tests/phpunit/maintenance/backupTextPassTest.php

   1 <?php
   2
   3 namespace MediaWiki\Tests\Maintenance;
   4
   5 use MediaWikiLangTestCase;
   6 use TextContentHandler;
   7 use TextPassDumper;
   8 use Title;
   9 use WikiExporter;
  10 use WikiPage;
  11
  12 require_once __DIR__ . "/../../../maintenance/dumpTextPass.php";
  13
  14 /**
  15  * Tests for TextPassDumper that rely on the database
  16  *
  17  * Some of these tests use the old constuctor for TextPassDumper
  18  * and the dump() function, while others use the new loadWithArgv( $args )
  19  * function and execute(). This is to ensure both the old and new methods
  20  * work properly.
  21  *
  22  * @group Database
  23  * @group Dump
  24  * @covers TextPassDumper
  25  */
  26 class TextPassDumperDatabaseTest extends DumpTestCase {
  27
  28         // We'll add several pages, revision and texts. The following variables hold the
  29         // corresponding ids.
  30         private $pageId1, $pageId2, $pageId3, $pageId4;
  31         private static $numOfPages = 4;
  32         private $revId1_1, $textId1_1;
  33         private $revId2_1, $textId2_1, $revId2_2, $textId2_2;
  34         private $revId2_3, $textId2_3, $revId2_4, $textId2_4;
  35         private $revId3_1, $textId3_1, $revId3_2, $textId3_2;
  36         private $revId4_1, $textId4_1;
  37         private static $numOfRevs = 8;
  38
  39         function addDBData() {
  40                 $this->tablesUsed[] = 'page';
  41                 $this->tablesUsed[] = 'revision';
  42                 $this->tablesUsed[] = 'ip_changes';
  43                 $this->tablesUsed[] = 'text';
  44
  45                 $this->mergeMwGlobalArrayValue( 'wgContentHandlers', [
  46                         "BackupTextPassTestModel" => BackupTextPassTestModelHandler::class,
  47                 ] );
  48
  49                 $ns = $this->getDefaultWikitextNS();
  50
  51                 try {
  52                         // Simple page
  53                         $title = Title::newFromText( 'BackupDumperTestP1', $ns );
  54                         $page = WikiPage::factory( $title );
  55                         list( $this->revId1_1, $this->textId1_1 ) = $this->addRevision( $page,
  56                                 "BackupDumperTestP1Text1", "BackupDumperTestP1Summary1" );
  57                         $this->pageId1 = $page->getId();
  58
  59                         // Page with more than one revision
  60                         $title = Title::newFromText( 'BackupDumperTestP2', $ns );
  61                         $page = WikiPage::factory( $title );
  62                         list( $this->revId2_1, $this->textId2_1 ) = $this->addRevision( $page,
  63                                 "BackupDumperTestP2Text1", "BackupDumperTestP2Summary1" );
  64                         list( $this->revId2_2, $this->textId2_2 ) = $this->addRevision( $page,
  65                                 "BackupDumperTestP2Text2", "BackupDumperTestP2Summary2" );
  66                         list( $this->revId2_3, $this->textId2_3 ) = $this->addRevision( $page,
  67                                 "BackupDumperTestP2Text3", "BackupDumperTestP2Summary3" );
  68                         list( $this->revId2_4, $this->textId2_4 ) = $this->addRevision( $page,
  69                                 "BackupDumperTestP2Text4 some additional Text  ",
  70                                 "BackupDumperTestP2Summary4 extra " );
  71                         $this->pageId2 = $page->getId();
  72
  73                         // Deleted page.
  74                         $title = Title::newFromText( 'BackupDumperTestP3', $ns );
  75                         $page = WikiPage::factory( $title );
  76                         list( $this->revId3_1, $this->textId3_1 ) = $this->addRevision( $page,
  77                                 "BackupDumperTestP3Text1", "BackupDumperTestP2Summary1" );
  78                         list( $this->revId3_2, $this->textId3_2 ) = $this->addRevision( $page,
  79                                 "BackupDumperTestP3Text2", "BackupDumperTestP2Summary2" );
  80                         $this->pageId3 = $page->getId();
  81                         $page->doDeleteArticle( "Testing ;)" );
  82
  83                         // Page from non-default namespace and model.
  84                         // ExportTransform applies.
  85
  86                         if ( $ns === NS_TALK ) {
  87                                 // @todo work around this.
  88                                 throw new MWException( "The default wikitext namespace is the talk namespace. "
  89                                         . " We can't currently deal with that." );
  90                         }
  91
  92                         $title = Title::newFromText( 'BackupDumperTestP1', NS_TALK );
  93                         $page = WikiPage::factory( $title );
  94                         list( $this->revId4_1, $this->textId4_1 ) = $this->addRevision( $page,
  95                                 "Talk about BackupDumperTestP1 Text1",
  96                                 "Talk BackupDumperTestP1 Summary1",
  97                                 "BackupTextPassTestModel" );
  98                         $this->pageId4 = $page->getId();
  99                 } catch ( Exception $e ) {
 100                         // We'd love to pass $e directly. However, ... see
 101                         // documentation of exceptionFromAddDBData in
 102                         // DumpTestCase
 103                         $this->exceptionFromAddDBData = $e;
 104                 }
 105         }
 106
 107         protected function setUp() {
 108                 parent::setUp();
 109
 110                 // Since we will restrict dumping by page ranges (to allow
 111                 // working tests, even if the db gets prepopulated by a base
 112                 // class), we have to assert, that the page id are consecutively
 113                 // increasing
 114                 $this->assertEquals(
 115                         [ $this->pageId2, $this->pageId3, $this->pageId4 ],
 116                         [ $this->pageId1 + 1, $this->pageId1 + 2, $this->pageId1 + 3 ],
 117                         "Page ids increasing without holes" );
 118         }
 119
 120         function testPlain() {
 121                 // Setting up the dump
 122                 $nameStub = $this->setUpStub();
 123                 $nameFull = $this->getNewTempFile();
 124                 $dumper = new TextPassDumper( [ "--stub=file:" . $nameStub,
 125                         "--output=file:" . $nameFull ] );
 126                 $dumper->reporting = false;
 127                 $dumper->setDB( $this->db );
 128
 129                 // Performing the dump
 130                 $dumper->dump( WikiExporter::FULL, WikiExporter::TEXT );
 131
 132                 // Checking for correctness of the dumped data
 133                 $this->assertDumpStart( $nameFull );
 134
 135                 // Page 1
 136                 $this->assertPageStart( $this->pageId1, NS_MAIN, "BackupDumperTestP1" );
 137                 $this->assertRevision( $this->revId1_1, "BackupDumperTestP1Summary1",
 138                         $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 139                         "BackupDumperTestP1Text1" );
 140                 $this->assertPageEnd();
 141
 142                 // Page 2
 143                 $this->assertPageStart( $this->pageId2, NS_MAIN, "BackupDumperTestP2" );
 144                 $this->assertRevision( $this->revId2_1, "BackupDumperTestP2Summary1",
 145                         $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 146                         "BackupDumperTestP2Text1" );
 147                 $this->assertRevision( $this->revId2_2, "BackupDumperTestP2Summary2",
 148                         $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 149                         "BackupDumperTestP2Text2", $this->revId2_1 );
 150                 $this->assertRevision( $this->revId2_3, "BackupDumperTestP2Summary3",
 151                         $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 152                         "BackupDumperTestP2Text3", $this->revId2_2 );
 153                 $this->assertRevision( $this->revId2_4, "BackupDumperTestP2Summary4 extra",
 154                         $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 155                         "BackupDumperTestP2Text4 some additional Text", $this->revId2_3 );
 156                 $this->assertPageEnd();
 157
 158                 // Page 3
 159                 // -> Page is marked deleted. Hence not visible
 160
 161                 // Page 4
 162                 $this->assertPageStart( $this->pageId4, NS_TALK, "Talk:BackupDumperTestP1" );
 163                 $this->assertRevision( $this->revId4_1, "Talk BackupDumperTestP1 Summary1",
 164                         $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 165                         "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 166                         false,
 167                         "BackupTextPassTestModel",
 168                         "text/plain" );
 169                 $this->assertPageEnd();
 170
 171                 $this->assertDumpEnd();
 172         }
 173
 174         function testPrefetchPlain() {
 175                 // The mapping between ids and text, for the hits of the prefetch mock
 176                 $prefetchMap = [
 177                         [ $this->pageId1, $this->revId1_1, "Prefetch_________1Text1" ],
 178                         [ $this->pageId2, $this->revId2_3, "Prefetch_________2Text3" ]
 179                 ];
 180
 181                 // The mock itself
 182                 $prefetchMock = $this->getMockBuilder( BaseDump::class )
 183                         ->setMethods( [ 'prefetch' ] )
 184                         ->disableOriginalConstructor()
 185                         ->getMock();
 186                 $prefetchMock->expects( $this->exactly( 6 ) )
 187                         ->method( 'prefetch' )
 188                         ->will( $this->returnValueMap( $prefetchMap ) );
 189
 190                 // Setting up of the dump
 191                 $nameStub = $this->setUpStub();
 192                 $nameFull = $this->getNewTempFile();
 193
 194                 $dumper = new TextPassDumper( [ "--stub=file:" . $nameStub,
 195                         "--output=file:" . $nameFull ] );
 196
 197                 $dumper->prefetch = $prefetchMock;
 198                 $dumper->reporting = false;
 199                 $dumper->setDB( $this->db );
 200
 201                 // Performing the dump
 202                 $dumper->dump( WikiExporter::FULL, WikiExporter::TEXT );
 203
 204                 // Checking for correctness of the dumped data
 205                 $this->assertDumpStart( $nameFull );
 206
 207                 // Page 1
 208                 $this->assertPageStart( $this->pageId1, NS_MAIN, "BackupDumperTestP1" );
 209                 // Prefetch kicks in. This is still the SHA-1 of the original text,
 210                 // But the actual text (with different SHA-1) comes from prefetch.
 211                 $this->assertRevision( $this->revId1_1, "BackupDumperTestP1Summary1",
 212                         $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 213                         "Prefetch_________1Text1" );
 214                 $this->assertPageEnd();
 215
 216                 // Page 2
 217                 $this->assertPageStart( $this->pageId2, NS_MAIN, "BackupDumperTestP2" );
 218                 $this->assertRevision( $this->revId2_1, "BackupDumperTestP2Summary1",
 219                         $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 220                         "BackupDumperTestP2Text1" );
 221                 $this->assertRevision( $this->revId2_2, "BackupDumperTestP2Summary2",
 222                         $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 223                         "BackupDumperTestP2Text2", $this->revId2_1 );
 224                 // Prefetch kicks in. This is still the SHA-1 of the original text,
 225                 // But the actual text (with different SHA-1) comes from prefetch.
 226                 $this->assertRevision( $this->revId2_3, "BackupDumperTestP2Summary3",
 227                         $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 228                         "Prefetch_________2Text3", $this->revId2_2 );
 229                 $this->assertRevision( $this->revId2_4, "BackupDumperTestP2Summary4 extra",
 230                         $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 231                         "BackupDumperTestP2Text4 some additional Text", $this->revId2_3 );
 232                 $this->assertPageEnd();
 233
 234                 // Page 3
 235                 // -> Page is marked deleted. Hence not visible
 236
 237                 // Page 4
 238                 $this->assertPageStart( $this->pageId4, NS_TALK, "Talk:BackupDumperTestP1" );
 239                 $this->assertRevision( $this->revId4_1, "Talk BackupDumperTestP1 Summary1",
 240                         $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 241                         "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 242                         false,
 243                         "BackupTextPassTestModel",
 244                         "text/plain" );
 245                 $this->assertPageEnd();
 246
 247                 $this->assertDumpEnd();
 248         }
 249
 250         /**
 251          * Ensures that checkpoint dumps are used and written, by successively increasing the
 252          * stub size and dumping until the duration crosses a threshold.
 253          *
 254          * @param string $checkpointFormat Either "file" for plain text or "gzip" for gzipped
 255          *   checkpoint files.
 256          */
 257         private function checkpointHelper( $checkpointFormat = "file" ) {
 258                 // Getting temporary names
 259                 $nameStub = $this->getNewTempFile();
 260                 $nameOutputDir = $this->getNewTempDirectory();
 261
 262                 $stderr = fopen( 'php://output', 'a' );
 263                 if ( $stderr === false ) {
 264                         $this->fail( "Could not open stream for stderr" );
 265                 }
 266
 267                 $iterations = 32; // We'll start with that many iterations of revisions
 268                 // in stub. Make sure that the generated volume is above the buffer size
 269                 // set below. Otherwise, the checkpointing does not trigger.
 270                 $lastDuration = 0;
 271                 $minDuration = 2; // We want the dump to take at least this many seconds
 272                 $checkpointAfter = 0.5; // Generate checkpoint after this many seconds
 273
 274                 // Until a dump takes at least $minDuration seconds, perform a dump and check
 275                 // duration. If the dump did not take long enough increase the iteration
 276                 // count, to generate a bigger stub file next time.
 277                 while ( $lastDuration < $minDuration ) {
 278                         // Setting up the dump
 279                         wfRecursiveRemoveDir( $nameOutputDir );
 280                         $this->assertTrue( wfMkdirParents( $nameOutputDir ),
 281                                 "Creating temporary output directory " );
 282                         $this->setUpStub( $nameStub, $iterations );
 283                         $dumper = new TextPassDumper();
 284                         $dumper->loadWithArgv( [ "--stub=file:" . $nameStub,
 285                                 "--output=" . $checkpointFormat . ":" . $nameOutputDir . "/full",
 286                                 "--maxtime=1" /*This is in minutes. Fixup is below*/,
 287                                 "--buffersize=32768", // The default of 32 iterations fill up 32KB about twice
 288                                 "--checkpointfile=checkpoint-%s-%s.xml.gz" ] );
 289                         $dumper->setDB( $this->db );
 290                         $dumper->maxTimeAllowed = $checkpointAfter; // Patching maxTime from 1 minute
 291                         $dumper->stderr = $stderr;
 292
 293                         // The actual dump and taking time
 294                         $ts_before = microtime( true );
 295                         $dumper->execute();
 296                         $ts_after = microtime( true );
 297                         $lastDuration = $ts_after - $ts_before;
 298
 299                         // Handling increasing the iteration count for the stubs
 300                         if ( $lastDuration < $minDuration ) {
 301                                 $old_iterations = $iterations;
 302                                 if ( $lastDuration > 0.2 ) {
 303                                         // lastDuration is big enough, to allow an educated guess
 304                                         $factor = ( $minDuration + 0.5 ) / $lastDuration;
 305                                         if ( ( $factor > 1.1 ) && ( $factor < 100 ) ) {
 306                                                 // educated guess is reasonable
 307                                                 $iterations = (int)( $iterations * $factor );
 308                                         }
 309                                 }
 310
 311                                 if ( $old_iterations == $iterations ) {
 312                                         // Heuristics were not applied, so we just *2.
 313                                         $iterations *= 2;
 314                                 }
 315
 316                                 $this->assertLessThan( 50000, $iterations,
 317                                         "Emergency stop against infinitely increasing iteration "
 318                                                 . "count ( last duration: $lastDuration )" );
 319                         }
 320                 }
 321
 322                 // The dump (hopefully) did take long enough to produce more than one
 323                 // checkpoint file.
 324                 // We now check all the checkpoint files for validity.
 325
 326                 $files = scandir( $nameOutputDir );
 327                 $this->assertTrue( asort( $files ), "Sorting files in temporary directory" );
 328                 $fileOpened = false;
 329                 $lookingForPage = 1;
 330                 $checkpointFiles = 0;
 331
 332                 // Each run of the following loop body tries to handle exactly 1 /page/ (not
 333                 // iteration of stub content). $i is only increased after having treated page 4.
 334                 for ( $i = 0; $i < $iterations; ) {
 335                         // 1. Assuring a file is opened and ready. Skipping across header if
 336                         //    necessary.
 337                         if ( !$fileOpened ) {
 338                                 $this->assertNotEmpty( $files, "No more existing dump files, "
 339                                         . "but not yet all pages found" );
 340                                 $fname = array_shift( $files );
 341                                 while ( $fname == "." || $fname == ".." ) {
 342                                         $this->assertNotEmpty( $files, "No more existing dump"
 343                                                 . " files, but not yet all pages found" );
 344                                         $fname = array_shift( $files );
 345                                 }
 346                                 if ( $checkpointFormat == "gzip" ) {
 347                                         $this->gunzip( $nameOutputDir . "/" . $fname );
 348                                 }
 349                                 $this->assertDumpStart( $nameOutputDir . "/" . $fname );
 350                                 $fileOpened = true;
 351                                 $checkpointFiles++;
 352                         }
 353
 354                         // 2. Performing a single page check
 355                         switch ( $lookingForPage ) {
 356                                 case 1:
 357                                         // Page 1
 358                                         $this->assertPageStart( $this->pageId1 + $i * self::$numOfPages, NS_MAIN,
 359                                                 "BackupDumperTestP1" );
 360                                         $this->assertRevision( $this->revId1_1 + $i * self::$numOfRevs, "BackupDumperTestP1Summary1",
 361                                                 $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 362                                                 "BackupDumperTestP1Text1" );
 363                                         $this->assertPageEnd();
 364
 365                                         $lookingForPage = 2;
 366                                         break;
 367
 368                                 case 2:
 369                                         // Page 2
 370                                         $this->assertPageStart( $this->pageId2 + $i * self::$numOfPages, NS_MAIN,
 371                                                 "BackupDumperTestP2" );
 372                                         $this->assertRevision( $this->revId2_1 + $i * self::$numOfRevs, "BackupDumperTestP2Summary1",
 373                                                 $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 374                                                 "BackupDumperTestP2Text1" );
 375                                         $this->assertRevision( $this->revId2_2 + $i * self::$numOfRevs, "BackupDumperTestP2Summary2",
 376                                                 $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 377                                                 "BackupDumperTestP2Text2", $this->revId2_1 + $i * self::$numOfRevs );
 378                                         $this->assertRevision( $this->revId2_3 + $i * self::$numOfRevs, "BackupDumperTestP2Summary3",
 379                                                 $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 380                                                 "BackupDumperTestP2Text3", $this->revId2_2 + $i * self::$numOfRevs );
 381                                         $this->assertRevision( $this->revId2_4 + $i * self::$numOfRevs,
 382                                                 "BackupDumperTestP2Summary4 extra",
 383                                                 $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 384                                                 "BackupDumperTestP2Text4 some additional Text",
 385                                                 $this->revId2_3 + $i * self::$numOfRevs );
 386                                         $this->assertPageEnd();
 387
 388                                         $lookingForPage = 4;
 389                                         break;
 390
 391                                 case 4:
 392                                         // Page 4
 393                                         $this->assertPageStart( $this->pageId4 + $i * self::$numOfPages, NS_TALK,
 394                                                 "Talk:BackupDumperTestP1" );
 395                                         $this->assertRevision( $this->revId4_1 + $i * self::$numOfRevs,
 396                                                 "Talk BackupDumperTestP1 Summary1",
 397                                                 $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 398                                                 "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 399                                                 false,
 400                                                 "BackupTextPassTestModel",
 401                                                 "text/plain" );
 402                                         $this->assertPageEnd();
 403
 404                                         $lookingForPage = 1;
 405
 406                                         // We dealt with the whole iteration.
 407                                         $i++;
 408                                         break;
 409
 410                                 default:
 411                                         $this->fail( "Bad setting for lookingForPage ($lookingForPage)" );
 412                         }
 413
 414                         // 3. Checking for the end of the current checkpoint file
 415                         if ( $this->xml->nodeType == XMLReader::END_ELEMENT
 416                                 && $this->xml->name == "mediawiki"
 417                         ) {
 418                                 $this->assertDumpEnd();
 419                                 $fileOpened = false;
 420                         }
 421                 }
 422
 423                 // Assuring we completely read all files ...
 424                 $this->assertFalse( $fileOpened, "Currently read file still open?" );
 425                 $this->assertEmpty( $files, "Remaining unchecked files" );
 426
 427                 // ... and have dealt with more than one checkpoint file
 428                 $this->assertGreaterThan(
 429                         1,
 430                         $checkpointFiles,
 431                         "expected more than 1 checkpoint to have been created. "
 432                                 . "Checkpoint interval is $checkpointAfter seconds, maybe your computer is too fast?"
 433                 );
 434
 435                 $this->expectETAOutput();
 436         }
 437
 438         /**
 439          * Broken per T70653.
 440          *
 441          * @group large
 442          * @group Broken
 443          */
 444         function testCheckpointPlain() {
 445                 $this->checkpointHelper();
 446         }
 447
 448         /**
 449          * tests for working checkpoint generation in gzip format work.
 450          *
 451          * We keep this test in addition to the simpler self::testCheckpointPlain, as there
 452          * were once problems when the used sinks were DumpPipeOutputs.
 453          *
 454          * xmldumps-backup typically uses bzip2 instead of gzip. However, as bzip2 requires
 455          * PHP extensions, we go for gzip instead, which triggers the same relevant code
 456          * paths while still being testable on more systems.
 457          *
 458          * Broken per T70653.
 459          *
 460          * @group large
 461          * @group Broken
 462          */
 463         function testCheckpointGzip() {
 464                 $this->checkHasGzip();
 465                 $this->checkpointHelper( "gzip" );
 466         }
 467
 468         /**
 469          * Creates a stub file that is used for testing the text pass of dumps
 470          *
 471          * @param string $fname (Optional) Absolute name of the file to write
 472          *   the stub into. If this parameter is null, a new temporary
 473          *   file is generated that is automatically removed upon tearDown.
 474          * @param int $iterations (Optional) specifies how often the block
 475          *   of 3 pages should go into the stub file. The page and
 476          *   revision id increase further and further, while the text
 477          *   id of the first iteration is reused. The pages and revision
 478          *   of iteration > 1 have no corresponding representation in the database.
 479          * @return string Absolute filename of the stub
 480          */
 481         private function setUpStub( $fname = null, $iterations = 1 ) {
 482                 if ( $fname === null ) {
 483                         $fname = $this->getNewTempFile();
 484                 }
 485                 $header = '<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.10/" '
 486                         . 'xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" '
 487                         . 'xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.10/ '
 488                         . 'http://www.mediawiki.org/xml/export-0.10.xsd" version="0.10" xml:lang="en">
 489   <siteinfo>
 490     <sitename>wikisvn</sitename>
 491     <base>http://localhost/wiki-svn/index.php/Main_Page</base>
 492     <generator>MediaWiki 1.21alpha</generator>
 493     <case>first-letter</case>
 494     <namespaces>
 495       <namespace key="-2" case="first-letter">Media</namespace>
 496       <namespace key="-1" case="first-letter">Special</namespace>
 497       <namespace key="0" case="first-letter" />
 498       <namespace key="1" case="first-letter">Talk</namespace>
 499       <namespace key="2" case="first-letter">User</namespace>
 500       <namespace key="3" case="first-letter">User talk</namespace>
 501       <namespace key="4" case="first-letter">Wikisvn</namespace>
 502       <namespace key="5" case="first-letter">Wikisvn talk</namespace>
 503       <namespace key="6" case="first-letter">File</namespace>
 504       <namespace key="7" case="first-letter">File talk</namespace>
 505       <namespace key="8" case="first-letter">MediaWiki</namespace>
 506       <namespace key="9" case="first-letter">MediaWiki talk</namespace>
 507       <namespace key="10" case="first-letter">Template</namespace>
 508       <namespace key="11" case="first-letter">Template talk</namespace>
 509       <namespace key="12" case="first-letter">Help</namespace>
 510       <namespace key="13" case="first-letter">Help talk</namespace>
 511       <namespace key="14" case="first-letter">Category</namespace>
 512       <namespace key="15" case="first-letter">Category talk</namespace>
 513     </namespaces>
 514   </siteinfo>
 515 ';
 516                 $tail = '</mediawiki>
 517 ';
 518
 519                 $content = $header;
 520                 $iterations = intval( $iterations );
 521                 for ( $i = 0; $i < $iterations; $i++ ) {
 522                         $page1 = '  <page>
 523     <title>BackupDumperTestP1</title>
 524     <ns>0</ns>
 525     <id>' . ( $this->pageId1 + $i * self::$numOfPages ) . '</id>
 526     <revision>
 527       <id>' . ( $this->revId1_1 + $i * self::$numOfRevs ) . '</id>
 528       <timestamp>2012-04-01T16:46:05Z</timestamp>
 529       <contributor>
 530         <ip>127.0.0.1</ip>
 531       </contributor>
 532       <comment>BackupDumperTestP1Summary1</comment>
 533       <model>wikitext</model>
 534       <format>text/x-wiki</format>
 535       <text id="' . $this->textId1_1 . '" bytes="23" />
 536       <sha1>0bolhl6ol7i6x0e7yq91gxgaan39j87</sha1>
 537     </revision>
 538   </page>
 539 ';
 540                         $page2 = '  <page>
 541     <title>BackupDumperTestP2</title>
 542     <ns>0</ns>
 543     <id>' . ( $this->pageId2 + $i * self::$numOfPages ) . '</id>
 544     <revision>
 545       <id>' . ( $this->revId2_1 + $i * self::$numOfRevs ) . '</id>
 546       <timestamp>2012-04-01T16:46:05Z</timestamp>
 547       <contributor>
 548         <ip>127.0.0.1</ip>
 549       </contributor>
 550       <comment>BackupDumperTestP2Summary1</comment>
 551       <model>wikitext</model>
 552       <format>text/x-wiki</format>
 553       <text id="' . $this->textId2_1 . '" bytes="23" />
 554       <sha1>jprywrymfhysqllua29tj3sc7z39dl2</sha1>
 555     </revision>
 556     <revision>
 557       <id>' . ( $this->revId2_2 + $i * self::$numOfRevs ) . '</id>
 558       <parentid>' . ( $this->revId2_1 + $i * self::$numOfRevs ) . '</parentid>
 559       <timestamp>2012-04-01T16:46:05Z</timestamp>
 560       <contributor>
 561         <ip>127.0.0.1</ip>
 562       </contributor>
 563       <comment>BackupDumperTestP2Summary2</comment>
 564       <model>wikitext</model>
 565       <format>text/x-wiki</format>
 566       <text id="' . $this->textId2_2 . '" bytes="23" />
 567       <sha1>b7vj5ks32po5m1z1t1br4o7scdwwy95</sha1>
 568     </revision>
 569     <revision>
 570       <id>' . ( $this->revId2_3 + $i * self::$numOfRevs ) . '</id>
 571       <parentid>' . ( $this->revId2_2 + $i * self::$numOfRevs ) . '</parentid>
 572       <timestamp>2012-04-01T16:46:05Z</timestamp>
 573       <contributor>
 574         <ip>127.0.0.1</ip>
 575       </contributor>
 576       <comment>BackupDumperTestP2Summary3</comment>
 577       <model>wikitext</model>
 578       <format>text/x-wiki</format>
 579       <text id="' . $this->textId2_3 . '" bytes="23" />
 580       <sha1>jfunqmh1ssfb8rs43r19w98k28gg56r</sha1>
 581     </revision>
 582     <revision>
 583       <id>' . ( $this->revId2_4 + $i * self::$numOfRevs ) . '</id>
 584       <parentid>' . ( $this->revId2_3 + $i * self::$numOfRevs ) . '</parentid>
 585       <timestamp>2012-04-01T16:46:05Z</timestamp>
 586       <contributor>
 587         <ip>127.0.0.1</ip>
 588       </contributor>
 589       <comment>BackupDumperTestP2Summary4 extra</comment>
 590       <model>wikitext</model>
 591       <format>text/x-wiki</format>
 592       <text id="' . $this->textId2_4 . '" bytes="44" />
 593       <sha1>6o1ciaxa6pybnqprmungwofc4lv00wv</sha1>
 594     </revision>
 595   </page>
 596 ';
 597                         // page 3 not in stub
 598
 599                         $page4 = '  <page>
 600     <title>Talk:BackupDumperTestP1</title>
 601     <ns>1</ns>
 602     <id>' . ( $this->pageId4 + $i * self::$numOfPages ) . '</id>
 603     <revision>
 604       <id>' . ( $this->revId4_1 + $i * self::$numOfRevs ) . '</id>
 605       <timestamp>2012-04-01T16:46:05Z</timestamp>
 606       <contributor>
 607         <ip>127.0.0.1</ip>
 608       </contributor>
 609       <comment>Talk BackupDumperTestP1 Summary1</comment>
 610       <model>BackupTextPassTestModel</model>
 611       <format>text/plain</format>
 612       <text id="' . $this->textId4_1 . '" bytes="35" />
 613       <sha1>nktofwzd0tl192k3zfepmlzxoax1lpe</sha1>
 614     </revision>
 615   </page>
 616 ';
 617                         $content .= $page1 . $page2 . $page4;
 618                 }
 619                 $content .= $tail;
 620                 $this->assertEquals( strlen( $content ), file_put_contents(
 621                         $fname, $content ), "Length of prepared stub" );
 622
 623                 return $fname;
 624         }
 625 }
 626
 627 class BackupTextPassTestModelHandler extends TextContentHandler {
 628
 629         public function __construct() {
 630                 parent::__construct( 'BackupTextPassTestModel' );
 631         }
 632
 633         public function exportTransform( $text, $format = null ) {
 634                 return strtoupper( $text );
 635         }
 636
 637 }
 638
 639 /**
 640  * Tests for TextPassDumper that do not rely on the database
 641  *
 642  * (As the Database group is only detected at class level (not method level), we
 643  * cannot bring this test case's tests into the above main test case.)
 644  *
 645  * @group Dump
 646  * @covers TextPassDumper
 647  */
 648 class TextPassDumperDatabaselessTest extends MediaWikiLangTestCase {
 649         /**
 650          * Ensures that setting the buffer size is effective.
 651          *
 652          * @dataProvider bufferSizeProvider
 653          */
 654         function testBufferSizeSetting( $expected, $size, $msg ) {
 655                 $dumper = new TextPassDumperAccessor();
 656                 $dumper->loadWithArgv( [ "--buffersize=" . $size ] );
 657                 $dumper->execute();
 658                 $this->assertEquals( $expected, $dumper->getBufferSize(), $msg );
 659         }
 660
 661         /**
 662          * Ensures that setting the buffer size is effective.
 663          *
 664          * @dataProvider bufferSizeProvider
 665          */
 666         function bufferSizeProvider() {
 667                 // expected, bufferSize to initialize with, message
 668                 return [
 669                         [ 512 * 1024, 512 * 1024, "Setting 512KB is not effective" ],
 670                         [ 8192, 8192, "Setting 8KB is not effective" ],
 671                         [ 4096, 2048, "Could set buffer size below lower bound" ]
 672                 ];
 673         }
 674 }
 675
 676 /**
 677  * Accessor for internal state of TextPassDumper
 678  *
 679  * Do not warrentless add getters here.
 680  */
 681 class TextPassDumperAccessor extends TextPassDumper {
 682         /**
 683          * Gets the bufferSize.
 684          *
 685          * If bufferSize setting does not work correctly, testCheckpoint... tests
 686          * fail and point in the wrong direction. To aid in troubleshooting when
 687          * testCheckpoint... tests break at some point in the future, we test the
 688          * bufferSize setting, hence need this accessor.
 689          *
 690          * (Yes, bufferSize is internal state of the TextPassDumper, but aiding
 691          * debugging of testCheckpoint... in the future seems to be worth testing
 692          * against it nonetheless.)
 693          */
 694         public function getBufferSize() {
 695                 return $this->bufferSize;
 696         }
 697
 698         function dump( $history, $text = null ) {
 699                 return true;
 700         }
 701 }