CRM/Dedupe/Finder.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | CiviCRM version 4.7                                                |
   5  +--------------------------------------------------------------------+
   6  | Copyright CiviCRM LLC (c) 2004-2016                                |
   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) 2004-2016
  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    * Return dupes of a given contact, using the default rule group (of a provided usage).
 184    *
 185    * @param int $cid
 186    *   Contact id of the given contact.
 187    * @param string $used
 188    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
 189    * @param string $ctype
 190    *   Contact type of the given contact.
 191    *
 192    * @return array
 193    *   array of dupe contact_ids
 194    */
 195   public static function dupesOfContact($cid, $used = 'Unsupervised', $ctype = NULL) {
 196     // if not provided, fetch the contact type from the database
 197     if (!$ctype) {
 198       $dao = new CRM_Contact_DAO_Contact();
 199       $dao->id = $cid;
 200       if (!$dao->find(TRUE)) {
 201         CRM_Core_Error::fatal("contact id of $cid does not exist");
 202       }
 203       $ctype = $dao->contact_type;
 204     }
 205     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 206     $rgBao->used = $used;
 207     $rgBao->contact_type = $ctype;
 208     if (!$rgBao->find(TRUE)) {
 209       CRM_Core_Error::fatal("$used rule for $ctype does not exist");
 210     }
 211     $dupes = self::dupes($rgBao->id, array($cid));
 212
 213     // get the dupes for this cid
 214     $result = array();
 215     foreach ($dupes as $dupe) {
 216       if ($dupe[0] == $cid) {
 217         $result[] = $dupe[1];
 218       }
 219       elseif ($dupe[1] == $cid) {
 220         $result[] = $dupe[0];
 221       }
 222     }
 223     return $result;
 224   }
 225
 226   /**
 227    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 228    * object into a valid $params array for dedupe
 229    *
 230    * @param array $fields
 231    *   Contact structure from formRule().
 232    * @param string $ctype
 233    *   Contact type of the given contact.
 234    *
 235    * @return array
 236    *   valid $params array for dedupe
 237    */
 238   public static function formatParams($fields, $ctype) {
 239     $flat = array();
 240     CRM_Utils_Array::flatten($fields, $flat);
 241
 242     // FIXME: This may no longer be necessary - check inputs
 243     $replace_these = array(
 244       'individual_prefix' => 'prefix_id',
 245       'individual_suffix' => 'suffix_id',
 246       'gender' => 'gender_id',
 247     );
 248     foreach (array('individual_suffix', 'individual_prefix', 'gender') as $name) {
 249       if (!empty($fields[$name])) {
 250         $flat[$replace_these[$name]] = $flat[$name];
 251         unset($flat[$name]);
 252       }
 253     }
 254
 255     // handle {birth,deceased}_date
 256     foreach (array(
 257       'birth_date',
 258       'deceased_date',
 259     ) as $date) {
 260       if (!empty($fields[$date])) {
 261         $flat[$date] = $fields[$date];
 262         if (is_array($flat[$date])) {
 263           $flat[$date] = CRM_Utils_Date::format($flat[$date]);
 264         }
 265         $flat[$date] = CRM_Utils_Date::processDate($flat[$date]);
 266       }
 267     }
 268
 269     if (!empty($flat['contact_source'])) {
 270       $flat['source'] = $flat['contact_source'];
 271       unset($flat['contact_source']);
 272     }
 273
 274     // handle preferred_communication_method
 275     if (!empty($fields['preferred_communication_method'])) {
 276       $methods = array_intersect($fields['preferred_communication_method'], array('1'));
 277       $methods = array_keys($methods);
 278       sort($methods);
 279       if ($methods) {
 280         $flat['preferred_communication_method'] = CRM_Core_DAO::VALUE_SEPARATOR . implode(CRM_Core_DAO::VALUE_SEPARATOR, $methods) . CRM_Core_DAO::VALUE_SEPARATOR;
 281       }
 282     }
 283
 284     // handle custom data
 285     $tree = CRM_Core_BAO_CustomGroup::getTree($ctype, CRM_Core_DAO::$_nullObject, NULL, -1);
 286     CRM_Core_BAO_CustomGroup::postProcess($tree, $fields, TRUE);
 287     foreach ($tree as $key => $cg) {
 288       if (!is_int($key)) {
 289         continue;
 290       }
 291       foreach ($cg['fields'] as $cf) {
 292         $flat[$cf['column_name']] = CRM_Utils_Array::value('data', $cf['customValue']);
 293       }
 294     }
 295
 296     // if the key is dotted, keep just the last part of it
 297     foreach ($flat as $key => $value) {
 298       if (substr_count($key, '.')) {
 299         $last = explode('.', $key);
 300         $last = array_pop($last);
 301         // make sure the first occurrence is kept, not the last
 302         if (!isset($flat[$last])) {
 303           $flat[$last] = $value;
 304         }
 305         unset($flat[$key]);
 306       }
 307     }
 308
 309     // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
 310     // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
 311     // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
 312     foreach ($flat as $key => $value) {
 313       $matches = array();
 314       if (preg_match('/(.*)-(Primary-[\d+])$|(.*)-(\d+|Primary)$/', $key, $matches)) {
 315         $return = array_values(array_filter($matches));
 316         $flat[$return[1]] = $value;
 317         unset($flat[$key]);
 318       }
 319     }
 320
 321     $params = array();
 322     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 323     if (is_array($supportedFields)) {
 324       foreach ($supportedFields as $table => $fields) {
 325         if ($table == 'civicrm_address') {
 326           // for matching on civicrm_address fields, we also need the location_type_id
 327           $fields['location_type_id'] = '';
 328           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 329           $fixes = array(
 330             'address_name' => 'name',
 331             'country' => 'country_id',
 332             'state_province' => 'state_province_id',
 333             'county' => 'county_id',
 334           );
 335           foreach ($fixes as $orig => $target) {
 336             if (!empty($flat[$orig])) {
 337               $params[$table][$target] = $flat[$orig];
 338             }
 339           }
 340         }
 341         foreach ($fields as $field => $title) {
 342           if (!empty($flat[$field])) {
 343             $params[$table][$field] = $flat[$field];
 344           }
 345         }
 346       }
 347     }
 348     return $params;
 349   }
 350
 351 }