| Index: trunk/extensions/MobileFrontend/MobileFormatter.php |
| — | — | @@ -3,16 +3,10 @@ |
| 4 | 4 | /** |
| 5 | 5 | * Converts HTML into a mobile-friendly version |
| 6 | 6 | */ |
| 7 | | -class MobileFormatter { |
| | 7 | +class MobileFormatter extends HtmlFormatter { |
| 8 | 8 | const WML_SECTION_SEPARATOR = '***************************************************************************'; |
| 9 | 9 | |
| 10 | | - /** |
| 11 | | - * @var DOMDocument |
| 12 | | - */ |
| 13 | | - protected $doc; |
| 14 | 10 | protected $format; |
| 15 | | - protected $removeImages = false; |
| 16 | | - protected $idWhitelist = array(); |
| 17 | 11 | |
| 18 | 12 | /** |
| 19 | 13 | * @var Ttile |
| — | — | @@ -69,9 +63,6 @@ |
| 70 | 64 | '.nomobile', |
| 71 | 65 | ); |
| 72 | 66 | |
| 73 | | - private $itemsToRemove = array(); |
| 74 | | - private $elementsToFlatten = array(); |
| 75 | | - |
| 76 | 67 | /** |
| 77 | 68 | * Constructor |
| 78 | 69 | * |
| — | — | @@ -81,7 +72,7 @@ |
| 82 | 73 | * @param WmlContext $wmlContext: Context for creation of WML cards, can be omitted if $format == 'XHTML' |
| 83 | 74 | */ |
| 84 | 75 | public function __construct( $html, $title, $format, WmlContext $wmlContext = null ) { |
| 85 | | - wfProfileIn( __METHOD__ ); |
| | 76 | + parent::__construct( $html ); |
| 86 | 77 | |
| 87 | 78 | $this->title = $title; |
| 88 | 79 | $this->format = $format; |
| — | — | @@ -89,27 +80,10 @@ |
| 90 | 81 | throw new MWException( __METHOD__ . '(): WML context not set' ); |
| 91 | 82 | } |
| 92 | 83 | $this->wmlContext = $wmlContext; |
| 93 | | - |
| 94 | | - $html = mb_convert_encoding( $html, 'HTML-ENTITIES', "UTF-8" ); |
| 95 | | - libxml_use_internal_errors( true ); |
| 96 | | - $this->doc = new DOMDocument(); |
| 97 | | - $this->doc->loadHTML( '<?xml encoding="UTF-8">' . $html ); |
| 98 | | - libxml_use_internal_errors( false ); |
| 99 | | - $this->doc->preserveWhiteSpace = false; |
| 100 | | - $this->doc->strictErrorChecking = false; |
| 101 | | - $this->doc->encoding = 'UTF-8'; |
| | 84 | + $this->flattenRedLinks(); |
| 102 | 85 | } |
| 103 | 86 | |
| 104 | 87 | /** |
| 105 | | - * Turns a chunk of HTML into a proper document |
| 106 | | - * @param string $html |
| 107 | | - * @return string |
| 108 | | - */ |
| 109 | | - public static function wrapHTML( $html ) { |
| 110 | | - return '<!doctype html><html><head></head><body>' . $html . '</body></html>'; |
| 111 | | - } |
| 112 | | - |
| 113 | | - /** |
| 114 | 88 | * Use the given message cache |
| 115 | 89 | * @param Array $messages |
| 116 | 90 | */ |
| — | — | @@ -118,13 +92,6 @@ |
| 119 | 93 | } |
| 120 | 94 | |
| 121 | 95 | /** |
| 122 | | - * @return DOMDocument: DOM to manipulate |
| 123 | | - */ |
| 124 | | - public function getDoc() { |
| 125 | | - return $this->doc; |
| 126 | | - } |
| 127 | | - |
| 128 | | - /** |
| 129 | 96 | * @return string: Output format |
| 130 | 97 | */ |
| 131 | 98 | public function getFormat() { |
| — | — | @@ -144,148 +111,37 @@ |
| 145 | 112 | } |
| 146 | 113 | |
| 147 | 114 | /** |
| 148 | | - * Sets whether images should be removed from output |
| 149 | | - * @param bool $flag |
| 150 | | - */ |
| 151 | | - public function removeImages( $flag = true ) { |
| 152 | | - $this->removeImages = $flag; |
| 153 | | - } |
| 154 | | - |
| 155 | | - /** |
| 156 | | - * Adds one or more selector of content to remove |
| 157 | | - * @param Array|string $selectors: Selector(s) of stuff to remove |
| 158 | | - */ |
| 159 | | - public function remove( $selectors ) { |
| 160 | | - $this->itemsToRemove = array_merge( $this->itemsToRemove, (array)$selectors ); |
| 161 | | - } |
| 162 | | - |
| 163 | | - /** |
| 164 | | - * Adds one or more element name to the list to flatten (remove tag, but not its content) |
| 165 | | - * @param Array|string $elements: Name(s) of tag(s) to flatten |
| 166 | | - */ |
| 167 | | - public function flatten( $elements ) { |
| 168 | | - $this->elementsToFlatten = array_merge( $this->elementsToFlatten, (array)$elements ); |
| 169 | | - } |
| 170 | | - |
| 171 | | - /** |
| 172 | | - * @param Array|string $ids: Id(s) of content to keep |
| 173 | | - */ |
| 174 | | - public function whitelistIds( $ids ) { |
| 175 | | - $this->idWhitelist = array_merge( $this->idWhitelist, array_flip( (array)$ids ) ); |
| 176 | | - } |
| 177 | | - |
| 178 | | - /** |
| 179 | 115 | * Removes content inappropriate for mobile devices |
| 180 | 116 | * @param bool $removeDefaults: Whether default settings at self::$defaultItemsToRemove should be used |
| 181 | 117 | */ |
| 182 | 118 | public function filterContent( $removeDefaults = true ) { |
| 183 | 119 | global $wgMFRemovableClasses; |
| 184 | 120 | |
| 185 | | - wfProfileIn(__METHOD__ ); |
| 186 | 121 | if ( $removeDefaults ) { |
| 187 | | - $this->itemsToRemove = array_merge( $this->itemsToRemove, |
| 188 | | - self::$defaultItemsToRemove, $wgMFRemovableClasses |
| 189 | | - ); |
| | 122 | + $this->remove( self::$defaultItemsToRemove ); |
| | 123 | + $this->remove( $wgMFRemovableClasses ); |
| 190 | 124 | } |
| 191 | | - $removals = $this->parseItemsToRemove(); |
| 192 | | - |
| 193 | | - // Remove tags |
| 194 | | - |
| 195 | | - // You can't remove DOMNodes from a DOMNodeList as you're iterating |
| 196 | | - // over them in a foreach loop. It will seemingly leave the internal |
| 197 | | - // iterator on the foreach out of wack and results will be quite |
| 198 | | - // strange. Though, making a queue of items to remove seems to work. |
| 199 | | - // For example: |
| 200 | | - |
| 201 | | - $domElemsToRemove = array(); |
| 202 | | - foreach ( $removals['TAG'] as $tagToRemove ) { |
| 203 | | - $tagToRemoveNodes = $this->doc->getElementsByTagName( $tagToRemove ); |
| 204 | | - foreach ( $tagToRemoveNodes as $tagToRemoveNode ) { |
| 205 | | - $tagToRemoveNodeIdAttributeValue = ''; |
| 206 | | - if ( $tagToRemoveNode ) { |
| 207 | | - $tagToRemoveNodeIdAttribute = $tagToRemoveNode->getAttributeNode( 'id' ); |
| 208 | | - if ( $tagToRemoveNodeIdAttribute ) { |
| 209 | | - $tagToRemoveNodeIdAttributeValue = $tagToRemoveNodeIdAttribute->value; |
| 210 | | - } |
| 211 | | - if ( !isset( $this->idWhitelist[$tagToRemoveNodeIdAttributeValue] ) ) { |
| 212 | | - $domElemsToRemove[] = $tagToRemoveNode; |
| 213 | | - } |
| 214 | | - } |
| 215 | | - } |
| 216 | | - } |
| 217 | | - |
| 218 | | - foreach ( $domElemsToRemove as $domElement ) { |
| 219 | | - $domElement->parentNode->removeChild( $domElement ); |
| 220 | | - } |
| 221 | | - |
| 222 | | - // Elements with named IDs |
| 223 | | - foreach ( $removals['ID'] as $itemToRemove ) { |
| 224 | | - $itemToRemoveNode = $this->doc->getElementById( $itemToRemove ); |
| 225 | | - if ( $itemToRemoveNode ) { |
| 226 | | - $itemToRemoveNode->parentNode->removeChild( $itemToRemoveNode ); |
| 227 | | - } |
| 228 | | - } |
| 229 | | - |
| 230 | | - // CSS Classes |
| 231 | | - $xpath = new DOMXpath( $this->doc ); |
| 232 | | - foreach ( $removals['CLASS'] as $classToRemove ) { |
| 233 | | - $elements = $xpath->query( '//*[@class="' . $classToRemove . '"]' ); |
| 234 | | - |
| 235 | | - foreach ( $elements as $element ) { |
| 236 | | - $element->parentNode->removeChild( $element ); |
| 237 | | - } |
| 238 | | - } |
| 239 | | - |
| 240 | | - // Tags with CSS Classes |
| 241 | | - foreach ( $removals['TAG_CLASS'] as $classToRemove ) { |
| 242 | | - $parts = explode( '.', $classToRemove ); |
| 243 | | - |
| 244 | | - $elements = $xpath->query( |
| 245 | | - '//' . $parts[0] . '[@class="' . $parts[1] . '"]' |
| 246 | | - ); |
| 247 | | - |
| 248 | | - foreach ( $elements as $element ) { |
| 249 | | - $removedElement = $element->parentNode->removeChild( $element ); |
| 250 | | - } |
| 251 | | - } |
| 252 | | - |
| 253 | | - // Handle red links with action equal to edit |
| 254 | | - $redLinks = $xpath->query( '//a[@class="new"]' ); |
| 255 | | - foreach ( $redLinks as $redLink ) { |
| 256 | | - // PHP Bug #36795 — Inappropriate "unterminated entity reference" |
| 257 | | - $spanNode = $this->doc->createElement( "span", str_replace( "&", "&", $redLink->nodeValue ) ); |
| 258 | | - |
| 259 | | - if ( $redLink->hasAttributes() ) { |
| 260 | | - $attributes = $redLink->attributes; |
| 261 | | - foreach ( $attributes as $i => $attribute ) { |
| 262 | | - if ( $attribute->name != 'href' ) { |
| 263 | | - $spanNode->setAttribute( $attribute->name, $attribute->value ); |
| 264 | | - } |
| 265 | | - } |
| 266 | | - } |
| 267 | | - |
| 268 | | - $redLink->parentNode->replaceChild( $spanNode, $redLink ); |
| 269 | | - } |
| 270 | | - wfProfileOut( __METHOD__ ); |
| | 125 | + parent::filterContent(); |
| 271 | 126 | } |
| 272 | 127 | |
| 273 | 128 | /** |
| 274 | 129 | * Performs final transformations to mobile format and returns resulting HTML/WML |
| 275 | 130 | * |
| 276 | | - * @param string|bool $id: ID of element to get HTML from or false to get it from the whole tree |
| 277 | | - * @param string $prependHtml: HTML to be prepended to result before final transformations |
| 278 | | - * @param string $appendHtml: HTML to be appended to result before final transformations |
| | 131 | + * @param DOMElement|string|null $element: ID of element to get HTML from or false to get it from the whole tree |
| | 132 | + |
| 279 | 133 | * @return string: Processed HTML |
| 280 | 134 | */ |
| 281 | | - public function getText( $id = false, $prependHtml = '', $appendHtml = '' ) { |
| | 135 | + public function getText( $element = false ) { |
| 282 | 136 | wfProfileIn( __METHOD__ ); |
| 283 | 137 | if ( $this->mainPage ) { |
| 284 | 138 | $element = $this->parseMainPage( $this->doc ); |
| 285 | | - } else { |
| 286 | | - $element = $id ? $this->doc->getElementById( $id ) : null; |
| 287 | 139 | } |
| 288 | | - $html = $prependHtml . $this->doc->saveXML( $element, LIBXML_NOEMPTYTAG ) . $appendHtml; |
| 289 | | - |
| | 140 | + $html = parent::getText( $element ); |
| | 141 | + wfProfileOut( __METHOD__ ); |
| | 142 | + return $html; |
| | 143 | + } |
| | 144 | + |
| | 145 | + protected function onHtmlReady( $html ) { |
| 290 | 146 | switch ( $this->format ) { |
| 291 | 147 | case 'XHTML': |
| 292 | 148 | if ( $this->expandableSections && !$this->mainPage && strlen( $html ) > 4000 ) { |
| — | — | @@ -300,15 +156,6 @@ |
| 301 | 157 | $html = $this->createWMLCard( $html ); |
| 302 | 158 | break; |
| 303 | 159 | } |
| 304 | | - if ( $this->elementsToFlatten ) { |
| 305 | | - $elements = implode( '|', $this->elementsToFlatten ); |
| 306 | | - $html = preg_replace( "#</?($elements)[^>]*>#is", '', $html ); |
| 307 | | - } |
| 308 | | - if ( !$element ) { |
| 309 | | - $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html ); |
| 310 | | - } |
| 311 | | - |
| 312 | | - wfProfileOut( __METHOD__ ); |
| 313 | 160 | return $html; |
| 314 | 161 | } |
| 315 | 162 | |
| — | — | @@ -496,40 +343,6 @@ |
| 497 | 344 | } |
| 498 | 345 | |
| 499 | 346 | /** |
| 500 | | - * Transforms CSS selectors into an internal representation suitable for processing |
| 501 | | - * @return array |
| 502 | | - */ |
| 503 | | - private function parseItemsToRemove() { |
| 504 | | - wfProfileIn( __METHOD__ ); |
| 505 | | - $removals = array( |
| 506 | | - 'ID' => array(), |
| 507 | | - 'TAG' => array(), |
| 508 | | - 'CLASS' => array(), |
| 509 | | - 'TAG_CLASS' => array(), |
| 510 | | - ); |
| 511 | | - |
| 512 | | - foreach ( $this->itemsToRemove as $itemToRemove ) { |
| 513 | | - $type = ''; |
| 514 | | - $rawName = ''; |
| 515 | | - CssDetection::detectIdCssOrTag( $itemToRemove, $type, $rawName ); |
| 516 | | - $removals[$type][] = $rawName; |
| 517 | | - } |
| 518 | | - |
| 519 | | - if ( $this->removeImages ) { |
| 520 | | - $removals['TAG'][] = "img"; |
| 521 | | - $removals['TAG'][] = "audio"; |
| 522 | | - $removals['TAG'][] = "video"; |
| 523 | | - $removals['CLASS'][] = "thumb tright"; |
| 524 | | - $removals['CLASS'][] = "thumb tleft"; |
| 525 | | - $removals['CLASS'][] = "thumbcaption"; |
| 526 | | - $removals['CLASS'][] = "gallery"; |
| 527 | | - } |
| 528 | | - |
| 529 | | - wfProfileOut( __METHOD__ ); |
| 530 | | - return $removals; |
| 531 | | - } |
| 532 | | - |
| 533 | | - /** |
| 534 | 347 | * Performs transformations specific to main page |
| 535 | 348 | * @param DOMDocument $mainPage: Tree to process |
| 536 | 349 | * @return DOMElement |
| Index: trunk/extensions/MobileFrontend/MobileFrontend.body.php |
| — | — | @@ -1068,24 +1068,25 @@ |
| 1069 | 1069 | |
| 1070 | 1070 | $formatter->setIsMainPage( self::$isMainPage ); |
| 1071 | 1071 | $prepend = ''; |
| 1072 | | - if ( $this->contentFormat == 'WML' ) { |
| 1073 | | - // Wml for searching |
| 1074 | | - $prepend = '<p><input emptyok="true" format="*M" type="text" name="search" value="" size="16" />' . |
| 1075 | | - '<do type="accept" label="' . self::$messages['mobile-frontend-search-submit'] . '">' . |
| 1076 | | - '<go href="' . $wgScript . '?title=Special%3ASearch&search=$(search)"></go></do></p>'; |
| 1077 | | - } elseif ( $this->contentFormat == 'XHTML' |
| | 1072 | + if ( $this->contentFormat == 'XHTML' |
| 1078 | 1073 | && self::$device['supports_javascript'] === true |
| 1079 | 1074 | && empty( self::$search ) ) |
| 1080 | 1075 | { |
| 1081 | 1076 | $formatter->enableExpandableSections(); |
| 1082 | 1077 | } |
| 1083 | | - $contentHtml = $formatter->getText( 'content', $prepend ); |
| | 1078 | + $contentHtml = $formatter->getText( 'content' ); |
| 1084 | 1079 | |
| 1085 | 1080 | $htmlTitle = htmlspecialchars( self::$htmlTitle ); |
| 1086 | 1081 | |
| 1087 | 1082 | if ( $this->contentFormat == 'WML' ) { |
| 1088 | 1083 | header( 'Content-Type: text/vnd.wap.wml' ); |
| 1089 | 1084 | |
| | 1085 | + // Wml for searching |
| | 1086 | + $prepend = '<p><input emptyok="true" format="*M" type="text" name="search" value="" size="16" />' . |
| | 1087 | + '<do type="accept" label="' . self::$messages['mobile-frontend-search-submit'] . '">' . |
| | 1088 | + '<go href="' . $wgScript . '?title=Special%3ASearch&search=$(search)"></go></do></p>'; |
| | 1089 | + $html = $prepend . $html; |
| | 1090 | + |
| 1090 | 1091 | $applicationWmlTemplate = new ApplicationWmlTemplate(); |
| 1091 | 1092 | $options = array( |
| 1092 | 1093 | 'mainPageUrl' => self::$mainPageUrl, |
| Index: trunk/extensions/MobileFrontend/MobileFrontend.php |
| — | — | @@ -46,6 +46,7 @@ |
| 47 | 47 | |
| 48 | 48 | 'CssDetection' => 'CssDetection', |
| 49 | 49 | 'DeviceDetection' => 'DeviceDetection', |
| | 50 | + 'HtmlFormatter' => 'HtmlFormatter', |
| 50 | 51 | 'MobileFormatter' => 'MobileFormatter', |
| 51 | 52 | 'WmlContext' => 'WmlContext', |
| 52 | 53 | |
| Index: trunk/extensions/MobileFrontend/HtmlFormatter.php |
| — | — | @@ -0,0 +1,250 @@ |
| | 2 | +<?php |
| | 3 | + |
| | 4 | +/** |
| | 5 | + * Performs transformations of HTML |
| | 6 | + */ |
| | 7 | +class HtmlFormatter { |
| | 8 | + /** |
| | 9 | + * @var DOMDocument |
| | 10 | + */ |
| | 11 | + protected $doc; |
| | 12 | + |
| | 13 | + private $itemsToRemove = array(); |
| | 14 | + private $elementsToFlatten = array(); |
| | 15 | + private $removeImages = false; |
| | 16 | + private $idWhitelist = array(); |
| | 17 | + private $flattenRedLinks = false; |
| | 18 | + |
| | 19 | + /** |
| | 20 | + * Constructor |
| | 21 | + * |
| | 22 | + * @param string $html: Text to process |
| | 23 | + */ |
| | 24 | + public function __construct( $html ) { |
| | 25 | + wfProfileIn( __METHOD__ ); |
| | 26 | + |
| | 27 | + $html = mb_convert_encoding( $html, 'HTML-ENTITIES', "UTF-8" ); |
| | 28 | + libxml_use_internal_errors( true ); |
| | 29 | + $this->doc = new DOMDocument(); |
| | 30 | + $this->doc->loadHTML( '<?xml encoding="UTF-8">' . $html ); |
| | 31 | + libxml_use_internal_errors( false ); |
| | 32 | + $this->doc->preserveWhiteSpace = false; |
| | 33 | + $this->doc->strictErrorChecking = false; |
| | 34 | + $this->doc->encoding = 'UTF-8'; |
| | 35 | + |
| | 36 | + wfProfileOut( __METHOD__ ); |
| | 37 | + } |
| | 38 | + |
| | 39 | + /** |
| | 40 | + * Turns a chunk of HTML into a proper document |
| | 41 | + * @param string $html |
| | 42 | + * @return string |
| | 43 | + */ |
| | 44 | + public static function wrapHTML( $html ) { |
| | 45 | + return '<!doctype html><html><head></head><body>' . $html . '</body></html>'; |
| | 46 | + } |
| | 47 | + |
| | 48 | + /** |
| | 49 | + * Override this in descendant class to modify HTML after it has been converted from DOM tree |
| | 50 | + * @param string $html: HTML to process |
| | 51 | + * @return string: Processed HTML |
| | 52 | + */ |
| | 53 | + protected function onHtmlReady( $html ) { |
| | 54 | + return $html; |
| | 55 | + } |
| | 56 | + |
| | 57 | + /** |
| | 58 | + * @return DOMDocument: DOM to manipulate |
| | 59 | + */ |
| | 60 | + public function getDoc() { |
| | 61 | + return $this->doc; |
| | 62 | + } |
| | 63 | + |
| | 64 | + /** |
| | 65 | + * Sets whether images should be removed from output |
| | 66 | + * @param bool $flag |
| | 67 | + */ |
| | 68 | + public function removeImages( $flag = true ) { |
| | 69 | + $this->removeImages = $flag; |
| | 70 | + } |
| | 71 | + |
| | 72 | + /** |
| | 73 | + * Adds one or more selector of content to remove |
| | 74 | + * @param Array|string $selectors: Selector(s) of stuff to remove |
| | 75 | + */ |
| | 76 | + public function remove( $selectors ) { |
| | 77 | + $this->itemsToRemove = array_merge( $this->itemsToRemove, (array)$selectors ); |
| | 78 | + } |
| | 79 | + |
| | 80 | + /** |
| | 81 | + * Adds one or more element name to the list to flatten (remove tag, but not its content) |
| | 82 | + * @param Array|string $elements: Name(s) of tag(s) to flatten |
| | 83 | + */ |
| | 84 | + public function flatten( $elements ) { |
| | 85 | + $this->elementsToFlatten = array_merge( $this->elementsToFlatten, (array)$elements ); |
| | 86 | + } |
| | 87 | + |
| | 88 | + /** |
| | 89 | + * Sets whether red links should be flattened |
| | 90 | + * @param bool $flag |
| | 91 | + */ |
| | 92 | + public function flattenRedLinks( $flag = true ) { |
| | 93 | + $this->flattenRedLinks = $flag; |
| | 94 | + } |
| | 95 | + |
| | 96 | + /** |
| | 97 | + * @param Array|string $ids: Id(s) of content to keep |
| | 98 | + */ |
| | 99 | + public function whitelistIds( $ids ) { |
| | 100 | + $this->idWhitelist = array_merge( $this->idWhitelist, array_flip( (array)$ids ) ); |
| | 101 | + } |
| | 102 | + |
| | 103 | + /** |
| | 104 | + * Removes content inappropriate for mobile devices |
| | 105 | + */ |
| | 106 | + public function filterContent() { |
| | 107 | + $removals = $this->parseItemsToRemove(); |
| | 108 | + |
| | 109 | + // Remove tags |
| | 110 | + |
| | 111 | + // You can't remove DOMNodes from a DOMNodeList as you're iterating |
| | 112 | + // over them in a foreach loop. It will seemingly leave the internal |
| | 113 | + // iterator on the foreach out of wack and results will be quite |
| | 114 | + // strange. Though, making a queue of items to remove seems to work. |
| | 115 | + // For example: |
| | 116 | + |
| | 117 | + $domElemsToRemove = array(); |
| | 118 | + foreach ( $removals['TAG'] as $tagToRemove ) { |
| | 119 | + $tagToRemoveNodes = $this->doc->getElementsByTagName( $tagToRemove ); |
| | 120 | + foreach ( $tagToRemoveNodes as $tagToRemoveNode ) { |
| | 121 | + $tagToRemoveNodeIdAttributeValue = ''; |
| | 122 | + if ( $tagToRemoveNode ) { |
| | 123 | + $tagToRemoveNodeIdAttribute = $tagToRemoveNode->getAttributeNode( 'id' ); |
| | 124 | + if ( $tagToRemoveNodeIdAttribute ) { |
| | 125 | + $tagToRemoveNodeIdAttributeValue = $tagToRemoveNodeIdAttribute->value; |
| | 126 | + } |
| | 127 | + if ( !isset( $this->idWhitelist[$tagToRemoveNodeIdAttributeValue] ) ) { |
| | 128 | + $domElemsToRemove[] = $tagToRemoveNode; |
| | 129 | + } |
| | 130 | + } |
| | 131 | + } |
| | 132 | + } |
| | 133 | + |
| | 134 | + foreach ( $domElemsToRemove as $domElement ) { |
| | 135 | + $domElement->parentNode->removeChild( $domElement ); |
| | 136 | + } |
| | 137 | + |
| | 138 | + // Elements with named IDs |
| | 139 | + foreach ( $removals['ID'] as $itemToRemove ) { |
| | 140 | + $itemToRemoveNode = $this->doc->getElementById( $itemToRemove ); |
| | 141 | + if ( $itemToRemoveNode ) { |
| | 142 | + $itemToRemoveNode->parentNode->removeChild( $itemToRemoveNode ); |
| | 143 | + } |
| | 144 | + } |
| | 145 | + |
| | 146 | + // CSS Classes |
| | 147 | + $xpath = new DOMXpath( $this->doc ); |
| | 148 | + foreach ( $removals['CLASS'] as $classToRemove ) { |
| | 149 | + $elements = $xpath->query( '//*[@class="' . $classToRemove . '"]' ); |
| | 150 | + |
| | 151 | + foreach ( $elements as $element ) { |
| | 152 | + $element->parentNode->removeChild( $element ); |
| | 153 | + } |
| | 154 | + } |
| | 155 | + |
| | 156 | + // Tags with CSS Classes |
| | 157 | + foreach ( $removals['TAG_CLASS'] as $classToRemove ) { |
| | 158 | + $parts = explode( '.', $classToRemove ); |
| | 159 | + |
| | 160 | + $elements = $xpath->query( |
| | 161 | + '//' . $parts[0] . '[@class="' . $parts[1] . '"]' |
| | 162 | + ); |
| | 163 | + |
| | 164 | + foreach ( $elements as $element ) { |
| | 165 | + $removedElement = $element->parentNode->removeChild( $element ); |
| | 166 | + } |
| | 167 | + } |
| | 168 | + |
| | 169 | + // Handle red links with action equal to edit |
| | 170 | + if ( $this->flattenRedLinks ) { |
| | 171 | + $redLinks = $xpath->query( '//a[@class="new"]' ); |
| | 172 | + foreach ( $redLinks as $redLink ) { |
| | 173 | + // PHP Bug #36795 — Inappropriate "unterminated entity reference" |
| | 174 | + $spanNode = $this->doc->createElement( "span", str_replace( "&", "&", $redLink->nodeValue ) ); |
| | 175 | + |
| | 176 | + if ( $redLink->hasAttributes() ) { |
| | 177 | + $attributes = $redLink->attributes; |
| | 178 | + foreach ( $attributes as $i => $attribute ) { |
| | 179 | + if ( $attribute->name != 'href' ) { |
| | 180 | + $spanNode->setAttribute( $attribute->name, $attribute->value ); |
| | 181 | + } |
| | 182 | + } |
| | 183 | + } |
| | 184 | + |
| | 185 | + $redLink->parentNode->replaceChild( $spanNode, $redLink ); |
| | 186 | + } |
| | 187 | + } |
| | 188 | + wfProfileOut( __METHOD__ ); |
| | 189 | + } |
| | 190 | + |
| | 191 | + /** |
| | 192 | + * Performs final transformations and returns resulting HTML |
| | 193 | + * |
| | 194 | + * @param DOMElement|string|null $element: ID of element to get HTML from or false to get it from the whole tree |
| | 195 | + * @return string: Processed HTML |
| | 196 | + */ |
| | 197 | + public function getText( $element = null ) { |
| | 198 | + wfProfileIn( __METHOD__ ); |
| | 199 | + |
| | 200 | + if ( $element !== null && !( $element instanceof DOMElement ) ) { |
| | 201 | + $element = $this->doc->getElementById( $element ); |
| | 202 | + } |
| | 203 | + $html = $this->doc->saveXML( $element, LIBXML_NOEMPTYTAG ); |
| | 204 | + if ( !$element ) { |
| | 205 | + $html = preg_replace( '/<!--.*?-->|^.*?<body>|<\/body>.*$/s', '', $html ); |
| | 206 | + } |
| | 207 | + $html = $this->onHtmlReady( $html ); |
| | 208 | + |
| | 209 | + if ( $this->elementsToFlatten ) { |
| | 210 | + $elements = implode( '|', $this->elementsToFlatten ); |
| | 211 | + $html = preg_replace( "#</?($elements)\\b[^>]*>#is", '', $html ); |
| | 212 | + } |
| | 213 | + |
| | 214 | + wfProfileOut( __METHOD__ ); |
| | 215 | + return $html; |
| | 216 | + } |
| | 217 | + |
| | 218 | + /** |
| | 219 | + * Transforms CSS selectors into an internal representation suitable for processing |
| | 220 | + * @return array |
| | 221 | + */ |
| | 222 | + protected function parseItemsToRemove() { |
| | 223 | + wfProfileIn( __METHOD__ ); |
| | 224 | + $removals = array( |
| | 225 | + 'ID' => array(), |
| | 226 | + 'TAG' => array(), |
| | 227 | + 'CLASS' => array(), |
| | 228 | + 'TAG_CLASS' => array(), |
| | 229 | + ); |
| | 230 | + |
| | 231 | + foreach ( $this->itemsToRemove as $itemToRemove ) { |
| | 232 | + $type = ''; |
| | 233 | + $rawName = ''; |
| | 234 | + CssDetection::detectIdCssOrTag( $itemToRemove, $type, $rawName ); |
| | 235 | + $removals[$type][] = $rawName; |
| | 236 | + } |
| | 237 | + |
| | 238 | + if ( $this->removeImages ) { |
| | 239 | + $removals['TAG'][] = "img"; |
| | 240 | + $removals['TAG'][] = "audio"; |
| | 241 | + $removals['TAG'][] = "video"; |
| | 242 | + $removals['CLASS'][] = "thumb tright"; |
| | 243 | + $removals['CLASS'][] = "thumb tleft"; |
| | 244 | + $removals['CLASS'][] = "thumbcaption"; |
| | 245 | + $removals['CLASS'][] = "gallery"; |
| | 246 | + } |
| | 247 | + |
| | 248 | + wfProfileOut( __METHOD__ ); |
| | 249 | + return $removals; |
| | 250 | + } |
| | 251 | +} |
| Property changes on: trunk/extensions/MobileFrontend/HtmlFormatter.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 252 | + native |