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