CRM/Dedupe/BAO/Rule.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_BAO_Rule extends CRM_Dedupe_DAO_Rule {
  41
  42   /**
  43    * Ids of the contacts to limit the SQL queries (whole-database queries otherwise)
  44    */
  45   var $contactIds = array();
  46
  47   /**
  48    * Params to dedupe against (queries against the whole contact set otherwise)
  49    */
  50   var $params = array();
  51
  52   /**
  53    * Return the SQL query for the given rule - either for finding matching
  54    * pairs of contacts, or for matching against the $params variable (if set).
  55    *
  56    * @return string
  57    *   SQL query performing the search
  58    */
  59   public function sql() {
  60     if ($this->params &&
  61       (!array_key_exists($this->rule_table, $this->params) ||
  62         !array_key_exists($this->rule_field, $this->params[$this->rule_table])
  63       )
  64     ) {
  65       // if params is present and doesn't have an entry for a field, don't construct the clause.
  66       return NULL;
  67     }
  68
  69     // we need to initialise WHERE, ON and USING here, as some table types
  70     // extend them; $where is an array of required conditions, $on and
  71     // $using are arrays of required field matchings (for substring and
  72     // full matches, respectively)
  73     $where = array();
  74     $on = array("SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})");
  75     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
  76     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
  77
  78     $innerJoinClauses = [
  79       "t1.{$this->rule_field} IS NOT NULL",
  80       "t2.{$this->rule_field} IS NOT NULL",
  81       "t1.{$this->rule_field} = t2.{$this->rule_field}"
  82     ];
  83     if ($fields[$this->rule_field]['type'] === CRM_Utils_Type::T_DATE) {
  84       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  85       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  86     }
  87     else {
  88       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  89       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  90     }
  91
  92     switch ($this->rule_table) {
  93       case 'civicrm_contact':
  94         $id = 'id';
  95         //we should restrict by contact type in the first step
  96         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  97         $ct = CRM_Core_DAO::singleValueQuery($sql);
  98         if ($this->params) {
  99           $where[] = "t1.contact_type = '{$ct}'";
 100         }
 101         else {
 102           $where[] = "t1.contact_type = '{$ct}'";
 103           $where[] = "t2.contact_type = '{$ct}'";
 104         }
 105         break;
 106
 107       case 'civicrm_address':
 108         $id = 'contact_id';
 109         $on[] = 't1.location_type_id = t2.location_type_id';
 110         $innerJoinClauses[] = ['t1.location_type_id = t2.location_type_id'];
 111         if ($this->params['civicrm_address']['location_type_id']) {
 112           $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE);
 113           if ($locTypeId) {
 114             $where[] = "t1.location_type_id = $locTypeId";
 115           }
 116         }
 117         break;
 118
 119       case 'civicrm_email':
 120       case 'civicrm_im':
 121       case 'civicrm_openid':
 122       case 'civicrm_phone':
 123         $id = 'contact_id';
 124         break;
 125
 126       case 'civicrm_note':
 127         $id = 'entity_id';
 128         if ($this->params) {
 129           $where[] = "t1.entity_table = 'civicrm_contact'";
 130         }
 131         else {
 132           $where[] = "t1.entity_table = 'civicrm_contact'";
 133           $where[] = "t2.entity_table = 'civicrm_contact'";
 134         }
 135         break;
 136
 137       default:
 138         // custom data tables
 139         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 140           $id = 'entity_id';
 141         }
 142         else {
 143           CRM_Core_Error::fatal("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 144         }
 145         break;
 146     }
 147
 148     // build SELECT based on the field names containing contact ids
 149     // if there are params provided, id1 should be 0
 150     if ($this->params) {
 151       $select = "t1.$id id1, {$this->rule_weight} weight";
 152       $subSelect = 'id1, weight';
 153     }
 154     else {
 155       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 156       $subSelect = 'id1, id2, weight';
 157     }
 158
 159     // build FROM (and WHERE, if it's a parametrised search)
 160     // based on whether the rule is about substrings or not
 161     if ($this->params) {
 162       $from = "{$this->rule_table} t1";
 163       $str = 'NULL';
 164       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 165         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 166       }
 167       if ($this->rule_length) {
 168         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 169         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 170       }
 171       else {
 172         $where[] = "t1.{$this->rule_field} = '$str'";
 173       }
 174     }
 175     else {
 176       if ($this->rule_length) {
 177         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 178       }
 179       else {
 180         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 181       }
 182     }
 183
 184     // finish building WHERE, also limit the results if requested
 185     if (!$this->params) {
 186       $where[] = "t1.$id < t2.$id";
 187     }
 188     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 189     if ($this->contactIds) {
 190       $cids = array();
 191       foreach ($this->contactIds as $cid) {
 192         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 193       }
 194       if (count($cids) == 1) {
 195         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 196       }
 197       else {
 198         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 199         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 200       }
 201       // The `weight` is ambiguous in the context of the union; put the whole
 202       // thing in a subquery.
 203       $query = "SELECT $subSelect FROM ($query) subunion";
 204     }
 205
 206     return $query;
 207   }
 208
 209   /**
 210    * find fields related to a rule group.
 211    *
 212    * @param array $params contains the rule group property to identify rule group
 213    *
 214    * @return array
 215    *   rule fields array associated to rule group
 216    */
 217   public static function dedupeRuleFields($params) {
 218     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 219     $rgBao->used = $params['used'];
 220     $rgBao->contact_type = $params['contact_type'];
 221     $rgBao->find(TRUE);
 222
 223     $ruleBao = new CRM_Dedupe_BAO_Rule();
 224     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 225     $ruleBao->find();
 226     $ruleFields = array();
 227     while ($ruleBao->fetch()) {
 228       $ruleFields[] = $ruleBao->rule_field;
 229     }
 230     return $ruleFields;
 231   }
 232
 233   /**
 234    * @param int $cid
 235    * @param int $oid
 236    *
 237    * @return bool
 238    */
 239   public static function validateContacts($cid, $oid) {
 240     if (!$cid || !$oid) {
 241       return NULL;
 242     }
 243     $exception = new CRM_Dedupe_DAO_Exception();
 244     $exception->contact_id1 = $cid;
 245     $exception->contact_id2 = $oid;
 246     //make sure contact2 > contact1.
 247     if ($cid > $oid) {
 248       $exception->contact_id1 = $oid;
 249       $exception->contact_id2 = $cid;
 250     }
 251
 252     return $exception->find(TRUE) ? FALSE : TRUE;
 253   }
 254
 255 }