| Index: trunk/phase3/includes/specials/SpecialExport.php |
| — | — | @@ -46,7 +46,8 @@ |
| 47 | 47 | $this->templates = $wgRequest->getCheck( 'templates' ); |
| 48 | 48 | $this->images = $wgRequest->getCheck( 'images' ); // Doesn't do anything yet |
| 49 | 49 | $this->pageLinkDepth = $this->validateLinkDepth( |
| 50 | | - $wgRequest->getIntOrNull( 'pagelink-depth' ) ); |
| | 50 | + $wgRequest->getIntOrNull( 'pagelink-depth' ) |
| | 51 | + ); |
| 51 | 52 | $nsindex = ''; |
| 52 | 53 | |
| 53 | 54 | if ( $wgRequest->getCheck( 'addcat' ) ) { |
| — | — | @@ -82,11 +83,13 @@ |
| 83 | 84 | $page = $wgRequest->getText( 'pages' ); |
| 84 | 85 | $this->curonly = $wgRequest->getCheck( 'curonly' ); |
| 85 | 86 | $rawOffset = $wgRequest->getVal( 'offset' ); |
| | 87 | + |
| 86 | 88 | if( $rawOffset ) { |
| 87 | 89 | $offset = wfTimestamp( TS_MW, $rawOffset ); |
| 88 | 90 | } else { |
| 89 | 91 | $offset = null; |
| 90 | 92 | } |
| | 93 | + |
| 91 | 94 | $limit = $wgRequest->getInt( 'limit' ); |
| 92 | 95 | $dir = $wgRequest->getVal( 'dir' ); |
| 93 | 96 | $history = array( |
| — | — | @@ -95,6 +98,7 @@ |
| 96 | 99 | 'limit' => $wgExportMaxHistory, |
| 97 | 100 | ); |
| 98 | 101 | $historyCheck = $wgRequest->getCheck( 'history' ); |
| | 102 | + |
| 99 | 103 | if ( $this->curonly ) { |
| 100 | 104 | $history = WikiExporter::CURRENT; |
| 101 | 105 | } elseif ( !$historyCheck ) { |
| — | — | @@ -111,9 +115,10 @@ |
| 112 | 116 | |
| 113 | 117 | if( $page != '' ) $this->doExport = true; |
| 114 | 118 | } else { |
| 115 | | - // Default to current-only for GET requests |
| | 119 | + // Default to current-only for GET requests. |
| 116 | 120 | $page = $wgRequest->getText( 'pages', $par ); |
| 117 | 121 | $historyCheck = $wgRequest->getCheck( 'history' ); |
| | 122 | + |
| 118 | 123 | if( $historyCheck ) { |
| 119 | 124 | $history = WikiExporter::FULL; |
| 120 | 125 | } else { |
| — | — | @@ -133,16 +138,20 @@ |
| 134 | 139 | |
| 135 | 140 | if ( $this->doExport ) { |
| 136 | 141 | $wgOut->disable(); |
| | 142 | + |
| 137 | 143 | // Cancel output buffering and gzipping if set |
| 138 | 144 | // This should provide safer streaming for pages with history |
| 139 | 145 | wfResetOutputBuffers(); |
| 140 | 146 | $wgRequest->response()->header( "Content-type: application/xml; charset=utf-8" ); |
| | 147 | + |
| 141 | 148 | if( $wgRequest->getCheck( 'wpDownload' ) ) { |
| 142 | 149 | // Provide a sane filename suggestion |
| 143 | 150 | $filename = urlencode( $wgSitename . '-' . wfTimestampNow() . '.xml' ); |
| 144 | 151 | $wgRequest->response()->header( "Content-disposition: attachment;filename={$filename}" ); |
| 145 | 152 | } |
| | 153 | + |
| 146 | 154 | $this->doExport( $page, $history, $list_authors ); |
| | 155 | + |
| 147 | 156 | return; |
| 148 | 157 | } |
| 149 | 158 | |
| — | — | @@ -166,7 +175,9 @@ |
| 167 | 176 | } else { |
| 168 | 177 | $wgOut->addHTML( wfMsgExt( 'exportnohistory', 'parse' ) ); |
| 169 | 178 | } |
| | 179 | + |
| 170 | 180 | $form .= Xml::checkLabel( wfMsg( 'export-templates' ), 'templates', 'wpExportTemplates', false ) . '<br />'; |
| | 181 | + |
| 171 | 182 | if( $wgExportMaxLinkDepth || $this->userCanOverrideExportDepth() ) { |
| 172 | 183 | $form .= Xml::inputLabel( wfMsg( 'export-pagelinks' ), 'pagelink-depth', 'pagelink-depth', 20, 0 ) . '<br />'; |
| 173 | 184 | } |
| — | — | @@ -176,12 +187,12 @@ |
| 177 | 188 | |
| 178 | 189 | $form .= Xml::submitButton( wfMsg( 'export-submit' ), $wgUser->getSkin()->tooltipAndAccessKeyAttribs( 'export' ) ); |
| 179 | 190 | $form .= Xml::closeElement( 'form' ); |
| | 191 | + |
| 180 | 192 | $wgOut->addHTML( $form ); |
| 181 | 193 | } |
| 182 | 194 | |
| 183 | 195 | private function userCanOverrideExportDepth() { |
| 184 | 196 | global $wgUser; |
| 185 | | - |
| 186 | 197 | return $wgUser->isAllowed( 'override-export-depth' ); |
| 187 | 198 | } |
| 188 | 199 | |
| — | — | @@ -231,6 +242,7 @@ |
| 232 | 243 | foreach( $pages as $k => $v ) { |
| 233 | 244 | $pages[$k] = str_replace( " ", "_", $v ); |
| 234 | 245 | } |
| | 246 | + |
| 235 | 247 | $pages = array_unique( $pages ); |
| 236 | 248 | |
| 237 | 249 | /* Ok, let's get to it... */ |
| — | — | @@ -249,9 +261,11 @@ |
| 250 | 262 | set_time_limit(0); |
| 251 | 263 | wfRestoreWarnings(); |
| 252 | 264 | } |
| | 265 | + |
| 253 | 266 | $exporter = new WikiExporter( $db, $history, $buffer ); |
| 254 | 267 | $exporter->list_authors = $list_authors; |
| 255 | 268 | $exporter->openStream(); |
| | 269 | + |
| 256 | 270 | foreach( $pages as $page ) { |
| 257 | 271 | /* |
| 258 | 272 | if( $wgExportMaxHistory && !$this->curonly ) { |
| — | — | @@ -274,6 +288,7 @@ |
| 275 | 289 | } |
| 276 | 290 | |
| 277 | 291 | $exporter->closeStream(); |
| | 292 | + |
| 278 | 293 | if( $lb ) { |
| 279 | 294 | $lb->closeAll(); |
| 280 | 295 | } |
| — | — | @@ -285,12 +300,16 @@ |
| 286 | 301 | $name = $title->getDBkey(); |
| 287 | 302 | |
| 288 | 303 | $dbr = wfGetDB( DB_SLAVE ); |
| 289 | | - $res = $dbr->select( array('page', 'categorylinks' ), |
| 290 | | - array( 'page_namespace', 'page_title' ), |
| 291 | | - array('cl_from=page_id', 'cl_to' => $name ), |
| 292 | | - __METHOD__, array('LIMIT' => '5000')); |
| | 304 | + $res = $dbr->select( |
| | 305 | + array( 'page', 'categorylinks' ), |
| | 306 | + array( 'page_namespace', 'page_title' ), |
| | 307 | + array( 'cl_from=page_id', 'cl_to' => $name ), |
| | 308 | + __METHOD__, |
| | 309 | + array( 'LIMIT' => '5000' ) |
| | 310 | + ); |
| 293 | 311 | |
| 294 | 312 | $pages = array(); |
| | 313 | + |
| 295 | 314 | while ( $row = $dbr->fetchObject( $res ) ) { |
| 296 | 315 | $n = $row->page_title; |
| 297 | 316 | if ($row->page_namespace) { |
| — | — | @@ -300,6 +319,7 @@ |
| 301 | 320 | |
| 302 | 321 | $pages[] = $n; |
| 303 | 322 | } |
| | 323 | + |
| 304 | 324 | $dbr->freeResult($res); |
| 305 | 325 | |
| 306 | 326 | return $pages; |
| — | — | @@ -309,21 +329,28 @@ |
| 310 | 330 | global $wgContLang; |
| 311 | 331 | |
| 312 | 332 | $dbr = wfGetDB( DB_SLAVE ); |
| 313 | | - $res = $dbr->select( 'page', array('page_namespace', 'page_title'), |
| 314 | | - array('page_namespace' => $nsindex), |
| 315 | | - __METHOD__, array('LIMIT' => '5000') ); |
| | 333 | + $res = $dbr->select( |
| | 334 | + 'page', |
| | 335 | + array( 'page_namespace', 'page_title' ), |
| | 336 | + array( 'page_namespace' => $nsindex ), |
| | 337 | + __METHOD__, |
| | 338 | + array( 'LIMIT' => '5000' ) |
| | 339 | + ); |
| 316 | 340 | |
| 317 | 341 | $pages = array(); |
| | 342 | + |
| 318 | 343 | while ( $row = $dbr->fetchObject( $res ) ) { |
| 319 | 344 | $n = $row->page_title; |
| 320 | | - if ($row->page_namespace) { |
| | 345 | + |
| | 346 | + if ( $row->page_namespace ) { |
| 321 | 347 | $ns = $wgContLang->getNsText( $row->page_namespace ); |
| 322 | 348 | $n = $ns . ':' . $n; |
| 323 | 349 | } |
| 324 | 350 | |
| 325 | 351 | $pages[] = $n; |
| 326 | 352 | } |
| 327 | | - $dbr->freeResult($res); |
| | 353 | + |
| | 354 | + $dbr->freeResult( $res ); |
| 328 | 355 | |
| 329 | 356 | return $pages; |
| 330 | 357 | } |
| — | — | @@ -336,9 +363,10 @@ |
| 337 | 364 | */ |
| 338 | 365 | private function getTemplates( $inputPages, $pageSet ) { |
| 339 | 366 | return $this->getLinks( $inputPages, $pageSet, |
| 340 | | - 'templatelinks', |
| 341 | | - array( 'tl_namespace AS namespace', 'tl_title AS title' ), |
| 342 | | - array( 'page_id=tl_from' ) ); |
| | 367 | + 'templatelinks', |
| | 368 | + array( 'tl_namespace AS namespace', 'tl_title AS title' ), |
| | 369 | + array( 'page_id=tl_from' ) |
| | 370 | + ); |
| 343 | 371 | } |
| 344 | 372 | |
| 345 | 373 | /** |
| — | — | @@ -346,14 +374,17 @@ |
| 347 | 375 | */ |
| 348 | 376 | private function validateLinkDepth( $depth ) { |
| 349 | 377 | global $wgExportMaxLinkDepth; |
| | 378 | + |
| 350 | 379 | if( $depth < 0 ) { |
| 351 | 380 | return 0; |
| 352 | 381 | } |
| | 382 | + |
| 353 | 383 | if ( !$this->userCanOverrideExportDepth() ) { |
| 354 | 384 | if( $depth > $wgExportMaxLinkDepth ) { |
| 355 | 385 | return $wgExportMaxLinkDepth; |
| 356 | 386 | } |
| 357 | 387 | } |
| | 388 | + |
| 358 | 389 | /* |
| 359 | 390 | * There's a HARD CODED limit of 5 levels of recursion here to prevent a |
| 360 | 391 | * crazy-big export from being done by someone setting the depth |
| — | — | @@ -364,55 +395,70 @@ |
| 365 | 396 | |
| 366 | 397 | /** Expand a list of pages to include pages linked to from that page. */ |
| 367 | 398 | private function getPageLinks( $inputPages, $pageSet, $depth ) { |
| 368 | | - for( $depth=$depth; $depth>0; --$depth ) { |
| 369 | | - $pageSet = $this->getLinks( $inputPages, $pageSet, 'pagelinks', |
| 370 | | - array( 'pl_namespace AS namespace', 'pl_title AS title' ), |
| 371 | | - array( 'page_id=pl_from' ) ); |
| | 399 | + for(; $depth > 0; --$depth ) { |
| | 400 | + $pageSet = $this->getLinks( |
| | 401 | + $inputPages, $pageSet, 'pagelinks', |
| | 402 | + array( 'pl_namespace AS namespace', 'pl_title AS title' ), |
| | 403 | + array( 'page_id=pl_from' ) |
| | 404 | + ); |
| 372 | 405 | $inputPages = array_keys( $pageSet ); |
| 373 | 406 | } |
| | 407 | + |
| 374 | 408 | return $pageSet; |
| 375 | 409 | } |
| 376 | 410 | |
| 377 | 411 | /** |
| 378 | 412 | * Expand a list of pages to include images used in those pages. |
| | 413 | + * |
| 379 | 414 | * @param $inputPages array, list of titles to look up |
| 380 | 415 | * @param $pageSet array, associative array indexed by titles for output |
| | 416 | + * |
| 381 | 417 | * @return array associative array index by titles |
| 382 | 418 | */ |
| 383 | 419 | private function getImages( $inputPages, $pageSet ) { |
| 384 | | - return $this->getLinks( $inputPages, $pageSet, |
| 385 | | - 'imagelinks', |
| 386 | | - array( NS_FILE . ' AS namespace', 'il_to AS title' ), |
| 387 | | - array( 'page_id=il_from' ) ); |
| | 420 | + return $this->getLinks( |
| | 421 | + $inputPages, |
| | 422 | + $pageSet, |
| | 423 | + 'imagelinks', |
| | 424 | + array( NS_FILE . ' AS namespace', 'il_to AS title' ), |
| | 425 | + array( 'page_id=il_from' ) |
| | 426 | + ); |
| 388 | 427 | } |
| 389 | 428 | |
| 390 | 429 | /** |
| 391 | 430 | * Expand a list of pages to include items used in those pages. |
| 392 | | - * @private |
| 393 | 431 | */ |
| 394 | 432 | private function getLinks( $inputPages, $pageSet, $table, $fields, $join ) { |
| 395 | 433 | $dbr = wfGetDB( DB_SLAVE ); |
| | 434 | + |
| 396 | 435 | foreach( $inputPages as $page ) { |
| 397 | 436 | $title = Title::newFromText( $page ); |
| | 437 | + |
| 398 | 438 | if( $title ) { |
| 399 | 439 | $pageSet[$title->getPrefixedText()] = true; |
| 400 | 440 | /// @todo Fixme: May or may not be more efficient to batch these |
| 401 | 441 | /// by namespace when given multiple input pages. |
| 402 | 442 | $result = $dbr->select( |
| 403 | | - array( 'page', $table ), |
| 404 | | - $fields, |
| 405 | | - array_merge( $join, |
| 406 | | - array( |
| 407 | | - 'page_namespace' => $title->getNamespace(), |
| 408 | | - 'page_title' => $title->getDBkey() ) ), |
| 409 | | - __METHOD__ ); |
| | 443 | + array( 'page', $table ), |
| | 444 | + $fields, |
| | 445 | + array_merge( |
| | 446 | + $join, |
| | 447 | + array( |
| | 448 | + 'page_namespace' => $title->getNamespace(), |
| | 449 | + 'page_title' => $title->getDBkey() |
| | 450 | + ) |
| | 451 | + ), |
| | 452 | + __METHOD__ |
| | 453 | + ); |
| | 454 | + |
| 410 | 455 | foreach( $result as $row ) { |
| 411 | 456 | $template = Title::makeTitle( $row->namespace, $row->title ); |
| 412 | 457 | $pageSet[$template->getPrefixedText()] = true; |
| 413 | 458 | } |
| 414 | 459 | } |
| 415 | 460 | } |
| | 461 | + |
| 416 | 462 | return $pageSet; |
| 417 | 463 | } |
| 418 | | -} |
| 419 | | - |
| | 464 | + |
| | 465 | +} |
| \ No newline at end of file |