mirror/minify
1
0
Fork 0
mirror of https://github.com/mrclay/minify.git synced 2025-08-17 19:37:22 +02:00
Code Issues Releases Wiki Activity

Minify/HTML.php : Made class dynamic to ease subclassing

Browse Source
This commit is contained in:
Steve Clay
2009-03-16 21:24:54 +00:00
parent 22ee97a934
commit e315788a43
1 changed files with 100 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

177
min/lib/Minify/HTML.php
View File

@@ -18,13 +18,6 @@
*/ */
class Minify_HTML { class Minify_HTML {
/**
* Defines which class to call as part of callbacks, change this
* if you extend Minify_HTML
* @var string
*/
protected static $className = 'Minify_HTML';
/** /**
* "Minify" an HTML page * "Minify" an HTML page
* *
@@ -44,122 +37,152 @@ class Minify_HTML {
* @return string * @return string
*/ */
public static function minify($html, $options = array()) { public static function minify($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($options['cssMinifier'])) { if (isset($options['cssMinifier'])) {
self::$_cssMinifier = $options['cssMinifier']; $this->_cssMinifier = $options['cssMinifier'];
} }
if (isset($options['jsMinifier'])) { if (isset($options['jsMinifier'])) {
self::$_jsMinifier = $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, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'));
} }
$html = str_replace("\r\n", "\n", trim($html)); $this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
$this->_placeholders = array();
self::$_isXhtml = (
isset($options['xhtml'])
? (bool)$options['xhtml']
: (false !== strpos($html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'))
);
self::$_replacementHash = 'MINIFYHTML' . md5(time());
self::$_placeholders = array();
// replace SCRIPTs (and minify) with placeholders // replace SCRIPTs (and minify) with placeholders
$html = preg_replace_callback( $this->_html = preg_replace_callback(
'/(\\s*)(<script\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i' '/(\\s*)(<script\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i'
,array(self::$className, '_removeScriptCB') ,array($this, '_removeScriptCB')
,$html); ,$this->_html);
// replace STYLEs (and minify) with placeholders // replace STYLEs (and minify) with placeholders
$html = preg_replace_callback( $this->_html = preg_replace_callback(
'/\\s*(<style\\b[^>]*?>)([\\s\\S]*?)<\\/style>\\s*/i' '/\\s*(<style\\b[^>]*?>)([\\s\\S]*?)<\\/style>\\s*/i'
,array(self::$className, '_removeStyleCB') ,array($this, '_removeStyleCB')
,$html); ,$this->_html);
// remove HTML comments (not containing IE conditional comments). // remove HTML comments (not containing IE conditional comments).
$html = preg_replace_callback( $this->_html = preg_replace_callback(
'/<!--([\\s\\S]*?)-->/' '/<!--([\\s\\S]*?)-->/'
,array(self::$className, '_commentCB') ,array($this, '_commentCB')
,$html); ,$this->_html);
// replace PREs with placeholders // replace PREs with placeholders
$html = preg_replace_callback('/\\s*(<pre\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i' $this->_html = preg_replace_callback('/\\s*(<pre\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i'
,array(self::$className, '_removePreCB') ,array($this, '_removePreCB')
, $html); ,$this->_html);
// replace TEXTAREAs with placeholders // replace TEXTAREAs with placeholders
$html = preg_replace_callback( $this->_html = preg_replace_callback(
'/\\s*(<textarea\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i' '/\\s*(<textarea\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i'
,array(self::$className, '_removeTaCB') ,array($this, '_removeTextareaCB')
, $html); ,$this->_html);
// trim each line. // trim each line.
// @todo take into account attribute values that span multiple lines. // @todo take into account attribute values that span multiple lines.
$html = preg_replace('/^\\s+|\\s+$/m', '', $html); $this->_html = preg_replace('/^\\s+|\\s+$/m', '', $this->_html);
// remove ws around block/undisplayed elements // remove ws around block/undisplayed elements
$html = preg_replace('/\\s+(<\\/?(?:area|base(?:font)?|blockquote|body' $this->_html = preg_replace('/\\s+(<\\/?(?:area|base(?:font)?|blockquote|body'
.'|caption|center|cite|col(?:group)?|dd|dir|div|dl|dt|fieldset|form' .'|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' .'|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)' .'|ol|opt(?:group|ion)|p|param|t(?:able|body|head|d|h||r|foot|itle)'
.'|ul)\\b[^>]*>)/i', '$1', $html); .'|ul)\\b[^>]*>)/i', '$1', $this->_html);
// remove ws outside of all elements // remove ws outside of all elements
$html = preg_replace_callback( $this->_html = preg_replace_callback(
'/>([^<]+)</' '/>([^<]+)</'
,array(self::$className, '_outsideTagCB') ,array($this, '_outsideTagCB')
,$html); ,$this->_html);
// use newlines before 1st attribute in open tags (to limit line lengths) // use newlines before 1st attribute in open tags (to limit line lengths)
$html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $html); $this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html);
// fill placeholders // fill placeholders
$html = str_replace( $this->_html = str_replace(
array_keys(self::$_placeholders) array_keys($this->_placeholders)
,array_values(self::$_placeholders) ,array_values($this->_placeholders)
,$html ,$this->_html
); );
self::$_placeholders = array(); return $this->_html;
self::$_cssMinifier = self::$_jsMinifier = null;
return $html;
} }
protected static function _commentCB($m) protected function _commentCB($m)
{ {
return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<![')) return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<!['))
? $m[0] ? $m[0]
: ''; : '';
} }
protected static function _reservePlace($content) protected function _reservePlace($content)
{ {
$placeholder = '%' . self::$_replacementHash . count(self::$_placeholders) . '%'; $placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%';
self::$_placeholders[$placeholder] = $content; $this->_placeholders[$placeholder] = $content;
return $placeholder; return $placeholder;
} }
protected static $_isXhtml = false; protected $_isXhtml = null;
protected static $_replacementHash = null; protected $_replacementHash = null;
protected static $_placeholders = array(); protected $_placeholders = array();
protected static $_cssMinifier = null; protected $_cssMinifier = null;
protected static $_jsMinifier = null; protected $_jsMinifier = null;
protected static function _outsideTagCB($m) protected function _outsideTagCB($m)
{ {
return '>' . preg_replace('/^\\s+|\\s+$/', ' ', $m[1]) . '<'; return '>' . preg_replace('/^\\s+|\\s+$/', ' ', $m[1]) . '<';
} }
protected static function _removePreCB($m) protected function _removePreCB($m)
{ {
return self::_reservePlace($m[1]); return $this->_reservePlace($m[1]);
} }
protected static function _removeTaCB($m) protected function _removeTextareaCB($m)
{ {
return self::_reservePlace($m[1]); return $this->_reservePlace($m[1]);
} }
protected static function _removeStyleCB($m) protected function _removeStyleCB($m)
{ {
$openStyle = $m[1]; $openStyle = $m[1];
$css = $m[2]; $css = $m[2];
@@ -167,21 +190,21 @@ class Minify_HTML {
$css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/', '', $css); $css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/', '', $css);
// remove CDATA section markers // remove CDATA section markers
$css = self::_removeCdata($css); $css = $this->_removeCdata($css);
// minify // minify
$minifier = self::$_cssMinifier $minifier = $this->_cssMinifier
? self::$_cssMinifier ? $this->_cssMinifier
: 'trim'; : 'trim';
$css = call_user_func($minifier, $css); $css = call_user_func($minifier, $css);
return self::_reservePlace(self::_needsCdata($css) return $this->_reservePlace($this->_needsCdata($css)
? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>" ? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>"
: "{$openStyle}{$css}</style>" : "{$openStyle}{$css}</style>"
); );
} }
protected static function _removeScriptCB($m) protected function _removeScriptCB($m)
{ {
$openScript = $m[2]; $openScript = $m[2];
$js = $m[3]; $js = $m[3];
@@ -194,29 +217,29 @@ class Minify_HTML {
$js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js); $js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js);
// remove CDATA section markers // remove CDATA section markers
$js = self::_removeCdata($js); $js = $this->_removeCdata($js);
// minify // minify
$minifier = self::$_jsMinifier $minifier = $this->_jsMinifier
? self::$_jsMinifier ? $this->_jsMinifier
: 'trim'; : 'trim';
$js = call_user_func($minifier, $js); $js = call_user_func($minifier, $js);
return self::_reservePlace(self::_needsCdata($js) return $this->_reservePlace($this->_needsCdata($js)
? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}" ? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}"
: "{$ws1}{$openScript}{$js}</script>{$ws2}" : "{$ws1}{$openScript}{$js}</script>{$ws2}"
); );
} }
protected static function _removeCdata($str) protected function _removeCdata($str)
{ {
return (false !== strpos($str, '<![CDATA[')) return (false !== strpos($str, '<![CDATA['))
? str_replace(array('<![CDATA[', ']]>'), '', $str) ? str_replace(array('<![CDATA[', ']]>'), '', $str)
: $str; : $str;
} }
protected static function _needsCdata($str) protected function _needsCdata($str)
{ {
return (self::$_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str)); return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str));
} }
} }