2012-03-14 06:44:41 -04:00
|
|
|
<?php
|
|
|
|
|
/**
|
2014-05-05 16:55:36 -04:00
|
|
|
* Class Minify_HTML
|
2012-03-14 06:44:41 -04:00
|
|
|
* @package Minify
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* Compress HTML
|
|
|
|
|
*
|
2014-05-05 16:55:36 -04:00
|
|
|
* This is a heavy regex-based removal of whitespace, unnecessary comments and
|
2012-03-14 06:44:41 -04:00
|
|
|
* tokens. IE conditional comments are preserved. There are also options to have
|
2014-05-05 16:55:36 -04:00
|
|
|
* STYLE and SCRIPT blocks compressed by callback functions.
|
|
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* A test suite is available.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* @package Minify
|
|
|
|
|
* @author Stephen Clay <steve@mrclay.org>
|
|
|
|
|
*/
|
|
|
|
|
class Minify_HTML {
|
2014-05-05 16:55:36 -04:00
|
|
|
/**
|
|
|
|
|
* @var boolean
|
|
|
|
|
*/
|
|
|
|
|
protected $_jsCleanComments = true;
|
2012-03-14 06:44:41 -04:00
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* "Minify" an HTML page
|
|
|
|
|
*
|
|
|
|
|
* @param string $html
|
|
|
|
|
*
|
|
|
|
|
* @param array $options
|
|
|
|
|
*
|
|
|
|
|
* 'cssMinifier' : (optional) callback function to process content of STYLE
|
|
|
|
|
* elements.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* 'jsMinifier' : (optional) callback function to process content of SCRIPT
|
|
|
|
|
* elements. Note: the type attribute is ignored.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
|
|
|
|
|
* unset, minify will sniff for an XHTML doctype.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* @return string
|
|
|
|
|
*/
|
|
|
|
|
public static function minify($html, $options = array()) {
|
2014-05-05 16:55:36 -04:00
|
|
|
$min = new self($html, $options);
|
2012-03-14 06:44:41 -04:00
|
|
|
return $min->process();
|
|
|
|
|
}
|
2014-05-05 16:55:36 -04:00
|
|
|
|
|
|
|
|
|
2012-03-14 06:44:41 -04:00
|
|
|
/**
|
|
|
|
|
* Create a minifier object
|
|
|
|
|
*
|
|
|
|
|
* @param string $html
|
|
|
|
|
*
|
|
|
|
|
* @param array $options
|
|
|
|
|
*
|
|
|
|
|
* 'cssMinifier' : (optional) callback function to process content of STYLE
|
|
|
|
|
* elements.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* 'jsMinifier' : (optional) callback function to process content of SCRIPT
|
|
|
|
|
* elements. Note: the type attribute is ignored.
|
2014-05-05 16:55:36 -04:00
|
|
|
*
|
|
|
|
|
* 'jsCleanComments' : (optional) whether to remove HTML comments beginning and end of script block
|
|
|
|
|
*
|
2012-03-14 06:44:41 -04:00
|
|
|
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
|
|
|
|
|
* unset, minify will sniff for an XHTML doctype.
|
|
|
|
|
*/
|
|
|
|
|
public function __construct($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 = $options['jsMinifier'];
|
|
|
|
|
}
|
2014-05-05 16:55:36 -04:00
|
|
|
if (isset($options['jsCleanComments'])) {
|
|
|
|
|
$this->_jsCleanComments = (bool)$options['jsCleanComments'];
|
|
|
|
|
}
|
2012-03-14 06:44:41 -04:00
|
|
|
}
|
2014-05-05 16:55:36 -04:00
|
|
|
|
|
|
|
|
|
2012-03-14 06:44:41 -04:00
|
|
|
/**
|
|
|
|
|
* Minify the markeup given in the constructor
|
|
|
|
|
*
|
|
|
|
|
* @return string
|
|
|
|
|
*/
|
|
|
|
|
public function process()
|
|
|
|
|
{
|
|
|
|
|
if ($this->_isXhtml === null) {
|
|
|
|
|
$this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
$this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
|
|
|
|
|
$this->_placeholders = array();
|
|
|
|
|
|
|
|
|
|
// replace SCRIPTs (and minify) with placeholders
|
|
|
|
|
$this->_html = preg_replace_callback(
|
|
|
|
|
'/(\\s*)<script(\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i'
|
|
|
|
|
,array($this, '_removeScriptCB')
|
|
|
|
|
,$this->_html);
|
|
|
|
|
|
|
|
|
|
// replace STYLEs (and minify) with placeholders
|
|
|
|
|
$this->_html = preg_replace_callback(
|
|
|
|
|
'/\\s*<style(\\b[^>]*>)([\\s\\S]*?)<\\/style>\\s*/i'
|
|
|
|
|
,array($this, '_removeStyleCB')
|
|
|
|
|
,$this->_html);
|
|
|
|
|
|
|
|
|
|
// remove HTML comments (not containing IE conditional comments).
|
|
|
|
|
$this->_html = preg_replace_callback(
|
|
|
|
|
'/<!--([\\s\\S]*?)-->/'
|
|
|
|
|
,array($this, '_commentCB')
|
|
|
|
|
,$this->_html);
|
|
|
|
|
|
|
|
|
|
// replace PREs with placeholders
|
|
|
|
|
$this->_html = preg_replace_callback('/\\s*<pre(\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i'
|
|
|
|
|
,array($this, '_removePreCB')
|
|
|
|
|
,$this->_html);
|
|
|
|
|
|
|
|
|
|
// replace TEXTAREAs with placeholders
|
|
|
|
|
$this->_html = preg_replace_callback(
|
|
|
|
|
'/\\s*<textarea(\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i'
|
|
|
|
|
,array($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'
|
2014-05-05 16:55:36 -04:00
|
|
|
.'|caption|center|col(?:group)?|dd|dir|div|dl|dt|fieldset|form'
|
2012-03-14 06:44:41 -04:00
|
|
|
.'|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(
|
|
|
|
|
'/>(\\s(?:\\s*))?([^<]+)(\\s(?:\s*))?</'
|
|
|
|
|
,'>$1$2$3<'
|
|
|
|
|
,$this->_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
|
|
|
|
|
);
|
|
|
|
|
// issue 229: multi-pass to catch scripts that didn't get replaced in textareas
|
|
|
|
|
$this->_html = str_replace(
|
|
|
|
|
array_keys($this->_placeholders)
|
|
|
|
|
,array_values($this->_placeholders)
|
|
|
|
|
,$this->_html
|
|
|
|
|
);
|
|
|
|
|
return $this->_html;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _commentCB($m)
|
|
|
|
|
{
|
|
|
|
|
return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<!['))
|
|
|
|
|
? $m[0]
|
|
|
|
|
: '';
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _reservePlace($content)
|
|
|
|
|
{
|
|
|
|
|
$placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%';
|
|
|
|
|
$this->_placeholders[$placeholder] = $content;
|
|
|
|
|
return $placeholder;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected $_isXhtml = null;
|
|
|
|
|
protected $_replacementHash = null;
|
|
|
|
|
protected $_placeholders = array();
|
|
|
|
|
protected $_cssMinifier = null;
|
|
|
|
|
protected $_jsMinifier = null;
|
|
|
|
|
|
|
|
|
|
protected function _removePreCB($m)
|
|
|
|
|
{
|
|
|
|
|
return $this->_reservePlace("<pre{$m[1]}");
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _removeTextareaCB($m)
|
|
|
|
|
{
|
|
|
|
|
return $this->_reservePlace("<textarea{$m[1]}");
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _removeStyleCB($m)
|
|
|
|
|
{
|
|
|
|
|
$openStyle = "<style{$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}/*<![CDATA[*/{$css}/*]]>*/</style>"
|
|
|
|
|
: "{$openStyle}{$css}</style>"
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _removeScriptCB($m)
|
|
|
|
|
{
|
|
|
|
|
$openScript = "<script{$m[2]}";
|
|
|
|
|
$js = $m[3];
|
|
|
|
|
|
|
|
|
|
// whitespace surrounding? preserve at least one space
|
|
|
|
|
$ws1 = ($m[1] === '') ? '' : ' ';
|
|
|
|
|
$ws2 = ($m[4] === '') ? '' : ' ';
|
2014-05-05 16:55:36 -04:00
|
|
|
|
2012-03-14 06:44:41 -04:00
|
|
|
// remove HTML comments (and ending "//" if present)
|
2014-05-05 16:55:36 -04:00
|
|
|
if ($this->_jsCleanComments) {
|
|
|
|
|
$js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js);
|
|
|
|
|
}
|
|
|
|
|
|
2012-03-14 06:44:41 -04:00
|
|
|
// remove CDATA section markers
|
|
|
|
|
$js = $this->_removeCdata($js);
|
|
|
|
|
|
|
|
|
|
// minify
|
|
|
|
|
$minifier = $this->_jsMinifier
|
|
|
|
|
? $this->_jsMinifier
|
2014-05-05 16:55:36 -04:00
|
|
|
: 'trim';
|
2012-03-14 06:44:41 -04:00
|
|
|
$js = call_user_func($minifier, $js);
|
|
|
|
|
|
|
|
|
|
return $this->_reservePlace($this->_needsCdata($js)
|
|
|
|
|
? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}"
|
|
|
|
|
: "{$ws1}{$openScript}{$js}</script>{$ws2}"
|
|
|
|
|
);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _removeCdata($str)
|
|
|
|
|
{
|
|
|
|
|
return (false !== strpos($str, '<![CDATA['))
|
|
|
|
|
? str_replace(array('<![CDATA[', ']]>'), '', $str)
|
|
|
|
|
: $str;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected function _needsCdata($str)
|
|
|
|
|
{
|
|
|
|
|
return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str));
|
|
|
|
|
}
|
|
|
|
|
}
|
