ???????? ?? ?????? ??????.
	 */
	public static function build_entity_map()
	{
		// Resources used:
		// http://en.wikipedia.org/wiki/List_of_XML_and_HTML_character_entity_references
		// http://www.intuitive.com/coolweb/entities.html

		// $base_url = './';
		$base_url = 'http://www.w3.org/TR/html4/';

		$fileNames = array('HTMLlat1.ent', 'HTMLspecial.ent', 'HTMLsymbol.ent');

		$result = array();
		foreach ($fileNames as $fileName) {
			$contents = file_get_contents($base_url . $fileName);
			if (preg_match_all('/<!ENTITY\s+(\S+)\s+CDATA\s+"([^"]+)"/', $contents, $m))
				foreach ($m[1] as $idx => $src) {
					if (!in_array($src, array('quot', 'apos', 'amp', 'lt', 'gt'))) {
						$dst = $m[2][$idx];
						$dst = html_entity_decode($dst, ENT_QUOTES, 'utf-8');
						$result['&' . $src . ';'] = $dst;
					}
				}
		}

		ksort($result);

		file_put_contents('entities.php', '<?php return ' . var_export($result, true) . ';');
	}

	/**
	 * ????? ?????????, ????????? ? ??????? build_entity_map().
	 *
	 * @var array
	 */
	protected static $entity_map = array(
		'&AElig;' => 'Ć',
		'&Aacute;' => 'Á',
		'&Acirc;' => 'Â',
		'&Agrave;' => 'Ŕ',
		'&Alpha;' => '?',
		'&Aring;' => 'Ĺ',
		'&Atilde;' => 'Ă',
		'&Auml;' => 'Ä',
		'&Beta;' => '?',
		'&Ccedil;' => 'Ç',
		'&Chi;' => '?',
		'&Dagger;' => '‡',
		'&Delta;' => '?',
		'&ETH;' => 'Đ',
		'&Eacute;' => 'É',
		'&Ecirc;' => 'Ę',
		'&Egrave;' => 'Č',
		'&Epsilon;' => '?',
		'&Eta;' => '?',
		'&Euml;' => 'Ë',
		'&Gamma;' => '?',
		'&Iacute;' => 'Í',
		'&Icirc;' => 'Î',
		'&Igrave;' => 'Ě',
		'&Iota;' => '?',
		'&Iuml;' => 'Ď',
		'&Kappa;' => '?',
		'&Lambda;' => '?',
		'&Mu;' => '?',
		'&Ntilde;' => 'Ń',
		'&Nu;' => '?',
		'&OElig;' => 'Œ',
		'&Oacute;' => 'Ó',
		'&Ocirc;' => 'Ô',
		'&Ograve;' => 'Ň',
		'&Omega;' => '?',
		'&Omicron;' => '?',
		'&Oslash;' => 'Ř',
		'&Otilde;' => 'Ő',
		'&Ouml;' => 'Ö',
		'&Phi;' => '?',
		'&Pi;' => '?',
		'&Prime;' => '?',
		'&Psi;' => '?',
		'&Rho;' => '?',
		'&Scaron;' => 'Š',
		'&Sigma;' => '?',
		'&THORN;' => 'Ţ',
		'&Tau;' => '?',
		'&Theta;' => '?',
		'&Uacute;' => 'Ú',
		'&Ucirc;' => 'Ű',
		'&Ugrave;' => 'Ů',
		'&Upsilon;' => '?',
		'&Uuml;' => 'Ü',
		'&Xi;' => '?',
		'&Yacute;' => 'Ý',
		'&Yuml;' => 'Ÿ',
		'&Zeta;' => '?',
		'&aacute;' => 'á',
		'&acirc;' => 'â',
		'&acute;' => '´',
		'&aelig;' => 'ć',
		'&agrave;' => 'ŕ',
		'&alefsym;' => '?',
		'&alpha;' => '?',
		'&and;' => '?',
		'&ang;' => '?',
		'&aring;' => 'ĺ',
		'&asymp;' => '?',
		'&atilde;' => 'ă',
		'&auml;' => 'ä',
		'&bdquo;' => '„',
		'&beta;' => '?',
		'&brvbar;' => 'Ś',
		'&bull;' => '•',
		'&cap;' => '?',
		'&ccedil;' => 'ç',
		'&cedil;' => '¸',
		'&cent;' => '˘',
		'&chi;' => '?',
		'&circ;' => 'ˆ',
		'&clubs;' => '?',
		'&cong;' => '?',
		'&copy;' => 'Š',
		'&crarr;' => '?',
		'&cup;' => '?',
		'&curren;' => '¤',
		'&dArr;' => '?',
		'&dagger;' => '†',
		'&darr;' => '?',
		'&deg;' => '°',
		'&delta;' => '?',
		'&diams;' => '?',
		'&divide;' => '÷',
		'&eacute;' => 'é',
		'&ecirc;' => 'ę',
		'&egrave;' => 'č',
		'&empty;' => '?',
		'&emsp;' => '?',
		'&ensp;' => '?',
		'&epsilon;' => '?',
		'&equiv;' => '?',
		'&eta;' => '?',
		'&eth;' => 'đ',
		'&euml;' => 'ë',
		'&euro;' => '€',
		'&exist;' => '?',
		'&fnof;' => 'ƒ',
		'&forall;' => '?',
		'&frac12;' => '˝',
		'&frac14;' => 'ź',
		'&frac34;' => 'ž',
		'&frasl;' => '?',
		'&gamma;' => '?',
		'&ge;' => '?',
		'&hArr;' => '?',
		'&harr;' => '?',
		'&hearts;' => '?',
		'&hellip;' => '…',
		'&iacute;' => 'í',
		'&icirc;' => 'î',
		'&iexcl;' => 'Ą',
		'&igrave;' => 'ě',
		'&image;' => '?',
		'&infin;' => '?',
		'&int;' => '?',
		'&iota;' => '?',
		'&iquest;' => 'ż',
		'&isin;' => '?',
		'&iuml;' => 'ď',
		'&kappa;' => '?',
		'&lArr;' => '?',
		'&lambda;' => '?',
		'&lang;' => '?',
		'&laquo;' => 'Ť',
		'&larr;' => '?',
		'&lceil;' => '?',
		'&ldquo;' => '“',
		'&le;' => '?',
		'&lfloor;' => '?',
		'&lowast;' => '?',
		'&loz;' => '?',
		'&lrm;' => '?',
		'&lsaquo;' => '‹',
		'&lsquo;' => '‘',
		'&macr;' => 'Ż',
		'&mdash;' => '—',
		'&micro;' => 'ľ',
		'&middot;' => 'ˇ',
		'&minus;' => '?',
		'&mu;' => '?',
		'&nabla;' => '?',
		'&nbsp;' => ' ',
		'&ndash;' => '–',
		'&ne;' => '?',
		'&ni;' => '?',
		'&not;' => 'Ź',
		'&notin;' => '?',
		'&nsub;' => '?',
		'&ntilde;' => 'ń',
		'&nu;' => '?',
		'&oacute;' => 'ó',
		'&ocirc;' => 'ô',
		'&oelig;' => 'œ',
		'&ograve;' => 'ň',
		'&oline;' => '?',
		'&omega;' => '?',
		'&omicron;' => '?',
		'&oplus;' => '?',
		'&or;' => '?',
		'&ordf;' => 'Ş',
		'&ordm;' => 'ş',
		'&oslash;' => 'ř',
		'&otilde;' => 'ő',
		'&otimes;' => '?',
		'&ouml;' => 'ö',
		'&para;' => 'ś',
		'&part;' => '?',
		'&permil;' => '‰',
		'&perp;' => '?',
		'&phi;' => '?',
		'&pi;' => '?',
		'&piv;' => '?',
		'&plusmn;' => 'ą',
		'&pound;' => 'Ł',
		'&prime;' => '?',
		'&prod;' => '?',
		'&prop;' => '?',
		'&psi;' => '?',
		'&rArr;' => '?',
		'&radic;' => '?',
		'&rang;' => '?',
		'&raquo;' => 'ť',
		'&rarr;' => '?',
		'&rceil;' => '?',
		'&rdquo;' => '”',
		'&real;' => '?',
		'&reg;' => 'Ž',
		'&rfloor;' => '?',
		'&rho;' => '?',
		'&rlm;' => '?',
		'&rsaquo;' => '›',
		'&rsquo;' => '’',
		'&sbquo;' => '‚',
		'&scaron;' => 'š',
		'&sdot;' => '?',
		'&sect;' => '§',
		'&shy;' => '­',
		'&sigma;' => '?',
		'&sigmaf;' => '?',
		'&sim;' => '?',
		'&spades;' => '?',
		'&sub;' => '?',
		'&sube;' => '?',
		'&sum;' => '?',
		'&sup1;' => 'š',
		'&sup2;' => '˛',
		'&sup3;' => 'ł',
		'&sup;' => '?',
		'&supe;' => '?',
		'&szlig;' => 'ß',
		'&tau;' => '?',
		'&there4;' => '?',
		'&theta;' => '?',
		'&thetasym;' => '?',
		'&thinsp;' => '?',
		'&thorn;' => 'ţ',
		'&tilde;' => '˜',
		'&times;' => '×',
		'&trade;' => '™',
		'&uArr;' => '?',
		'&uacute;' => 'ú',
		'&uarr;' => '?',
		'&ucirc;' => 'ű',
		'&ugrave;' => 'ů',
		'&uml;' => '¨',
		'&upsih;' => '?',
		'&upsilon;' => '?',
		'&uuml;' => 'ü',
		'&weierp;' => '?',
		'&xi;' => '?',
		'&yacute;' => 'ý',
		'&yen;' => 'Ľ',
		'&yuml;' => '˙',
		'&zeta;' => '?',
		'&zwj;' => '?',
		'&zwnj;' => '?',
		);
}