CRM/Dedupe/Finder.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | CiviCRM version 4.6                                                |
   5  +--------------------------------------------------------------------+
   6  | Copyright CiviCRM LLC (c) 2004-2014                                |
   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) 2004-2014
  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 $cids
  49    *   Contact ids to limit the search to.
  50    *
  51    * @return array  array of (cid1, cid2, weight) dupe triples
  52    */
  53   public static function dupes($rgid, $cids = array()) {
  54     $rgBao             = new CRM_Dedupe_BAO_RuleGroup();
  55     $rgBao->id         = $rgid;
  56     $rgBao->contactIds = $cids;
  57     if (!$rgBao->find(TRUE)) {
  58       CRM_Core_Error::fatal("Dedupe rule not found for selected contacts");
  59     }
  60
  61     $rgBao->fillTable();
  62     $dao = new CRM_Core_DAO();
  63     $dao->query($rgBao->thresholdQuery());
  64     $dupes = array();
  65     while ($dao->fetch()) {
  66       $dupes[] = array($dao->id1, $dao->id2, $dao->weight);
  67     }
  68     $dao->query($rgBao->tableDropQuery());
  69
  70     return $dupes;
  71   }
  72
  73   /**
  74    * Return an array of possible dupes, based on the provided array of
  75    * params, using the default rule group for the given contact type and
  76    * usage.
  77    *
  78    * check_permission is a boolean flag to indicate if permission should be considered.
  79    * default is to always check permissioning but public pages for example might not want
  80    * permission to be checked for anonymous users. Refer CRM-6211. We might be beaking
  81    * Multi-Site dedupe for public pages.
  82    *
  83    * @param array $params
  84    *   Array of params of the form $params[$table][$field] == $value.
  85    * @param string $ctype
  86    *   Contact type to match against.
  87    * @param string $used
  88    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
  89    * @param array $except
  90    *   Array of contacts that shouldn't be considered dupes.
  91    * @param int $ruleGroupID
  92    *   The id of the dedupe rule we should be using.
  93    *
  94    * @return array  matching contact ids
  95    */
  96   static function dupesByParams(
  97     $params,
  98     $ctype,
  99     $used = 'Unsupervised',
 100     $except = array(),
 101     $ruleGroupID = NULL
 102   ) {
 103     // If $params is empty there is zero reason to proceed.
 104     if (!$params) {
 105       return array();
 106     }
 107
 108     $foundByID = FALSE;
 109     if ($ruleGroupID) {
 110       $rgBao               = new CRM_Dedupe_BAO_RuleGroup();
 111       $rgBao->id           = $ruleGroupID;
 112       $rgBao->contact_type = $ctype;
 113       if ($rgBao->find(TRUE)) {
 114         $foundByID = TRUE;
 115       }
 116     }
 117
 118     if (!$foundByID) {
 119       $rgBao               = new CRM_Dedupe_BAO_RuleGroup();
 120       $rgBao->contact_type = $ctype;
 121       $rgBao->used         = $used;
 122       if (!$rgBao->find(TRUE)) {
 123         CRM_Core_Error::fatal("$used rule for $ctype does not exist");
 124       }
 125     }
 126     $params['check_permission'] = CRM_Utils_Array::value('check_permission', $params, TRUE);
 127
 128     $rgBao->params = $params;
 129     $rgBao->fillTable();
 130     $dao = new CRM_Core_DAO();
 131     $dao->query($rgBao->thresholdQuery($params['check_permission']));
 132     $dupes = array();
 133     while ($dao->fetch()) {
 134       if (isset($dao->id) && $dao->id) {
 135         $dupes[] = $dao->id;
 136       }
 137     }
 138     $dao->query($rgBao->tableDropQuery());
 139     return array_diff($dupes, $except);
 140   }
 141
 142   /**
 143    * Return a contact_id-keyed array of arrays of possible dupes in the given group.
 144    *
 145    * @param int $rgid
 146    *   Rule group id.
 147    * @param int $gid
 148    *   Contact group id (currently, works only with non-smart groups).
 149    *
 150    * @return array  array of (cid1, cid2, weight) dupe triples
 151    */
 152   public static function dupesInGroup($rgid, $gid) {
 153     $cids = array_keys(CRM_Contact_BAO_Group::getMember($gid));
 154     if (!empty($cids)) {
 155       return self::dupes($rgid, $cids);
 156     }
 157     return array();
 158   }
 159
 160   /**
 161    * Return dupes of a given contact, using the default rule group (of a provided usage).
 162    *
 163    * @param int $cid
 164    *   Contact id of the given contact.
 165    * @param string $used
 166    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
 167    * @param string $ctype
 168    *   Contact type of the given contact.
 169    *
 170    * @return array  array of dupe contact_ids
 171    */
 172   public static function dupesOfContact($cid, $used = 'Unsupervised', $ctype = NULL) {
 173     // if not provided, fetch the contact type from the database
 174     if (!$ctype) {
 175       $dao = new CRM_Contact_DAO_Contact();
 176       $dao->id = $cid;
 177       if (!$dao->find(TRUE)) {
 178         CRM_Core_Error::fatal("contact id of $cid does not exist");
 179       }
 180       $ctype = $dao->contact_type;
 181     }
 182     $rgBao               = new CRM_Dedupe_BAO_RuleGroup();
 183     $rgBao->used         = $used;
 184     $rgBao->contact_type = $ctype;
 185     if (!$rgBao->find(TRUE)) {
 186       CRM_Core_Error::fatal("$used rule for $ctype does not exist");
 187     }
 188     $dupes = self::dupes($rgBao->id, array($cid));
 189
 190     // get the dupes for this cid
 191     $result = array();
 192     foreach ($dupes as $dupe) {
 193       if ($dupe[0] == $cid) {
 194         $result[] = $dupe[1];
 195       }
 196       elseif ($dupe[1] == $cid) {
 197         $result[] = $dupe[0];
 198       }
 199     }
 200     return $result;
 201   }
 202
 203   /**
 204    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 205    * object into a valid $params array for dedupe
 206    *
 207    * @param array $fields
 208    *   Contact structure from formRule().
 209    * @param string $ctype
 210    *   Contact type of the given contact.
 211    *
 212    * @return array  valid $params array for dedupe
 213    */
 214   public static function formatParams($fields, $ctype) {
 215     $flat = array();
 216     CRM_Utils_Array::flatten($fields, $flat);
 217
 218     // FIXME: This may no longer be necessary - check inputs
 219     $replace_these = array(
 220       'individual_prefix' => 'prefix_id',
 221       'individual_suffix' => 'suffix_id',
 222       'gender' => 'gender_id',
 223     );
 224     foreach (array('individual_suffix', 'individual_prefix', 'gender') as $name) {
 225       if (!empty($fields[$name])) {
 226         $flat[$replace_these[$name]] = $flat[$name];
 227         unset($flat[$name]);
 228       }
 229     }
 230
 231     // handle {birth,deceased}_date
 232     foreach (array(
 233       'birth_date', 'deceased_date') as $date) {
 234       if (!empty($fields[$date])) {
 235         $flat[$date] = $fields[$date];
 236         if (is_array($flat[$date])) {
 237           $flat[$date] = CRM_Utils_Date::format($flat[$date]);
 238         }
 239         $flat[$date] = CRM_Utils_Date::processDate($flat[$date]);
 240       }
 241     }
 242
 243     if (!empty($flat['contact_source'])) {
 244       $flat['source'] = $flat['contact_source'];
 245       unset($flat['contact_source']);
 246     }
 247
 248     // handle preferred_communication_method
 249     if (!empty($fields['preferred_communication_method'])) {
 250       $methods = array_intersect($fields['preferred_communication_method'], array('1'));
 251       $methods = array_keys($methods);
 252       sort($methods);
 253       if ($methods) {
 254         $flat['preferred_communication_method'] = CRM_Core_DAO::VALUE_SEPARATOR . implode(CRM_Core_DAO::VALUE_SEPARATOR, $methods) . CRM_Core_DAO::VALUE_SEPARATOR;
 255       }
 256     }
 257
 258     // handle custom data
 259     $tree = CRM_Core_BAO_CustomGroup::getTree($ctype, CRM_Core_DAO::$_nullObject, NULL, -1);
 260     CRM_Core_BAO_CustomGroup::postProcess($tree, $fields, TRUE);
 261     foreach ($tree as $key => $cg) {
 262       if (!is_int($key)) {
 263         continue;
 264       }
 265       foreach ($cg['fields'] as $cf) {
 266         $flat[$cf['column_name']] = CRM_Utils_Array::value('data', $cf['customValue']);
 267       }
 268     }
 269
 270     // if the key is dotted, keep just the last part of it
 271     foreach ($flat as $key => $value) {
 272       if (substr_count($key, '.')) {
 273         $last = explode('.', $key);
 274         $last = array_pop($last);
 275         // make sure the first occurence is kept, not the last
 276         if (!isset($flat[$last])) {
 277           $flat[$last] = $value;
 278         }
 279         unset($flat[$key]);
 280       }
 281     }
 282
 283     // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
 284     // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
 285     // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
 286     foreach ($flat as $key => $value) {
 287       $matches = array();
 288       if (preg_match('/(.*)-(\d+|Primary)$/', $key, $matches)) {
 289         $flat[$matches[1]] = $value;
 290         unset($flat[$key]);
 291       }
 292     }
 293
 294     $params = array();
 295     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 296     if (is_array($supportedFields)) {
 297       foreach ($supportedFields as $table => $fields) {
 298         if ($table == 'civicrm_address') {
 299           // for matching on civicrm_address fields, we also need the location_type_id
 300           $fields['location_type_id'] = '';
 301           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 302           $fixes = array(
 303             'address_name' => 'name',
 304           'country' => 'country_id',
 305             'state_province' => 'state_province_id',
 306           'county' => 'county_id',
 307           );
 308           foreach ($fixes as $orig => $target) {
 309             if (!empty($flat[$orig])) {
 310               $params[$table][$target] = $flat[$orig];
 311             }
 312           }
 313         }
 314         foreach ($fields as $field => $title) {
 315           if (!empty($flat[$field])) {
 316             $params[$table][$field] = $flat[$field];
 317           }
 318         }
 319       }
 320     }
 321     return $params;
 322   }
 323 }