www/plugins-dist/compresseur/lib/minify_html/class.minify_html.php

   1 <?php
   2
   3 /**
   4  * Class Minify_HTML
   5  * @package Minify
   6  */
   7
   8
   9 /**
  10  * Surcharge du minifieur HTML
  11  *
  12  * Surcharge pour ne pas manger les commentaires <!--extra-->
  13  * qui servent parfois aux plugins, et parfois même après
  14  * le passage du compacteur HTML
  15  *
  16  * C'était le cas du bouton statistiques du formulaire admin par exemple
  17  *
  18  * On permet de conserver également tout commentaire commençant par <!--keepme: -->
  19  *
  20  * @package SPIP\Compresseur\Minifieur
  21 **/
  22 class Minify_HTML_SPIP extends Minify_HTML {
  23
  24     /**
  25      * {@inheritdoc}
  26      *
  27      *
  28      * @param string $html
  29      *     HTML à minifier
  30      * @param array $options
  31      *     Tableau d'option avec les index possibles
  32      *     - 'cssMinifier' : (optional) callback function to process content of STYLE
  33      *        elements.
  34      *     - 'jsMinifier' : (optional) callback function to process content of SCRIPT
  35      *        elements. Note: the type attribute is ignored.
  36      *     - 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
  37      *        unset, minify will sniff for an XHTML doctype.
  38      * @return string
  39      *     HTML minifié
  40     **/
  41     public static function minify($html, $options = array()) {
  42         $min = new Minify_HTML_SPIP($html, $options);
  43         return $min->process();
  44     }
  45
  46     /**
  47      * Minification des commentaires
  48      *
  49      * Le cas <!--extra--> doit être conservé dans les commentaires,
  50      * tout comme <!--keepme: xxx -->
  51      *
  52      * @param array $m    Matches du preg_match d'un commentaire HTML
  53      * @return string     Contenu minifié
  54      */
  55     protected function _commentCB($m)
  56     {
  57         if ($m[1] === 'extra') return $m[0];
  58         if ($m[1] AND $m[1][0] === 'k' AND substr($m[1],0,7) === 'keepme:') return $m[0];
  59         return parent::_commentCB($m);
  60     }
  61 }
  62
  63
  64 /**
  65  * Compress HTML
  66  *
  67  * This is a heavy regex-based removal of whitespace, unnecessary comments and
  68  * tokens. IE conditional comments are preserved. There are also options to have
  69  * STYLE and SCRIPT blocks compressed by callback functions.
  70  *
  71  * A test suite is available.
  72  *
  73  * @link https://github.com/mrclay/minify/blob/master/min/lib/Minify/HTML.php
  74  * @version 9e4176f1
  75  * @note Librairie modifiée pour prendre en compte les attributs avec multilignes.
  76  *
  77  * @package Minify
  78  * @author Stephen Clay <steve@mrclay.org>
  79  */
  80 class Minify_HTML {
  81     /**
  82      * @var boolean
  83      */
  84     protected $_jsCleanComments = true;
  85
  86     /**
  87      * "Minify" an HTML page
  88      *
  89      * @param string $html
  90      *
  91      * @param array $options
  92      *
  93      * 'cssMinifier' : (optional) callback function to process content of STYLE
  94      * elements.
  95      *
  96      * 'jsMinifier' : (optional) callback function to process content of SCRIPT
  97      * elements. Note: the type attribute is ignored.
  98      *
  99      * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
 100      * unset, minify will sniff for an XHTML doctype.
 101      *
 102      * @return string
 103      */
 104     public static function minify($html, $options = array()) {
 105         $min = new self($html, $options);
 106         return $min->process();
 107     }
 108
 109
 110     /**
 111      * Create a minifier object
 112      *
 113      * @param string $html
 114      *
 115      * @param array $options
 116      *
 117      * 'cssMinifier' : (optional) callback function to process content of STYLE
 118      * elements.
 119      *
 120      * 'jsMinifier' : (optional) callback function to process content of SCRIPT
 121      * elements. Note: the type attribute is ignored.
 122      *
 123      * 'jsCleanComments' : (optional) whether to remove HTML comments beginning and end of script block
 124      *
 125      * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If
 126      * unset, minify will sniff for an XHTML doctype.
 127      */
 128     public function __construct($html, $options = array())
 129     {
 130         $this->_html = str_replace("\r\n", "\n", trim($html));
 131         if (isset($options['xhtml'])) {
 132             $this->_isXhtml = (bool)$options['xhtml'];
 133         }
 134         if (isset($options['cssMinifier'])) {
 135             $this->_cssMinifier = $options['cssMinifier'];
 136         }
 137         if (isset($options['jsMinifier'])) {
 138             $this->_jsMinifier = $options['jsMinifier'];
 139         }
 140         if (isset($options['jsCleanComments'])) {
 141             $this->_jsCleanComments = (bool)$options['jsCleanComments'];
 142         }
 143     }
 144
 145
 146     /**
 147      * Minify the markeup given in the constructor
 148      *
 149      * @return string
 150      */
 151     public function process()
 152     {
 153         if ($this->_isXhtml === null) {
 154             $this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML'));
 155         }
 156
 157         $this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']);
 158         $this->_placeholders = array();
 159
 160         // replace SCRIPTs (and minify) with placeholders
 161         $this->_html = preg_replace_callback(
 162             '/(\\s*)<script(\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i'
 163             ,array($this, '_removeScriptCB')
 164             ,$this->_html);
 165
 166         // replace STYLEs (and minify) with placeholders
 167         $this->_html = preg_replace_callback(
 168             '/\\s*<style(\\b[^>]*>)([\\s\\S]*?)<\\/style>\\s*/i'
 169             ,array($this, '_removeStyleCB')
 170             ,$this->_html);
 171
 172         // remove HTML comments (not containing IE conditional comments).
 173         $this->_html = preg_replace_callback(
 174             '/<!--([\\s\\S]*?)-->/'
 175             ,array($this, '_commentCB')
 176             ,$this->_html);
 177
 178         // replace PREs with placeholders
 179         $this->_html = preg_replace_callback('/\\s*<pre(\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i'
 180             ,array($this, '_removePreCB')
 181             ,$this->_html);
 182
 183         // replace TEXTAREAs with placeholders
 184         $this->_html = preg_replace_callback(
 185             '/\\s*<textarea(\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i'
 186             ,array($this, '_removeTextareaCB')
 187             ,$this->_html);
 188
 189         // replace INPUTs with placeholders
 190         $this->_html = preg_replace_callback(
 191             '/\\s*<input(\\b[^>]*?>)\\s*/i'
 192             ,array($this, '_removeInputCB')
 193             ,$this->_html);
 194
 195         // trim each line.
 196         // @todo take into account attribute values that span multiple lines.
 197               // 2 regexp because merging un /^\\s+|\\s+$/m also del a lot of newline chars ???
 198         $this->_html = preg_replace('/\\s+$/m', '', $this->_html);
 199         $this->_html = preg_replace('/^\\s+/m', '', $this->_html);
 200
 201         // remove ws around block/undisplayed elements
 202         $this->_html = preg_replace('/\\s+(<\\/?(?:area|base(?:font)?|blockquote|body'
 203             .'|caption|center|col(?:group)?|dd|dir|div|dl|dt|fieldset|form'
 204             .'|frame(?:set)?|h[1-6]|head|hr|html|legend|li|link|map|menu|meta'
 205             .'|ol|opt(?:group|ion)|p|param|t(?:able|body|head|d|h||r|foot|itle)'
 206             .'|ul)\\b[^>]*>)/i', '$1', $this->_html);
 207
 208         // remove ws outside of all elements
 209         $this->_html = preg_replace(
 210             '/>(\\s(?:\\s*))?([^<]+)(\\s(?:\s*))?</'
 211             ,'>$1$2$3<'
 212             ,$this->_html);
 213
 214         // use newlines before 1st attribute in open tags (to limit line lengths)
 215         $this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html);
 216
 217         // fill placeholders
 218         $this->_html = str_replace(
 219             array_keys($this->_placeholders)
 220             ,array_values($this->_placeholders)
 221             ,$this->_html
 222         );
 223         // issue 229: multi-pass to catch scripts that didn't get replaced in textareas
 224         $this->_html = str_replace(
 225             array_keys($this->_placeholders)
 226             ,array_values($this->_placeholders)
 227             ,$this->_html
 228         );
 229         return $this->_html;
 230     }
 231
 232     protected function _commentCB($m)
 233     {
 234         return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<!['))
 235             ? $m[0]
 236             : '';
 237     }
 238
 239     protected function _reservePlace($content)
 240     {
 241         $placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%';
 242         $this->_placeholders[$placeholder] = $content;
 243         return $placeholder;
 244     }
 245
 246     protected $_isXhtml = null;
 247     protected $_replacementHash = null;
 248     protected $_placeholders = array();
 249     protected $_cssMinifier = null;
 250     protected $_jsMinifier = null;
 251
 252     protected function _removePreCB($m)
 253     {
 254         return $this->_reservePlace("<pre{$m[1]}");
 255     }
 256
 257                 protected function _removeInputCB($m)
 258                 {
 259              return $this->_reservePlace("<input{$m[1]}");
 260                 }
 261
 262     protected function _removeTextareaCB($m)
 263     {
 264         return $this->_reservePlace("<textarea{$m[1]}");
 265     }
 266
 267     protected function _removeStyleCB($m)
 268     {
 269         $openStyle = "<style{$m[1]}";
 270         $css = $m[2];
 271         // remove HTML comments
 272         $css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/', '', $css);
 273
 274         // remove CDATA section markers
 275         $css = $this->_removeCdata($css);
 276
 277         // minify
 278         $minifier = $this->_cssMinifier
 279             ? $this->_cssMinifier
 280             : 'trim';
 281         $css = call_user_func($minifier, $css);
 282
 283         return $this->_reservePlace($this->_needsCdata($css)
 284             ? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>"
 285             : "{$openStyle}{$css}</style>"
 286         );
 287     }
 288
 289     protected function _removeScriptCB($m)
 290     {
 291         $openScript = "<script{$m[2]}";
 292         $js = $m[3];
 293
 294         // whitespace surrounding? preserve at least one space
 295         $ws1 = ($m[1] === '') ? '' : ' ';
 296         $ws2 = ($m[4] === '') ? '' : ' ';
 297
 298         // remove HTML comments (and ending "//" if present)
 299         if ($this->_jsCleanComments) {
 300             $js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js);
 301         }
 302
 303         // remove CDATA section markers
 304         $js = $this->_removeCdata($js);
 305
 306         // minify
 307         $minifier = $this->_jsMinifier
 308             ? $this->_jsMinifier
 309             : 'trim';
 310         $js = call_user_func($minifier, $js);
 311
 312         return $this->_reservePlace($this->_needsCdata($js)
 313             ? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}"
 314             : "{$ws1}{$openScript}{$js}</script>{$ws2}"
 315         );
 316     }
 317
 318     protected function _removeCdata($str)
 319     {
 320         return (false !== strpos($str, '<![CDATA['))
 321             ? str_replace(array('<![CDATA[', ']]>'), '', $str)
 322             : $str;
 323     }
 324
 325     protected function _needsCdata($str)
 326     {
 327         return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str));
 328     }
 329 }