| 1 |
<?php |
| 2 |
/** |
| 3 |
* Class Minify_HTML |
| 4 |
* @package Minify |
| 5 |
*/ |
| 6 |
|
| 7 |
/** |
| 8 |
* Compress HTML |
| 9 |
* |
| 10 |
* This is a heavy regex-based removal of whitespace, unnecessary comments and |
| 11 |
* tokens. IE conditional comments are preserved. There are also options to have |
| 12 |
* STYLE and SCRIPT blocks compressed by callback functions. |
| 13 |
* |
| 14 |
* A test suite is available. |
| 15 |
* |
| 16 |
* @package Minify |
| 17 |
* @author Stephen Clay <steve@mrclay.org> |
| 18 |
*/ |
| 19 |
class Minify_HTML |
| 20 |
{ |
| 21 |
/** |
| 22 |
* @var boolean |
| 23 |
*/ |
| 24 |
protected $_jsCleanComments = true; |
| 25 |
|
| 26 |
/** |
| 27 |
* @var string |
| 28 |
*/ |
| 29 |
protected $_html; |
| 30 |
|
| 31 |
/** |
| 32 |
* "Minify" an HTML page |
| 33 |
* |
| 34 |
* @param string $html |
| 35 |
* |
| 36 |
* @param array $options |
| 37 |
* |
| 38 |
* 'cssMinifier' : (optional) callback function to process content of STYLE |
| 39 |
* elements. |
| 40 |
* |
| 41 |
* 'jsMinifier' : (optional) callback function to process content of SCRIPT |
| 42 |
* elements. Note: the type attribute is ignored. |
| 43 |
* |
| 44 |
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If |
| 45 |
* unset, minify will sniff for an XHTML doctype. |
| 46 |
* |
| 47 |
* @return string |
| 48 |
*/ |
| 49 |
public static function minify($html, $options = array()) |
| 50 |
{ |
| 51 |
$min = new self($html, $options); |
| 52 |
|
| 53 |
return $min->process(); |
| 54 |
} |
| 55 |
|
| 56 |
/** |
| 57 |
* Create a minifier object |
| 58 |
* |
| 59 |
* @param string $html |
| 60 |
* |
| 61 |
* @param array $options |
| 62 |
* |
| 63 |
* 'cssMinifier' : (optional) callback function to process content of STYLE |
| 64 |
* elements. |
| 65 |
* |
| 66 |
* 'jsMinifier' : (optional) callback function to process content of SCRIPT |
| 67 |
* elements. Note: the type attribute is ignored. |
| 68 |
* |
| 69 |
* 'jsCleanComments' : (optional) whether to remove HTML comments beginning and end of script block |
| 70 |
* |
| 71 |
* 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If |
| 72 |
* unset, minify will sniff for an XHTML doctype. |
| 73 |
*/ |
| 74 |
public function __construct($html, $options = array()) |
| 75 |
{ |
| 76 |
$this->_html = str_replace("\r\n", "\n", trim($html)); |
| 77 |
if (isset($options['xhtml'])) { |
| 78 |
$this->_isXhtml = (bool)$options['xhtml']; |
| 79 |
} |
| 80 |
if (isset($options['cssMinifier'])) { |
| 81 |
$this->_cssMinifier = $options['cssMinifier']; |
| 82 |
} |
| 83 |
if (isset($options['jsMinifier'])) { |
| 84 |
$this->_jsMinifier = $options['jsMinifier']; |
| 85 |
} |
| 86 |
if (isset($options['jsCleanComments'])) { |
| 87 |
$this->_jsCleanComments = (bool)$options['jsCleanComments']; |
| 88 |
} |
| 89 |
} |
| 90 |
|
| 91 |
/** |
| 92 |
* Minify the markeup given in the constructor |
| 93 |
* |
| 94 |
* @return string |
| 95 |
*/ |
| 96 |
public function process() |
| 97 |
{ |
| 98 |
if ($this->_isXhtml === null) { |
| 99 |
$this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML')); |
| 100 |
} |
| 101 |
|
| 102 |
$this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']); |
| 103 |
$this->_placeholders = array(); |
| 104 |
|
| 105 |
// replace SCRIPTs (and minify) with placeholders |
| 106 |
$this->_html = preg_replace_callback( |
| 107 |
'/(\\s*)<script(\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/iu', |
| 108 |
array($this, '_removeScriptCB'), |
| 109 |
$this->_html |
| 110 |
); |
| 111 |
|
| 112 |
// replace STYLEs (and minify) with placeholders |
| 113 |
$this->_html = preg_replace_callback( |
| 114 |
'/\\s*<style(\\b[^>]*>)([\\s\\S]*?)<\\/style>\\s*/iu', |
| 115 |
array($this, '_removeStyleCB'), |
| 116 |
$this->_html |
| 117 |
); |
| 118 |
|
| 119 |
// remove HTML comments (not containing IE conditional comments). |
| 120 |
$this->_html = preg_replace_callback( |
| 121 |
'/<!--([\\s\\S]*?)-->/u', |
| 122 |
array($this, '_commentCB'), |
| 123 |
$this->_html |
| 124 |
); |
| 125 |
|
| 126 |
// replace PREs with placeholders |
| 127 |
$this->_html = preg_replace_callback('/\\s*<pre(\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/iu', array($this, '_removePreCB'), $this->_html); |
| 128 |
|
| 129 |
// replace TEXTAREAs with placeholders |
| 130 |
$this->_html = preg_replace_callback( |
| 131 |
'/\\s*<textarea(\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/iu', |
| 132 |
array($this, '_removeTextareaCB'), |
| 133 |
$this->_html |
| 134 |
); |
| 135 |
|
| 136 |
// trim each line. |
| 137 |
// @todo take into account attribute values that span multiple lines. |
| 138 |
$this->_html = preg_replace('/^\\s+|\\s+$/mu', '', $this->_html); |
| 139 |
|
| 140 |
// remove ws around block/undisplayed elements |
| 141 |
$this->_html = preg_replace('/\\s+(<\\/?(?:area|article|aside|base(?:font)?|blockquote|body' |
| 142 |
.'|canvas|caption|center|col(?:group)?|dd|dir|div|dl|dt|fieldset|figcaption|figure|footer|form' |
| 143 |
.'|frame(?:set)?|h[1-6]|head|header|hgroup|hr|html|legend|li|link|main|map|menu|meta|nav' |
| 144 |
.'|ol|opt(?:group|ion)|output|p|param|section|t(?:able|body|head|d|h||r|foot|itle)' |
| 145 |
.'|ul|video)\\b[^>]*>)/iu', '$1', $this->_html); |
| 146 |
|
| 147 |
// remove ws outside of all elements |
| 148 |
$this->_html = preg_replace( |
| 149 |
'/>(\\s(?:\\s*))?([^<]+)(\\s(?:\s*))?</u', |
| 150 |
'>$1$2$3<', |
| 151 |
$this->_html |
| 152 |
); |
| 153 |
|
| 154 |
// use newlines before 1st attribute in open tags (to limit line lengths) |
| 155 |
$this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/iu', "$1\n$2", $this->_html); |
| 156 |
|
| 157 |
// fill placeholders |
| 158 |
$this->_html = str_replace( |
| 159 |
array_keys($this->_placeholders), |
| 160 |
array_values($this->_placeholders), |
| 161 |
$this->_html |
| 162 |
); |
| 163 |
// issue 229: multi-pass to catch scripts that didn't get replaced in textareas |
| 164 |
$this->_html = str_replace( |
| 165 |
array_keys($this->_placeholders), |
| 166 |
array_values($this->_placeholders), |
| 167 |
$this->_html |
| 168 |
); |
| 169 |
|
| 170 |
return $this->_html; |
| 171 |
} |
| 172 |
|
| 173 |
protected function _commentCB($m) |
| 174 |
{ |
| 175 |
return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<![') || 0 === strpos($m[1], '#')) |
| 176 |
? $m[0] |
| 177 |
: ''; |
| 178 |
} |
| 179 |
|
| 180 |
protected function _reservePlace($content) |
| 181 |
{ |
| 182 |
$placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%'; |
| 183 |
$this->_placeholders[$placeholder] = $content; |
| 184 |
|
| 185 |
return $placeholder; |
| 186 |
} |
| 187 |
|
| 188 |
protected $_isXhtml; |
| 189 |
protected $_replacementHash; |
| 190 |
protected $_placeholders = array(); |
| 191 |
protected $_cssMinifier; |
| 192 |
protected $_jsMinifier; |
| 193 |
|
| 194 |
protected function _removePreCB($m) |
| 195 |
{ |
| 196 |
return $this->_reservePlace("<pre{$m[1]}"); |
| 197 |
} |
| 198 |
|
| 199 |
protected function _removeTextareaCB($m) |
| 200 |
{ |
| 201 |
return $this->_reservePlace("<textarea{$m[1]}"); |
| 202 |
} |
| 203 |
|
| 204 |
protected function _removeStyleCB($m) |
| 205 |
{ |
| 206 |
$openStyle = "<style{$m[1]}"; |
| 207 |
$css = $m[2]; |
| 208 |
// remove HTML comments |
| 209 |
$css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/u', '', $css); |
| 210 |
|
| 211 |
// remove CDATA section markers |
| 212 |
$css = $this->_removeCdata($css); |
| 213 |
|
| 214 |
// minify |
| 215 |
$minifier = $this->_cssMinifier |
| 216 |
? $this->_cssMinifier |
| 217 |
: 'trim'; |
| 218 |
$css = call_user_func($minifier, $css); |
| 219 |
|
| 220 |
return $this->_reservePlace( |
| 221 |
$this->_needsCdata($css) |
| 222 |
? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>" |
| 223 |
: "{$openStyle}{$css}</style>" |
| 224 |
); |
| 225 |
} |
| 226 |
|
| 227 |
protected function _removeScriptCB($m) |
| 228 |
{ |
| 229 |
$openScript = "<script{$m[2]}"; |
| 230 |
$js = $m[3]; |
| 231 |
|
| 232 |
// whitespace surrounding? preserve at least one space |
| 233 |
$ws1 = ($m[1] === '') ? '' : ' '; |
| 234 |
$ws2 = ($m[4] === '') ? '' : ' '; |
| 235 |
|
| 236 |
// remove HTML comments (and ending "//" if present) |
| 237 |
if ($this->_jsCleanComments) { |
| 238 |
$js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/u', '', $js); |
| 239 |
} |
| 240 |
|
| 241 |
// remove CDATA section markers |
| 242 |
$js = $this->_removeCdata($js); |
| 243 |
|
| 244 |
// minify |
| 245 |
$minifier = $this->_jsMinifier |
| 246 |
? $this->_jsMinifier |
| 247 |
: 'trim'; |
| 248 |
$js = call_user_func($minifier, $js); |
| 249 |
|
| 250 |
return $this->_reservePlace( |
| 251 |
$this->_needsCdata($js) |
| 252 |
? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}" |
| 253 |
: "{$ws1}{$openScript}{$js}</script>{$ws2}" |
| 254 |
); |
| 255 |
} |
| 256 |
|
| 257 |
protected function _removeCdata($str) |
| 258 |
{ |
| 259 |
return (false !== strpos($str, '<![CDATA[')) |
| 260 |
? str_replace(array('<![CDATA[', ']]>'), '', $str) |
| 261 |
: $str; |
| 262 |
} |
| 263 |
|
| 264 |
protected function _needsCdata($str) |
| 265 |
{ |
| 266 |
return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/u', $str)); |
| 267 |
} |
| 268 |
} |
| 269 |
|
