tests/phpunit/CRM/Dedupe/MergerTest.php

   1 <?php
   2
   3 /**
   4  * Class CRM_Dedupe_DedupeMergerTest
   5  * @group headless
   6  */
   7 class CRM_Dedupe_MergerTest extends CiviUnitTestCase {
   8
   9   protected $_groupId;
  10   protected $_contactIds = array();
  11
  12   public function tearDown() {
  13     $this->quickCleanup(array('civicrm_contact', 'civicrm_group_contact', 'civicrm_group'));
  14     parent::tearDown();
  15   }
  16
  17   public function createDupeContacts() {
  18     // create a group to hold contacts, so that dupe checks don't consider any other contacts in the DB
  19     $params = array(
  20       'name'       => 'Test Dupe Merger Group',
  21       'title'      => 'Test Dupe Merger Group',
  22       'domain_id'  => 1,
  23       'is_active'  => 1,
  24       'visibility' => 'Public Pages',
  25       'version'    => 3,
  26     );
  27     // TODO: This is not an API test!!
  28     $result = civicrm_api('group', 'create', $params);
  29     $this->_groupId = $result['id'];
  30
  31     // contact data set
  32
  33     // make dupe checks based on based on following contact sets:
  34     // FIRST - LAST - EMAIL
  35     // ---------------------------------
  36     // robin  - hood - robin@example.com
  37     // robin  - hood - robin@example.com
  38     // robin  - hood - hood@example.com
  39     // robin  - dale - robin@example.com
  40     // little - dale - dale@example.com
  41     // little - dale - dale@example.com
  42     // will   - dale - dale@example.com
  43     // will   - dale - will@example.com
  44     // will   - dale - will@example.com
  45     $params = array(
  46       array(
  47         'first_name' => 'robin',
  48         'last_name' => 'hood',
  49         'email' => 'robin@example.com',
  50         'contact_type' => 'Individual',
  51       ),
  52       array(
  53         'first_name' => 'robin',
  54         'last_name' => 'hood',
  55         'email' => 'robin@example.com',
  56         'contact_type' => 'Individual',
  57       ),
  58       array(
  59         'first_name' => 'robin',
  60         'last_name' => 'hood',
  61         'email' => 'hood@example.com',
  62         'contact_type' => 'Individual',
  63       ),
  64       array(
  65         'first_name' => 'robin',
  66         'last_name' => 'dale',
  67         'email' => 'robin@example.com',
  68         'contact_type' => 'Individual',
  69       ),
  70       array(
  71         'first_name' => 'little',
  72         'last_name' => 'dale',
  73         'email' => 'dale@example.com',
  74         'contact_type' => 'Individual',
  75       ),
  76       array(
  77         'first_name' => 'little',
  78         'last_name' => 'dale',
  79         'email' => 'dale@example.com',
  80         'contact_type' => 'Individual',
  81       ),
  82       array(
  83         'first_name' => 'will',
  84         'last_name' => 'dale',
  85         'email' => 'dale@example.com',
  86         'contact_type' => 'Individual',
  87       ),
  88       array(
  89         'first_name' => 'will',
  90         'last_name' => 'dale',
  91         'email' => 'will@example.com',
  92         'contact_type' => 'Individual',
  93       ),
  94       array(
  95         'first_name' => 'will',
  96         'last_name' => 'dale',
  97         'email' => 'will@example.com',
  98         'contact_type' => 'Individual',
  99       ),
 100     );
 101
 102     $count = 1;
 103     foreach ($params as $param) {
 104       $param['version'] = 3;
 105       $contact = civicrm_api('contact', 'create', $param);
 106       $this->_contactIds[$count++] = $contact['id'];
 107
 108       $grpParams = array(
 109         'contact_id' => $contact['id'],
 110         'group_id'   => $this->_groupId,
 111         'version'    => 3,
 112       );
 113       $res = civicrm_api('group_contact', 'create', $grpParams);
 114     }
 115   }
 116
 117   /**
 118    * Delete all created contacts.
 119    */
 120   public function deleteDupeContacts() {
 121     foreach ($this->_contactIds as $contactId) {
 122       $this->contactDelete($contactId);
 123     }
 124
 125     // delete dupe group
 126     $params = array('id' => $this->_groupId, 'version' => 3);
 127     civicrm_api('group', 'delete', $params);
 128   }
 129
 130   /**
 131    * Test the batch merge.
 132    */
 133   public function testBatchMergeSelectedDuplicates() {
 134     $this->createDupeContacts();
 135
 136     // verify that all contacts have been created separately
 137     $this->assertEquals(count($this->_contactIds), 9, 'Check for number of contacts.');
 138
 139     $dao = new CRM_Dedupe_DAO_RuleGroup();
 140     $dao->contact_type = 'Individual';
 141     $dao->name = 'IndividualSupervised';
 142     $dao->is_default = 1;
 143     $dao->find(TRUE);
 144
 145     $foundDupes = CRM_Dedupe_Finder::dupesInGroup($dao->id, $this->_groupId);
 146
 147     // -------------------------------------------------------------------------
 148     // Name and Email (reserved) Matches ( 3 pairs )
 149     // --------------------------------------------------------------------------
 150     // robin  - hood - robin@example.com
 151     // robin  - hood - robin@example.com
 152     // little - dale - dale@example.com
 153     // little - dale - dale@example.com
 154     // will   - dale - will@example.com
 155     // will   - dale - will@example.com
 156     // so 3 pairs for - first + last + mail
 157     $this->assertEquals(count($foundDupes), 3, 'Check Individual-Supervised dupe rule for dupesInGroup().');
 158
 159     // Run dedupe finder as the browser would
 160     $_SERVER['REQUEST_METHOD'] = 'GET'; //avoid invalid key error
 161     $object = new CRM_Contact_Page_DedupeFind();
 162     $object->set('gid', $this->_groupId);
 163     $object->set('rgid', $dao->id);
 164     $object->set('action', CRM_Core_Action::UPDATE);
 165     @$object->run();
 166
 167     // Retrieve pairs from prev next cache table
 168     $select = array('pn.is_selected' => 'is_selected');
 169     $cacheKeyString = "merge Individual_{$dao->id}_{$this->_groupId}";
 170     $pnDupePairs = CRM_Core_BAO_PrevNextCache::retrieve($cacheKeyString, NULL, NULL, 0, 0, $select);
 171
 172     $this->assertEquals(count($foundDupes), count($pnDupePairs), 'Check number of dupe pairs in prev next cache.');
 173
 174     // mark first two pairs as selected
 175     CRM_Core_DAO::singleValueQuery("UPDATE civicrm_prevnext_cache SET is_selected = 1 WHERE id IN ({$pnDupePairs[0]['prevnext_id']}, {$pnDupePairs[1]['prevnext_id']})");
 176
 177     $pnDupePairs = CRM_Core_BAO_PrevNextCache::retrieve($cacheKeyString, NULL, NULL, 0, 0, $select);
 178     $this->assertEquals($pnDupePairs[0]['is_selected'], 1, 'Check if first record in dupe pairs is marked as selected.');
 179     $this->assertEquals($pnDupePairs[0]['is_selected'], 1, 'Check if second record in dupe pairs is marked as selected.');
 180
 181     // batch merge selected dupes
 182     $result = CRM_Dedupe_Merger::batchMerge($dao->id, $this->_groupId, 'safe', TRUE, 5, 1);
 183     $this->assertEquals(count($result['merged']), 2, 'Check number of merged pairs.');
 184
 185     // retrieve pairs from prev next cache table
 186     $pnDupePairs = CRM_Core_BAO_PrevNextCache::retrieve($cacheKeyString, NULL, NULL, 0, 0, $select);
 187     $this->assertEquals(count($pnDupePairs), 1, 'Check number of remaining dupe pairs in prev next cache.');
 188
 189     $this->deleteDupeContacts();
 190   }
 191
 192   /**
 193    * Test the batch merge.
 194    */
 195   public function testBatchMergeAllDuplicates() {
 196     $this->createDupeContacts();
 197
 198     // verify that all contacts have been created separately
 199     $this->assertEquals(count($this->_contactIds), 9, 'Check for number of contacts.');
 200
 201     $dao = new CRM_Dedupe_DAO_RuleGroup();
 202     $dao->contact_type = 'Individual';
 203     $dao->name = 'IndividualSupervised';
 204     $dao->is_default = 1;
 205     $dao->find(TRUE);
 206
 207     $foundDupes = CRM_Dedupe_Finder::dupesInGroup($dao->id, $this->_groupId);
 208
 209     // -------------------------------------------------------------------------
 210     // Name and Email (reserved) Matches ( 3 pairs )
 211     // --------------------------------------------------------------------------
 212     // robin  - hood - robin@example.com
 213     // robin  - hood - robin@example.com
 214     // little - dale - dale@example.com
 215     // little - dale - dale@example.com
 216     // will   - dale - will@example.com
 217     // will   - dale - will@example.com
 218     // so 3 pairs for - first + last + mail
 219     $this->assertEquals(count($foundDupes), 3, 'Check Individual-Supervised dupe rule for dupesInGroup().');
 220
 221     // Run dedupe finder as the browser would
 222     $_SERVER['REQUEST_METHOD'] = 'GET'; //avoid invalid key error
 223     $object = new CRM_Contact_Page_DedupeFind();
 224     $object->set('gid', $this->_groupId);
 225     $object->set('rgid', $dao->id);
 226     $object->set('action', CRM_Core_Action::UPDATE);
 227     @$object->run();
 228
 229     // Retrieve pairs from prev next cache table
 230     $select = array('pn.is_selected' => 'is_selected');
 231     $cacheKeyString = "merge Individual_{$dao->id}_{$this->_groupId}";
 232     $pnDupePairs = CRM_Core_BAO_PrevNextCache::retrieve($cacheKeyString, NULL, NULL, 0, 0, $select);
 233
 234     $this->assertEquals(count($foundDupes), count($pnDupePairs), 'Check number of dupe pairs in prev next cache.');
 235
 236     // batch merge all dupes
 237     $result = CRM_Dedupe_Merger::batchMerge($dao->id, $this->_groupId, 'safe', TRUE, 5, 2);
 238     $this->assertEquals(count($result['merged']), 3, 'Check number of merged pairs.');
 239
 240     // retrieve pairs from prev next cache table
 241     $pnDupePairs = CRM_Core_BAO_PrevNextCache::retrieve($cacheKeyString, NULL, NULL, 0, 0, $select);
 242     $this->assertEquals(count($pnDupePairs), 0, 'Check number of remaining dupe pairs in prev next cache.');
 243
 244     $this->deleteDupeContacts();
 245   }
 246
 247   /**
 248    * The goal of this function is to test that all required tables are returned.
 249    */
 250   public function testGetCidRefs() {
 251     $this->entityCustomGroupWithSingleFieldCreate(__FUNCTION__, 'Contacts');
 252     $this->assertEquals(array_merge($this->getStaticCIDRefs(), $this->getHackedInCIDRef()), CRM_Dedupe_Merger::cidRefs());
 253     $this->assertEquals(array_merge($this->getCalculatedCIDRefs(), $this->getHackedInCIDRef()), CRM_Dedupe_Merger::cidRefs());
 254   }
 255
 256   /**
 257    * Get the list of not-really-cid-refs that are currently hacked in.
 258    *
 259    * This is hacked into getCIDs function.
 260    *
 261    * @return array
 262    */
 263   public function getHackedInCIDRef() {
 264     return array(
 265       'civicrm_entity_tag' => array(
 266         0 => 'entity_id',
 267       ),
 268     );
 269   }
 270
 271   /**
 272    * Test function that gets duplicate pairs.
 273    *
 274    * It turns out there are 2 code paths retrieving this data so my initial focus is on ensuring
 275    * they match.
 276    */
 277   public function testGetMatches() {
 278     $this->setupMatchData();
 279     $pairs = CRM_Dedupe_Merger::getDuplicatePairs(
 280       1,
 281       NULL,
 282       TRUE,
 283       25,
 284       FALSE
 285     );
 286
 287     $this->assertEquals(array(
 288       0 => array(
 289         'srcID' => $this->contacts[0]['id'],
 290         'srcName' => 'Mr. Mickey Mouse II',
 291         'dstID' => $this->contacts[1]['id'],
 292         'dstName' => 'Mr. Mickey Mouse II',
 293         'weight' => 20,
 294         'canMerge' => TRUE,
 295       ),
 296       1 => array(
 297         'srcID' => $this->contacts[2]['id'],
 298         'srcName' => 'Mr. Minnie Mouse II',
 299         'dstID' => $this->contacts[3]['id'],
 300         'dstName' => 'Mr. Minnie Mouse II',
 301         'weight' => 20,
 302         'canMerge' => TRUE,
 303       ),
 304     ), $pairs);
 305   }
 306
 307   /**
 308    * Test function that gets duplicate pairs.
 309    *
 310    * It turns out there are 2 code paths retrieving this data so my initial focus is on ensuring
 311    * they match.
 312    */
 313   public function testGetMatchesInGroup() {
 314     $this->setupMatchData();
 315
 316     $groupID = $this->groupCreate(array('title' => 'she-mice'));
 317
 318     $this->callAPISuccess('GroupContact', 'create', array('group_id' => $groupID, 'contact_id' => $this->contacts[3]['id']));
 319
 320     $pairs = CRM_Dedupe_Merger::getDuplicatePairs(
 321       1,
 322       $groupID,
 323       TRUE,
 324       25,
 325       FALSE
 326     );
 327
 328     $this->assertEquals(array(
 329       0 => array(
 330         'srcID' => $this->contacts[3]['id'],
 331         'srcName' => 'Mr. Minnie Mouse II',
 332         'dstID' => $this->contacts[2]['id'],
 333         'dstName' => 'Mr. Minnie Mouse II',
 334         'weight' => 20,
 335         'canMerge' => TRUE,
 336       ),
 337     ), $pairs);
 338   }
 339
 340   /**
 341    * Set up some contacts for our matching.
 342    */
 343   public function setupMatchData() {
 344     $fixtures = array(
 345       array(
 346         'first_name' => 'Mickey',
 347         'last_name' => 'Mouse',
 348         'email' => 'mickey@mouse.com',
 349       ),
 350       array(
 351         'first_name' => 'Mickey',
 352         'last_name' => 'Mouse',
 353         'email' => 'mickey@mouse.com',
 354       ),
 355       array(
 356         'first_name' => 'Minnie',
 357         'last_name' => 'Mouse',
 358         'email' => 'mickey@mouse.com',
 359       ),
 360       array(
 361         'first_name' => 'Minnie',
 362         'last_name' => 'Mouse',
 363         'email' => 'mickey@mouse.com',
 364       ),
 365     );
 366     foreach ($fixtures as $fixture) {
 367       $contactID = $this->individualCreate($fixture);
 368       $this->contacts[] = array_merge($fixture, array('id' => $contactID));
 369     }
 370   }
 371
 372
 373   /**
 374    * Get the list of tables that refer to the CID.
 375    *
 376    * This is a statically maintained (in this test list).
 377    *
 378    * There is also a check against an automated list but having both seems to add extra stability to me. They do
 379    * not change often.
 380    */
 381   public function getStaticCIDRefs() {
 382     return array(
 383       'civicrm_acl_cache' => array(
 384         0 => 'contact_id',
 385       ),
 386       'civicrm_acl_contact_cache' => array(
 387         0 => 'user_id',
 388         1 => 'contact_id',
 389       ),
 390       'civicrm_action_log' => array(
 391         0 => 'contact_id',
 392       ),
 393       'civicrm_activity_contact' => array(
 394         0 => 'contact_id',
 395       ),
 396       'civicrm_address' => array(
 397         0 => 'contact_id',
 398       ),
 399       'civicrm_batch' => array(
 400         0 => 'created_id',
 401         1 => 'modified_id',
 402       ),
 403       'civicrm_campaign' => array(
 404         0 => 'created_id',
 405         1 => 'last_modified_id',
 406       ),
 407       'civicrm_case_contact' => array(
 408         0 => 'contact_id',
 409       ),
 410       'civicrm_contact' => array(
 411         0 => 'primary_contact_id',
 412         1 => 'employer_id',
 413       ),
 414       'civicrm_contribution' => array(
 415         0 => 'contact_id',
 416       ),
 417       'civicrm_contribution_page' => array(
 418         0 => 'created_id',
 419       ),
 420       'civicrm_contribution_recur' => array(
 421         0 => 'contact_id',
 422       ),
 423       'civicrm_contribution_soft' => array(
 424         0 => 'contact_id',
 425       ),
 426       'civicrm_custom_group' => array(
 427         0 => 'created_id',
 428       ),
 429       'civicrm_dashboard_contact' => array(
 430         0 => 'contact_id',
 431       ),
 432       'civicrm_dedupe_exception' => array(
 433         0 => 'contact_id1',
 434         1 => 'contact_id2',
 435       ),
 436       'civicrm_domain' => array(
 437         0 => 'contact_id',
 438       ),
 439       'civicrm_email' => array(
 440         0 => 'contact_id',
 441       ),
 442       'civicrm_event' => array(
 443         0 => 'created_id',
 444       ),
 445       'civicrm_event_carts' => array(
 446         0 => 'user_id',
 447       ),
 448       'civicrm_financial_account' => array(
 449         0 => 'contact_id',
 450       ),
 451       'civicrm_financial_item' => array(
 452         0 => 'contact_id',
 453       ),
 454       'civicrm_grant' => array(
 455         0 => 'contact_id',
 456       ),
 457       'civicrm_group' => array(
 458         0 => 'created_id',
 459         1 => 'modified_id',
 460       ),
 461       'civicrm_group_contact' => array(
 462         0 => 'contact_id',
 463       ),
 464       'civicrm_group_contact_cache' => array(
 465         0 => 'contact_id',
 466       ),
 467       'civicrm_group_organization' => array(
 468         0 => 'organization_id',
 469       ),
 470       'civicrm_im' => array(
 471         0 => 'contact_id',
 472       ),
 473       'civicrm_log' => array(
 474         0 => 'modified_id',
 475       ),
 476       'civicrm_mailing' => array(
 477         0 => 'created_id',
 478         1 => 'scheduled_id',
 479         2 => 'approver_id',
 480       ),
 481       'civicrm_mailing_abtest' => array(
 482         0 => 'created_id',
 483       ),
 484       'civicrm_mailing_event_queue' => array(
 485         0 => 'contact_id',
 486       ),
 487       'civicrm_mailing_event_subscribe' => array(
 488         0 => 'contact_id',
 489       ),
 490       'civicrm_mailing_recipients' => array(
 491         0 => 'contact_id',
 492       ),
 493       'civicrm_membership' => array(
 494         0 => 'contact_id',
 495       ),
 496       'civicrm_membership_log' => array(
 497         0 => 'modified_id',
 498       ),
 499       'civicrm_membership_type' => array(
 500         0 => 'member_of_contact_id',
 501       ),
 502       'civicrm_note' => array(
 503         0 => 'contact_id',
 504       ),
 505       'civicrm_openid' => array(
 506         0 => 'contact_id',
 507       ),
 508       'civicrm_participant' => array(
 509         0 => 'contact_id',
 510         1 => 'transferred_to_contact_id', //CRM-16761
 511       ),
 512       'civicrm_payment_token' => array(
 513         0 => 'contact_id',
 514         1 => 'created_id',
 515       ),
 516       'civicrm_pcp' => array(
 517         0 => 'contact_id',
 518       ),
 519       'civicrm_phone' => array(
 520         0 => 'contact_id',
 521       ),
 522       'civicrm_pledge' => array(
 523         0 => 'contact_id',
 524       ),
 525       'civicrm_print_label' => array(
 526         0 => 'created_id',
 527       ),
 528       'civicrm_relationship' => array(
 529         0 => 'contact_id_a',
 530         1 => 'contact_id_b',
 531       ),
 532       'civicrm_report_instance' => array(
 533         0 => 'created_id',
 534         1 => 'owner_id',
 535       ),
 536       'civicrm_setting' => array(
 537         0 => 'contact_id',
 538         1 => 'created_id',
 539       ),
 540       'civicrm_subscription_history' => array(
 541         0 => 'contact_id',
 542       ),
 543       'civicrm_survey' => array(
 544         0 => 'created_id',
 545         1 => 'last_modified_id',
 546       ),
 547       'civicrm_tag' => array(
 548         0 => 'created_id',
 549       ),
 550       'civicrm_uf_group' => array(
 551         0 => 'created_id',
 552       ),
 553       'civicrm_uf_match' => array(
 554         0 => 'contact_id',
 555       ),
 556       'civicrm_value_testgetcidref_1' => array(
 557         0 => 'entity_id',
 558       ),
 559       'civicrm_website' => array(
 560         0 => 'contact_id',
 561       ),
 562     );
 563   }
 564
 565   /**
 566    * Get a list of CIDs that is calculated off the schema.
 567    *
 568    * Note this is an expensive and table locking query. Should be safe in tests though.
 569    */
 570   public function getCalculatedCIDRefs() {
 571     $cidRefs = array();
 572     $sql = "
 573 SELECT
 574     table_name,
 575     column_name
 576 FROM information_schema.key_column_usage
 577 WHERE
 578     referenced_table_schema = database() AND
 579     referenced_table_name = 'civicrm_contact' AND
 580     referenced_column_name = 'id';
 581       ";
 582     $dao = CRM_Core_DAO::executeQuery($sql);
 583     while ($dao->fetch()) {
 584       $cidRefs[$dao->table_name][] = $dao->column_name;
 585     }
 586     // Do specific re-ordering changes to make this the same as the ref validated one.
 587     // The above query orders by FK alphabetically.
 588     // There might be cleverer ways to do this but it shouldn't change much.
 589     $cidRefs['civicrm_contact'][0] = 'primary_contact_id';
 590     $cidRefs['civicrm_contact'][1] = 'employer_id';
 591     $cidRefs['civicrm_acl_contact_cache'][0] = 'user_id';
 592     $cidRefs['civicrm_acl_contact_cache'][1] = 'contact_id';
 593     $cidRefs['civicrm_mailing'][0] = 'created_id';
 594     $cidRefs['civicrm_mailing'][1] = 'scheduled_id';
 595     $cidRefs['civicrm_mailing'][2] = 'approver_id';
 596     return $cidRefs;
 597   }
 598
 599 }