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