CRM/Dedupe/Finder.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | CiviCRM version 5                                                  |
   5  +--------------------------------------------------------------------+
   6  | Copyright CiviCRM LLC (c) 2004-2019                                |
   7  +--------------------------------------------------------------------+
   8  | This file is a part of CiviCRM.                                    |
   9  |                                                                    |
  10  | CiviCRM is free software; you can copy, modify, and distribute it  |
  11  | under the terms of the GNU Affero General Public License           |
  12  | Version 3, 19 November 2007 and the CiviCRM Licensing Exception.   |
  13  |                                                                    |
  14  | CiviCRM is distributed in the hope that it will be useful, but     |
  15  | WITHOUT ANY WARRANTY; without even the implied warranty of         |
  16  | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.               |
  17  | See the GNU Affero General Public License for more details.        |
  18  |                                                                    |
  19  | You should have received a copy of the GNU Affero General Public   |
  20  | License and the CiviCRM Licensing Exception along                  |
  21  | with this program; if not, contact CiviCRM LLC                     |
  22  | at info[AT]civicrm[DOT]org. If you have questions about the        |
  23  | GNU Affero General Public License or the licensing of CiviCRM,     |
  24  | see the CiviCRM license FAQ at http://civicrm.org/licensing        |
  25  +--------------------------------------------------------------------+
  26  */
  27
  28 /**
  29  *
  30  * @package CRM
  31  * @copyright CiviCRM LLC (c) 2004-2019
  32  * $Id$
  33  *
  34  */
  35
  36 /**
  37  * The CiviCRM duplicate discovery engine is based on an
  38  * algorithm designed by David Strauss <david@fourkitchens.com>.
  39  */
  40 class CRM_Dedupe_Finder {
  41
  42   /**
  43    * Return a contact_id-keyed array of arrays of possible dupes
  44    * (of the key contact_id) - limited to dupes of $cids if provided.
  45    *
  46    * @param int $rgid
  47    *   Rule group id.
  48    * @param array $cids
  49    *   Contact ids to limit the search to.
  50    *
  51    * @param bool $checkPermissions
  52    *   Respect logged in user permissions.
  53    *
  54    * @param int $searchLimit
  55    *  Limit for the number of contacts to be used for comparison.
  56    *  The search methodology finds all matches for the searchedContacts so this limits
  57    *  the number of searched contacts, not the matches found.
  58    *
  59    * @return array
  60    *   Array of (cid1, cid2, weight) dupe triples
  61    *
  62    * @throws CiviCRM_API3_Exception
  63    * @throws Exception
  64    */
  65   public static function dupes($rgid, $cids = [], $checkPermissions = TRUE, $searchLimit = 0) {
  66     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
  67     $rgBao->id = $rgid;
  68     $rgBao->contactIds = $cids;
  69     if (!$rgBao->find(TRUE)) {
  70       CRM_Core_Error::fatal("Dedupe rule not found for selected contacts");
  71     }
  72     if (empty($rgBao->contactIds) && !empty($searchLimit)) {
  73       $limitedContacts = civicrm_api3('Contact', 'get', [
  74         'return' => 'id',
  75         'contact_type' => $rgBao->contact_type,
  76         'options' => ['limit' => $searchLimit],
  77       ]);
  78       $rgBao->contactIds = array_keys($limitedContacts['values']);
  79     }
  80
  81     $rgBao->fillTable();
  82     $dao = new CRM_Core_DAO();
  83     $dao->query($rgBao->thresholdQuery($checkPermissions));
  84     $dupes = [];
  85     while ($dao->fetch()) {
  86       $dupes[] = [$dao->id1, $dao->id2, $dao->weight];
  87     }
  88     $dao->query($rgBao->tableDropQuery());
  89
  90     return $dupes;
  91   }
  92
  93   /**
  94    * Return an array of possible dupes, based on the provided array of
  95    * params, using the default rule group for the given contact type and
  96    * usage.
  97    *
  98    * check_permission is a boolean flag to indicate if permission should be considered.
  99    * default is to always check permissioning but public pages for example might not want
 100    * permission to be checked for anonymous users. Refer CRM-6211. We might be breaking
 101    * Multi-Site dedupe for public pages.
 102    *
 103    * @param array $params
 104    *   Array of params of the form $params[$table][$field] == $value.
 105    * @param string $ctype
 106    *   Contact type to match against.
 107    * @param string $used
 108    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
 109    * @param array $except
 110    *   Array of contacts that shouldn't be considered dupes.
 111    * @param int $ruleGroupID
 112    *   The id of the dedupe rule we should be using.
 113    *
 114    * @return array
 115    *   matching contact ids
 116    */
 117   public static function dupesByParams(
 118     $params,
 119     $ctype,
 120     $used = 'Unsupervised',
 121     $except = [],
 122     $ruleGroupID = NULL
 123   ) {
 124     // If $params is empty there is zero reason to proceed.
 125     if (!$params) {
 126       return [];
 127     }
 128     $checkPermission = CRM_Utils_Array::value('check_permission', $params, TRUE);
 129     // This may no longer be required - see https://github.com/civicrm/civicrm-core/pull/13176
 130     $params = array_filter($params);
 131
 132     $foundByID = FALSE;
 133     if ($ruleGroupID) {
 134       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 135       $rgBao->id = $ruleGroupID;
 136       $rgBao->contact_type = $ctype;
 137       if ($rgBao->find(TRUE)) {
 138         $foundByID = TRUE;
 139       }
 140     }
 141
 142     if (!$foundByID) {
 143       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 144       $rgBao->contact_type = $ctype;
 145       $rgBao->used = $used;
 146       if (!$rgBao->find(TRUE)) {
 147         CRM_Core_Error::fatal("$used rule for $ctype does not exist");
 148       }
 149     }
 150
 151     if (isset($params['civicrm_phone']['phone_numeric'])) {
 152       $orig = $params['civicrm_phone']['phone_numeric'];
 153       $params['civicrm_phone']['phone_numeric'] = preg_replace('/[^\d]/', '', $orig);
 154     }
 155     $rgBao->params = $params;
 156     $rgBao->fillTable();
 157     $dao = new CRM_Core_DAO();
 158     $dao->query($rgBao->thresholdQuery($checkPermission));
 159     $dupes = [];
 160     while ($dao->fetch()) {
 161       if (isset($dao->id) && $dao->id) {
 162         $dupes[] = $dao->id;
 163       }
 164     }
 165     $dao->query($rgBao->tableDropQuery());
 166     return array_diff($dupes, $except);
 167   }
 168
 169   /**
 170    * Return a contact_id-keyed array of arrays of possible dupes in the given group.
 171    *
 172    * @param int $rgid
 173    *   Rule group id.
 174    * @param int $gid
 175    *   Contact group id.
 176    *
 177    * @param int $searchLimit
 178    *  Limit for the number of contacts to be used for comparison.
 179    *  The search methodology finds all matches for the searchedContacts so this limits
 180    *  the number of searched contacts, not the matches found.
 181    *
 182    * @return array
 183    *   array of (cid1, cid2, weight) dupe triples
 184    */
 185   public static function dupesInGroup($rgid, $gid, $searchLimit = 0) {
 186     $cids = array_keys(CRM_Contact_BAO_Group::getMember($gid, TRUE, $searchLimit));
 187     if (!empty($cids)) {
 188       return self::dupes($rgid, $cids);
 189     }
 190     return [];
 191   }
 192
 193   /**
 194    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 195    * object into a valid $params array for dedupe
 196    *
 197    * @param array $fields
 198    *   Contact structure from formRule().
 199    * @param string $ctype
 200    *   Contact type of the given contact.
 201    *
 202    * @return array
 203    *   valid $params array for dedupe
 204    */
 205   public static function formatParams($fields, $ctype) {
 206     $flat = [];
 207     CRM_Utils_Array::flatten($fields, $flat);
 208
 209     // FIXME: This may no longer be necessary - check inputs
 210     $replace_these = [
 211       'individual_prefix' => 'prefix_id',
 212       'individual_suffix' => 'suffix_id',
 213       'gender' => 'gender_id',
 214     ];
 215     foreach (['individual_suffix', 'individual_prefix', 'gender'] as $name) {
 216       if (!empty($fields[$name])) {
 217         $flat[$replace_these[$name]] = $flat[$name];
 218         unset($flat[$name]);
 219       }
 220     }
 221
 222     // handle {birth,deceased}_date
 223     foreach ([
 224       'birth_date',
 225       'deceased_date',
 226     ] as $date) {
 227       if (!empty($fields[$date])) {
 228         $flat[$date] = $fields[$date];
 229         if (is_array($flat[$date])) {
 230           $flat[$date] = CRM_Utils_Date::format($flat[$date]);
 231         }
 232         $flat[$date] = CRM_Utils_Date::processDate($flat[$date]);
 233       }
 234     }
 235
 236     if (!empty($flat['contact_source'])) {
 237       $flat['source'] = $flat['contact_source'];
 238       unset($flat['contact_source']);
 239     }
 240
 241     // handle preferred_communication_method
 242     if (!empty($fields['preferred_communication_method'])) {
 243       $methods = array_intersect($fields['preferred_communication_method'], ['1']);
 244       $methods = array_keys($methods);
 245       sort($methods);
 246       if ($methods) {
 247         $flat['preferred_communication_method'] = CRM_Core_DAO::VALUE_SEPARATOR . implode(CRM_Core_DAO::VALUE_SEPARATOR, $methods) . CRM_Core_DAO::VALUE_SEPARATOR;
 248       }
 249     }
 250
 251     // handle custom data
 252     $tree = CRM_Core_BAO_CustomGroup::getTree($ctype, NULL, NULL, -1);
 253     CRM_Core_BAO_CustomGroup::postProcess($tree, $fields, TRUE);
 254     foreach ($tree as $key => $cg) {
 255       if (!is_int($key)) {
 256         continue;
 257       }
 258       foreach ($cg['fields'] as $cf) {
 259         $flat[$cf['column_name']] = CRM_Utils_Array::value('data', $cf['customValue']);
 260       }
 261     }
 262
 263     // if the key is dotted, keep just the last part of it
 264     foreach ($flat as $key => $value) {
 265       if (substr_count($key, '.')) {
 266         $last = explode('.', $key);
 267         $last = array_pop($last);
 268         // make sure the first occurrence is kept, not the last
 269         if (!isset($flat[$last])) {
 270           $flat[$last] = $value;
 271         }
 272         unset($flat[$key]);
 273       }
 274     }
 275
 276     // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
 277     // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
 278     // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
 279     foreach ($flat as $key => $value) {
 280       $matches = [];
 281       if (preg_match('/(.*)-(Primary-[\d+])$|(.*)-(\d+|Primary)$/', $key, $matches)) {
 282         $return = array_values(array_filter($matches));
 283         // make sure the first occurrence is kept, not the last
 284         $flat[$return[1]] = empty($flat[$return[1]]) ? $value : $flat[$return[1]];
 285         unset($flat[$key]);
 286       }
 287     }
 288
 289     $params = [];
 290     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 291     if (is_array($supportedFields)) {
 292       foreach ($supportedFields as $table => $fields) {
 293         if ($table == 'civicrm_address') {
 294           // for matching on civicrm_address fields, we also need the location_type_id
 295           $fields['location_type_id'] = '';
 296           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 297           $fixes = [
 298             'address_name' => 'name',
 299             'country' => 'country_id',
 300             'state_province' => 'state_province_id',
 301             'county' => 'county_id',
 302           ];
 303           foreach ($fixes as $orig => $target) {
 304             if (!empty($flat[$orig])) {
 305               $params[$table][$target] = $flat[$orig];
 306             }
 307           }
 308         }
 309         if ($table == 'civicrm_phone') {
 310           $fixes = [
 311             'phone' => 'phone_numeric',
 312           ];
 313           foreach ($fixes as $orig => $target) {
 314             if (!empty($flat[$orig])) {
 315               $params[$table][$target] = $flat[$orig];
 316             }
 317           }
 318         }
 319         foreach ($fields as $field => $title) {
 320           if (!empty($flat[$field])) {
 321             $params[$table][$field] = $flat[$field];
 322           }
 323         }
 324       }
 325     }
 326     return $params;
 327   }
 328
 329   /**
 330    * Parse duplicate pairs into a standardised array and store in the prev_next_cache.
 331    *
 332    * @param array $foundDupes
 333    * @param string $cacheKeyString
 334    *
 335    * @return array
 336    *   Dupe pairs with the keys
 337    *   -srcID
 338    *   -srcName
 339    *   -dstID
 340    *   -dstName
 341    *   -weight
 342    *   -canMerge
 343    *
 344    * @throws CRM_Core_Exception
 345    */
 346   public static function parseAndStoreDupePairs($foundDupes, $cacheKeyString) {
 347     $cids = [];
 348     foreach ($foundDupes as $dupe) {
 349       $cids[$dupe[0]] = 1;
 350       $cids[$dupe[1]] = 1;
 351     }
 352     $cidString = implode(', ', array_keys($cids));
 353
 354     $dao = CRM_Core_DAO::executeQuery("SELECT id, display_name FROM civicrm_contact WHERE id IN ($cidString) ORDER BY sort_name");
 355     $displayNames = [];
 356     while ($dao->fetch()) {
 357       $displayNames[$dao->id] = $dao->display_name;
 358     }
 359
 360     $userId = CRM_Core_Session::getLoggedInContactID();
 361     foreach ($foundDupes as $dupes) {
 362       $srcID = $dupes[1];
 363       $dstID = $dupes[0];
 364       // The logged in user should never be the src (ie. the contact to be removed).
 365       if ($srcID == $userId) {
 366         $srcID = $dstID;
 367         $dstID = $userId;
 368       }
 369
 370       $mainContacts[] = $row = [
 371         'dstID' => $dstID,
 372         'dstName' => $displayNames[$dstID],
 373         'srcID' => $srcID,
 374         'srcName' => $displayNames[$srcID],
 375         'weight' => $dupes[2],
 376         'canMerge' => TRUE,
 377       ];
 378
 379       $data = CRM_Core_DAO::escapeString(serialize($row));
 380       CRM_Core_BAO_PrevNextCache::setItem('civicrm_contact', $dstID, $srcID, $cacheKeyString, $data);
 381     }
 382     return $mainContacts;
 383   }
 384
 385 }