| Index: trunk/extensions/MobileFrontend/api/ApiQueryExtracts.php |
| — | — | @@ -21,9 +21,10 @@ |
| 22 | 22 | wfProfileOut( __METHOD__ ); |
| 23 | 23 | return; |
| 24 | 24 | } |
| 25 | | - $isXml = $this->getMain()->getPrinter()->getFormat() == 'XML'; |
| | 25 | + $isXml = $this->getMain()->isInternalMode() || $this->getMain()->getPrinter()->getFormat() == 'XML'; |
| 26 | 26 | $result = $this->getResult(); |
| 27 | 27 | $params = $this->params = $this->extractRequestParams(); |
| | 28 | + $this->requireMaxOneParameter( $params, 'chars', 'sentences' ); |
| 28 | 29 | $continue = 0; |
| 29 | 30 | $limit = intval( $params['limit'] ); |
| 30 | 31 | if ( $limit > 1 && !$params['intro'] ) { |
| — | — | @@ -45,9 +46,8 @@ |
| 46 | 47 | break; |
| 47 | 48 | } |
| 48 | 49 | $text = $this->getExtract( $t ); |
| 49 | | - if ( isset( $params['length'] ) ) { |
| 50 | | - $text = $this->trimText( $text ); |
| 51 | | - } |
| | 50 | + $text = $this->truncate( $text ); |
| | 51 | + |
| 52 | 52 | if ( $isXml ) { |
| 53 | 53 | $fit = $result->addValue( array( 'query', 'pages', $id ), 'extract', array( '*' => $text ) ); |
| 54 | 54 | } else { |
| — | — | @@ -74,7 +74,7 @@ |
| 75 | 75 | $api = new ApiMain( new FauxRequest( |
| 76 | 76 | array( |
| 77 | 77 | 'action' => 'query', |
| 78 | | - 'prop' => 'excerpts', |
| | 78 | + 'prop' => 'extracts', |
| 79 | 79 | 'explaintext' => true, |
| 80 | 80 | 'exlimit' => count( $results ), |
| 81 | 81 | 'pageids' => implode( '|', $pageIds ), |
| — | — | @@ -83,8 +83,8 @@ |
| 84 | 84 | $api->execute(); |
| 85 | 85 | $data = $api->getResultData(); |
| 86 | 86 | foreach ( $pageIds as $id ) { |
| 87 | | - if ( isset( $data['query']['pages'][$id]['excerpts'][0] ) ) { |
| 88 | | - $results[$id]['extract'] = $data['query']['pages'][$id]['extract'][0]; |
| | 87 | + if ( isset( $data['query']['pages'][$id]['extract']['*'] ) ) { |
| | 88 | + $results[$id]['extract'] = $data['query']['pages'][$id]['extract']['*']; |
| 89 | 89 | $results[$id]['extract trimmed'] = false; |
| 90 | 90 | } |
| 91 | 91 | } |
| — | — | @@ -92,7 +92,7 @@ |
| 93 | 93 | } |
| 94 | 94 | |
| 95 | 95 | /** |
| 96 | | - * Returns a processed, but not trimmed excerpt |
| | 96 | + * Returns a processed, but not trimmed extract |
| 97 | 97 | * @param Title $title |
| 98 | 98 | * @return string |
| 99 | 99 | */ |
| — | — | @@ -188,10 +188,8 @@ |
| 189 | 189 | } |
| 190 | 190 | |
| 191 | 191 | /** |
| 192 | | - * Converts page HTML into an excerpt |
| | 192 | + * Converts page HTML into an extract |
| 193 | 193 | * @param string $text |
| 194 | | - * @param Title $title |
| 195 | | - * @param bool $plainText |
| 196 | 194 | * @return string |
| 197 | 195 | */ |
| 198 | 196 | private function convertText( $text ) { |
| — | — | @@ -203,16 +201,22 @@ |
| 204 | 202 | return trim( $text ); |
| 205 | 203 | } |
| 206 | 204 | |
| | 205 | + private function truncate( $text ) { |
| | 206 | + if ( $this->params['chars'] ) { |
| | 207 | + return $this->getFirstChars( $text, $this->params['chars'] ); |
| | 208 | + } elseif ( $this->params['sentences'] ) { |
| | 209 | + return $this->getFirstSentences( $text, $this->params['sentences'] ); |
| | 210 | + } |
| | 211 | + return $text; |
| | 212 | + } |
| | 213 | + |
| 207 | 214 | /** |
| 208 | 215 | * |
| 209 | 216 | * @param string $text |
| 210 | 217 | * @param int $requestedLength |
| 211 | | - * @param bool $plainText |
| 212 | 218 | * @return string |
| 213 | 219 | */ |
| 214 | | - private function trimText( $text, $requestedLength, $plainText ) { |
| 215 | | - global $wgUseTidy; |
| 216 | | - |
| | 220 | + private function getFirstChars( $text, $requestedLength ) { |
| 217 | 221 | wfProfileIn( __METHOD__ ); |
| 218 | 222 | $length = mb_strlen( $text ); |
| 219 | 223 | if ( $length <= $requestedLength ) { |
| — | — | @@ -223,20 +227,69 @@ |
| 224 | 228 | preg_match( $pattern, $text, $m ); |
| 225 | 229 | $text = $m[0]; |
| 226 | 230 | // Fix possibly unclosed tags |
| 227 | | - if ( $wgUseTidy && !$plainText ) { |
| | 231 | + $text = $this->tidy( $text ); |
| | 232 | + $text .= wfMessage( 'ellipsis' )->inContentLanguage()->text(); |
| | 233 | + wfProfileOut( __METHOD__ ); |
| | 234 | + return $text; |
| | 235 | + } |
| | 236 | + |
| | 237 | + /** |
| | 238 | + * |
| | 239 | + * @param string $text |
| | 240 | + * @param int $requestedSentenceCount |
| | 241 | + */ |
| | 242 | + private function getFirstSentences( $text, $requestedSentenceCount ) { |
| | 243 | + wfProfileIn( __METHOD__ ); |
| | 244 | + // Based on code from OpenSearchXml by Brion Vibber |
| | 245 | + $endchars = array( |
| | 246 | + '([^\d])\.\s', '\!\s', '\?\s', // regular ASCII |
| | 247 | + '。', // full-width ideographic full-stop |
| | 248 | + '．', '！', '？', // double-width roman forms |
| | 249 | + '｡', // half-width ideographic full stop |
| | 250 | + ); |
| | 251 | + |
| | 252 | + $endgroup = implode( '|', $endchars ); |
| | 253 | + $end = "(?:$endgroup)"; |
| | 254 | + $sentence = ".+?$end+"; |
| | 255 | + $regexp = "/^($sentence){{$requestedSentenceCount}}/u"; |
| | 256 | + $matches = array(); |
| | 257 | + if( preg_match( $regexp, $text, $matches ) ) { |
| | 258 | + return $matches[0]; |
| | 259 | + } else { |
| | 260 | + // Just return the first line |
| | 261 | + $lines = explode( "\n", $text ); |
| | 262 | + return trim( $lines[0] ); |
| | 263 | + } |
| | 264 | + $text = $this->tidy( $text ); |
| | 265 | + wfProfileOut( __METHOD__ ); |
| | 266 | + return $text; |
| | 267 | + } |
| | 268 | + |
| | 269 | + /** |
| | 270 | + * A simple wrapper around tidy |
| | 271 | + * @param string $text |
| | 272 | + */ |
| | 273 | + private function tidy( $text ) { |
| | 274 | + global $wgUseTidy; |
| | 275 | + |
| | 276 | + wfProfileIn( __METHOD__ ); |
| | 277 | + if ( $wgUseTidy && !$this->params['plaintext'] ) { |
| 228 | 278 | $text = trim ( MWTidy::tidy( $text ) ); |
| 229 | 279 | } |
| 230 | | - $text .= wfMessage( 'ellipsis' )->inContentLanguage()->text(); |
| 231 | 280 | wfProfileOut( __METHOD__ ); |
| 232 | 281 | return $text; |
| 233 | 282 | } |
| 234 | 283 | |
| 235 | 284 | public function getAllowedParams() { |
| 236 | 285 | return array( |
| 237 | | - 'length' => array( |
| | 286 | + 'chars' => array( |
| 238 | 287 | ApiBase::PARAM_TYPE => 'integer', |
| 239 | 288 | ApiBase::PARAM_MIN => 1, |
| 240 | 289 | ), |
| | 290 | + 'sentences' => array( |
| | 291 | + ApiBase::PARAM_TYPE => 'integer', |
| | 292 | + ApiBase::PARAM_MIN => 1, |
| | 293 | + ), |
| 241 | 294 | 'limit' => array( |
| 242 | 295 | ApiBase::PARAM_DFLT => 1, |
| 243 | 296 | ApiBase::PARAM_TYPE => 'limit', |
| — | — | @@ -258,7 +311,8 @@ |
| 259 | 312 | |
| 260 | 313 | public function getParamDescription() { |
| 261 | 314 | return array( |
| 262 | | - 'length' => 'How many characters to return, actual text returned might be slightly longer.', |
| | 315 | + 'chars' => 'How many characters to return, actual text returned might be slightly longer.', |
| | 316 | + 'sentences' => 'How many sentences to return', |
| 263 | 317 | 'limit' => 'How many extracts to return. ', |
| 264 | 318 | 'intro' => 'Return only content before the first section', |
| 265 | 319 | 'plaintext' => 'Return extracts as plaintext instead of limited HTML', |
| — | — | @@ -284,7 +338,7 @@ |
| 285 | 339 | |
| 286 | 340 | public function getExamples() { |
| 287 | 341 | return array( |
| 288 | | - 'api.php?action=query&prop=extracts&exlength=175&titles=Therion' => 'Get a 175-character extract', |
| | 342 | + 'api.php?action=query&prop=extracts&exchars=175&titles=Therion' => 'Get a 175-character extract', |
| 289 | 343 | ); |
| 290 | 344 | } |
| 291 | 345 | |
| — | — | @@ -329,8 +383,8 @@ |
| 330 | 384 | $text = parent::getText(); |
| 331 | 385 | if ( $this->plainText ) { |
| 332 | 386 | $text = html_entity_decode( $text ); |
| 333 | | - $text = str_replace( "\r", "\n", $text ); |
| 334 | | - $text = preg_replace( "/\n{3,}/", "\n\n", $text ); |
| | 387 | + $text = str_replace( "\r", "\n", $text ); // for Windows |
| | 388 | + $text = preg_replace( "/\n{3,}/", "\n\n", $text ); // normalise newlines |
| 335 | 389 | $text = preg_replace_callback( |
| 336 | 390 | "/" . ApiQueryExtracts::SECTION_MARKER_START . '(\d)'. ApiQueryExtracts::SECTION_MARKER_END . "(.*?)$/m", |
| 337 | 391 | array( $this, 'sectionCallback' ), |
| Index: trunk/extensions/MobileFrontend/MobileFrontend.php |
| — | — | @@ -132,7 +132,7 @@ |
| 133 | 133 | $wgHooks['APIAfterExecute'][] = 'ApiParseExtender::onAPIAfterExecute'; |
| 134 | 134 | $wgHooks['APIGetParamDescription'][] = 'ApiParseExtender::onAPIGetParamDescription'; |
| 135 | 135 | $wgHooks['APIGetDescription'][] = 'ApiParseExtender::onAPIGetDescription'; |
| 136 | | -$wgHooks['OpenSearchXml'][] = 'ApiQueryExcerpts::onOpenSearchXml'; |
| | 136 | +$wgHooks['OpenSearchXml'][] = 'ApiQueryExtracts::onOpenSearchXml'; |
| 137 | 137 | |
| 138 | 138 | function efMobileFrontend_Setup() { |
| 139 | 139 | global $wgExtMobileFrontend, $wgHooks; |
| — | — | @@ -174,6 +174,6 @@ |
| 175 | 175 | } |
| 176 | 176 | |
| 177 | 177 | /** |
| 178 | | - * Whether this extension should provide its excerpts to OpenSearchXml extension |
| | 178 | + * Whether this extension should provide its extracts to OpenSearchXml extension |
| 179 | 179 | */ |
| 180 | 180 | $wgMFExtendOpenSearchXml = false; |