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 $params variable (if set).
  41    *
  42    * @return string
  43    *   SQL query performing the search
  44    *
  45    * @throws \CRM_Core_Exception
  46    * @throws \CiviCRM_API3_Exception
  47    */
  48   public function sql() {
  49     if ($this->params &&
  50       (!array_key_exists($this->rule_table, $this->params) ||
  51         !array_key_exists($this->rule_field, $this->params[$this->rule_table])
  52       )
  53     ) {
  54       // if params is present and doesn't have an entry for a field, don't construct the clause.
  55       return NULL;
  56     }
  57
  58     // we need to initialise WHERE, ON and USING here, as some table types
  59     // extend them; $where is an array of required conditions, $on and
  60     // $using are arrays of required field matchings (for substring and
  61     // full matches, respectively)
  62     $where = [];
  63     $on = ["SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})"];
  64     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
  65     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
  66
  67     $innerJoinClauses = [
  68       "t1.{$this->rule_field} IS NOT NULL",
  69       "t2.{$this->rule_field} IS NOT NULL",
  70       "t1.{$this->rule_field} = t2.{$this->rule_field}",
  71     ];
  72     if ($fields[$this->rule_field]['type'] === CRM_Utils_Type::T_DATE) {
  73       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  74       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  75     }
  76     else {
  77       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  78       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  79     }
  80
  81     switch ($this->rule_table) {
  82       case 'civicrm_contact':
  83         $id = 'id';
  84         //we should restrict by contact type in the first step
  85         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  86         $ct = CRM_Core_DAO::singleValueQuery($sql);
  87         if ($this->params) {
  88           $where[] = "t1.contact_type = '{$ct}'";
  89         }
  90         else {
  91           $where[] = "t1.contact_type = '{$ct}'";
  92           $where[] = "t2.contact_type = '{$ct}'";
  93         }
  94         break;
  95
  96       case 'civicrm_address':
  97         $id = 'contact_id';
  98         $on[] = 't1.location_type_id = t2.location_type_id';
  99         $innerJoinClauses[] = 't1.location_type_id = t2.location_type_id';
 100         if (!empty($this->params['civicrm_address']['location_type_id'])) {
 101           $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE);
 102           if ($locTypeId) {
 103             $where[] = "t1.location_type_id = $locTypeId";
 104           }
 105         }
 106         break;
 107
 108       case 'civicrm_email':
 109       case 'civicrm_im':
 110       case 'civicrm_openid':
 111       case 'civicrm_phone':
 112         $id = 'contact_id';
 113         break;
 114
 115       case 'civicrm_note':
 116         $id = 'entity_id';
 117         if ($this->params) {
 118           $where[] = "t1.entity_table = 'civicrm_contact'";
 119         }
 120         else {
 121           $where[] = "t1.entity_table = 'civicrm_contact'";
 122           $where[] = "t2.entity_table = 'civicrm_contact'";
 123         }
 124         break;
 125
 126       default:
 127         // custom data tables
 128         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 129           $id = 'entity_id';
 130         }
 131         else {
 132           throw new CRM_Core_Exception("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 133         }
 134         break;
 135     }
 136
 137     // build SELECT based on the field names containing contact ids
 138     // if there are params provided, id1 should be 0
 139     if ($this->params) {
 140       $select = "t1.$id id1, {$this->rule_weight} weight";
 141       $subSelect = 'id1, weight';
 142     }
 143     else {
 144       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 145       $subSelect = 'id1, id2, weight';
 146     }
 147
 148     // build FROM (and WHERE, if it's a parametrised search)
 149     // based on whether the rule is about substrings or not
 150     if ($this->params) {
 151       $from = "{$this->rule_table} t1";
 152       $str = 'NULL';
 153       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 154         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 155       }
 156       if ($this->rule_length) {
 157         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 158         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 159       }
 160       else {
 161         $where[] = "t1.{$this->rule_field} = '$str'";
 162       }
 163     }
 164     else {
 165       if ($this->rule_length) {
 166         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 167       }
 168       else {
 169         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 170       }
 171     }
 172
 173     // finish building WHERE, also limit the results if requested
 174     if (!$this->params) {
 175       $where[] = "t1.$id < t2.$id";
 176     }
 177     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 178     if ($this->contactIds) {
 179       $cids = [];
 180       foreach ($this->contactIds as $cid) {
 181         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 182       }
 183       if (count($cids) == 1) {
 184         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 185       }
 186       else {
 187         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 188         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 189       }
 190       // The `weight` is ambiguous in the context of the union; put the whole
 191       // thing in a subquery.
 192       $query = "SELECT $subSelect FROM ($query) subunion";
 193     }
 194
 195     return $query;
 196   }
 197
 198   /**
 199    * find fields related to a rule group.
 200    *
 201    * @param array $params contains the rule group property to identify rule group
 202    *
 203    * @return array
 204    *   rule fields array associated to rule group
 205    */
 206   public static function dedupeRuleFields($params) {
 207     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 208     $rgBao->used = $params['used'];
 209     $rgBao->contact_type = $params['contact_type'];
 210     $rgBao->find(TRUE);
 211
 212     $ruleBao = new CRM_Dedupe_BAO_Rule();
 213     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 214     $ruleBao->find();
 215     $ruleFields = [];
 216     while ($ruleBao->fetch()) {
 217       $ruleFields[] = $ruleBao->rule_field;
 218     }
 219     return $ruleFields;
 220   }
 221
 222   /**
 223    * @param int $cid
 224    * @param int $oid
 225    *
 226    * @return bool
 227    */
 228   public static function validateContacts($cid, $oid) {
 229     if (!$cid || !$oid) {
 230       return NULL;
 231     }
 232     $exception = new CRM_Dedupe_DAO_Exception();
 233     $exception->contact_id1 = $cid;
 234     $exception->contact_id2 = $oid;
 235     //make sure contact2 > contact1.
 236     if ($cid > $oid) {
 237       $exception->contact_id1 = $oid;
 238       $exception->contact_id2 = $cid;
 239     }
 240
 241     return $exception->find(TRUE) ? FALSE : TRUE;
 242   }
 243
 244 }