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    * @var array
  45    */
  46   public $contactIds = [];
  47
  48   /**
  49    * Params to dedupe against (queries against the whole contact set otherwise)
  50    * @var array
  51    */
  52   public $params = [];
  53
  54   /**
  55    * Return the SQL query for the given rule - either for finding matching
  56    * pairs of contacts, or for matching against the $params variable (if set).
  57    *
  58    * @return string
  59    *   SQL query performing the search
  60    */
  61   public function sql() {
  62     if ($this->params &&
  63       (!array_key_exists($this->rule_table, $this->params) ||
  64         !array_key_exists($this->rule_field, $this->params[$this->rule_table])
  65       )
  66     ) {
  67       // if params is present and doesn't have an entry for a field, don't construct the clause.
  68       return NULL;
  69     }
  70
  71     // we need to initialise WHERE, ON and USING here, as some table types
  72     // extend them; $where is an array of required conditions, $on and
  73     // $using are arrays of required field matchings (for substring and
  74     // full matches, respectively)
  75     $where = [];
  76     $on = ["SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})"];
  77     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
  78     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
  79
  80     $innerJoinClauses = [
  81       "t1.{$this->rule_field} IS NOT NULL",
  82       "t2.{$this->rule_field} IS NOT NULL",
  83       "t1.{$this->rule_field} = t2.{$this->rule_field}",
  84     ];
  85     if ($fields[$this->rule_field]['type'] === CRM_Utils_Type::T_DATE) {
  86       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  87       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  88     }
  89     else {
  90       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  91       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  92     }
  93
  94     switch ($this->rule_table) {
  95       case 'civicrm_contact':
  96         $id = 'id';
  97         //we should restrict by contact type in the first step
  98         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  99         $ct = CRM_Core_DAO::singleValueQuery($sql);
 100         if ($this->params) {
 101           $where[] = "t1.contact_type = '{$ct}'";
 102         }
 103         else {
 104           $where[] = "t1.contact_type = '{$ct}'";
 105           $where[] = "t2.contact_type = '{$ct}'";
 106         }
 107         break;
 108
 109       case 'civicrm_address':
 110         $id = 'contact_id';
 111         $on[] = 't1.location_type_id = t2.location_type_id';
 112         $innerJoinClauses[] = 't1.location_type_id = t2.location_type_id';
 113         if (!empty($this->params['civicrm_address']['location_type_id'])) {
 114           $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE);
 115           if ($locTypeId) {
 116             $where[] = "t1.location_type_id = $locTypeId";
 117           }
 118         }
 119         break;
 120
 121       case 'civicrm_email':
 122       case 'civicrm_im':
 123       case 'civicrm_openid':
 124       case 'civicrm_phone':
 125         $id = 'contact_id';
 126         break;
 127
 128       case 'civicrm_note':
 129         $id = 'entity_id';
 130         if ($this->params) {
 131           $where[] = "t1.entity_table = 'civicrm_contact'";
 132         }
 133         else {
 134           $where[] = "t1.entity_table = 'civicrm_contact'";
 135           $where[] = "t2.entity_table = 'civicrm_contact'";
 136         }
 137         break;
 138
 139       default:
 140         // custom data tables
 141         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 142           $id = 'entity_id';
 143         }
 144         else {
 145           CRM_Core_Error::fatal("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 146         }
 147         break;
 148     }
 149
 150     // build SELECT based on the field names containing contact ids
 151     // if there are params provided, id1 should be 0
 152     if ($this->params) {
 153       $select = "t1.$id id1, {$this->rule_weight} weight";
 154       $subSelect = 'id1, weight';
 155     }
 156     else {
 157       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 158       $subSelect = 'id1, id2, weight';
 159     }
 160
 161     // build FROM (and WHERE, if it's a parametrised search)
 162     // based on whether the rule is about substrings or not
 163     if ($this->params) {
 164       $from = "{$this->rule_table} t1";
 165       $str = 'NULL';
 166       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 167         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 168       }
 169       if ($this->rule_length) {
 170         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 171         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 172       }
 173       else {
 174         $where[] = "t1.{$this->rule_field} = '$str'";
 175       }
 176     }
 177     else {
 178       if ($this->rule_length) {
 179         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 180       }
 181       else {
 182         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 183       }
 184     }
 185
 186     // finish building WHERE, also limit the results if requested
 187     if (!$this->params) {
 188       $where[] = "t1.$id < t2.$id";
 189     }
 190     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 191     if ($this->contactIds) {
 192       $cids = [];
 193       foreach ($this->contactIds as $cid) {
 194         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 195       }
 196       if (count($cids) == 1) {
 197         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 198       }
 199       else {
 200         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 201         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 202       }
 203       // The `weight` is ambiguous in the context of the union; put the whole
 204       // thing in a subquery.
 205       $query = "SELECT $subSelect FROM ($query) subunion";
 206     }
 207
 208     return $query;
 209   }
 210
 211   /**
 212    * find fields related to a rule group.
 213    *
 214    * @param array $params contains the rule group property to identify rule group
 215    *
 216    * @return array
 217    *   rule fields array associated to rule group
 218    */
 219   public static function dedupeRuleFields($params) {
 220     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 221     $rgBao->used = $params['used'];
 222     $rgBao->contact_type = $params['contact_type'];
 223     $rgBao->find(TRUE);
 224
 225     $ruleBao = new CRM_Dedupe_BAO_Rule();
 226     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 227     $ruleBao->find();
 228     $ruleFields = [];
 229     while ($ruleBao->fetch()) {
 230       $ruleFields[] = $ruleBao->rule_field;
 231     }
 232     return $ruleFields;
 233   }
 234
 235   /**
 236    * @param int $cid
 237    * @param int $oid
 238    *
 239    * @return bool
 240    */
 241   public static function validateContacts($cid, $oid) {
 242     if (!$cid || !$oid) {
 243       return NULL;
 244     }
 245     $exception = new CRM_Dedupe_DAO_Exception();
 246     $exception->contact_id1 = $cid;
 247     $exception->contact_id2 = $oid;
 248     //make sure contact2 > contact1.
 249     if ($cid > $oid) {
 250       $exception->contact_id1 = $oid;
 251       $exception->contact_id2 = $cid;
 252     }
 253
 254     return $exception->find(TRUE) ? FALSE : TRUE;
 255   }
 256
 257 }