CRM/Dedupe/Finder.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | CiviCRM version 4.7                                                |
   5  +--------------------------------------------------------------------+
   6  | Copyright CiviCRM LLC (c) 2004-2017                                |
   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) 2004-2017
  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 $limit
  55    *   Optional limit. This limits the number of contacts for which the code will
  56    *   attempt to find matches.
  57    *
  58    * @return array
  59    *   Array of (cid1, cid2, weight) dupe triples
  60    *
  61    * @throws CiviCRM_API3_Exception
  62    * @throws Exception
  63    */
  64   public static function dupes($rgid, $cids = array(), $checkPermissions = TRUE, $limit = NULL) {
  65     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
  66     $rgBao->id = $rgid;
  67     $rgBao->contactIds = $cids;
  68     if (!$rgBao->find(TRUE)) {
  69       CRM_Core_Error::fatal("Dedupe rule not found for selected contacts");
  70     }
  71     if (empty($rgBao->contactIds) && !empty($limit)) {
  72       $limitedContacts = civicrm_api3('Contact', 'get', array(
  73         'return' => 'id',
  74         'contact_type' => $rgBao->contact_type,
  75         'options' => array('limit' => $limit),
  76       ));
  77       $rgBao->contactIds = array_keys($limitedContacts['values']);
  78     }
  79
  80     $rgBao->fillTable();
  81     $dao = new CRM_Core_DAO();
  82     $dao->query($rgBao->thresholdQuery($checkPermissions));
  83     $dupes = array();
  84     while ($dao->fetch()) {
  85       $dupes[] = array($dao->id1, $dao->id2, $dao->weight);
  86     }
  87     $dao->query($rgBao->tableDropQuery());
  88
  89     return $dupes;
  90   }
  91
  92   /**
  93    * Return an array of possible dupes, based on the provided array of
  94    * params, using the default rule group for the given contact type and
  95    * usage.
  96    *
  97    * check_permission is a boolean flag to indicate if permission should be considered.
  98    * default is to always check permissioning but public pages for example might not want
  99    * permission to be checked for anonymous users. Refer CRM-6211. We might be beaking
 100    * Multi-Site dedupe for public pages.
 101    *
 102    * @param array $params
 103    *   Array of params of the form $params[$table][$field] == $value.
 104    * @param string $ctype
 105    *   Contact type to match against.
 106    * @param string $used
 107    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
 108    * @param array $except
 109    *   Array of contacts that shouldn't be considered dupes.
 110    * @param int $ruleGroupID
 111    *   The id of the dedupe rule we should be using.
 112    *
 113    * @return array
 114    *   matching contact ids
 115    */
 116   public static function dupesByParams(
 117     $params,
 118     $ctype,
 119     $used = 'Unsupervised',
 120     $except = array(),
 121     $ruleGroupID = NULL
 122   ) {
 123     // If $params is empty there is zero reason to proceed.
 124     if (!$params) {
 125       return array();
 126     }
 127
 128     $foundByID = FALSE;
 129     if ($ruleGroupID) {
 130       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 131       $rgBao->id = $ruleGroupID;
 132       $rgBao->contact_type = $ctype;
 133       if ($rgBao->find(TRUE)) {
 134         $foundByID = TRUE;
 135       }
 136     }
 137
 138     if (!$foundByID) {
 139       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 140       $rgBao->contact_type = $ctype;
 141       $rgBao->used = $used;
 142       if (!$rgBao->find(TRUE)) {
 143         CRM_Core_Error::fatal("$used rule for $ctype does not exist");
 144       }
 145     }
 146     $params['check_permission'] = CRM_Utils_Array::value('check_permission', $params, TRUE);
 147
 148     $rgBao->params = $params;
 149     $rgBao->fillTable();
 150     $dao = new CRM_Core_DAO();
 151     $dao->query($rgBao->thresholdQuery($params['check_permission']));
 152     $dupes = array();
 153     while ($dao->fetch()) {
 154       if (isset($dao->id) && $dao->id) {
 155         $dupes[] = $dao->id;
 156       }
 157     }
 158     $dao->query($rgBao->tableDropQuery());
 159     return array_diff($dupes, $except);
 160   }
 161
 162   /**
 163    * Return a contact_id-keyed array of arrays of possible dupes in the given group.
 164    *
 165    * @param int $rgid
 166    *   Rule group id.
 167    * @param int $gid
 168    *   Contact group id (currently, works only with non-smart groups).
 169    *
 170    * @param int $limit
 171    * @return array
 172    *   array of (cid1, cid2, weight) dupe triples
 173    */
 174   public static function dupesInGroup($rgid, $gid, $limit = NULL) {
 175     $cids = array_keys(CRM_Contact_BAO_Group::getMember($gid, $limit));
 176     if (!empty($cids)) {
 177       return self::dupes($rgid, $cids);
 178     }
 179     return array();
 180   }
 181
 182   /**
 183    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 184    * object into a valid $params array for dedupe
 185    *
 186    * @param array $fields
 187    *   Contact structure from formRule().
 188    * @param string $ctype
 189    *   Contact type of the given contact.
 190    *
 191    * @return array
 192    *   valid $params array for dedupe
 193    */
 194   public static function formatParams($fields, $ctype) {
 195     $flat = array();
 196     CRM_Utils_Array::flatten($fields, $flat);
 197
 198     // FIXME: This may no longer be necessary - check inputs
 199     $replace_these = array(
 200       'individual_prefix' => 'prefix_id',
 201       'individual_suffix' => 'suffix_id',
 202       'gender' => 'gender_id',
 203     );
 204     foreach (array('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 (array(
 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'], array('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, CRM_Core_DAO::$_nullObject, NULL, -1);
 242     CRM_Core_BAO_CustomGroup::postProcess($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       $matches = array();
 270       if (preg_match('/(.*)-(Primary-[\d+])$|(.*)-(\d+|Primary)$/', $key, $matches)) {
 271         $return = array_values(array_filter($matches));
 272         $flat[$return[1]] = $value;
 273         unset($flat[$key]);
 274       }
 275     }
 276
 277     $params = array();
 278     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 279     if (is_array($supportedFields)) {
 280       foreach ($supportedFields as $table => $fields) {
 281         if ($table == 'civicrm_address') {
 282           // for matching on civicrm_address fields, we also need the location_type_id
 283           $fields['location_type_id'] = '';
 284           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 285           $fixes = array(
 286             'address_name' => 'name',
 287             'country' => 'country_id',
 288             'state_province' => 'state_province_id',
 289             'county' => 'county_id',
 290           );
 291           foreach ($fixes as $orig => $target) {
 292             if (!empty($flat[$orig])) {
 293               $params[$table][$target] = $flat[$orig];
 294             }
 295           }
 296         }
 297         foreach ($fields as $field => $title) {
 298           if (!empty($flat[$field])) {
 299             $params[$table][$field] = $flat[$field];
 300           }
 301         }
 302       }
 303     }
 304     return $params;
 305   }
 306
 307   /**
 308    * Parse duplicate pairs into a standardised array and store in the prev_next_cache.
 309    *
 310    * @param array $foundDupes
 311    * @param string $cacheKeyString
 312    *
 313    * @return array Dupe pairs with the keys
 314    *   Dupe pairs with the keys
 315    *   -srcID
 316    *   -srcName
 317    *   -dstID
 318    *   -dstName
 319    *   -weight
 320    *   -canMerge
 321    *
 322    * @throws CRM_Core_Exception
 323    */
 324   public static function parseAndStoreDupePairs($foundDupes, $cacheKeyString) {
 325     $cids = array();
 326     foreach ($foundDupes as $dupe) {
 327       $cids[$dupe[0]] = 1;
 328       $cids[$dupe[1]] = 1;
 329     }
 330     $cidString = implode(', ', array_keys($cids));
 331
 332     $dao = CRM_Core_DAO::executeQuery("SELECT id, display_name FROM civicrm_contact WHERE id IN ($cidString) ORDER BY sort_name");
 333     $displayNames = array();
 334     while ($dao->fetch()) {
 335       $displayNames[$dao->id] = $dao->display_name;
 336     }
 337
 338     $userId = CRM_Core_Session::getLoggedInContactID();
 339     foreach ($foundDupes as $dupes) {
 340       $srcID = $dupes[1];
 341       $dstID = $dupes[0];
 342       // The logged in user should never be the src (ie. the contact to be removed).
 343       if ($srcID == $userId) {
 344         $srcID = $dstID;
 345         $dstID = $userId;
 346       }
 347
 348       $mainContacts[] = $row = array(
 349         'dstID' => $dstID,
 350         'dstName' => $displayNames[$dstID],
 351         'srcID' => $srcID,
 352         'srcName' => $displayNames[$srcID],
 353         'weight' => $dupes[2],
 354         'canMerge' => TRUE,
 355       );
 356
 357       $data = CRM_Core_DAO::escapeString(serialize($row));
 358       $values[] = " ( 'civicrm_contact', $dstID, $srcID, '$cacheKeyString', '$data' ) ";
 359     }
 360     CRM_Core_BAO_PrevNextCache::setItem($values);
 361     return $mainContacts;
 362   }
 363
 364 }