CRM/Dedupe/BAO/Rule.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | Copyright CiviCRM LLC. All rights reserved.                        |
   5  |                                                                    |
   6  | This work is published under the GNU AGPLv3 license with some      |
   7  | permitted exceptions and without any warranty. For full license    |
   8  | and copyright information, see https://civicrm.org/licensing       |
   9  +--------------------------------------------------------------------+
  10  */
  11
  12 /**
  13  *
  14  * @package CRM
  15  * @copyright CiviCRM LLC https://civicrm.org/licensing
  16  * $Id$
  17  *
  18  */
  19
  20 /**
  21  * The CiviCRM duplicate discovery engine is based on an
  22  * algorithm designed by David Strauss <david@fourkitchens.com>.
  23  */
  24 class CRM_Dedupe_BAO_Rule extends CRM_Dedupe_DAO_Rule {
  25
  26   /**
  27    * Ids of the contacts to limit the SQL queries (whole-database queries otherwise)
  28    * @var array
  29    */
  30   public $contactIds = [];
  31
  32   /**
  33    * Params to dedupe against (queries against the whole contact set otherwise)
  34    * @var array
  35    */
  36   public $params = [];
  37
  38   /**
  39    * Return the SQL query for the given rule - either for finding matching
  40    * pairs of contacts, or for matching against the $params variable (if set).
  41    *
  42    * @return string
  43    *   SQL query performing the search
  44    */
  45   public function sql() {
  46     if ($this->params &&
  47       (!array_key_exists($this->rule_table, $this->params) ||
  48         !array_key_exists($this->rule_field, $this->params[$this->rule_table])
  49       )
  50     ) {
  51       // if params is present and doesn't have an entry for a field, don't construct the clause.
  52       return NULL;
  53     }
  54
  55     // we need to initialise WHERE, ON and USING here, as some table types
  56     // extend them; $where is an array of required conditions, $on and
  57     // $using are arrays of required field matchings (for substring and
  58     // full matches, respectively)
  59     $where = [];
  60     $on = ["SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})"];
  61     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
  62     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
  63
  64     $innerJoinClauses = [
  65       "t1.{$this->rule_field} IS NOT NULL",
  66       "t2.{$this->rule_field} IS NOT NULL",
  67       "t1.{$this->rule_field} = t2.{$this->rule_field}",
  68     ];
  69     if ($fields[$this->rule_field]['type'] === CRM_Utils_Type::T_DATE) {
  70       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  71       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  72     }
  73     else {
  74       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  75       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  76     }
  77
  78     switch ($this->rule_table) {
  79       case 'civicrm_contact':
  80         $id = 'id';
  81         //we should restrict by contact type in the first step
  82         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  83         $ct = CRM_Core_DAO::singleValueQuery($sql);
  84         if ($this->params) {
  85           $where[] = "t1.contact_type = '{$ct}'";
  86         }
  87         else {
  88           $where[] = "t1.contact_type = '{$ct}'";
  89           $where[] = "t2.contact_type = '{$ct}'";
  90         }
  91         break;
  92
  93       case 'civicrm_address':
  94         $id = 'contact_id';
  95         $on[] = 't1.location_type_id = t2.location_type_id';
  96         $innerJoinClauses[] = 't1.location_type_id = t2.location_type_id';
  97         if (!empty($this->params['civicrm_address']['location_type_id'])) {
  98           $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE);
  99           if ($locTypeId) {
 100             $where[] = "t1.location_type_id = $locTypeId";
 101           }
 102         }
 103         break;
 104
 105       case 'civicrm_email':
 106       case 'civicrm_im':
 107       case 'civicrm_openid':
 108       case 'civicrm_phone':
 109         $id = 'contact_id';
 110         break;
 111
 112       case 'civicrm_note':
 113         $id = 'entity_id';
 114         if ($this->params) {
 115           $where[] = "t1.entity_table = 'civicrm_contact'";
 116         }
 117         else {
 118           $where[] = "t1.entity_table = 'civicrm_contact'";
 119           $where[] = "t2.entity_table = 'civicrm_contact'";
 120         }
 121         break;
 122
 123       default:
 124         // custom data tables
 125         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 126           $id = 'entity_id';
 127         }
 128         else {
 129           CRM_Core_Error::fatal("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 130         }
 131         break;
 132     }
 133
 134     // build SELECT based on the field names containing contact ids
 135     // if there are params provided, id1 should be 0
 136     if ($this->params) {
 137       $select = "t1.$id id1, {$this->rule_weight} weight";
 138       $subSelect = 'id1, weight';
 139     }
 140     else {
 141       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 142       $subSelect = 'id1, id2, weight';
 143     }
 144
 145     // build FROM (and WHERE, if it's a parametrised search)
 146     // based on whether the rule is about substrings or not
 147     if ($this->params) {
 148       $from = "{$this->rule_table} t1";
 149       $str = 'NULL';
 150       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 151         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 152       }
 153       if ($this->rule_length) {
 154         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 155         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 156       }
 157       else {
 158         $where[] = "t1.{$this->rule_field} = '$str'";
 159       }
 160     }
 161     else {
 162       if ($this->rule_length) {
 163         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 164       }
 165       else {
 166         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 167       }
 168     }
 169
 170     // finish building WHERE, also limit the results if requested
 171     if (!$this->params) {
 172       $where[] = "t1.$id < t2.$id";
 173     }
 174     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 175     if ($this->contactIds) {
 176       $cids = [];
 177       foreach ($this->contactIds as $cid) {
 178         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 179       }
 180       if (count($cids) == 1) {
 181         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 182       }
 183       else {
 184         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 185         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 186       }
 187       // The `weight` is ambiguous in the context of the union; put the whole
 188       // thing in a subquery.
 189       $query = "SELECT $subSelect FROM ($query) subunion";
 190     }
 191
 192     return $query;
 193   }
 194
 195   /**
 196    * find fields related to a rule group.
 197    *
 198    * @param array $params contains the rule group property to identify rule group
 199    *
 200    * @return array
 201    *   rule fields array associated to rule group
 202    */
 203   public static function dedupeRuleFields($params) {
 204     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 205     $rgBao->used = $params['used'];
 206     $rgBao->contact_type = $params['contact_type'];
 207     $rgBao->find(TRUE);
 208
 209     $ruleBao = new CRM_Dedupe_BAO_Rule();
 210     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 211     $ruleBao->find();
 212     $ruleFields = [];
 213     while ($ruleBao->fetch()) {
 214       $ruleFields[] = $ruleBao->rule_field;
 215     }
 216     return $ruleFields;
 217   }
 218
 219   /**
 220    * @param int $cid
 221    * @param int $oid
 222    *
 223    * @return bool
 224    */
 225   public static function validateContacts($cid, $oid) {
 226     if (!$cid || !$oid) {
 227       return NULL;
 228     }
 229     $exception = new CRM_Dedupe_DAO_Exception();
 230     $exception->contact_id1 = $cid;
 231     $exception->contact_id2 = $oid;
 232     //make sure contact2 > contact1.
 233     if ($cid > $oid) {
 234       $exception->contact_id1 = $oid;
 235       $exception->contact_id2 = $cid;
 236     }
 237
 238     return $exception->find(TRUE) ? FALSE : TRUE;
 239   }
 240
 241 }