tests/phpunit/maintenance/categoryChangesAsRdfTest.php

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