Further separation of legacy dedupe finder from code to be cleaned up

CRM/Dedupe/BAO/DedupeRuleGroup.php

@@ -14,6 +14,7 @@
  * @package CRM
  * @copyright CiviCRM LLC https://civicrm.org/licensing
  */
+use Civi\Api4\DedupeRuleGroup;
 use Civi\Core\Event\GenericHookEvent;
 
 /**
@@ -150,14 +151,26 @@ public static function hook_civicrm_findExistingDuplicates(GenericHookEvent $eve
     if ($event->tableName) {
       $contactIDs = explode(',', CRM_Core_DAO::singleValueQuery('SELECT GROUP_CONCAT(id) FROM ' . $event->tableName));
     }
-    $tempTable = $ruleGroup->fillTable($ruleGroup->id, $contactIDs, [], FALSE);
+
+    $optimizer = new CRM_Dedupe_FinderQueryOptimizer($ruleGroup->id, $contactIDs, []);
+    $tableQueries = $optimizer->getRuleQueries();
+    if (empty($tableQueries)) {
+      return;
+    }
+
+    $ruleGroup = DedupeRuleGroup::get(FALSE)
+      ->addWhere('id', '=', $ruleGroup->id)
+      ->execute()
+      ->first();
+    $self = new self();
+    $tempTable = $self->runTablesQuery([], $tableQueries, $ruleGroup['threshold']);
     if (!$tempTable) {
       return;
     }
     $aclFrom = $aclWhere = '';
     $dedupeTable = $tempTable;
-    $contactType = $ruleGroup->contact_type;
-    $threshold = $ruleGroup->threshold;
+    $contactType = $ruleGroup['contact_type'];
+    $threshold = $ruleGroup['threshold'];
 
     if ($event->checkPermissions) {
       [$aclFrom, $aclWhere] = CRM_Contact_BAO_Contact_Permission::cacheClause(['c1', 'c2']);
@@ -183,24 +196,38 @@ public static function hook_civicrm_findExistingDuplicates(GenericHookEvent $eve
       $duplicates[] = ['entity_id_1' => $dao->id1, 'entity_id_2' => $dao->id2, 'weight' => $dao->weight];
     }
     $event->duplicates = $duplicates;
-    \CRM_Core_DAO::executeQuery($ruleGroup->tableDropQuery());
+    \CRM_Core_DAO::executeQuery('DROP TEMPORARY TABLE IF EXISTS ' . $dedupeTable);
   }
 
+  /**
+   * @throws \CRM_Core_Exception
+   * @throws \Civi\Core\Exception\DBQueryException
+   * @throws \Civi\API\Exception\UnauthorizedException
+   */
   public static function hook_civicrm_findDuplicates(GenericHookEvent $event): void {
     if (!empty($event->dedupeResults['handled'])) {
       // @todo - in time we can deprecate this & expect them to use stopPropagation().
       return;
     }
-    $rgBao = new CRM_Dedupe_BAO_DedupeRuleGroup();
-    $dedupeTable = $rgBao->fillTable($event->dedupeParams['rule_group_id'], [], $event->dedupeParams['match_params'], FALSE);
+
+    $optimizer = new CRM_Dedupe_FinderQueryOptimizer($event->dedupeParams['rule_group_id'], [], $event->dedupeParams['match_params']);
+    $tableQueries = $optimizer->getRuleQueries();
+    if (empty($tableQueries)) {
+      return;
+    }
+
+    $ruleGroup = DedupeRuleGroup::get(FALSE)
+      ->addWhere('id', '=', $event->dedupeParams['rule_group_id'])
+      ->execute()
+      ->first();
+    $self = new self();
+    $dedupeTable = $self->runTablesQuery($event->dedupeParams['match_params'], $tableQueries, $ruleGroup['threshold']);
     if (!$dedupeTable) {
       $event->dedupeResults['ids'] = [];
       return;
     }
     $aclFrom = '';
     $aclWhere = '';
-    $contactType = $rgBao->contact_type;
-    $threshold = $rgBao->threshold;
     if ($event->dedupeParams['check_permission']) {
       [$aclFrom, $aclWhere] = CRM_Contact_BAO_Contact_Permission::cacheClause('civicrm_contact');
       $aclWhere = $aclWhere ? "AND {$aclWhere}" : '';
@@ -212,67 +239,19 @@ public static function hook_civicrm_findDuplicates(GenericHookEvent $event): voi
                 AND weight >= %2";
 
     $dao = CRM_Core_DAO::executeQuery($query, [
-      1 => [$contactType, 'String'],
-      2 => [$threshold, 'Integer'],
+      1 => [$ruleGroup['contact_type'], 'String'],
+      2 => [$ruleGroup['threshold'], 'Integer'],
     ]);
     $dupes = [];
     while ($dao->fetch()) {
       if (isset($dao->id) && $dao->id) {
         $dupes[] = $dao->id;
       }
     }
-    CRM_Core_DAO::executeQuery($rgBao->tableDropQuery());
+    CRM_Core_DAO::executeQuery('DROP TEMPORARY TABLE IF EXISTS ' . $dedupeTable);
     $event->dedupeResults['ids'] = array_diff($dupes, $event->dedupeParams['excluded_contact_ids']);
   }
 
-  /**
-   * Fill the dedupe finder table.
-   *
-   * @internal do not access from outside core.
-   *
-   * @param int $id
-   * @param array $contactIDs
-   * @param array $params
-   * @param bool $legacyMode
-   *   Legacy mode is called to give backward hook compatibility, in the legacydedupefinder
-   *   extension. It is intended to be transitional, with the non-legacy mode being
-   *   separated out and optimized once it no longer has to comply with the legacy
-   *   hook and reserved query methodology.
-   *
-   * @return false|string
-   * @throws \Civi\Core\Exception\DBQueryException
-   */
-  public function fillTable(int $id, array $contactIDs, array $params, $legacyMode = TRUE) {
-    $ruleGroup = $this;
-    $ruleGroup->id = $id;
-    // make sure we've got a fetched dbrecord, not sure if this is enforced
-    $ruleGroup->find(TRUE);
-    $optimizer = new CRM_Dedupe_FinderQueryOptimizer($id, $contactIDs, $params);
-    // Reserved Rule Groups can optionally get special treatment by
-    // implementing an optimization class and returning a query array.
-    if ($legacyMode && $optimizer->isUseReservedQuery()) {
-      $tableQueries = $optimizer->getReservedQuery();
-    }
-    else {
-      $tableQueries = $optimizer->getRuleQueries();
-    }
-    // if there are no rules in this rule group
-    // add an empty query fulfilling the pattern
-    if ($legacyMode) {
-      if (!$tableQueries) {
-        // Just for the hook.... (which is deprecated).
-        $ruleGroup->noRules = TRUE;
-      }
-      CRM_Utils_Hook::dupeQuery($ruleGroup, 'table', $tableQueries);
-    }
-    if (empty($tableQueries)) {
-      return FALSE;
-    }
-    $threshold = $ruleGroup->threshold;
-
-    return $this->runTablesQuery($params, $tableQueries, $threshold);
-  }
-
   /**
    * Function to determine if a given query set contains inclusive or exclusive set of weights.
    * The function assumes that the query set is already ordered by weight in desc order.

CRM/Dedupe/FinderQueryOptimizer.php

@@ -69,30 +69,6 @@ public function __construct(int $dedupeRuleGroupID, array $contactIDs, array $pa
     }
   }
 
-  /**
-   * Is a file based reserved query configured.
-   *
-   * File based reserved queries were an early idea about how to optimise the dedupe queries.
-   *
-   * In theory extensions could implement them although there is no evidence any of them have.
-   * However, if these are implemented by core or by extensions we should not attempt to optimise
-   * the query by (e.g.) combining queries.
-   *
-   * In practice the queries implemented only return one query anyway
-   *
-   * @internal for core use only.
-   *
-   * @return bool
-   * @throws \CRM_Core_Exception
-   *
-   * @see \CRM_Dedupe_BAO_QueryBuilder_IndividualGeneral
-   * @see \CRM_Dedupe_BAO_QueryBuilder_IndividualSupervised
-   */
-  public function isUseReservedQuery(): bool {
-    return $this->lookup('RuleGroup', 'is_reserved') &&
-      CRM_Utils_File::isIncludable('CRM/Dedupe/BAO/QueryBuilder/' . $this->lookup('RuleGroup', 'name') . '.php');
-  }
-
   /**
    * Return the SQL query for the given rule - either for finding matching
    * pairs of contacts, or for matching against the $params variable (if set).
@@ -228,30 +204,6 @@ private function getContactIDFieldName(string $tableName): string {
     throw new CRM_Core_Exception('invalid field');
   }
 
-  /**
-   * Get the reserved query based on a static class.
-   *
-   * This was an early idea about optimisation & extendability. It is likely
-   * there are no implementations of rules this way outside the 3 core files.
-   *
-   * It is also likely the core files can go once we are optimising the queries based on the
-   * rule.
-   *
-   * @internal  Do not call from outside of core.
-   *
-   * @return array
-   * @throws \CRM_Core_Exception
-   */
-  public function getReservedQuery(): array {
-    $bao = new CRM_Dedupe_BAO_DedupeRuleGroup();
-    $bao->id = $this->lookup('RuleGroup', 'id');
-    $bao->find(TRUE);
-    $bao->params = $this->lookupParameters;
-    $bao->contactIds = $this->contactIDs;
-    $command = empty($this->lookupParameters) ? 'internal' : 'record';
-    return call_user_func(["CRM_Dedupe_BAO_QueryBuilder_" . $this->lookup('RuleGroup', 'name'), $command], $bao);
-  }
-
   /**
    * Get the queries to fill the table for the various rules.
    *

ext/legacydedupefinder/Civi/LegacyFinder/Finder.php

@@ -29,7 +29,9 @@ public static function findExistingDuplicates(GenericHookEvent $event): void {
       $contactIDs = explode(',', \CRM_Core_DAO::singleValueQuery('SELECT GROUP_CONCAT(id) FROM ' . $event->tableName));
     }
     $ruleGroup->contactIds = $contactIDs;
-    $tempTable = $ruleGroup->fillTable($ruleGroup->id, $contactIDs, []);
+    // make sure we've got a fetched db record, not sure if this is enforced
+    $ruleGroup->find(TRUE);
+    $tempTable = self::fillTable($ruleGroup, $ruleGroup->id, $contactIDs, []);
     if (!$tempTable) {
       return;
     }
@@ -79,7 +81,9 @@ public static function findDuplicates(GenericHookEvent $event): void {
     }
     $rgBao = new \CRM_Dedupe_BAO_DedupeRuleGroup();
     $rgBao->params = $event->dedupeParams['match_params'];
-    $dedupeTable = $rgBao->fillTable($event->dedupeParams['rule_group_id'], [], $event->dedupeParams['match_params'], TRUE);
+    // make sure we've got a fetched dbrecord, not sure if this is enforced
+    $rgBao->find(TRUE);
+    $dedupeTable = self::fillTable($rgBao, $event->dedupeParams['rule_group_id'], [], $event->dedupeParams['match_params'], TRUE);
     if (!$dedupeTable) {
       $event->dedupeResults['ids'] = [];
       return;
@@ -112,4 +116,43 @@ public static function findDuplicates(GenericHookEvent $event): void {
     $event->dedupeResults['ids'] = array_diff($dupes, $event->dedupeParams['excluded_contact_ids']);
   }
 
+  /**
+   * Fill the dedupe finder table.
+   *
+   * @internal do not access from outside core
+   *
+   * @param \CRM_Dedupe_BAO_DedupeRuleGroup $ruleGroup
+   * @param int $id
+   * @param array $contactIDs
+   * @param array $params
+   *
+   * @return false|string
+   * @throws \CRM_Core_Exception
+   */
+  private static function fillTable($ruleGroup, int $id, array $contactIDs, array $params) {
+    $optimizer = new FinderQueryOptimizer($id, $contactIDs, $params);
+    // Reserved Rule Groups can optionally get special treatment by
+    // implementing an optimization class and returning a query array.
+    if ($optimizer->isUseReservedQuery()) {
+      $tableQueries = $optimizer->getReservedQuery();
+    }
+    else {
+      $tableQueries = $optimizer->getRuleQueries();
+    }
+    // if there are no rules in this rule group
+    // add an empty query fulfilling the pattern
+    if (!$tableQueries) {
+      // Just for the hook.... (which is deprecated).
+      $ruleGroup->noRules = TRUE;
+    }
+    \CRM_Utils_Hook::dupeQuery($ruleGroup, 'table', $tableQueries);
+
+    if (empty($tableQueries)) {
+      return FALSE;
+    }
+    $threshold = $ruleGroup->threshold;
+
+    return $ruleGroup->runTablesQuery($params, $tableQueries, $threshold);
+  }
+
 }