*/
class Minify_HTML {
/**
* "Minify" an HTML page
*
* @param string $html
*
* @param array $options
*
* 'cssMinifier' : (optional) callback function to process content of STYLE
* elements.
*
* 'jsMinifier' : (optional) callback function to process content of SCRIPT
* elements. Note: the type attribute is ignored.
*
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
* unset, minify will sniff for an XHTML doctype.
*
* @return string
*/
public static function minify($html, $options = array()) {
$min = new Minify_HTML($html, $options);
return $min->process();
}
/**
* Create a minifier object
*
* @param string $html
*
* @param array $options
*
* 'cssMinifier' : (optional) callback function to process content of STYLE
* elements.
*
* 'jsMinifier' : (optional) callback function to process content of SCRIPT
* elements. Note: the type attribute is ignored.
*
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
* unset, minify will sniff for an XHTML doctype.
*
* @return null
*/
public function __construct($html, $options = array())
{
$this->_html = str_replace("\r\n", "\n", trim($html));
if (isset($options['xhtml'])) {
$this->_isXhtml = (bool)$options['xhtml'];
}
if (isset($options['cssMinifier'])) {
$this->_cssMinifier = $options['cssMinifier'];
}
if (isset($options['jsMinifier'])) {
$this->_jsMinifier = $options['jsMinifier'];
}
}
/**
* Minify the markeup given in the constructor
*
* @return string
*/
public function process()
{
if ($this->_isXhtml === null) {
$this->_isXhtml = (false !== strpos($this->_html, '_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
$this->_placeholders = array();
// replace SCRIPTs (and minify) with placeholders
$this->_html = preg_replace_callback(
'/(\\s*)(]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i'
,array($this, '_removeScriptCB')
,$this->_html);
// replace STYLEs (and minify) with placeholders
$this->_html = preg_replace_callback(
'/\\s*(]*?>)([\\s\\S]*?)<\\/style>\\s*/i'
,array($this, '_removeStyleCB')
,$this->_html);
// remove HTML comments (not containing IE conditional comments).
$this->_html = preg_replace_callback(
'//'
,array($this, '_commentCB')
,$this->_html);
// replace PREs with placeholders
$this->_html = preg_replace_callback('/\\s*(]*?>[\\s\\S]*?<\\/pre>)\\s*/i'
,array($this, '_removePreCB')
,$this->_html);
// replace TEXTAREAs with placeholders
$this->_html = preg_replace_callback(
'/\\s*(]*?>[\\s\\S]*?<\\/textarea>)\\s*/i'
,array($this, '_removeTextareaCB')
,$this->_html);
// trim each line.
// @todo take into account attribute values that span multiple lines.
$this->_html = preg_replace('/^\\s+|\\s+$/m', '', $this->_html);
// remove ws around block/undisplayed elements
$this->_html = preg_replace('/\\s+(<\\/?(?:area|base(?:font)?|blockquote|body'
.'|caption|center|cite|col(?:group)?|dd|dir|div|dl|dt|fieldset|form'
.'|frame(?:set)?|h[1-6]|head|hr|html|legend|li|link|map|menu|meta'
.'|ol|opt(?:group|ion)|p|param|t(?:able|body|head|d|h||r|foot|itle)'
.'|ul)\\b[^>]*>)/i', '$1', $this->_html);
// remove ws outside of all elements
$this->_html = preg_replace_callback(
'/>([^<]+)_html);
// use newlines before 1st attribute in open tags (to limit line lengths)
$this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html);
// fill placeholders
$this->_html = str_replace(
array_keys($this->_placeholders)
,array_values($this->_placeholders)
,$this->_html
);
return $this->_html;
}
protected function _commentCB($m)
{
return (0 === strpos($m[1], '[') || false !== strpos($m[1], '_replacementHash . count($this->_placeholders) . '%';
$this->_placeholders[$placeholder] = $content;
return $placeholder;
}
protected $_isXhtml = null;
protected $_replacementHash = null;
protected $_placeholders = array();
protected $_cssMinifier = null;
protected $_jsMinifier = null;
protected function _outsideTagCB($m)
{
return '>' . preg_replace('/^\\s+|\\s+$/', ' ', $m[1]) . '<';
}
protected function _removePreCB($m)
{
return $this->_reservePlace($m[1]);
}
protected function _removeTextareaCB($m)
{
return $this->_reservePlace($m[1]);
}
protected function _removeStyleCB($m)
{
$openStyle = $m[1];
$css = $m[2];
// remove HTML comments
$css = preg_replace('/(?:^\\s*\\s*$)/', '', $css);
// remove CDATA section markers
$css = $this->_removeCdata($css);
// minify
$minifier = $this->_cssMinifier
? $this->_cssMinifier
: 'trim';
$css = call_user_func($minifier, $css);
return $this->_reservePlace($this->_needsCdata($css)
? "{$openStyle}/**/"
: "{$openStyle}{$css}"
);
}
protected function _removeScriptCB($m)
{
$openScript = $m[2];
$js = $m[3];
// whitespace surrounding? preserve at least one space
$ws1 = ($m[1] === '') ? '' : ' ';
$ws2 = ($m[4] === '') ? '' : ' ';
// remove HTML comments (and ending "//" if present)
$js = preg_replace('/(?:^\\s*\\s*$)/', '', $js);
// remove CDATA section markers
$js = $this->_removeCdata($js);
// minify
$minifier = $this->_jsMinifier
? $this->_jsMinifier
: 'trim';
$js = call_user_func($minifier, $js);
return $this->_reservePlace($this->_needsCdata($js)
? "{$ws1}{$openScript}/**/{$ws2}"
: "{$ws1}{$openScript}{$js}{$ws2}"
);
}
protected function _removeCdata($str)
{
return (false !== strpos($str, ''), '', $str)
: $str;
}
protected function _needsCdata($str)
{
return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str));
}
}