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  */
  17
  18 /**
  19  * The CiviCRM duplicate discovery engine is based on an
  20  * algorithm designed by David Strauss <david@fourkitchens.com>.
  21  */
  22 class CRM_Dedupe_BAO_Rule extends CRM_Dedupe_DAO_Rule {
  23
  24   /**
  25    * Ids of the contacts to limit the SQL queries (whole-database queries otherwise)
  26    * @var array
  27    */
  28   public $contactIds = [];
  29
  30   /**
  31    * Params to dedupe against (queries against the whole contact set otherwise)
  32    * @var array
  33    */
  34   public $params = [];
  35
  36   /**
  37    * Return the SQL query for the given rule - either for finding matching
  38    * pairs of contacts, or for matching against the $params variable (if set).
  39    *
  40    * @return string
  41    *   SQL query performing the search
  42    *
  43    * @throws \CRM_Core_Exception
  44    * @throws \CiviCRM_API3_Exception
  45    */
  46   public function sql() {
  47     if ($this->params &&
  48       (!array_key_exists($this->rule_table, $this->params) ||
  49         !array_key_exists($this->rule_field, $this->params[$this->rule_table])
  50       )
  51     ) {
  52       // if params is present and doesn't have an entry for a field, don't construct the clause.
  53       return NULL;
  54     }
  55
  56     // we need to initialise WHERE, ON and USING here, as some table types
  57     // extend them; $where is an array of required conditions, $on and
  58     // $using are arrays of required field matchings (for substring and
  59     // full matches, respectively)
  60     $where = [];
  61     $on = ["SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR(t2.{$this->rule_field}, 1, {$this->rule_length})"];
  62
  63     $innerJoinClauses = [
  64       "t1.{$this->rule_field} IS NOT NULL",
  65       "t2.{$this->rule_field} IS NOT NULL",
  66       "t1.{$this->rule_field} = t2.{$this->rule_field}",
  67     ];
  68     if ($this->getFieldType($this->rule_field) === CRM_Utils_Type::T_DATE) {
  69       $innerJoinClauses[] = "t1.{$this->rule_field} > '1000-01-01'";
  70       $innerJoinClauses[] = "t2.{$this->rule_field} > '1000-01-01'";
  71     }
  72     else {
  73       $innerJoinClauses[] = "t1.{$this->rule_field} <> ''";
  74       $innerJoinClauses[] = "t2.{$this->rule_field} <> ''";
  75     }
  76
  77     switch ($this->rule_table) {
  78       case 'civicrm_contact':
  79         $id = 'id';
  80         //we should restrict by contact type in the first step
  81         $sql = "SELECT contact_type FROM civicrm_dedupe_rule_group WHERE id = {$this->dedupe_rule_group_id};";
  82         $ct = CRM_Core_DAO::singleValueQuery($sql);
  83         if ($this->params) {
  84           $where[] = "t1.contact_type = '{$ct}'";
  85         }
  86         else {
  87           $where[] = "t1.contact_type = '{$ct}'";
  88           $where[] = "t2.contact_type = '{$ct}'";
  89         }
  90         break;
  91
  92       case 'civicrm_address':
  93       case 'civicrm_email':
  94       case 'civicrm_im':
  95       case 'civicrm_openid':
  96       case 'civicrm_phone':
  97         $id = 'contact_id';
  98         break;
  99
 100       case 'civicrm_note':
 101         $id = 'entity_id';
 102         if ($this->params) {
 103           $where[] = "t1.entity_table = 'civicrm_contact'";
 104         }
 105         else {
 106           $where[] = "t1.entity_table = 'civicrm_contact'";
 107           $where[] = "t2.entity_table = 'civicrm_contact'";
 108         }
 109         break;
 110
 111       default:
 112         // custom data tables
 113         if (preg_match('/^civicrm_value_/', $this->rule_table) || preg_match('/^custom_value_/', $this->rule_table)) {
 114           $id = 'entity_id';
 115         }
 116         else {
 117           throw new CRM_Core_Exception("Unsupported rule_table for civicrm_dedupe_rule.id of {$this->id}");
 118         }
 119         break;
 120     }
 121
 122     // build SELECT based on the field names containing contact ids
 123     // if there are params provided, id1 should be 0
 124     if ($this->params) {
 125       $select = "t1.$id id1, {$this->rule_weight} weight";
 126       $subSelect = 'id1, weight';
 127     }
 128     else {
 129       $select = "t1.$id id1, t2.$id id2, {$this->rule_weight} weight";
 130       $subSelect = 'id1, id2, weight';
 131     }
 132
 133     // build FROM (and WHERE, if it's a parametrised search)
 134     // based on whether the rule is about substrings or not
 135     if ($this->params) {
 136       $from = "{$this->rule_table} t1";
 137       $str = 'NULL';
 138       if (isset($this->params[$this->rule_table][$this->rule_field])) {
 139         $str = CRM_Utils_Type::escape($this->params[$this->rule_table][$this->rule_field], 'String');
 140       }
 141       if ($this->rule_length) {
 142         $where[] = "SUBSTR(t1.{$this->rule_field}, 1, {$this->rule_length}) = SUBSTR('$str', 1, {$this->rule_length})";
 143         $where[] = "t1.{$this->rule_field} IS NOT NULL";
 144       }
 145       else {
 146         $where[] = "t1.{$this->rule_field} = '$str'";
 147       }
 148     }
 149     else {
 150       if ($this->rule_length) {
 151         $from = "{$this->rule_table} t1 JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $on) . ")";
 152       }
 153       else {
 154         $from = "{$this->rule_table} t1 INNER JOIN {$this->rule_table} t2 ON (" . implode(' AND ', $innerJoinClauses) . ")";
 155       }
 156     }
 157
 158     // finish building WHERE, also limit the results if requested
 159     if (!$this->params) {
 160       $where[] = "t1.$id < t2.$id";
 161     }
 162     $query = "SELECT $select FROM $from WHERE " . implode(' AND ', $where);
 163     if ($this->contactIds) {
 164       $cids = [];
 165       foreach ($this->contactIds as $cid) {
 166         $cids[] = CRM_Utils_Type::escape($cid, 'Integer');
 167       }
 168       if (count($cids) == 1) {
 169         $query .= " AND (t1.$id = {$cids[0]}) UNION $query AND t2.$id = {$cids[0]}";
 170       }
 171       else {
 172         $query .= " AND t1.$id IN (" . implode(',', $cids) . ")
 173         UNION $query AND  t2.$id IN (" . implode(',', $cids) . ")";
 174       }
 175       // The `weight` is ambiguous in the context of the union; put the whole
 176       // thing in a subquery.
 177       $query = "SELECT $subSelect FROM ($query) subunion";
 178     }
 179
 180     return $query;
 181   }
 182
 183   /**
 184    * find fields related to a rule group.
 185    *
 186    * @param array $params contains the rule group property to identify rule group
 187    *
 188    * @return array
 189    *   rule fields array associated to rule group
 190    */
 191   public static function dedupeRuleFields($params) {
 192     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 193     $rgBao->used = $params['used'];
 194     $rgBao->contact_type = $params['contact_type'];
 195     $rgBao->find(TRUE);
 196
 197     $ruleBao = new CRM_Dedupe_BAO_Rule();
 198     $ruleBao->dedupe_rule_group_id = $rgBao->id;
 199     $ruleBao->find();
 200     $ruleFields = [];
 201     while ($ruleBao->fetch()) {
 202       $field_name = $ruleBao->rule_field;
 203       if ($field_name == 'phone_numeric') {
 204         $field_name = 'phone';
 205       }
 206       $ruleFields[] = $field_name;
 207     }
 208     return $ruleFields;
 209   }
 210
 211   /**
 212    * @param int $cid
 213    * @param int $oid
 214    *
 215    * @return bool
 216    */
 217   public static function validateContacts($cid, $oid) {
 218     if (!$cid || !$oid) {
 219       return NULL;
 220     }
 221     $exception = new CRM_Dedupe_DAO_Exception();
 222     $exception->contact_id1 = $cid;
 223     $exception->contact_id2 = $oid;
 224     //make sure contact2 > contact1.
 225     if ($cid > $oid) {
 226       $exception->contact_id1 = $oid;
 227       $exception->contact_id2 = $cid;
 228     }
 229
 230     return !$exception->find(TRUE);
 231   }
 232
 233   /**
 234    * Get the specification for the given field.
 235    *
 236    * @param string $fieldName
 237    *
 238    * @return array
 239    * @throws \CiviCRM_API3_Exception
 240    */
 241   public function getFieldType($fieldName) {
 242     $entity = CRM_Core_DAO_AllCoreTables::getBriefName(CRM_Core_DAO_AllCoreTables::getClassForTable($this->rule_table));
 243     if (!$entity) {
 244       // This means we have stored a custom field rather than an entity name in rule_table, figure out the entity.
 245       $entity = civicrm_api3('CustomGroup', 'getvalue', ['table_name' => $this->rule_table, 'return' => 'extends']);
 246       if (in_array($entity, ['Individual', 'Household', 'Organization'])) {
 247         $entity = 'Contact';
 248       }
 249       $fieldName = 'custom_' . civicrm_api3('CustomField', 'getvalue', ['column_name' => $fieldName, 'return' => 'id']);
 250     }
 251     $fields = civicrm_api3($entity, 'getfields', ['action' => 'create'])['values'];
 252     return $fields[$fieldName]['type'];
 253   }
 254
 255 }