tests/phpunit/maintenance/CategoryChangesAsRdfTest.php

   1 <?php
   2
   3 use MediaWiki\MainConfigNames;
   4 use MediaWiki\Title\Title;
   5 use MediaWiki\Utils\MWTimestamp;
   6 use Wikimedia\Rdbms\IDatabase;
   7 use Wikimedia\Timestamp\ConvertibleTimestamp;
   8
   9 /**
  10  * Tests for CategoryChangesAsRdf recent changes exporter.
  11  * @covers \CategoryChangesAsRdf
  12  * @group Database
  13  */
  14 class CategoryChangesAsRdfTest extends MediaWikiLangTestCase {
  15
  16         protected function setUp(): void {
  17                 parent::setUp();
  18                 $this->overrideConfigValues( [
  19                         MainConfigNames::Server => 'http://acme.test',
  20                         MainConfigNames::CanonicalServer => 'http://acme.test',
  21                 ] );
  22         }
  23
  24         public static function provideCategoryData() {
  25                 return [
  26                         'delete category' => [
  27                                 __DIR__ . "/../data/categoriesrdf/delete.sparql",
  28                                 'getDeletedCatsIterator',
  29                                 'handleDeletes',
  30                                 [
  31                                         (object)[ 'rc_title' => 'Test', 'rc_cur_id' => 1, '_processed' => 1 ],
  32                                         (object)[ 'rc_title' => 'Test 2', 'rc_cur_id' => 2, '_processed' => 2 ],
  33                                 ],
  34                         ],
  35                         'move category' => [
  36                                 __DIR__ . "/../data/categoriesrdf/move.sparql",
  37                                 'getMovedCatsIterator',
  38                                 'handleMoves',
  39                                 [
  40                                         (object)[
  41                                                 'rc_title' => 'Test',
  42                                                 'rc_cur_id' => 4,
  43                                                 'page_title' => 'MovedTo',
  44                                                 'page_namespace' => NS_CATEGORY,
  45                                                 '_processed' => 4,
  46                                                 'pp_propname' => null,
  47                                                 'cat_pages' => 10,
  48                                                 'cat_subcats' => 2,
  49                                                 'cat_files' => 1,
  50                                         ],
  51                                         (object)[
  52                                                 'rc_title' => 'MovedTo',
  53                                                 'rc_cur_id' => 4,
  54                                                 'page_title' => 'MovedAgain',
  55                                                 'page_namespace' => NS_CATEGORY,
  56                                                 'pp_propname' => 'hiddencat',
  57                                                 'cat_pages' => 10,
  58                                                 'cat_subcats' => 2,
  59                                                 'cat_files' => 1,
  60                                         ],
  61                                         (object)[
  62                                                 'rc_title' => 'Test 2',
  63                                                 'rc_cur_id' => 5,
  64                                                 'page_title' => 'AlsoMoved',
  65                                                 'page_namespace' => NS_CATEGORY,
  66                                                 '_processed' => 5,
  67                                                 'pp_propname' => null,
  68                                                 'cat_pages' => 10,
  69                                                 'cat_subcats' => 2,
  70                                                 'cat_files' => 1,
  71                                         ],
  72                                         (object)[
  73                                                 'rc_title' => 'Test 3',
  74                                                 'rc_cur_id' => 6,
  75                                                 'page_title' => 'MovedOut',
  76                                                 'page_namespace' => NS_MAIN,
  77                                                 'pp_propname' => null,
  78                                                 'cat_pages' => 10,
  79                                                 'cat_subcats' => 2,
  80                                                 'cat_files' => 1,
  81                                         ],
  82                                         (object)[
  83                                                 'rc_title' => 'Test 4',
  84                                                 'rc_cur_id' => 7,
  85                                                 'page_title' => 'Already Done',
  86                                                 'page_namespace' => NS_CATEGORY,
  87                                                 'pp_propname' => null,
  88                                                 'cat_pages' => 10,
  89                                                 'cat_subcats' => 2,
  90                                                 'cat_files' => 1,
  91                                         ],
  92                                 ],
  93                                 [ 7 => true ],
  94                         ],
  95                         'restore deleted category' => [
  96                                 __DIR__ . "/../data/categoriesrdf/restore.sparql",
  97                                 'getRestoredCatsIterator',
  98                                 'handleRestores',
  99                                 [
 100                                         (object)[
 101                                                 'rc_title' => 'Restored cat',
 102                                                 'rc_cur_id' => 10,
 103                                                 '_processed' => 10,
 104                                                 'pp_propname' => null,
 105                                                 'cat_pages' => 10,
 106                                                 'cat_subcats' => 2,
 107                                                 'cat_files' => 1,
 108                                         ],
 109                                         (object)[
 110                                                 'rc_title' => 'Restored again',
 111                                                 'rc_cur_id' => 10,
 112                                                 'pp_propname' => null,
 113                                                 'cat_pages' => 10,
 114                                                 'cat_subcats' => 2,
 115                                                 'cat_files' => 1,
 116                                         ],
 117                                         (object)[
 118                                                 'rc_title' => 'Already seen',
 119                                                 'rc_cur_id' => 11,
 120                                                 'pp_propname' => null,
 121                                                 'cat_pages' => 10,
 122                                                 'cat_subcats' => 2,
 123                                                 'cat_files' => 1,
 124                                         ],
 125                                 ],
 126                                 [ 11 => true ],
 127                         ],
 128                         'new page' => [
 129                                 __DIR__ . "/../data/categoriesrdf/new.sparql",
 130                                 'getNewCatsIterator',
 131                                 'handleAdds',
 132                                 [
 133                                         (object)[
 134                                                 'rc_title' => 'New category',
 135                                                 'rc_cur_id' => 20,
 136                                                 '_processed' => 20,
 137                                                 'pp_propname' => null,
 138                                                 'cat_pages' => 10,
 139                                                 'cat_subcats' => 2,
 140                                                 'cat_files' => 1,
 141                                         ],
 142                                         (object)[
 143                                                 'rc_title' => 'Новая категория 😃',
 144                                                 'rc_cur_id' => 21,
 145                                                 '_processed' => 21,
 146                                                 'pp_propname' => 'hiddencat',
 147                                                 'cat_pages' => 10,
 148                                                 'cat_subcats' => 2,
 149                                                 'cat_files' => 1,
 150                                         ],
 151                                         (object)[
 152                                                 'rc_title' => 'Processed already',
 153                                                 'rc_cur_id' => 22,
 154                                         ],
 155                                 ],
 156                                 [ 22 => true ],
 157                         ],
 158                         'edit category' => [
 159                                 __DIR__ . "/../data/categoriesrdf/edit.sparql",
 160                                 'getChangedCatsIterator',
 161                                 'handleEdits',
 162                                 [
 163                                         (object)[
 164                                                 'rc_title' => 'Changed category',
 165                                                 'rc_cur_id' => 30,
 166                                                 '_processed' => 30,
 167                                                 'pp_propname' => null,
 168                                                 'cat_pages' => 10,
 169                                                 'cat_subcats' => 2,
 170                                                 'cat_files' => 1,
 171                                         ],
 172                                         (object)[
 173                                                 'rc_title' => 'Changed again',
 174                                                 'rc_cur_id' => 30,
 175                                                 'pp_propname' => null,
 176                                                 'cat_pages' => 12,
 177                                                 'cat_subcats' => 2,
 178                                                 'cat_files' => 1,
 179                                         ],
 180                                         (object)[
 181                                                 'rc_title' => 'Processed already',
 182                                                 'rc_cur_id' => 31,
 183                                                 'pp_propname' => null,
 184                                                 'cat_pages' => 10,
 185                                                 'cat_subcats' => 2,
 186                                                 'cat_files' => 1,
 187                                         ],
 188                                 ],
 189                                 [ 31 => true ],
 190                         ],
 191                         // TODO: not sure how to test categorization changes, it uses the database select...
 192                 ];
 193         }
 194
 195         /**
 196          * Mock category links iterator.
 197          * @param IDatabase $dbr
 198          * @param array $ids
 199          * @return array
 200          */
 201         public function getCategoryLinksIterator( $dbr, array $ids ) {
 202                 $res = [];
 203                 foreach ( $ids as $pageid ) {
 204                         $res[] = (object)[ 'cl_from' => $pageid, 'cl_to' => "Parent of $pageid" ];
 205                 }
 206                 return $res;
 207         }
 208
 209         /**
 210          * @dataProvider provideCategoryData
 211          * @param string $testFileName Name of the test, defines filename with expected results.
 212          * @param string $iterator Iterator method name to mock
 213          * @param string $handler Handler method to call
 214          * @param array $result Result to be returned from mock iterator
 215          * @param array $preProcessed List of pre-processed items
 216          */
 217         public function testSparqlUpdate( $testFileName, $iterator, $handler, $result,
 218                         array $preProcessed = [] ) {
 219                 $dumpScript =
 220                         $this->getMockBuilder( CategoryChangesAsRdf::class )
 221                                 ->onlyMethods( [ $iterator, 'getCategoryLinksIterator' ] )
 222                                 ->getMock();
 223
 224                 $dumpScript->method( 'getCategoryLinksIterator' )
 225                         ->willReturnCallback( [ $this, 'getCategoryLinksIterator' ] );
 226
 227                 $dumpScript->expects( $this->once() )
 228                         ->method( $iterator )
 229                         ->willReturn( [ $result ] );
 230
 231                 $ref = new ReflectionObject( $dumpScript );
 232                 $processedProperty = $ref->getProperty( 'processed' );
 233                 $processedProperty->setAccessible( true );
 234                 $processedProperty->setValue( $dumpScript, $preProcessed );
 235
 236                 $output = fopen( "php://memory", "w+b" );
 237                 $dbr = $this->getDb();
 238                 /** @var CategoryChangesAsRdf $dumpScript */
 239                 $dumpScript->initialize();
 240                 $dumpScript->getRdf();
 241                 $dumpScript->$handler( $dbr, $output );
 242
 243                 rewind( $output );
 244                 $sparql = stream_get_contents( $output );
 245                 $this->assertFileContains( $testFileName, $sparql );
 246
 247                 $processed = $processedProperty->getValue( $dumpScript );
 248                 $expectedProcessed = array_keys( $preProcessed );
 249                 foreach ( $result as $row ) {
 250                         if ( isset( $row->_processed ) ) {
 251                                 $this->assertArrayHasKey( $row->_processed, $processed,
 252                                         "ID {$row->_processed} was not processed!" );
 253                                 $expectedProcessed[] = $row->_processed;
 254                         }
 255                 }
 256                 $this->assertSame( $expectedProcessed, array_keys( $processed ),
 257                         'Processed array has wrong items' );
 258         }
 259
 260         public function testUpdateTs() {
 261                 $dumpScript = new CategoryChangesAsRdf();
 262                 $dumpScript->initialize();
 263                 $update = $dumpScript->updateTS( 1503620949 );
 264                 $outFile = __DIR__ . '/../data/categoriesrdf/updatets.txt';
 265                 $this->assertFileContains( $outFile, $update );
 266         }
 267
 268         public function testCategorization() {
 269                 $this->overrideConfigValue(
 270                         MainConfigNames::RCWatchCategoryMembership,
 271                         true
 272                 );
 273                 $start = new MWTimestamp( "2020-07-31T10:00:00" );
 274                 $end = new MWTimestamp( "2020-07-31T10:01:00" );
 275                 ConvertibleTimestamp::setFakeTime( "2020-07-31T10:00:00" );
 276                 $l1 = Title::makeTitle( NS_CATEGORY, __CLASS__ . "_L1" );
 277                 $l2 = Title::makeTitle( NS_CATEGORY, __CLASS__ . "_L2" );
 278                 $pageInL2 = Title::makeTitle( NS_CATEGORY, __CLASS__ . "_Page" );
 279                 $this->editPage( $l1, "" );
 280                 $this->editPage( $l2, "[[{$l1->getPrefixedText()}]]" );
 281                 $this->editPage( $pageInL2, "[[{$l2->getPrefixedText()}]]" );
 282
 283                 $output = fopen( "php://memory", "w+b" );
 284
 285                 $this->runJobs( [], [
 286                         'type' => 'categoryMembershipChange'
 287                 ] );
 288
 289                 $dbr = $this->getDb();
 290                 $categoryChangesAsRdf = new CategoryChangesAsRdf();
 291
 292                 $ref = new ReflectionObject( $categoryChangesAsRdf );
 293                 $startTSProp = $ref->getProperty( 'startTS' );
 294                 $startTSProp->setAccessible( true );
 295                 $startTSProp->setValue( $categoryChangesAsRdf, $start->getTimestamp() );
 296                 $endTSProp = $ref->getProperty( 'endTS' );
 297                 $endTSProp->setAccessible( true );
 298                 $endTSProp->setValue( $categoryChangesAsRdf, $end->getTimestamp() );
 299
 300                 $categoryChangesAsRdf->initialize();
 301                 $categoryChangesAsRdf->getRdf();
 302                 $categoryChangesAsRdf->handleCategorization( $dbr, $output );
 303                 rewind( $output );
 304                 $sparql = stream_get_contents( $output );
 305                 $outFile = __DIR__ . '/../data/categoriesrdf/categorization.txt';
 306                 $this->assertFileContains( $outFile, $sparql );
 307         }
 308
 309 }