| @@ -1,7 +1,7 @@ | ||
| 1 | 1 | <?php |
| 2 | 2 | /** |
| 3 | - * Class Minify_HTML | |
| 3 | + * Class Minify_HTML | |
| 4 | 4 | * @package Minify |
| 5 | 5 | */ |
| 6 | 6 | |
| 7 | 7 | /** |
| @@ -6,22 +6,19 @@ | ||
| 6 | 6 | |
| 7 | 7 | /** |
| 8 | 8 | * Compress HTML |
| 9 | 9 | * |
| 10 | - * This is a heavy regex-based removal of whitespace, unnecessary comments and | |
| 10 | + * This is a heavy regex-based removal of whitespace, unnecessary comments and | |
| 11 | 11 | * tokens. IE conditional comments are preserved. There are also options to have |
| 12 | - * STYLE and SCRIPT blocks compressed by callback functions. | |
| 13 | - * | |
| 12 | + * STYLE and SCRIPT blocks compressed by callback functions. | |
| 13 | + * | |
| 14 | 14 | * A test suite is available. |
| 15 | - * | |
| 15 | + * | |
| 16 | 16 | * @package Minify |
| 17 | 17 | * @author Stephen Clay <steve@mrclay.org> |
| 18 | 18 | */ |
| 19 | 19 | class Minify_HTML { |
| 20 | 20 | |
| 21 | - /** @var string */ | |
| 22 | - private $_html; | |
| 23 | - | |
| 24 | 21 | /** |
| 25 | 22 | * "Minify" an HTML page |
| 26 | 23 | * |
| 27 | 24 | * @param string $html |
| @@ -29,18 +26,18 @@ | ||
| 29 | 26 | * @param array $options |
| 30 | 27 | * |
| 31 | 28 | * 'cssMinifier' : (optional) callback function to process content of STYLE |
| 32 | 29 | * elements. |
| 33 | - * | |
| 30 | + * | |
| 34 | 31 | * 'jsMinifier' : (optional) callback function to process content of SCRIPT |
| 35 | 32 | * elements. Note: the type attribute is ignored. |
| 36 | - * | |
| 33 | + * | |
| 37 | 34 | * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If |
| 38 | 35 | * unset, minify will sniff for an XHTML doctype. |
| 39 | - * | |
| 36 | + * | |
| 40 | 37 | * 'keepComments' : (optional boolean) should the HTML comments be kept |
| 41 | 38 | * in the HTML Code? |
| 42 | - * | |
| 39 | + * | |
| 43 | 40 | * @return string |
| 44 | 41 | */ |
| 45 | 42 | public static function minify($html, $options = array()) { |
| 46 | 43 | $min = new Minify_HTML($html, $options); |
| @@ -45,10 +42,10 @@ | ||
| 45 | 42 | public static function minify($html, $options = array()) { |
| 46 | 43 | $min = new Minify_HTML($html, $options); |
| 47 | 44 | return $min->process(); |
| 48 | 45 | } |
| 49 | - | |
| 50 | - | |
| 46 | + | |
| 47 | + | |
| 51 | 48 | /** |
| 52 | 49 | * Create a minifier object |
| 53 | 50 | * |
| 54 | 51 | * @param string $html |
| @@ -56,18 +53,18 @@ | ||
| 56 | 53 | * @param array $options |
| 57 | 54 | * |
| 58 | 55 | * 'cssMinifier' : (optional) callback function to process content of STYLE |
| 59 | 56 | * elements. |
| 60 | - * | |
| 57 | + * | |
| 61 | 58 | * 'jsMinifier' : (optional) callback function to process content of SCRIPT |
| 62 | 59 | * elements. Note: the type attribute is ignored. |
| 63 | - * | |
| 60 | + * | |
| 64 | 61 | * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If |
| 65 | 62 | * unset, minify will sniff for an XHTML doctype. |
| 66 | - * | |
| 63 | + * | |
| 67 | 64 | * 'xhtml' : (optional boolean) should content be treated as XHTML1.0? If |
| 68 | 65 | * unset, minify will sniff for an XHTML doctype. |
| 69 | - * | |
| 66 | + * | |
| 70 | 67 | * @return null |
| 71 | 68 | */ |
| 72 | 69 | public function __construct($html, $options = array()) |
| 73 | 70 | { |
| @@ -84,13 +81,13 @@ | ||
| 84 | 81 | if (isset($options['keepComments'])) { |
| 85 | 82 | $this->_keepComments = $options['keepComments']; |
| 86 | 83 | } |
| 87 | 84 | } |
| 88 | - | |
| 89 | - | |
| 85 | + | |
| 86 | + | |
| 90 | 87 | /** |
| 91 | 88 | * Minify the markeup given in the constructor |
| 92 | - * | |
| 89 | + * | |
| 93 | 90 | * @return string |
| 94 | 91 | */ |
| 95 | 92 | public function process() |
| 96 | 93 | { |
| @@ -96,72 +93,69 @@ | ||
| 96 | 93 | { |
| 97 | 94 | if ($this->_isXhtml === null) { |
| 98 | 95 | $this->_isXhtml = (false !== strpos($this->_html, '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML')); |
| 99 | 96 | } |
| 100 | - | |
| 97 | + | |
| 101 | 98 | $this->_replacementHash = 'MINIFYHTML' . md5($_SERVER['REQUEST_TIME']); |
| 102 | 99 | $this->_placeholders = array(); |
| 103 | - | |
| 100 | + | |
| 104 | 101 | // replace SCRIPTs (and minify) with placeholders |
| 105 | 102 | $this->_html = preg_replace_callback( |
| 106 | 103 | '/(\\s*)(<script\\b[^>]*?>)([\\s\\S]*?)<\\/script>(\\s*)/i' |
| 107 | 104 | ,array($this, '_removeScriptCB') |
| 108 | 105 | ,$this->_html); |
| 109 | - | |
| 106 | + | |
| 110 | 107 | // replace STYLEs (and minify) with placeholders |
| 111 | 108 | $this->_html = preg_replace_callback( |
| 112 | 109 | '/\\s*(<style\\b[^>]*?>)([\\s\\S]*?)<\\/style>\\s*/i' |
| 113 | 110 | ,array($this, '_removeStyleCB') |
| 114 | 111 | ,$this->_html); |
| 115 | - | |
| 112 | + | |
| 116 | 113 | // remove HTML comments (not containing IE conditional comments). |
| 117 | - if ($this->_keepComments == false) { | |
| 118 | - $this->_html = preg_replace_callback( | |
| 119 | - '/<!--([\\s\\S]*?)-->/' | |
| 120 | - ,array($this, '_commentCB') | |
| 121 | - ,$this->_html); | |
| 114 | + if ($this->_keepComments == false) { | |
| 115 | + $this->_html = preg_replace_callback( | |
| 116 | + '/<!--([\\s\\S]*?)-->/' | |
| 117 | + ,array($this, '_commentCB') | |
| 118 | + ,$this->_html); | |
| 122 | 119 | } |
| 123 | - | |
| 120 | + | |
| 124 | 121 | // replace PREs with placeholders |
| 125 | 122 | $this->_html = preg_replace_callback('/\\s*(<pre\\b[^>]*?>[\\s\\S]*?<\\/pre>)\\s*/i' |
| 126 | 123 | ,array($this, '_removePreCB') |
| 127 | 124 | ,$this->_html); |
| 128 | - | |
| 125 | + | |
| 129 | 126 | // replace TEXTAREAs with placeholders |
| 130 | 127 | $this->_html = preg_replace_callback( |
| 131 | 128 | '/\\s*(<textarea\\b[^>]*?>[\\s\\S]*?<\\/textarea>)\\s*/i' |
| 132 | 129 | ,array($this, '_removeTextareaCB') |
| 133 | 130 | ,$this->_html); |
| 134 | - | |
| 135 | - // replace data: URIs with placeholders | |
| 131 | + | |
| 132 | + // replace data: URIs with placeholders | |
| 136 | 133 | $this->_html = preg_replace_callback( |
| 137 | 134 | '/(=("|\')data:.*\\2)/Ui' |
| 138 | 135 | ,array($this, '_removeDataURICB') |
| 139 | 136 | ,$this->_html); |
| 140 | - | |
| 137 | + | |
| 141 | 138 | // trim each line. |
| 142 | - // replace by space instead of '' to avoid newline after opening tag getting zapped | |
| 143 | - $this->_html = preg_replace('/^\s+|\s+$/m', ' ', $this->_html); | |
| 144 | - | |
| 139 | + // @todo take into account attribute values that span multiple lines. | |
| 140 | + $this->_html = preg_replace('/^\\s+|\\s+$/m', '', $this->_html); | |
| 141 | + | |
| 145 | 142 | // remove ws around block/undisplayed elements |
| 146 | - $this->_html = preg_replace('/\\s+(<\\/?(?:area|article|aside|base(?:font)?|blockquote|body' | |
| 147 | - .'|canvas|caption|center|col(?:group)?|dd|dir|div|dl|dt|fieldset|figcaption|figure|footer|form' | |
| 148 | - .'|frame(?:set)?|h[1-6]|head|header|hgroup|hr|html|legend|li|link|main|map|menu|meta|nav' | |
| 149 | - .'|ol|opt(?:group|ion)|output|p|param|section|t(?:able|body|head|d|h||r|foot|itle)' | |
| 150 | - .'|ul|video)\\b[^>]*>)/i', '$1', $this->_html); | |
| 151 | - | |
| 143 | + $this->_html = preg_replace('/\\s+(<\\/?(?:area|base(?:font)?|blockquote|body' | |
| 144 | + .'|caption|center|cite|col(?:group)?|dd|dir|div|dl|dt|fieldset|form' | |
| 145 | + .'|frame(?:set)?|h[1-6]|head|hr|html|legend|li|link|map|menu|meta' | |
| 146 | + .'|ol|opt(?:group|ion)|p|param|t(?:able|body|head|d|h||r|foot|itle)' | |
| 147 | + .'|ul)\\b[^>]*>)/i', '$1', $this->_html); | |
| 148 | + | |
| 152 | 149 | // remove ws outside of all elements |
| 153 | 150 | $this->_html = preg_replace_callback( |
| 154 | 151 | '/>([^<]+)</' |
| 155 | 152 | ,array($this, '_outsideTagCB') |
| 156 | 153 | ,$this->_html); |
| 157 | - | |
| 154 | + | |
| 158 | 155 | // use newlines before 1st attribute in open tags (to limit line lengths) |
| 159 | 156 | //$this->_html = preg_replace('/(<[a-z\\-]+)\\s+([^>]+>)/i', "$1\n$2", $this->_html); |
| 160 | - | |
| 161 | - // reverse order while preserving keys to ensure the last replacement is done first, etc ... | |
| 162 | - $this->_placeholders = array_reverse( $this->_placeholders, true ); | |
| 163 | - | |
| 157 | + | |
| 164 | 158 | // fill placeholders |
| 165 | 159 | $this->_html = str_replace( |
| 166 | 160 | array_keys($this->_placeholders) |
| 167 | 161 | ,array_values($this->_placeholders) |
| @@ -168,9 +162,9 @@ | ||
| 168 | 162 | ,$this->_html |
| 169 | 163 | ); |
| 170 | 164 | return $this->_html; |
| 171 | 165 | } |
| 172 | - | |
| 166 | + | |
| 173 | 167 | protected function _commentCB($m) |
| 174 | 168 | { |
| 175 | 169 | return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<![')) |
| 176 | 170 | ? $m[0] |
| @@ -175,9 +169,9 @@ | ||
| 175 | 169 | return (0 === strpos($m[1], '[') || false !== strpos($m[1], '<![')) |
| 176 | 170 | ? $m[0] |
| 177 | 171 | : ''; |
| 178 | 172 | } |
| 179 | - | |
| 173 | + | |
| 180 | 174 | protected function _reservePlace($content) |
| 181 | 175 | { |
| 182 | 176 | $placeholder = '%' . $this->_replacementHash . count($this->_placeholders) . '%'; |
| 183 | 177 | $this->_placeholders[$placeholder] = $content; |
| @@ -188,26 +182,26 @@ | ||
| 188 | 182 | protected $_replacementHash = null; |
| 189 | 183 | protected $_placeholders = array(); |
| 190 | 184 | protected $_cssMinifier = null; |
| 191 | 185 | protected $_jsMinifier = null; |
| 192 | - protected $_keepComments = false; | |
| 186 | + protected $_keepComments = false; | |
| 193 | 187 | |
| 194 | 188 | protected function _outsideTagCB($m) |
| 195 | 189 | { |
| 196 | 190 | return '>' . preg_replace('/^\\s+|\\s+$/', ' ', $m[1]) . '<'; |
| 197 | 191 | } |
| 198 | - | |
| 192 | + | |
| 199 | 193 | protected function _removePreCB($m) |
| 200 | 194 | { |
| 201 | 195 | return $this->_reservePlace($m[1]); |
| 202 | 196 | } |
| 203 | - | |
| 197 | + | |
| 204 | 198 | protected function _removeTextareaCB($m) |
| 205 | 199 | { |
| 206 | 200 | return $this->_reservePlace($m[1]); |
| 207 | 201 | } |
| 208 | - | |
| 209 | - protected function _removeDataURICB($m) | |
| 202 | + | |
| 203 | + protected function _removeDataURICB($m) | |
| 210 | 204 | { |
| 211 | 205 | return $this->_reservePlace($m[1]); |
| 212 | 206 | } |
| 213 | 207 | |
| @@ -216,18 +210,18 @@ | ||
| 216 | 210 | $openStyle = $m[1]; |
| 217 | 211 | $css = $m[2]; |
| 218 | 212 | // remove HTML comments |
| 219 | 213 | $css = preg_replace('/(?:^\\s*<!--|-->\\s*$)/', '', $css); |
| 220 | - | |
| 214 | + | |
| 221 | 215 | // remove CDATA section markers |
| 222 | 216 | $css = $this->_removeCdata($css); |
| 223 | - | |
| 217 | + | |
| 224 | 218 | // minify |
| 225 | 219 | $minifier = $this->_cssMinifier |
| 226 | 220 | ? $this->_cssMinifier |
| 227 | 221 | : 'trim'; |
| 228 | 222 | $css = call_user_func($minifier, $css); |
| 229 | - | |
| 223 | + | |
| 230 | 224 | return $this->_reservePlace($this->_needsCdata($css) |
| 231 | 225 | ? "{$openStyle}/*<![CDATA[*/{$css}/*]]>*/</style>" |
| 232 | 226 | : "{$openStyle}{$css}</style>" |
| 233 | 227 | ); |
| @@ -236,27 +230,25 @@ | ||
| 236 | 230 | protected function _removeScriptCB($m) |
| 237 | 231 | { |
| 238 | 232 | $openScript = $m[2]; |
| 239 | 233 | $js = $m[3]; |
| 240 | - | |
| 234 | + | |
| 241 | 235 | // whitespace surrounding? preserve at least one space |
| 242 | 236 | $ws1 = ($m[1] === '') ? '' : ' '; |
| 243 | 237 | $ws2 = ($m[4] === '') ? '' : ' '; |
| 244 | - | |
| 245 | - if ($this->_keepComments == false) { | |
| 246 | - // remove HTML comments (and ending "//" if present) | |
| 247 | - $js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js); | |
| 248 | - | |
| 249 | - // remove CDATA section markers | |
| 250 | - $js = $this->_removeCdata($js); | |
| 251 | - } | |
| 252 | - | |
| 238 | + | |
| 239 | + // remove HTML comments (and ending "//" if present) | |
| 240 | + $js = preg_replace('/(?:^\\s*<!--\\s*|\\s*(?:\\/\\/)?\\s*-->\\s*$)/', '', $js); | |
| 241 | + | |
| 242 | + // remove CDATA section markers | |
| 243 | + $js = $this->_removeCdata($js); | |
| 244 | + | |
| 253 | 245 | // minify |
| 254 | 246 | $minifier = $this->_jsMinifier |
| 255 | 247 | ? $this->_jsMinifier |
| 256 | - : 'trim'; | |
| 248 | + : 'trim'; | |
| 257 | 249 | $js = call_user_func($minifier, $js); |
| 258 | - | |
| 250 | + | |
| 259 | 251 | return $this->_reservePlace($this->_needsCdata($js) |
| 260 | 252 | ? "{$ws1}{$openScript}/*<![CDATA[*/{$js}/*]]>*/</script>{$ws2}" |
| 261 | 253 | : "{$ws1}{$openScript}{$js}</script>{$ws2}" |
| 262 | 254 | ); |
| @@ -264,12 +256,12 @@ | ||
| 264 | 256 | |
| 265 | 257 | protected function _removeCdata($str) |
| 266 | 258 | { |
| 267 | 259 | return (false !== strpos($str, '<![CDATA[')) |
| 268 | - ? str_replace(array('/* <![CDATA[ */','/* ]]> */','/*<![CDATA[*/','/*]]>*/','<![CDATA[', ']]>'), '', $str) | |
| 260 | + ? str_replace(array('<![CDATA[', ']]>'), '', $str) | |
| 269 | 261 | : $str; |
| 270 | 262 | } |
| 271 | - | |
| 263 | + | |
| 272 | 264 | protected function _needsCdata($str) |
| 273 | 265 | { |
| 274 | 266 | return ($this->_isXhtml && preg_match('/(?:[<&]|\\-\\-|\\]\\]>)/', $str)); |
| 275 | 267 | } |
