CRM/Dedupe/Finder.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_Finder {
  25
  26   /**
  27    * Return a contact_id-keyed array of arrays of possible dupes
  28    * (of the key contact_id) - limited to dupes of $cids if provided.
  29    *
  30    * @param int $rgid
  31    *   Rule group id.
  32    * @param array $cids
  33    *   Contact ids to limit the search to.
  34    *
  35    * @param bool $checkPermissions
  36    *   Respect logged in user permissions.
  37    *
  38    * @return array
  39    *   Array of (cid1, cid2, weight) dupe triples
  40    *
  41    * @throws \CRM_Core_Exception
  42    */
  43   public static function dupes($rgid, $cids = [], $checkPermissions = TRUE) {
  44     $rgBao = new CRM_Dedupe_BAO_RuleGroup();
  45     $rgBao->id = $rgid;
  46     $rgBao->contactIds = $cids;
  47     if (!$rgBao->find(TRUE)) {
  48       throw new CRM_Core_Exception('Dedupe rule not found for selected contacts');
  49     }
  50
  51     $rgBao->fillTable();
  52     $dao = new CRM_Core_DAO();
  53     $dao->query($rgBao->thresholdQuery($checkPermissions));
  54     $dupes = [];
  55     while ($dao->fetch()) {
  56       $dupes[] = [$dao->id1, $dao->id2, $dao->weight];
  57     }
  58     $dao->query($rgBao->tableDropQuery());
  59
  60     return $dupes;
  61   }
  62
  63   /**
  64    * Return an array of possible dupes, based on the provided array of
  65    * params, using the default rule group for the given contact type and
  66    * usage.
  67    *
  68    * check_permission is a boolean flag to indicate if permission should be considered.
  69    * default is to always check permissioning but public pages for example might not want
  70    * permission to be checked for anonymous users. Refer CRM-6211. We might be breaking
  71    * Multi-Site dedupe for public pages.
  72    *
  73    * @param array $params
  74    *   Array of params of the form $params[$table][$field] == $value.
  75    * @param string $ctype
  76    *   Contact type to match against.
  77    * @param string $used
  78    *   Dedupe rule group usage ('Unsupervised' or 'Supervised' or 'General').
  79    * @param array $except
  80    *   Array of contacts that shouldn't be considered dupes.
  81    * @param int $ruleGroupID
  82    *   The id of the dedupe rule we should be using.
  83    *
  84    * @return array
  85    *   matching contact ids
  86    * @throws \CRM_Core_Exception
  87    */
  88   public static function dupesByParams(
  89     $params,
  90     $ctype,
  91     $used = 'Unsupervised',
  92     $except = [],
  93     $ruleGroupID = NULL
  94   ) {
  95     // If $params is empty there is zero reason to proceed.
  96     if (!$params) {
  97       return [];
  98     }
  99     $checkPermission = CRM_Utils_Array::value('check_permission', $params, TRUE);
 100     // This may no longer be required - see https://github.com/civicrm/civicrm-core/pull/13176
 101     $params = array_filter($params);
 102
 103     $foundByID = FALSE;
 104     if ($ruleGroupID) {
 105       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 106       $rgBao->id = $ruleGroupID;
 107       $rgBao->contact_type = $ctype;
 108       if ($rgBao->find(TRUE)) {
 109         $foundByID = TRUE;
 110       }
 111     }
 112
 113     if (!$foundByID) {
 114       $rgBao = new CRM_Dedupe_BAO_RuleGroup();
 115       $rgBao->contact_type = $ctype;
 116       $rgBao->used = $used;
 117       if (!$rgBao->find(TRUE)) {
 118         throw new CRM_Core_Exception("$used rule for $ctype does not exist");
 119       }
 120     }
 121
 122     if (isset($params['civicrm_phone']['phone_numeric'])) {
 123       $orig = $params['civicrm_phone']['phone_numeric'];
 124       $params['civicrm_phone']['phone_numeric'] = preg_replace('/[^\d]/', '', $orig);
 125     }
 126     $rgBao->params = $params;
 127     $rgBao->fillTable();
 128     $dao = new CRM_Core_DAO();
 129     $dao->query($rgBao->thresholdQuery($checkPermission));
 130     $dupes = [];
 131     while ($dao->fetch()) {
 132       if (isset($dao->id) && $dao->id) {
 133         $dupes[] = $dao->id;
 134       }
 135     }
 136     $dao->query($rgBao->tableDropQuery());
 137     return array_diff($dupes, $except);
 138   }
 139
 140   /**
 141    * Return a contact_id-keyed array of arrays of possible dupes in the given group.
 142    *
 143    * @param int $rgid
 144    *   Rule group id.
 145    * @param int $gid
 146    *   Contact group id.
 147    *
 148    * @param int $searchLimit
 149    *  Limit for the number of contacts to be used for comparison.
 150    *  The search methodology finds all matches for the searchedContacts so this limits
 151    *  the number of searched contacts, not the matches found.
 152    *
 153    * @return array
 154    *   array of (cid1, cid2, weight) dupe triples
 155    *
 156    * @throws \CRM_Core_Exception
 157    */
 158   public static function dupesInGroup($rgid, $gid, $searchLimit = 0) {
 159     $cids = array_keys(CRM_Contact_BAO_Group::getMember($gid, TRUE, $searchLimit));
 160     if (!empty($cids)) {
 161       return self::dupes($rgid, $cids);
 162     }
 163     return [];
 164   }
 165
 166   /**
 167    * A hackish function needed to massage CRM_Contact_Form_$ctype::formRule()
 168    * object into a valid $params array for dedupe
 169    *
 170    * @param array $fields
 171    *   Contact structure from formRule().
 172    * @param string $ctype
 173    *   Contact type of the given contact.
 174    *
 175    * @return array
 176    *   valid $params array for dedupe
 177    * @throws \CRM_Core_Exception
 178    */
 179   public static function formatParams($fields, $ctype) {
 180     $flat = [];
 181     CRM_Utils_Array::flatten($fields, $flat);
 182
 183     // FIXME: This may no longer be necessary - check inputs
 184     $replace_these = [
 185       'individual_prefix' => 'prefix_id',
 186       'individual_suffix' => 'suffix_id',
 187       'gender' => 'gender_id',
 188     ];
 189     foreach (['individual_suffix', 'individual_prefix', 'gender'] as $name) {
 190       if (!empty($fields[$name])) {
 191         $flat[$replace_these[$name]] = $flat[$name];
 192         unset($flat[$name]);
 193       }
 194     }
 195
 196     // handle {birth,deceased}_date
 197     foreach ([
 198       'birth_date',
 199       'deceased_date',
 200     ] as $date) {
 201       if (!empty($fields[$date])) {
 202         $flat[$date] = $fields[$date];
 203         if (is_array($flat[$date])) {
 204           $flat[$date] = CRM_Utils_Date::format($flat[$date]);
 205         }
 206         $flat[$date] = CRM_Utils_Date::processDate($flat[$date]);
 207       }
 208     }
 209
 210     if (!empty($flat['contact_source'])) {
 211       $flat['source'] = $flat['contact_source'];
 212       unset($flat['contact_source']);
 213     }
 214
 215     // handle preferred_communication_method
 216     if (!empty($fields['preferred_communication_method'])) {
 217       $methods = array_intersect($fields['preferred_communication_method'], ['1']);
 218       $methods = array_keys($methods);
 219       sort($methods);
 220       if ($methods) {
 221         $flat['preferred_communication_method'] = CRM_Core_DAO::VALUE_SEPARATOR . implode(CRM_Core_DAO::VALUE_SEPARATOR, $methods) . CRM_Core_DAO::VALUE_SEPARATOR;
 222       }
 223     }
 224
 225     // handle custom data
 226     $tree = CRM_Core_BAO_CustomGroup::getTree($ctype, NULL, NULL, -1);
 227     CRM_Core_BAO_CustomGroup::postProcess($tree, $fields, TRUE);
 228     foreach ($tree as $key => $cg) {
 229       if (!is_int($key)) {
 230         continue;
 231       }
 232       foreach ($cg['fields'] as $cf) {
 233         $flat[$cf['column_name']] = $cf['customValue']['data'] ?? NULL;
 234       }
 235     }
 236
 237     // if the key is dotted, keep just the last part of it
 238     foreach ($flat as $key => $value) {
 239       if (substr_count($key, '.')) {
 240         $last = explode('.', $key);
 241         $last = array_pop($last);
 242         // make sure the first occurrence is kept, not the last
 243         if (!isset($flat[$last])) {
 244           $flat[$last] = $value;
 245         }
 246         unset($flat[$key]);
 247       }
 248     }
 249
 250     // drop the -digit (and -Primary, for CRM-3902) postfixes (so event registration's $flat['email-5'] becomes $flat['email'])
 251     // FIXME: CRM-5026 should be fixed here; the below clobbers all address info; we should split off address fields and match
 252     // the -digit to civicrm_address.location_type_id and -Primary to civicrm_address.is_primary
 253     foreach ($flat as $key => $value) {
 254       $matches = [];
 255       if (preg_match('/(.*)-(Primary-[\d+])$|(.*)-(\d+|Primary)$/', $key, $matches)) {
 256         $return = array_values(array_filter($matches));
 257         // make sure the first occurrence is kept, not the last
 258         $flat[$return[1]] = empty($flat[$return[1]]) ? $value : $flat[$return[1]];
 259         unset($flat[$key]);
 260       }
 261     }
 262
 263     $params = [];
 264     $supportedFields = CRM_Dedupe_BAO_RuleGroup::supportedFields($ctype);
 265     if (is_array($supportedFields)) {
 266       foreach ($supportedFields as $table => $fields) {
 267         if ($table === 'civicrm_address') {
 268           // for matching on civicrm_address fields, we also need the location_type_id
 269           $fields['location_type_id'] = '';
 270           // FIXME: we also need to do some hacking for id and name fields, see CRM-3902’s comments
 271           $fixes = [
 272             'address_name' => 'name',
 273             'country' => 'country_id',
 274             'state_province' => 'state_province_id',
 275             'county' => 'county_id',
 276           ];
 277           foreach ($fixes as $orig => $target) {
 278             if (!empty($flat[$orig])) {
 279               $params[$table][$target] = $flat[$orig];
 280             }
 281           }
 282         }
 283         if ($table === 'civicrm_phone') {
 284           $fixes = [
 285             'phone' => 'phone_numeric',
 286           ];
 287           foreach ($fixes as $orig => $target) {
 288             if (!empty($flat[$orig])) {
 289               $params[$table][$target] = $flat[$orig];
 290             }
 291           }
 292         }
 293         foreach ($fields as $field => $title) {
 294           if (!empty($flat[$field])) {
 295             $params[$table][$field] = $flat[$field];
 296           }
 297         }
 298       }
 299     }
 300     return $params;
 301   }
 302
 303   /**
 304    * Parse duplicate pairs into a standardised array and store in the prev_next_cache.
 305    *
 306    * @param array $foundDupes
 307    * @param string $cacheKeyString
 308    *
 309    * @return array
 310    *   Dupe pairs with the keys
 311    *   -srcID
 312    *   -srcName
 313    *   -dstID
 314    *   -dstName
 315    *   -weight
 316    *   -canMerge
 317    */
 318   public static function parseAndStoreDupePairs($foundDupes, $cacheKeyString) {
 319     $cids = [];
 320     foreach ($foundDupes as $dupe) {
 321       $cids[$dupe[0]] = 1;
 322       $cids[$dupe[1]] = 1;
 323     }
 324     $cidString = implode(', ', array_keys($cids));
 325
 326     $dao = CRM_Core_DAO::executeQuery("SELECT id, display_name FROM civicrm_contact WHERE id IN ($cidString) ORDER BY sort_name");
 327     $displayNames = [];
 328     while ($dao->fetch()) {
 329       $displayNames[$dao->id] = $dao->display_name;
 330     }
 331
 332     $userId = CRM_Core_Session::getLoggedInContactID();
 333     foreach ($foundDupes as $dupes) {
 334       $srcID = $dupes[1];
 335       $dstID = $dupes[0];
 336       // The logged in user should never be the src (ie. the contact to be removed).
 337       if ($srcID == $userId) {
 338         $srcID = $dstID;
 339         $dstID = $userId;
 340       }
 341
 342       $mainContacts[] = $row = [
 343         'dstID' => (int) $dstID,
 344         'dstName' => $displayNames[$dstID],
 345         'srcID' => (int) $srcID,
 346         'srcName' => $displayNames[$srcID],
 347         'weight' => $dupes[2],
 348         'canMerge' => TRUE,
 349       ];
 350
 351       CRM_Core_DAO::executeQuery("INSERT INTO civicrm_prevnext_cache (entity_table, entity_id1, entity_id2, cacheKey, data) VALUES
 352         ('civicrm_contact', %1, %2, %3, %4)", [
 353           1 => [$dstID, 'Integer'],
 354           2 => [$srcID, 'Integer'],
 355           3 => [$cacheKeyString, 'String'],
 356           4 => [serialize($row), 'String'],
 357         ]
 358       );
 359     }
 360     return $mainContacts;
 361   }
 362
 363 }