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
  65     $innerJoinClauses = [
  66       "t1.{$this->rule_field} IS NOT NULL",
  67       "t2.{$this->rule_field} IS NOT NULL",
  68       "t1.{$this->rule_field} = t2.{$this->rule_field}",
  69     ];
  70     if ($this->getFieldType($this->rule_field) === CRM_Utils_Type::T_DATE) {
  71       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  72       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  73     }
  74     else {
  75       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  76       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  77     }
  78
  79     switch ($this->rule_table) {
  80       case 'civicrm_contact':
  81         $id = 'id';
  82         //we should restrict by contact type in the first step
  83         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  84         $ct = CRM_Core_DAO::singleValueQuery($sql);
  85         if ($this->params) {
  86           $where[] = "t1.contact_type = '{$ct}'";
  87         }
  88         else {
  89           $where[] = "t1.contact_type = '{$ct}'";
  90           $where[] = "t2.contact_type = '{$ct}'";
  91         }
  92         break;
  93
  94       case 'civicrm_address':
  95         $id = 'contact_id';
  96         $on[] = 't1.location_type_id = t2.location_type_id';
  97         $innerJoinClauses[] = 't1.location_type_id = t2.location_type_id';
  98         if (!empty($this->params['civicrm_address']['location_type_id'])) {
  99           $locTypeId = CRM_Utils_Type::escape($this->params['civicrm_address']['location_type_id'], 'Integer', FALSE);
 100           if ($locTypeId) {
 101             $where[] = "t1.location_type_id = $locTypeId";
 102           }
 103         }
 104         break;
 105
 106       case 'civicrm_email':
 107       case 'civicrm_im':
 108       case 'civicrm_openid':
 109       case 'civicrm_phone':
 110         $id = 'contact_id';
 111         break;
 112
 113       case 'civicrm_note':
 114         $id = 'entity_id';
 115         if ($this->params) {
 116           $where[] = "t1.entity_table = 'civicrm_contact'";
 117         }
 118         else {
 119           $where[] = "t1.entity_table = 'civicrm_contact'";
 120           $where[] = "t2.entity_table = 'civicrm_contact'";
 121         }
 122         break;
 123
 124       default:
 125         // custom data tables
 126         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 127           $id = 'entity_id';
 128         }
 129         else {
 130           throw new CRM_Core_Exception("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 131         }
 132         break;
 133     }
 134
 135     // build SELECT based on the field names containing contact ids
 136     // if there are params provided, id1 should be 0
 137     if ($this->params) {
 138       $select = "t1.$id id1, {$this->rule_weight} weight";
 139       $subSelect = 'id1, weight';
 140     }
 141     else {
 142       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 143       $subSelect = 'id1, id2, weight';
 144     }
 145
 146     // build FROM (and WHERE, if it's a parametrised search)
 147     // based on whether the rule is about substrings or not
 148     if ($this->params) {
 149       $from = "{$this->rule_table} t1";
 150       $str = 'NULL';
 151       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 152         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 153       }
 154       if ($this->rule_length) {
 155         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 156         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 157       }
 158       else {
 159         $where[] = "t1.{$this->rule_field} = '$str'";
 160       }
 161     }
 162     else {
 163       if ($this->rule_length) {
 164         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 165       }
 166       else {
 167         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 168       }
 169     }
 170
 171     // finish building WHERE, also limit the results if requested
 172     if (!$this->params) {
 173       $where[] = "t1.$id < t2.$id";
 174     }
 175     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 176     if ($this->contactIds) {
 177       $cids = [];
 178       foreach ($this->contactIds as $cid) {
 179         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 180       }
 181       if (count($cids) == 1) {
 182         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 183       }
 184       else {
 185         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 186         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 187       }
 188       // The `weight` is ambiguous in the context of the union; put the whole
 189       // thing in a subquery.
 190       $query = "SELECT $subSelect FROM ($query) subunion";
 191     }
 192
 193     return $query;
 194   }
 195
 196   /**
 197    * find fields related to a rule group.
 198    *
 199    * @param array $params contains the rule group property to identify rule group
 200    *
 201    * @return array
 202    *   rule fields array associated to rule group
 203    */
 204   public static function dedupeRuleFields($params) {
 205     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 206     $rgBao->used = $params['used'];
 207     $rgBao->contact_type = $params['contact_type'];
 208     $rgBao->find(TRUE);
 209
 210     $ruleBao = new CRM_Dedupe_BAO_Rule();
 211     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 212     $ruleBao->find();
 213     $ruleFields = [];
 214     while ($ruleBao->fetch()) {
 215       $field_name = $ruleBao->rule_field;
 216       if ($field_name == 'phone_numeric') {
 217         $field_name = 'phone';
 218       }
 219       $ruleFields[] = $field_name;
 220     }
 221     return $ruleFields;
 222   }
 223
 224   /**
 225    * @param int $cid
 226    * @param int $oid
 227    *
 228    * @return bool
 229    */
 230   public static function validateContacts($cid, $oid) {
 231     if (!$cid || !$oid) {
 232       return NULL;
 233     }
 234     $exception = new CRM_Dedupe_DAO_Exception();
 235     $exception->contact_id1 = $cid;
 236     $exception->contact_id2 = $oid;
 237     //make sure contact2 > contact1.
 238     if ($cid > $oid) {
 239       $exception->contact_id1 = $oid;
 240       $exception->contact_id2 = $cid;
 241     }
 242
 243     return $exception->find(TRUE) ? FALSE : TRUE;
 244   }
 245
 246   /**
 247    * Get the specification for the given field.
 248    *
 249    * @param string $fieldName
 250    *
 251    * @return array
 252    * @throws \CiviCRM_API3_Exception
 253    */
 254   public function getFieldType($fieldName) {
 255     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
 256     if (!$entity) {
 257       // This means we have stored a custom field rather than an entity name in rule_table, figure out the entity.
 258       $entity = civicrm_api3('CustomGroup', 'getvalue', ['table_name' => $this->rule_table, 'return' => 'extends']);
 259       if (in_array($entity, ['Individual', 'Household', 'Organization'])) {
 260         $entity = 'Contact';
 261       }
 262       $fieldName = 'custom_' . civicrm_api3('CustomField', 'getvalue', ['column_name' => $fieldName, 'return' => 'id']);
 263     }
 264     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
 265     return $fields[$fieldName]['type'];
 266   }
 267
 268 }