CRM/Dedupe/Finder.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | Copyright CiviCRM LLC. All rights reserved.                        |
   5  |                                                                    |
   6  | This work is published under the GNU AGPLv3 license with some      |
   7  | permitted exceptions and without any warranty. For full license    |
   8  | and copyright information, see https://civicrm.org/licensing       |
   9  +--------------------------------------------------------------------+
  10  */
  11
  12 /**
  13  *
  14  * @package CRM
  15  * @copyright CiviCRM LLC https://civicrm.org/licensing
  16  * $Id$
  17  *
  18  */
  19
  20 /**
  21  * The CiviCRM duplicate discovery engine is based on an
  22  * algorithm designed by David Strauss <david@fourkitchens.com>.
  23  */
  24 class CRM_Dedupe_Finder {
  25
  26   /**
  27    * Return a contact_id-keyed array of arrays of possible dupes
  28    * (of the key contact_id) - limited to dupes of $cids if provided.
  29    *
  30    * @param int $rgid
  31    *   Rule group id.
  32    * @param array $cids
  33    *   Contact ids to limit the search to.
  34    *
  35    * @param bool $checkPermissions
  36    *   Respect logged in user permissions.
  37    *
  38    * @return array
  39    *   Array of (cid1, cid2, weight) dupe triples
  40    *
  41    * @throws Exception
  42    */
  43   public static function dupes($rgid, $cids = [], $checkPermissions = TRUE) {
  44     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
  45     $rgBao->id = $rgid;
  46     $rgBao->contactIds = $cids;
  47     if (!$rgBao->find(TRUE)) {
  48       throw new CRM_Core_Exception('Dedupe rule not found for selected contacts');
  49     }
  50
  51     $rgBao->fillTable();
  52     $dao = new CRM_Core_DAO();
  53     $dao->query($rgBao->thresholdQuery($checkPermissions));
  54     $dupes = [];
  55     while ($dao->fetch()) {
  56       $dupes[] = [$dao->id1, $dao->id2, $dao->weight];
  57     }
  58     $dao->query($rgBao->tableDropQuery());
  59
  60     return $dupes;
  61   }
  62
  63   /**
  64    * Return an array of possible dupes, based on the provided array of
  65    * params, using the default rule group for the given contact type and
  66    * usage.
  67    *
  68    * check_permission is a boolean flag to indicate if permission should be considered.
  69    * default is to always check permissioning but public pages for example might not want
  70    * permission to be checked for anonymous users. Refer CRM-6211. We might be breaking
  71    * Multi-Site dedupe for public pages.
  72    *
  73    * @param array $params
  74    *   Array of params of the form $params[$table][$field] == $value.
  75    * @param string $ctype
  76    *   Contact type to match against.
  77    * @param string $used
  78    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
  79    * @param array $except
  80    *   Array of contacts that shouldn't be considered dupes.
  81    * @param int $ruleGroupID
  82    *   The id of the dedupe rule we should be using.
  83    *
  84    * @return array
  85    *   matching contact ids
  86    * @throws \CRM_Core_Exception
  87    */
  88   public static function dupesByParams(
  89     $params,
  90     $ctype,
  91     $used = 'Unsupervised',
  92     $except = [],
  93     $ruleGroupID = NULL
  94   ) {
  95     // If $params is empty there is zero reason to proceed.
  96     if (!$params) {
  97       return [];
  98     }
  99     $checkPermission = CRM_Utils_Array::value('check_permission', $params, TRUE);
 100     // This may no longer be required - see https://github.com/civicrm/civicrm-core/pull/13176
 101     $params = array_filter($params);
 102
 103     $foundByID = FALSE;
 104     if ($ruleGroupID) {
 105       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 106       $rgBao->id = $ruleGroupID;
 107       $rgBao->contact_type = $ctype;
 108       if ($rgBao->find(TRUE)) {
 109         $foundByID = TRUE;
 110       }
 111     }
 112
 113     if (!$foundByID) {
 114       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 115       $rgBao->contact_type = $ctype;
 116       $rgBao->used = $used;
 117       if (!$rgBao->find(TRUE)) {
 118         throw new CRM_Core_Exception("$used rule for $ctype does not exist");
 119       }
 120     }
 121
 122     if (isset($params['civicrm_phone']['phone_numeric'])) {
 123       $orig = $params['civicrm_phone']['phone_numeric'];
 124       $params['civicrm_phone']['phone_numeric'] = preg_replace('/[^\d]/', '', $orig);
 125     }
 126     $rgBao->params = $params;
 127     $rgBao->fillTable();
 128     $dao = new CRM_Core_DAO();
 129     $dao->query($rgBao->thresholdQuery($checkPermission));
 130     $dupes = [];
 131     while ($dao->fetch()) {
 132       if (isset($dao->id) && $dao->id) {
 133         $dupes[] = $dao->id;
 134       }
 135     }
 136     $dao->query($rgBao->tableDropQuery());
 137     return array_diff($dupes, $except);
 138   }
 139
 140   /**
 141    * Return a contact_id-keyed array of arrays of possible dupes in the given group.
 142    *
 143    * @param int $rgid
 144    *   Rule group id.
 145    * @param int $gid
 146    *   Contact group id.
 147    *
 148    * @param int $searchLimit
 149    *  Limit for the number of contacts to be used for comparison.
 150    *  The search methodology finds all matches for the searchedContacts so this limits
 151    *  the number of searched contacts, not the matches found.
 152    *
 153    * @return array
 154    *   array of (cid1, cid2, weight) dupe triples
 155    * @throws \CiviCRM_API3_Exception
 156    */
 157   public static function dupesInGroup($rgid, $gid, $searchLimit = 0) {
 158     $cids = array_keys(CRM_Contact_BAO_Group::getMember($gid, TRUE, $searchLimit));
 159     if (!empty($cids)) {
 160       return self::dupes($rgid, $cids);
 161     }
 162     return [];
 163   }
 164
 165   /**
 166    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 167    * object into a valid $params array for dedupe
 168    *
 169    * @param array $fields
 170    *   Contact structure from formRule().
 171    * @param string $ctype
 172    *   Contact type of the given contact.
 173    *
 174    * @return array
 175    *   valid $params array for dedupe
 176    * @throws \CRM_Core_Exception
 177    */
 178   public static function formatParams($fields, $ctype) {
 179     $flat = [];
 180     CRM_Utils_Array::flatten($fields, $flat);
 181
 182     // FIXME: This may no longer be necessary - check inputs
 183     $replace_these = [
 184       'individual_prefix' => 'prefix_id',
 185       'individual_suffix' => 'suffix_id',
 186       'gender' => 'gender_id',
 187     ];
 188     foreach (['individual_suffix', 'individual_prefix', 'gender'] as $name) {
 189       if (!empty($fields[$name])) {
 190         $flat[$replace_these[$name]] = $flat[$name];
 191         unset($flat[$name]);
 192       }
 193     }
 194
 195     // handle {birth,deceased}_date
 196     foreach ([
 197       'birth_date',
 198       'deceased_date',
 199     ] as $date) {
 200       if (!empty($fields[$date])) {
 201         $flat[$date] = $fields[$date];
 202         if (is_array($flat[$date])) {
 203           $flat[$date] = CRM_Utils_Date::format($flat[$date]);
 204         }
 205         $flat[$date] = CRM_Utils_Date::processDate($flat[$date]);
 206       }
 207     }
 208
 209     if (!empty($flat['contact_source'])) {
 210       $flat['source'] = $flat['contact_source'];
 211       unset($flat['contact_source']);
 212     }
 213
 214     // handle preferred_communication_method
 215     if (!empty($fields['preferred_communication_method'])) {
 216       $methods = array_intersect($fields['preferred_communication_method'], ['1']);
 217       $methods = array_keys($methods);
 218       sort($methods);
 219       if ($methods) {
 220         $flat['preferred_communication_method'] = CRM_Core_DAO::VALUE_SEPARATOR . implode(CRM_Core_DAO::VALUE_SEPARATOR, $methods) . CRM_Core_DAO::VALUE_SEPARATOR;
 221       }
 222     }
 223
 224     // handle custom data
 225     $tree = CRM_Core_BAO_CustomGroup::getTree($ctype, NULL, NULL, -1);
 226     CRM_Core_BAO_CustomGroup::postProcess($tree, $fields, TRUE);
 227     foreach ($tree as $key => $cg) {
 228       if (!is_int($key)) {
 229         continue;
 230       }
 231       foreach ($cg['fields'] as $cf) {
 232         $flat[$cf['column_name']] = CRM_Utils_Array::value('data', $cf['customValue']);
 233       }
 234     }
 235
 236     // if the key is dotted, keep just the last part of it
 237     foreach ($flat as $key => $value) {
 238       if (substr_count($key, '.')) {
 239         $last = explode('.', $key);
 240         $last = array_pop($last);
 241         // make sure the first occurrence is kept, not the last
 242         if (!isset($flat[$last])) {
 243           $flat[$last] = $value;
 244         }
 245         unset($flat[$key]);
 246       }
 247     }
 248
 249     // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
 250     // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
 251     // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
 252     foreach ($flat as $key => $value) {
 253       $matches = [];
 254       if (preg_match('/(.*)-(Primary-[\d+])$|(.*)-(\d+|Primary)$/', $key, $matches)) {
 255         $return = array_values(array_filter($matches));
 256         // make sure the first occurrence is kept, not the last
 257         $flat[$return[1]] = empty($flat[$return[1]]) ? $value : $flat[$return[1]];
 258         unset($flat[$key]);
 259       }
 260     }
 261
 262     $params = [];
 263     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 264     if (is_array($supportedFields)) {
 265       foreach ($supportedFields as $table => $fields) {
 266         if ($table == 'civicrm_address') {
 267           // for matching on civicrm_address fields, we also need the location_type_id
 268           $fields['location_type_id'] = '';
 269           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 270           $fixes = [
 271             'address_name' => 'name',
 272             'country' => 'country_id',
 273             'state_province' => 'state_province_id',
 274             'county' => 'county_id',
 275           ];
 276           foreach ($fixes as $orig => $target) {
 277             if (!empty($flat[$orig])) {
 278               $params[$table][$target] = $flat[$orig];
 279             }
 280           }
 281         }
 282         if ($table === 'civicrm_phone') {
 283           $fixes = [
 284             'phone' => 'phone_numeric',
 285           ];
 286           foreach ($fixes as $orig => $target) {
 287             if (!empty($flat[$orig])) {
 288               $params[$table][$target] = $flat[$orig];
 289             }
 290           }
 291         }
 292         foreach ($fields as $field => $title) {
 293           if (!empty($flat[$field])) {
 294             $params[$table][$field] = $flat[$field];
 295           }
 296         }
 297       }
 298     }
 299     return $params;
 300   }
 301
 302   /**
 303    * Parse duplicate pairs into a standardised array and store in the prev_next_cache.
 304    *
 305    * @param array $foundDupes
 306    * @param string $cacheKeyString
 307    *
 308    * @return array
 309    *   Dupe pairs with the keys
 310    *   -srcID
 311    *   -srcName
 312    *   -dstID
 313    *   -dstName
 314    *   -weight
 315    *   -canMerge
 316    */
 317   public static function parseAndStoreDupePairs($foundDupes, $cacheKeyString) {
 318     $cids = [];
 319     foreach ($foundDupes as $dupe) {
 320       $cids[$dupe[0]] = 1;
 321       $cids[$dupe[1]] = 1;
 322     }
 323     $cidString = implode(', ', array_keys($cids));
 324
 325     $dao = CRM_Core_DAO::executeQuery("SELECT id, display_name FROM civicrm_contact WHERE id IN ($cidString) ORDER BY sort_name");
 326     $displayNames = [];
 327     while ($dao->fetch()) {
 328       $displayNames[$dao->id] = $dao->display_name;
 329     }
 330
 331     $userId = CRM_Core_Session::getLoggedInContactID();
 332     foreach ($foundDupes as $dupes) {
 333       $srcID = $dupes[1];
 334       $dstID = $dupes[0];
 335       // The logged in user should never be the src (ie. the contact to be removed).
 336       if ($srcID == $userId) {
 337         $srcID = $dstID;
 338         $dstID = $userId;
 339       }
 340
 341       $mainContacts[] = $row = [
 342         'dstID' => $dstID,
 343         'dstName' => $displayNames[$dstID],
 344         'srcID' => $srcID,
 345         'srcName' => $displayNames[$srcID],
 346         'weight' => $dupes[2],
 347         'canMerge' => TRUE,
 348       ];
 349
 350       $data = CRM_Core_DAO::escapeString(serialize($row));
 351       CRM_Core_BAO_PrevNextCache::setItem('civicrm_contact', $dstID, $srcID, $cacheKeyString, $data);
 352     }
 353     return $mainContacts;
 354   }
 355
 356 }