tests/phpunit/maintenance/backupTextPassTest.php

   1 <?php
   2
   3 namespace MediaWiki\Tests\Maintenance;
   4
   5 use Exception;
   6 use MediaWikiLangTestCase;
   7 use MWException;
   8 use TextContentHandler;
   9 use TextPassDumper;
  10 use Title;
  11 use WikiExporter;
  12 use WikiPage;
  13
  14 require_once __DIR__ . "/../../../maintenance/dumpTextPass.php";
  15
  16 /**
  17  * Tests for TextPassDumper that rely on the database
  18  *
  19  * Some of these tests use the old constuctor for TextPassDumper
  20  * and the dump() function, while others use the new loadWithArgv( $args )
  21  * function and execute(). This is to ensure both the old and new methods
  22  * work properly.
  23  *
  24  * @group Database
  25  * @group Dump
  26  * @covers TextPassDumper
  27  */
  28 class TextPassDumperDatabaseTest extends DumpTestCase {
  29
  30         // We'll add several pages, revision and texts. The following variables hold the
  31         // corresponding ids.
  32         private $pageId1, $pageId2, $pageId3, $pageId4;
  33         private static $numOfPages = 4;
  34         private $revId1_1, $textId1_1;
  35         private $revId2_1, $textId2_1, $revId2_2, $textId2_2;
  36         private $revId2_3, $textId2_3, $revId2_4, $textId2_4;
  37         private $revId3_1, $textId3_1, $revId3_2, $textId3_2;
  38         private $revId4_1, $textId4_1;
  39         private static $numOfRevs = 8;
  40
  41         function addDBData() {
  42                 $this->tablesUsed[] = 'page';
  43                 $this->tablesUsed[] = 'revision';
  44                 $this->tablesUsed[] = 'ip_changes';
  45                 $this->tablesUsed[] = 'text';
  46
  47                 $this->mergeMwGlobalArrayValue( 'wgContentHandlers', [
  48                         "BackupTextPassTestModel" => BackupTextPassTestModelHandler::class,
  49                 ] );
  50
  51                 $ns = $this->getDefaultWikitextNS();
  52
  53                 try {
  54                         // Simple page
  55                         $title = Title::newFromText( 'BackupDumperTestP1', $ns );
  56                         $page = WikiPage::factory( $title );
  57                         list( $this->revId1_1, $this->textId1_1 ) = $this->addRevision( $page,
  58                                 "BackupDumperTestP1Text1", "BackupDumperTestP1Summary1" );
  59                         $this->pageId1 = $page->getId();
  60
  61                         // Page with more than one revision
  62                         $title = Title::newFromText( 'BackupDumperTestP2', $ns );
  63                         $page = WikiPage::factory( $title );
  64                         list( $this->revId2_1, $this->textId2_1 ) = $this->addRevision( $page,
  65                                 "BackupDumperTestP2Text1", "BackupDumperTestP2Summary1" );
  66                         list( $this->revId2_2, $this->textId2_2 ) = $this->addRevision( $page,
  67                                 "BackupDumperTestP2Text2", "BackupDumperTestP2Summary2" );
  68                         list( $this->revId2_3, $this->textId2_3 ) = $this->addRevision( $page,
  69                                 "BackupDumperTestP2Text3", "BackupDumperTestP2Summary3" );
  70                         list( $this->revId2_4, $this->textId2_4 ) = $this->addRevision( $page,
  71                                 "BackupDumperTestP2Text4 some additional Text  ",
  72                                 "BackupDumperTestP2Summary4 extra " );
  73                         $this->pageId2 = $page->getId();
  74
  75                         // Deleted page.
  76                         $title = Title::newFromText( 'BackupDumperTestP3', $ns );
  77                         $page = WikiPage::factory( $title );
  78                         list( $this->revId3_1, $this->textId3_1 ) = $this->addRevision( $page,
  79                                 "BackupDumperTestP3Text1", "BackupDumperTestP2Summary1" );
  80                         list( $this->revId3_2, $this->textId3_2 ) = $this->addRevision( $page,
  81                                 "BackupDumperTestP3Text2", "BackupDumperTestP2Summary2" );
  82                         $this->pageId3 = $page->getId();
  83                         $page->doDeleteArticle( "Testing ;)" );
  84
  85                         // Page from non-default namespace and model.
  86                         // ExportTransform applies.
  87
  88                         if ( $ns === NS_TALK ) {
  89                                 // @todo work around this.
  90                                 throw new MWException( "The default wikitext namespace is the talk namespace. "
  91                                         . " We can't currently deal with that." );
  92                         }
  93
  94                         $title = Title::newFromText( 'BackupDumperTestP1', NS_TALK );
  95                         $page = WikiPage::factory( $title );
  96                         list( $this->revId4_1, $this->textId4_1 ) = $this->addRevision( $page,
  97                                 "Talk about BackupDumperTestP1 Text1",
  98                                 "Talk BackupDumperTestP1 Summary1",
  99                                 "BackupTextPassTestModel" );
 100                         $this->pageId4 = $page->getId();
 101                 } catch ( Exception $e ) {
 102                         // We'd love to pass $e directly. However, ... see
 103                         // documentation of exceptionFromAddDBData in
 104                         // DumpTestCase
 105                         $this->exceptionFromAddDBData = $e;
 106                 }
 107         }
 108
 109         protected function setUp() {
 110                 parent::setUp();
 111
 112                 // Since we will restrict dumping by page ranges (to allow
 113                 // working tests, even if the db gets prepopulated by a base
 114                 // class), we have to assert, that the page id are consecutively
 115                 // increasing
 116                 $this->assertEquals(
 117                         [ $this->pageId2, $this->pageId3, $this->pageId4 ],
 118                         [ $this->pageId1 + 1, $this->pageId1 + 2, $this->pageId1 + 3 ],
 119                         "Page ids increasing without holes" );
 120         }
 121
 122         function testPlain() {
 123                 // Setting up the dump
 124                 $nameStub = $this->setUpStub();
 125                 $nameFull = $this->getNewTempFile();
 126                 $dumper = new TextPassDumper( [ "--stub=file:" . $nameStub,
 127                         "--output=file:" . $nameFull ] );
 128                 $dumper->reporting = false;
 129                 $dumper->setDB( $this->db );
 130
 131                 // Performing the dump
 132                 $dumper->dump( WikiExporter::FULL, WikiExporter::TEXT );
 133
 134                 // Checking for correctness of the dumped data
 135                 $this->assertDumpStart( $nameFull );
 136
 137                 // Page 1
 138                 $this->assertPageStart( $this->pageId1, NS_MAIN, "BackupDumperTestP1" );
 139                 $this->assertRevision( $this->revId1_1, "BackupDumperTestP1Summary1",
 140                         $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 141                         "BackupDumperTestP1Text1" );
 142                 $this->assertPageEnd();
 143
 144                 // Page 2
 145                 $this->assertPageStart( $this->pageId2, NS_MAIN, "BackupDumperTestP2" );
 146                 $this->assertRevision( $this->revId2_1, "BackupDumperTestP2Summary1",
 147                         $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 148                         "BackupDumperTestP2Text1" );
 149                 $this->assertRevision( $this->revId2_2, "BackupDumperTestP2Summary2",
 150                         $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 151                         "BackupDumperTestP2Text2", $this->revId2_1 );
 152                 $this->assertRevision( $this->revId2_3, "BackupDumperTestP2Summary3",
 153                         $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 154                         "BackupDumperTestP2Text3", $this->revId2_2 );
 155                 $this->assertRevision( $this->revId2_4, "BackupDumperTestP2Summary4 extra",
 156                         $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 157                         "BackupDumperTestP2Text4 some additional Text", $this->revId2_3 );
 158                 $this->assertPageEnd();
 159
 160                 // Page 3
 161                 // -> Page is marked deleted. Hence not visible
 162
 163                 // Page 4
 164                 $this->assertPageStart( $this->pageId4, NS_TALK, "Talk:BackupDumperTestP1" );
 165                 $this->assertRevision( $this->revId4_1, "Talk BackupDumperTestP1 Summary1",
 166                         $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 167                         "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 168                         false,
 169                         "BackupTextPassTestModel",
 170                         "text/plain" );
 171                 $this->assertPageEnd();
 172
 173                 $this->assertDumpEnd();
 174         }
 175
 176         function testPrefetchPlain() {
 177                 // The mapping between ids and text, for the hits of the prefetch mock
 178                 $prefetchMap = [
 179                         [ $this->pageId1, $this->revId1_1, "Prefetch_________1Text1" ],
 180                         [ $this->pageId2, $this->revId2_3, "Prefetch_________2Text3" ]
 181                 ];
 182
 183                 // The mock itself
 184                 $prefetchMock = $this->getMockBuilder( BaseDump::class )
 185                         ->setMethods( [ 'prefetch' ] )
 186                         ->disableOriginalConstructor()
 187                         ->getMock();
 188                 $prefetchMock->expects( $this->exactly( 6 ) )
 189                         ->method( 'prefetch' )
 190                         ->will( $this->returnValueMap( $prefetchMap ) );
 191
 192                 // Setting up of the dump
 193                 $nameStub = $this->setUpStub();
 194                 $nameFull = $this->getNewTempFile();
 195
 196                 $dumper = new TextPassDumper( [ "--stub=file:" . $nameStub,
 197                         "--output=file:" . $nameFull ] );
 198
 199                 $dumper->prefetch = $prefetchMock;
 200                 $dumper->reporting = false;
 201                 $dumper->setDB( $this->db );
 202
 203                 // Performing the dump
 204                 $dumper->dump( WikiExporter::FULL, WikiExporter::TEXT );
 205
 206                 // Checking for correctness of the dumped data
 207                 $this->assertDumpStart( $nameFull );
 208
 209                 // Page 1
 210                 $this->assertPageStart( $this->pageId1, NS_MAIN, "BackupDumperTestP1" );
 211                 // Prefetch kicks in. This is still the SHA-1 of the original text,
 212                 // But the actual text (with different SHA-1) comes from prefetch.
 213                 $this->assertRevision( $this->revId1_1, "BackupDumperTestP1Summary1",
 214                         $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 215                         "Prefetch_________1Text1" );
 216                 $this->assertPageEnd();
 217
 218                 // Page 2
 219                 $this->assertPageStart( $this->pageId2, NS_MAIN, "BackupDumperTestP2" );
 220                 $this->assertRevision( $this->revId2_1, "BackupDumperTestP2Summary1",
 221                         $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 222                         "BackupDumperTestP2Text1" );
 223                 $this->assertRevision( $this->revId2_2, "BackupDumperTestP2Summary2",
 224                         $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 225                         "BackupDumperTestP2Text2", $this->revId2_1 );
 226                 // Prefetch kicks in. This is still the SHA-1 of the original text,
 227                 // But the actual text (with different SHA-1) comes from prefetch.
 228                 $this->assertRevision( $this->revId2_3, "BackupDumperTestP2Summary3",
 229                         $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 230                         "Prefetch_________2Text3", $this->revId2_2 );
 231                 $this->assertRevision( $this->revId2_4, "BackupDumperTestP2Summary4 extra",
 232                         $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 233                         "BackupDumperTestP2Text4 some additional Text", $this->revId2_3 );
 234                 $this->assertPageEnd();
 235
 236                 // Page 3
 237                 // -> Page is marked deleted. Hence not visible
 238
 239                 // Page 4
 240                 $this->assertPageStart( $this->pageId4, NS_TALK, "Talk:BackupDumperTestP1" );
 241                 $this->assertRevision( $this->revId4_1, "Talk BackupDumperTestP1 Summary1",
 242                         $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 243                         "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 244                         false,
 245                         "BackupTextPassTestModel",
 246                         "text/plain" );
 247                 $this->assertPageEnd();
 248
 249                 $this->assertDumpEnd();
 250         }
 251
 252         /**
 253          * Ensures that checkpoint dumps are used and written, by successively increasing the
 254          * stub size and dumping until the duration crosses a threshold.
 255          *
 256          * @param string $checkpointFormat Either "file" for plain text or "gzip" for gzipped
 257          *   checkpoint files.
 258          */
 259         private function checkpointHelper( $checkpointFormat = "file" ) {
 260                 // Getting temporary names
 261                 $nameStub = $this->getNewTempFile();
 262                 $nameOutputDir = $this->getNewTempDirectory();
 263
 264                 $stderr = fopen( 'php://output', 'a' );
 265                 if ( $stderr === false ) {
 266                         $this->fail( "Could not open stream for stderr" );
 267                 }
 268
 269                 $iterations = 32; // We'll start with that many iterations of revisions
 270                 // in stub. Make sure that the generated volume is above the buffer size
 271                 // set below. Otherwise, the checkpointing does not trigger.
 272                 $lastDuration = 0;
 273                 $minDuration = 2; // We want the dump to take at least this many seconds
 274                 $checkpointAfter = 0.5; // Generate checkpoint after this many seconds
 275
 276                 // Until a dump takes at least $minDuration seconds, perform a dump and check
 277                 // duration. If the dump did not take long enough increase the iteration
 278                 // count, to generate a bigger stub file next time.
 279                 while ( $lastDuration < $minDuration ) {
 280                         // Setting up the dump
 281                         wfRecursiveRemoveDir( $nameOutputDir );
 282                         $this->assertTrue( wfMkdirParents( $nameOutputDir ),
 283                                 "Creating temporary output directory " );
 284                         $this->setUpStub( $nameStub, $iterations );
 285                         $dumper = new TextPassDumper();
 286                         $dumper->loadWithArgv( [ "--stub=file:" . $nameStub,
 287                                 "--output=" . $checkpointFormat . ":" . $nameOutputDir . "/full",
 288                                 "--maxtime=1" /*This is in minutes. Fixup is below*/,
 289                                 "--buffersize=32768", // The default of 32 iterations fill up 32KB about twice
 290                                 "--checkpointfile=checkpoint-%s-%s.xml.gz" ] );
 291                         $dumper->setDB( $this->db );
 292                         $dumper->maxTimeAllowed = $checkpointAfter; // Patching maxTime from 1 minute
 293                         $dumper->stderr = $stderr;
 294
 295                         // The actual dump and taking time
 296                         $ts_before = microtime( true );
 297                         $dumper->execute();
 298                         $ts_after = microtime( true );
 299                         $lastDuration = $ts_after - $ts_before;
 300
 301                         // Handling increasing the iteration count for the stubs
 302                         if ( $lastDuration < $minDuration ) {
 303                                 $old_iterations = $iterations;
 304                                 if ( $lastDuration > 0.2 ) {
 305                                         // lastDuration is big enough, to allow an educated guess
 306                                         $factor = ( $minDuration + 0.5 ) / $lastDuration;
 307                                         if ( ( $factor > 1.1 ) && ( $factor < 100 ) ) {
 308                                                 // educated guess is reasonable
 309                                                 $iterations = (int)( $iterations * $factor );
 310                                         }
 311                                 }
 312
 313                                 if ( $old_iterations == $iterations ) {
 314                                         // Heuristics were not applied, so we just *2.
 315                                         $iterations *= 2;
 316                                 }
 317
 318                                 $this->assertLessThan( 50000, $iterations,
 319                                         "Emergency stop against infinitely increasing iteration "
 320                                                 . "count ( last duration: $lastDuration )" );
 321                         }
 322                 }
 323
 324                 // The dump (hopefully) did take long enough to produce more than one
 325                 // checkpoint file.
 326                 // We now check all the checkpoint files for validity.
 327
 328                 $files = scandir( $nameOutputDir );
 329                 $this->assertTrue( asort( $files ), "Sorting files in temporary directory" );
 330                 $fileOpened = false;
 331                 $lookingForPage = 1;
 332                 $checkpointFiles = 0;
 333
 334                 // Each run of the following loop body tries to handle exactly 1 /page/ (not
 335                 // iteration of stub content). $i is only increased after having treated page 4.
 336                 for ( $i = 0; $i < $iterations; ) {
 337                         // 1. Assuring a file is opened and ready. Skipping across header if
 338                         //    necessary.
 339                         if ( !$fileOpened ) {
 340                                 $this->assertNotEmpty( $files, "No more existing dump files, "
 341                                         . "but not yet all pages found" );
 342                                 $fname = array_shift( $files );
 343                                 while ( $fname == "." || $fname == ".." ) {
 344                                         $this->assertNotEmpty( $files, "No more existing dump"
 345                                                 . " files, but not yet all pages found" );
 346                                         $fname = array_shift( $files );
 347                                 }
 348                                 if ( $checkpointFormat == "gzip" ) {
 349                                         $this->gunzip( $nameOutputDir . "/" . $fname );
 350                                 }
 351                                 $this->assertDumpStart( $nameOutputDir . "/" . $fname );
 352                                 $fileOpened = true;
 353                                 $checkpointFiles++;
 354                         }
 355
 356                         // 2. Performing a single page check
 357                         switch ( $lookingForPage ) {
 358                                 case 1:
 359                                         // Page 1
 360                                         $this->assertPageStart( $this->pageId1 + $i * self::$numOfPages, NS_MAIN,
 361                                                 "BackupDumperTestP1" );
 362                                         $this->assertRevision( $this->revId1_1 + $i * self::$numOfRevs, "BackupDumperTestP1Summary1",
 363                                                 $this->textId1_1, false, "0bolhl6ol7i6x0e7yq91gxgaan39j87",
 364                                                 "BackupDumperTestP1Text1" );
 365                                         $this->assertPageEnd();
 366
 367                                         $lookingForPage = 2;
 368                                         break;
 369
 370                                 case 2:
 371                                         // Page 2
 372                                         $this->assertPageStart( $this->pageId2 + $i * self::$numOfPages, NS_MAIN,
 373                                                 "BackupDumperTestP2" );
 374                                         $this->assertRevision( $this->revId2_1 + $i * self::$numOfRevs, "BackupDumperTestP2Summary1",
 375                                                 $this->textId2_1, false, "jprywrymfhysqllua29tj3sc7z39dl2",
 376                                                 "BackupDumperTestP2Text1" );
 377                                         $this->assertRevision( $this->revId2_2 + $i * self::$numOfRevs, "BackupDumperTestP2Summary2",
 378                                                 $this->textId2_2, false, "b7vj5ks32po5m1z1t1br4o7scdwwy95",
 379                                                 "BackupDumperTestP2Text2", $this->revId2_1 + $i * self::$numOfRevs );
 380                                         $this->assertRevision( $this->revId2_3 + $i * self::$numOfRevs, "BackupDumperTestP2Summary3",
 381                                                 $this->textId2_3, false, "jfunqmh1ssfb8rs43r19w98k28gg56r",
 382                                                 "BackupDumperTestP2Text3", $this->revId2_2 + $i * self::$numOfRevs );
 383                                         $this->assertRevision( $this->revId2_4 + $i * self::$numOfRevs,
 384                                                 "BackupDumperTestP2Summary4 extra",
 385                                                 $this->textId2_4, false, "6o1ciaxa6pybnqprmungwofc4lv00wv",
 386                                                 "BackupDumperTestP2Text4 some additional Text",
 387                                                 $this->revId2_3 + $i * self::$numOfRevs );
 388                                         $this->assertPageEnd();
 389
 390                                         $lookingForPage = 4;
 391                                         break;
 392
 393                                 case 4:
 394                                         // Page 4
 395                                         $this->assertPageStart( $this->pageId4 + $i * self::$numOfPages, NS_TALK,
 396                                                 "Talk:BackupDumperTestP1" );
 397                                         $this->assertRevision( $this->revId4_1 + $i * self::$numOfRevs,
 398                                                 "Talk BackupDumperTestP1 Summary1",
 399                                                 $this->textId4_1, false, "nktofwzd0tl192k3zfepmlzxoax1lpe",
 400                                                 "TALK ABOUT BACKUPDUMPERTESTP1 TEXT1",
 401                                                 false,
 402                                                 "BackupTextPassTestModel",
 403                                                 "text/plain" );
 404                                         $this->assertPageEnd();
 405
 406                                         $lookingForPage = 1;
 407
 408                                         // We dealt with the whole iteration.
 409                                         $i++;
 410                                         break;
 411
 412                                 default:
 413                                         $this->fail( "Bad setting for lookingForPage ($lookingForPage)" );
 414                         }
 415
 416                         // 3. Checking for the end of the current checkpoint file
 417                         if ( $this->xml->nodeType == XMLReader::END_ELEMENT
 418                                 && $this->xml->name == "mediawiki"
 419                         ) {
 420                                 $this->assertDumpEnd();
 421                                 $fileOpened = false;
 422                         }
 423                 }
 424
 425                 // Assuring we completely read all files ...
 426                 $this->assertFalse( $fileOpened, "Currently read file still open?" );
 427                 $this->assertEmpty( $files, "Remaining unchecked files" );
 428
 429                 // ... and have dealt with more than one checkpoint file
 430                 $this->assertGreaterThan(
 431                         1,
 432                         $checkpointFiles,
 433                         "expected more than 1 checkpoint to have been created. "
 434                                 . "Checkpoint interval is $checkpointAfter seconds, maybe your computer is too fast?"
 435                 );
 436
 437                 $this->expectETAOutput();
 438         }
 439
 440         /**
 441          * Broken per T70653.
 442          *
 443          * @group large
 444          * @group Broken
 445          */
 446         function testCheckpointPlain() {
 447                 $this->checkpointHelper();
 448         }
 449
 450         /**
 451          * tests for working checkpoint generation in gzip format work.
 452          *
 453          * We keep this test in addition to the simpler self::testCheckpointPlain, as there
 454          * were once problems when the used sinks were DumpPipeOutputs.
 455          *
 456          * xmldumps-backup typically uses bzip2 instead of gzip. However, as bzip2 requires
 457          * PHP extensions, we go for gzip instead, which triggers the same relevant code
 458          * paths while still being testable on more systems.
 459          *
 460          * Broken per T70653.
 461          *
 462          * @group large
 463          * @group Broken
 464          */
 465         function testCheckpointGzip() {
 466                 $this->checkHasGzip();
 467                 $this->checkpointHelper( "gzip" );
 468         }
 469
 470         /**
 471          * Creates a stub file that is used for testing the text pass of dumps
 472          *
 473          * @param string $fname (Optional) Absolute name of the file to write
 474          *   the stub into. If this parameter is null, a new temporary
 475          *   file is generated that is automatically removed upon tearDown.
 476          * @param int $iterations (Optional) specifies how often the block
 477          *   of 3 pages should go into the stub file. The page and
 478          *   revision id increase further and further, while the text
 479          *   id of the first iteration is reused. The pages and revision
 480          *   of iteration > 1 have no corresponding representation in the database.
 481          * @return string Absolute filename of the stub
 482          */
 483         private function setUpStub( $fname = null, $iterations = 1 ) {
 484                 if ( $fname === null ) {
 485                         $fname = $this->getNewTempFile();
 486                 }
 487                 $header = '<mediawiki xmlns="http://www.mediawiki.org/xml/export-0.10/" '
 488                         . 'xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" '
 489                         . 'xsi:schemaLocation="http://www.mediawiki.org/xml/export-0.10/ '
 490                         . 'http://www.mediawiki.org/xml/export-0.10.xsd" version="0.10" xml:lang="en">
 491   <siteinfo>
 492     <sitename>wikisvn</sitename>
 493     <base>http://localhost/wiki-svn/index.php/Main_Page</base>
 494     <generator>MediaWiki 1.21alpha</generator>
 495     <case>first-letter</case>
 496     <namespaces>
 497       <namespace key="-2" case="first-letter">Media</namespace>
 498       <namespace key="-1" case="first-letter">Special</namespace>
 499       <namespace key="0" case="first-letter" />
 500       <namespace key="1" case="first-letter">Talk</namespace>
 501       <namespace key="2" case="first-letter">User</namespace>
 502       <namespace key="3" case="first-letter">User talk</namespace>
 503       <namespace key="4" case="first-letter">Wikisvn</namespace>
 504       <namespace key="5" case="first-letter">Wikisvn talk</namespace>
 505       <namespace key="6" case="first-letter">File</namespace>
 506       <namespace key="7" case="first-letter">File talk</namespace>
 507       <namespace key="8" case="first-letter">MediaWiki</namespace>
 508       <namespace key="9" case="first-letter">MediaWiki talk</namespace>
 509       <namespace key="10" case="first-letter">Template</namespace>
 510       <namespace key="11" case="first-letter">Template talk</namespace>
 511       <namespace key="12" case="first-letter">Help</namespace>
 512       <namespace key="13" case="first-letter">Help talk</namespace>
 513       <namespace key="14" case="first-letter">Category</namespace>
 514       <namespace key="15" case="first-letter">Category talk</namespace>
 515     </namespaces>
 516   </siteinfo>
 517 ';
 518                 $tail = '</mediawiki>
 519 ';
 520
 521                 $content = $header;
 522                 $iterations = intval( $iterations );
 523                 for ( $i = 0; $i < $iterations; $i++ ) {
 524                         $page1 = '  <page>
 525     <title>BackupDumperTestP1</title>
 526     <ns>0</ns>
 527     <id>' . ( $this->pageId1 + $i * self::$numOfPages ) . '</id>
 528     <revision>
 529       <id>' . ( $this->revId1_1 + $i * self::$numOfRevs ) . '</id>
 530       <timestamp>2012-04-01T16:46:05Z</timestamp>
 531       <contributor>
 532         <ip>127.0.0.1</ip>
 533       </contributor>
 534       <comment>BackupDumperTestP1Summary1</comment>
 535       <model>wikitext</model>
 536       <format>text/x-wiki</format>
 537       <text id="' . $this->textId1_1 . '" bytes="23" />
 538       <sha1>0bolhl6ol7i6x0e7yq91gxgaan39j87</sha1>
 539     </revision>
 540   </page>
 541 ';
 542                         $page2 = '  <page>
 543     <title>BackupDumperTestP2</title>
 544     <ns>0</ns>
 545     <id>' . ( $this->pageId2 + $i * self::$numOfPages ) . '</id>
 546     <revision>
 547       <id>' . ( $this->revId2_1 + $i * self::$numOfRevs ) . '</id>
 548       <timestamp>2012-04-01T16:46:05Z</timestamp>
 549       <contributor>
 550         <ip>127.0.0.1</ip>
 551       </contributor>
 552       <comment>BackupDumperTestP2Summary1</comment>
 553       <model>wikitext</model>
 554       <format>text/x-wiki</format>
 555       <text id="' . $this->textId2_1 . '" bytes="23" />
 556       <sha1>jprywrymfhysqllua29tj3sc7z39dl2</sha1>
 557     </revision>
 558     <revision>
 559       <id>' . ( $this->revId2_2 + $i * self::$numOfRevs ) . '</id>
 560       <parentid>' . ( $this->revId2_1 + $i * self::$numOfRevs ) . '</parentid>
 561       <timestamp>2012-04-01T16:46:05Z</timestamp>
 562       <contributor>
 563         <ip>127.0.0.1</ip>
 564       </contributor>
 565       <comment>BackupDumperTestP2Summary2</comment>
 566       <model>wikitext</model>
 567       <format>text/x-wiki</format>
 568       <text id="' . $this->textId2_2 . '" bytes="23" />
 569       <sha1>b7vj5ks32po5m1z1t1br4o7scdwwy95</sha1>
 570     </revision>
 571     <revision>
 572       <id>' . ( $this->revId2_3 + $i * self::$numOfRevs ) . '</id>
 573       <parentid>' . ( $this->revId2_2 + $i * self::$numOfRevs ) . '</parentid>
 574       <timestamp>2012-04-01T16:46:05Z</timestamp>
 575       <contributor>
 576         <ip>127.0.0.1</ip>
 577       </contributor>
 578       <comment>BackupDumperTestP2Summary3</comment>
 579       <model>wikitext</model>
 580       <format>text/x-wiki</format>
 581       <text id="' . $this->textId2_3 . '" bytes="23" />
 582       <sha1>jfunqmh1ssfb8rs43r19w98k28gg56r</sha1>
 583     </revision>
 584     <revision>
 585       <id>' . ( $this->revId2_4 + $i * self::$numOfRevs ) . '</id>
 586       <parentid>' . ( $this->revId2_3 + $i * self::$numOfRevs ) . '</parentid>
 587       <timestamp>2012-04-01T16:46:05Z</timestamp>
 588       <contributor>
 589         <ip>127.0.0.1</ip>
 590       </contributor>
 591       <comment>BackupDumperTestP2Summary4 extra</comment>
 592       <model>wikitext</model>
 593       <format>text/x-wiki</format>
 594       <text id="' . $this->textId2_4 . '" bytes="44" />
 595       <sha1>6o1ciaxa6pybnqprmungwofc4lv00wv</sha1>
 596     </revision>
 597   </page>
 598 ';
 599                         // page 3 not in stub
 600
 601                         $page4 = '  <page>
 602     <title>Talk:BackupDumperTestP1</title>
 603     <ns>1</ns>
 604     <id>' . ( $this->pageId4 + $i * self::$numOfPages ) . '</id>
 605     <revision>
 606       <id>' . ( $this->revId4_1 + $i * self::$numOfRevs ) . '</id>
 607       <timestamp>2012-04-01T16:46:05Z</timestamp>
 608       <contributor>
 609         <ip>127.0.0.1</ip>
 610       </contributor>
 611       <comment>Talk BackupDumperTestP1 Summary1</comment>
 612       <model>BackupTextPassTestModel</model>
 613       <format>text/plain</format>
 614       <text id="' . $this->textId4_1 . '" bytes="35" />
 615       <sha1>nktofwzd0tl192k3zfepmlzxoax1lpe</sha1>
 616     </revision>
 617   </page>
 618 ';
 619                         $content .= $page1 . $page2 . $page4;
 620                 }
 621                 $content .= $tail;
 622                 $this->assertEquals( strlen( $content ), file_put_contents(
 623                         $fname, $content ), "Length of prepared stub" );
 624
 625                 return $fname;
 626         }
 627 }
 628
 629 class BackupTextPassTestModelHandler extends TextContentHandler {
 630
 631         public function __construct() {
 632                 parent::__construct( 'BackupTextPassTestModel' );
 633         }
 634
 635         public function exportTransform( $text, $format = null ) {
 636                 return strtoupper( $text );
 637         }
 638
 639 }
 640
 641 /**
 642  * Tests for TextPassDumper that do not rely on the database
 643  *
 644  * (As the Database group is only detected at class level (not method level), we
 645  * cannot bring this test case's tests into the above main test case.)
 646  *
 647  * @group Dump
 648  * @covers TextPassDumper
 649  */
 650 class TextPassDumperDatabaselessTest extends MediaWikiLangTestCase {
 651         /**
 652          * Ensures that setting the buffer size is effective.
 653          *
 654          * @dataProvider bufferSizeProvider
 655          */
 656         function testBufferSizeSetting( $expected, $size, $msg ) {
 657                 $dumper = new TextPassDumperAccessor();
 658                 $dumper->loadWithArgv( [ "--buffersize=" . $size ] );
 659                 $dumper->execute();
 660                 $this->assertEquals( $expected, $dumper->getBufferSize(), $msg );
 661         }
 662
 663         /**
 664          * Ensures that setting the buffer size is effective.
 665          *
 666          * @dataProvider bufferSizeProvider
 667          */
 668         function bufferSizeProvider() {
 669                 // expected, bufferSize to initialize with, message
 670                 return [
 671                         [ 512 * 1024, 512 * 1024, "Setting 512KB is not effective" ],
 672                         [ 8192, 8192, "Setting 8KB is not effective" ],
 673                         [ 4096, 2048, "Could set buffer size below lower bound" ]
 674                 ];
 675         }
 676 }
 677
 678 /**
 679  * Accessor for internal state of TextPassDumper
 680  *
 681  * Do not warrentless add getters here.
 682  */
 683 class TextPassDumperAccessor extends TextPassDumper {
 684         /**
 685          * Gets the bufferSize.
 686          *
 687          * If bufferSize setting does not work correctly, testCheckpoint... tests
 688          * fail and point in the wrong direction. To aid in troubleshooting when
 689          * testCheckpoint... tests break at some point in the future, we test the
 690          * bufferSize setting, hence need this accessor.
 691          *
 692          * (Yes, bufferSize is internal state of the TextPassDumper, but aiding
 693          * debugging of testCheckpoint... in the future seems to be worth testing
 694          * against it nonetheless.)
 695          */
 696         public function getBufferSize() {
 697                 return $this->bufferSize;
 698         }
 699
 700         function dump( $history, $text = null ) {
 701                 return true;
 702         }
 703 }