Your IP : 216.73.216.43
<?php namespace Pelago;
class Emogrifier
{
const ENCODING = 'UTF-8';
const CACHE_KEY_CSS = 0;
const CACHE_KEY_SELECTOR = 1;
const CACHE_KEY_XPATH = 2;
const CACHE_KEY_CSS_DECLARATION_BLOCK = 3;
const INDEX = 0;
const MULTIPLIER = 1;
const ID_ATTRIBUTE_MATCHER = '/(\\w+)?\\#([\\w\\-]+)/';
const CLASS_ATTRIBUTE_MATCHER = '/(\\w+|[\\*\\]])?((\\.[\\w\\-]+)+)/';
private $html = '';
private $css = '';
private $unprocessableHtmlTags = array('wbr');
private $caches = array(self::CACHE_KEY_CSS => array(), self::CACHE_KEY_SELECTOR => array(), self::CACHE_KEY_XPATH => array(), self::CACHE_KEY_CSS_DECLARATION_BLOCK => array());
private $visitedNodes = array();
private $styleAttributesForNodes = array();
public $preserveEncoding = false;
public function __construct($html = '', $css = '')
{
$this->setHtml($html);
$this->setCss($css);
}
public function __destruct()
{
$this->purgeVisitedNodes();
}
public function setHtml($html)
{
$this->html = $html;
}
public function setCss($css)
{
$this->css = $css;
}
public function emogrify()
{
if ($this->html === '')
{
throw new \BadMethodCallException('Please set some HTML first before calling emogrify.', 1390393096);
}
$xmlDocument = $this->createXmlDocument();
$xpath = new \DOMXPath($xmlDocument);
$this->clearAllCaches();
$this->purgeVisitedNodes();
$nodesWithStyleAttributes = $xpath->query('//*[@style]');
if ($nodesWithStyleAttributes !== false)
{
foreach ($nodesWithStyleAttributes as $node)
{
$normalizedOriginalStyle = preg_replace_callback('/[A-z\\-]+(?=\\:)/S', function ($m)
{
return strtolower($m[0]);
}, $node->getAttribute('style'));
$nodePath = $node->getNodePath();
if (!isset($this->styleAttributesForNodes[$nodePath]))
{
$this->styleAttributesForNodes[$nodePath] = $this->parseCssDeclarationBlock($normalizedOriginalStyle);
$this->visitedNodes[$nodePath] = $node;
}
$node->setAttribute('style', $normalizedOriginalStyle);
}
}
$allCss = $this->css;
$allCss .= $this->getCssFromAllStyleNodes($xpath);
$cssParts = $this->splitCssAndMediaQuery($allCss);
$cssKey = md5($cssParts['css']);
if (!isset($this->caches[self::CACHE_KEY_CSS][$cssKey]))
{
preg_match_all('/(?:^|[\\s^{}]*)([^{]+){([^}]*)}/mis', $cssParts['css'], $matches, PREG_SET_ORDER);
$allSelectors = array();
foreach ($matches as $key => $selectorString)
{
if (!strlen(trim($selectorString[2])))
{
continue;
}
$selectors = explode(',', $selectorString[1]);
foreach ($selectors as $selector)
{
if (strpos($selector, ':') !== false && !preg_match('/:\\S+\\-(child|type)\\(/i', $selector))
{
continue;
}
$allSelectors[] = array('selector' => trim($selector), 'attributes' => trim($selectorString[2]), 'line' => $key);
}
}
usort($allSelectors, array($this, 'sortBySelectorPrecedence'));
$this->caches[self::CACHE_KEY_CSS][$cssKey] = $allSelectors;
}
foreach ($this->caches[self::CACHE_KEY_CSS][$cssKey] as $value)
{
$nodesMatchingCssSelectors = $xpath->query($this->translateCssToXpath($value['selector']));
foreach ($nodesMatchingCssSelectors as $node)
{
if ($node->hasAttribute('style'))
{
$oldStyleDeclarations = $this->parseCssDeclarationBlock($node->getAttribute('style'));
}
else
{
$oldStyleDeclarations = array();
}
$newStyleDeclarations = $this->parseCssDeclarationBlock($value['attributes']);
$node->setAttribute('style', $this->generateStyleStringFromDeclarationsArrays($oldStyleDeclarations, $newStyleDeclarations));
}
}
foreach ($this->styleAttributesForNodes as $nodePath => $styleAttributesForNode)
{
$node = $this->visitedNodes[$nodePath];
$currentStyleAttributes = $this->parseCssDeclarationBlock($node->getAttribute('style'));
$node->setAttribute('style', $this->generateStyleStringFromDeclarationsArrays($currentStyleAttributes, $styleAttributesForNode));
}
$nodesWithStyleDisplayNone = $xpath->query('//*[contains(translate(translate(@style," ",""),"NOE","noe"),"display:none")]');
if ($nodesWithStyleDisplayNone->length > 0)
{
foreach ($nodesWithStyleDisplayNone as $node)
{
if ($node->parentNode && is_callable(array($node->parentNode, 'removeChild')))
{
$node->parentNode->removeChild($node);
}
}
}
$this->copyCssWithMediaToStyleNode($cssParts, $xmlDocument);
if ($this->preserveEncoding)
{
return mb_convert_encoding($xmlDocument->saveHTML(), self::ENCODING, 'HTML-ENTITIES');
}
else
{
return $xmlDocument->saveHTML();
}
}
private function clearAllCaches()
{
$this->clearCache(self::CACHE_KEY_CSS);
$this->clearCache(self::CACHE_KEY_SELECTOR);
$this->clearCache(self::CACHE_KEY_XPATH);
$this->clearCache(self::CACHE_KEY_CSS_DECLARATION_BLOCK);
}
private function clearCache($key)
{
$allowedCacheKeys = array(self::CACHE_KEY_CSS, self::CACHE_KEY_SELECTOR, self::CACHE_KEY_XPATH, self::CACHE_KEY_CSS_DECLARATION_BLOCK);
if (!in_array($key, $allowedCacheKeys, true))
{
throw new \InvalidArgumentException('Invalid cache key: ' . $key, 1391822035);
}
$this->caches[$key] = array();
}
private function purgeVisitedNodes()
{
$this->visitedNodes = array();
$this->styleAttributesForNodes = array();
}
public function addUnprocessableHtmlTag($tagName)
{
$this->unprocessableHtmlTags[] = $tagName;
}
public function removeUnprocessableHtmlTag($tagName)
{
$key = array_search($tagName, $this->unprocessableHtmlTags, true);
if ($key !== false)
{
unset($this->unprocessableHtmlTags[$key]);
}
}
private function generateStyleStringFromDeclarationsArrays($oldStyles, $newStyles)
{
$combinedStyles = array_merge($oldStyles, $newStyles);
$style = '';
foreach ($combinedStyles as $attributeName => $attributeValue)
{
$style .= strtolower(trim($attributeName)) . ': ' . trim($attributeValue) . '; ';
}
return trim($style);
}
public function copyCssWithMediaToStyleNode($cssParts, $xmlDocument)
{
if (isset($cssParts['media']) && $cssParts['media'] !== '')
{
$this->addStyleElementToDocument($xmlDocument, $cssParts['media']);
}
}
private function getCssFromAllStyleNodes($xpath)
{
$styleNodes = $xpath->query('//style');
if ($styleNodes === false)
{
return '';
}
$css = '';
foreach ($styleNodes as $styleNode)
{
$css .= "\n\n" . $styleNode->nodeValue;
$styleNode->parentNode->removeChild($styleNode);
}
return $css;
}
protected function addStyleElementToDocument($document, $css)
{
$styleElement = $document->createElement('style', $css);
$styleAttribute = $document->createAttribute('type');
$styleAttribute->value = 'text/css';
$styleElement->appendChild($styleAttribute);
$head = $this->getOrCreateHeadElement($document);
$head->appendChild($styleElement);
}
private function getOrCreateHeadElement($document)
{
$head = $document->getElementsByTagName('head')->item(0);
if ($head === null)
{
$head = $document->createElement('head');
$html = $document->getElementsByTagName('html')->item(0);
$html->insertBefore($head, $document->getElementsByTagName('body')->item(0));
}
return $head;
}
private function splitCssAndMediaQuery($css)
{
$media = '';
$css = preg_replace_callback('#@media\\s+(?:only\\s)?(?:[\\s{\\(]|screen|all)\\s?[^{]+{.*}\\s*}\\s*#misU', function ($matches) use(&$media)
{
$media .= $matches[0];
}, $css);
$search = array('/\\/\\*.*\\*\\//sU', '/^\\s*@import\\s[^;]+;/misU', '/^\\s*@media\\s[^{]+{(.*)}\\s*}\\s/misU');
$replace = array('', '', '');
$css = preg_replace($search, $replace, $css);
return array('css' => $css, 'media' => $media);
}
private function createXmlDocument()
{
$xmlDocument = new \DOMDocument();
$xmlDocument->encoding = self::ENCODING;
$xmlDocument->strictErrorChecking = false;
$xmlDocument->formatOutput = true;
$libXmlState = libxml_use_internal_errors(true);
$xmlDocument->loadHTML($this->getUnifiedHtml());
libxml_clear_errors();
libxml_use_internal_errors($libXmlState);
$xmlDocument->normalizeDocument();
return $xmlDocument;
}
private function getUnifiedHtml()
{
if (!empty($this->unprocessableHtmlTags))
{
$unprocessableHtmlTags = implode('|', $this->unprocessableHtmlTags);
$bodyWithoutUnprocessableTags = preg_replace('/<\\/?(' . $unprocessableHtmlTags . ')[^>]*>/i', '', $this->html);
}
else
{
$bodyWithoutUnprocessableTags = $this->html;
}
return mb_convert_encoding($bodyWithoutUnprocessableTags, 'HTML-ENTITIES', self::ENCODING);
}
private function sortBySelectorPrecedence($a, $b)
{
$precedenceA = $this->getCssSelectorPrecedence($a['selector']);
$precedenceB = $this->getCssSelectorPrecedence($b['selector']);
$precedenceForEquals = $a['line'] < $b['line'] ? -1 : 1;
$precedenceForNotEquals = $precedenceA < $precedenceB ? -1 : 1;
return $precedenceA === $precedenceB ? $precedenceForEquals : $precedenceForNotEquals;
}
private function getCssSelectorPrecedence($selector)
{
$selectorKey = md5($selector);
if (!isset($this->caches[self::CACHE_KEY_SELECTOR][$selectorKey]))
{
$precedence = 0;
$value = 100;
$search = array('\\#', '\\.', '');
foreach ($search as $s)
{
if (trim($selector) === '')
{
break;
}
$number = 0;
$selector = preg_replace('/' . $s . '\\w+/', '', $selector, -1, $number);
$precedence += $value * $number;
$value /= 10;
}
$this->caches[self::CACHE_KEY_SELECTOR][$selectorKey] = $precedence;
}
return $this->caches[self::CACHE_KEY_SELECTOR][$selectorKey];
}
private function translateCssToXpath($paramCssSelector)
{
$cssSelector = ' ' . $paramCssSelector . ' ';
$cssSelector = preg_replace_callback('/\\s+\\w+\\s+/', function ($matches)
{
return strtolower($matches[0]);
}, $cssSelector);
$cssSelector = trim($cssSelector);
$xpathKey = md5($cssSelector);
if (!isset($this->caches[self::CACHE_KEY_XPATH][$xpathKey]))
{
$search = array('/\\s+>\\s+/', '/\\s+\\+\\s+/', '/\\s+/', '/([^\\/]+):first-child/i', '/([^\\/]+):last-child/i', '/^\\[(\\w+)\\]/', '/(\\w)\\[(\\w+)\\]/', '/(\\w)\\[(\\w+)\\=[\'"]?(\\w+)[\'"]?\\]/');
$replace = array('/', '/following-sibling::*[1]/self::', '//', '*[1]/self::\\1', '*[last()]/self::\\1', '*[@\\1]', '\\1[@\\2]', '\\1[@\\2="\\3"]');
$cssSelector = '//' . preg_replace($search, $replace, $cssSelector);
$cssSelector = preg_replace_callback(self::ID_ATTRIBUTE_MATCHER, array($this, 'matchIdAttributes'), $cssSelector);
$cssSelector = preg_replace_callback(self::CLASS_ATTRIBUTE_MATCHER, array($this, 'matchClassAttributes'), $cssSelector);
$cssSelector = preg_replace_callback('/([^\\/]+):nth-child\\(\\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', array($this, 'translateNthChild'), $cssSelector);
$cssSelector = preg_replace_callback('/([^\\/]+):nth-of-type\\(\\s*(odd|even|[+\\-]?\\d|[+\\-]?\\d?n(\\s*[+\\-]\\s*\\d)?)\\s*\\)/i', array($this, 'translateNthOfType'), $cssSelector);
$this->caches[self::CACHE_KEY_SELECTOR][$xpathKey] = $cssSelector;
}
return $this->caches[self::CACHE_KEY_SELECTOR][$xpathKey];
}
private function matchIdAttributes($match)
{
return (strlen($match[1]) ? $match[1] : '*') . '[@id="' . $match[2] . '"]';
}
private function matchClassAttributes($match)
{
return (strlen($match[1]) ? $match[1] : '*') . '[contains(concat(" ",@class," "),concat(" ","' . implode('"," "))][contains(concat(" ",@class," "),concat(" ","', explode('.', substr($match[2], 1))) . '"," "))]';
}
private function translateNthChild($match)
{
$result = $this->parseNth($match);
if (isset($result[self::MULTIPLIER]))
{
if ($result[self::MULTIPLIER] < 0)
{
$result[self::MULTIPLIER] = abs($result[self::MULTIPLIER]);
return sprintf('*[(last() - position()) mod %u = %u]/self::%s', $result[self::MULTIPLIER], $result[self::INDEX], $match[1]);
}
else
{
return sprintf('*[position() mod %u = %u]/self::%s', $result[self::MULTIPLIER], $result[self::INDEX], $match[1]);
}
}
else
{
return sprintf('*[%u]/self::%s', $result[self::INDEX], $match[1]);
}
}
private function translateNthOfType($match)
{
$result = $this->parseNth($match);
if (isset($result[self::MULTIPLIER]))
{
if ($result[self::MULTIPLIER] < 0)
{
$result[self::MULTIPLIER] = abs($result[self::MULTIPLIER]);
return sprintf('%s[(last() - position()) mod %u = %u]', $match[1], $result[self::MULTIPLIER], $result[self::INDEX]);
}
else
{
return sprintf('%s[position() mod %u = %u]', $match[1], $result[self::MULTIPLIER], $result[self::INDEX]);
}
}
else
{
return sprintf('%s[%u]', $match[1], $result[self::INDEX]);
}
}
private function parseNth($match)
{
if (in_array(strtolower($match[2]), array('even', 'odd'), true))
{
$index = strtolower($match[2]) === 'even' ? 0 : 1;
return array(self::MULTIPLIER => 2, self::INDEX => $index);
}
elseif (stripos($match[2], 'n') === false)
{
$index = (int) str_replace(' ', '', $match[2]);
return array(self::INDEX => $index);
}
else
{
if (isset($match[3]))
{
$multipleTerm = str_replace($match[3], '', $match[2]);
$index = (int) str_replace(' ', '', $match[3]);
}
else
{
$multipleTerm = $match[2];
$index = 0;
}
$multiplier = (int) str_ireplace('n', '', $multipleTerm);
if (!strlen($multiplier))
{
$multiplier = 1;
}
elseif ($multiplier === 0)
{
return array(self::INDEX => $index);
}
else
{
$multiplier = (int) $multiplier;
}
while ($index < 0)
{
$index += abs($multiplier);
}
return array(self::MULTIPLIER => $multiplier, self::INDEX => $index);
}
}
private function parseCssDeclarationBlock($cssDeclarationBlock)
{
if (isset($this->caches[self::CACHE_KEY_CSS_DECLARATION_BLOCK][$cssDeclarationBlock]))
{
return $this->caches[self::CACHE_KEY_CSS_DECLARATION_BLOCK][$cssDeclarationBlock];
}
$properties = array();
$declarations = explode(';', $cssDeclarationBlock);
foreach ($declarations as $declaration)
{
$matches = array();
if (!preg_match('/ *([A-Za-z\\-]+) *: *([^;]+) */', $declaration, $matches))
{
continue;
}
$propertyName = strtolower($matches[1]);
$propertyValue = $matches[2];
$properties[$propertyName] = $propertyValue;
}
$this->caches[self::CACHE_KEY_CSS_DECLARATION_BLOCK][$cssDeclarationBlock] = $properties;
return $properties;
}
}