Merge pull request #15196 from eileenmcnaughton/dedupe_api

[civicrm-core.git] / CRM / Dedupe / Merger.php

diff --git a/CRM/Dedupe/Merger.php b/CRM/Dedupe/Merger.php
index 9e501f35f60c2a07f11f11f03d049543429ea0f0..9e22a66feca07f210c1f1c489fcf4ff8b6d6505e 100644 (file)
--- a/CRM/Dedupe/Merger.php
+++ b/CRM/Dedupe/Merger.php
@@ -691,9 +691,17 @@ INNER JOIN  civicrm_membership membership2 ON membership1.membership_type_id = m
    *  If not set explicitly this is calculated but it is preferred that it be set
    *  per comments on isSelected above.
    *
+   * @param int $searchLimit
+   *   Limit on number of contacts to search for duplicates for.
+   *   This means that if the limit is 1000 then only duplicates for the first 1000 contacts
+   *   matching criteria will be found and batchMerged (the number of merges could be less than or greater than 100)
+   *
    * @return array|bool
+   *
+   * @throws \CRM_Core_Exception
+   * @throws \CiviCRM_API3_Exception
    */
-  public static function batchMerge($rgid, $gid = NULL, $mode = 'safe', $batchLimit = 1, $isSelected = 2, $criteria = [], $checkPermissions = TRUE, $reloadCacheIfEmpty = NULL) {
+  public static function batchMerge($rgid, $gid = NULL, $mode = 'safe', $batchLimit = 1, $isSelected = 2, $criteria = [], $checkPermissions = TRUE, $reloadCacheIfEmpty = NULL, $searchLimit = 0) {
     $redirectForPerformance = ($batchLimit > 1) ? TRUE : FALSE;
 
     if (!isset($reloadCacheIfEmpty)) {
@@ -706,7 +714,7 @@ INNER JOIN  civicrm_membership membership2 ON membership1.membership_type_id = m
     $dupePairs = self::getDuplicatePairs($rgid, $gid, $reloadCacheIfEmpty, $batchLimit, $isSelected, ($mode == 'aggressive'), $criteria, $checkPermissions);
 
     $cacheParams = [
-      'cache_key_string' => self::getMergeCacheKeyString($rgid, $gid, $criteria, $checkPermissions),
+      'cache_key_string' => self::getMergeCacheKeyString($rgid, $gid, $criteria, $checkPermissions, $searchLimit),
       // @todo stop passing these parameters in & instead calculate them in the merge function based
       // on the 'real' params like $isRespectExclusions $batchLimit and $isSelected.
       'join' => self::getJoinOnDedupeTable(),
@@ -1833,20 +1841,23 @@ INNER JOIN  civicrm_membership membership2 ON membership1.membership_type_id = m
    * @param int $searchLimit
    *   Limit to searching for matches against this many contacts.
    *
+   * @param int $isForceNewSearch
+   *   Should a new search be forced, bypassing any cache retrieval.
+   *
    * @return array
    *   Array of matches meeting the criteria.
    *
    * @throws \CRM_Core_Exception
    * @throws \CiviCRM_API3_Exception
    */
-  public static function getDuplicatePairs($rule_group_id, $group_id, $reloadCacheIfEmpty, $batchLimit, $isSelected, $includeConflicts = TRUE, $criteria = [], $checkPermissions = TRUE, $searchLimit = 0) {
-    $dupePairs = self::getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, $includeConflicts, $criteria, $checkPermissions);
+  public static function getDuplicatePairs($rule_group_id, $group_id, $reloadCacheIfEmpty, $batchLimit, $isSelected, $includeConflicts = TRUE, $criteria = [], $checkPermissions = TRUE, $searchLimit = 0, $isForceNewSearch = 0) {
+    $dupePairs = $isForceNewSearch ? [] : self::getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, $includeConflicts, $criteria, $checkPermissions, $searchLimit);
     if (empty($dupePairs) && $reloadCacheIfEmpty) {
       // If we haven't found any dupes, probably cache is empty.
       // Try filling cache and give another try. We don't need to specify include conflicts here are there will not be any
       // until we have done some processing.
       CRM_Core_BAO_PrevNextCache::refillCache($rule_group_id, $group_id, $criteria, $checkPermissions, $searchLimit);
-      return self::getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, FALSE, $criteria, $checkPermissions);
+      return self::getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, FALSE, $criteria, $checkPermissions, $searchLimit);
     }
     return $dupePairs;
   }
@@ -1859,17 +1870,21 @@ INNER JOIN  civicrm_membership membership2 ON membership1.membership_type_id = m
    * @param array $criteria
    *   Additional criteria to narrow down the merge group.
    *   Currently we are only supporting the key 'contact' within it.
-   *
    * @param bool $checkPermissions
    *   Respect the users permissions.
+   * @param int $searchLimit
+   *   Number of contacts to seek dupes for (we need this because if
+   *   we change it the results won't be refreshed otherwise. Changing the limit
+   *   from 100 to 1000 SHOULD result in a new dedupe search).
    *
    * @return string
    */
-  public static function getMergeCacheKeyString($rule_group_id, $group_id, $criteria = [], $checkPermissions = TRUE) {
+  public static function getMergeCacheKeyString($rule_group_id, $group_id, $criteria, $checkPermissions, $searchLimit) {
     $contactType = CRM_Dedupe_BAO_RuleGroup::getContactTypeForRuleGroup($rule_group_id);
     $cacheKeyString = "merge_{$contactType}";
     $cacheKeyString .= $rule_group_id ? "_{$rule_group_id}" : '_0';
     $cacheKeyString .= $group_id ? "_{$group_id}" : '_0';
+    $cacheKeyString .= '_' . (int) $searchLimit;
     $cacheKeyString .= !empty($criteria) ? md5(serialize($criteria)) : '_0';
     if ($checkPermissions) {
       $contactID = CRM_Core_Session::getLoggedInContactID();
@@ -2487,12 +2502,13 @@ INNER JOIN  civicrm_membership membership2 ON membership1.membership_type_id = m
    * @param bool $includeConflicts
    * @param array $criteria
    * @param int $checkPermissions
+   * @param int $searchLimit
    *
    * @return array
    */
-  protected static function getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, $includeConflicts, $criteria, $checkPermissions) {
+  protected static function getCachedDuplicateMatches($rule_group_id, $group_id, $batchLimit, $isSelected, $includeConflicts, $criteria, $checkPermissions, $searchLimit = 0) {
     return CRM_Core_BAO_PrevNextCache::retrieve(
-      self::getMergeCacheKeyString($rule_group_id, $group_id, $criteria, $checkPermissions),
+      self::getMergeCacheKeyString($rule_group_id, $group_id, $criteria, $checkPermissions, $searchLimit),
       self::getJoinOnDedupeTable(),
       self::getWhereString($isSelected),
       0, $batchLimit,