Prioritize newer files during background scanning

GDATASoftwareAG · May 21, 2024 · 172bad4 · 172bad4
1 parent 969ee0d
commit 172bad4
Show file tree

Hide file tree

Showing 2 changed files with 46 additions and 15 deletions.
diff --git a/lib/Db/DbFileMapper.php b/lib/Db/DbFileMapper.php
@@ -47,6 +47,37 @@ public function getFileIdsWithoutTags(array $excludedTagIds, int $limit): array
         return $fileIds;
     }
 
+    /**
+     * Get file ids that have at least one of the given tags
+     * @param array $includedTagIds
+     * @param int $limit
+     * @return array of file ids
+     * @throws Exception if the database platform is not supported
+     */
+    public function getFileIdsWithTags(array $includedTagIds, int $limit): array
+    {
+        $qb = $this->db->getQueryBuilder();
+        $qb->automaticTablePrefix(true);
+
+        $qb->select('f.fileid')
+            ->from($this->getTableName(), 'f')
+            ->leftJoin('f', 'systemtag_object_mapping', 'o', $qb->expr()->eq('f.fileid', $qb->createFunction($this->getPlatformSpecificCast())))
+            ->leftJoin('f', 'mimetypes', 'm', $qb->expr()->eq('f.mimetype', 'm.id'))
+            ->where($qb->expr()->in('o.systemtagid', $qb->createNamedParameter($includedTagIds, IQueryBuilder::PARAM_INT_ARRAY)))
+            ->andWhere($qb->expr()->notLike('m.mimetype', $qb->createNamedParameter('%unix-directory%')))
+            ->andWhere($qb->expr()->lte('f.size', $qb->createNamedParameter(VerdictService::MAX_FILE_SIZE)))
+            ->andWhere($qb->expr()->like('f.path', $qb->createNamedParameter('files/%')))
+            ->orderBy('f.fileid', 'DESC')
+            ->setMaxResults($limit);
+
+        $fileIds = [];
+        $result = $qb->executeQuery();
+        while ($row = $result->fetch()) {
+            $fileIds[] = $row['fileid'];
+        }
+        return $fileIds;
+    }
+
     /**
      * Create a platform-specific cast function
      * @return string the database platform-specific cast function

diff --git a/lib/Service/TagService.php b/lib/Service/TagService.php
@@ -110,17 +110,17 @@ public function hasUnscannedTag(int $fileId): bool
     /**
      * @param string $tagName
      * @param int $limit Count of object ids you want to get
-     * @param string $offset The last object id you already received
      * @return array
+     * @throws Exception if the database platform is not supported
      */
-    public function getFileIdsWithTag(string $tagName, int $limit, string $offset): array
+    public function getFileIdsWithTag(string $tagName, int $limit): array
     {
         try {
             $tag = $this->getTag($tagName, false);
         } catch (TagNotFoundException) {
             return [];
         }
-        return $this->tagMapper->getObjectIdsForTags([$tag->getId()], 'files', $limit, $offset);
+        return $this->dbFileMapper->getFileIdsWithTags([$tag->getId()], $limit);
     }
 
     /**
@@ -138,24 +138,24 @@ public function getFileIdsWithoutTags(array $excludedTagIds, int $limit): array
      * Get file ids that have any of the given tags
      * @param array $tagIds The tags to get the file ids for
      * @param int $limit The count of file ids you want to get
-     * @param ISystemTag|null $priorTagId Tag id to prioritize over the others
+     * @param ISystemTag|null $priorityTagId Tag id to prioritize over the others
      * @return array
      * @throws TagNotFoundException if a tag does not exist
+     * @throws Exception If the database platform is not supported
      */
-    public function getRandomTaggedFileIds(array $tagIds, int $limit, ?ISystemTag $priorTagId = null): array
+    public function getRandomTaggedFileIds(array $tagIds, int $limit, ?ISystemTag $priorityTagId = null): array
     {
-        if ($priorTagId === null) {
-            $objectIds = $this->tagMapper->getObjectIdsForTags($tagIds, 'files');
-            shuffle($objectIds);
-            return array_slice($objectIds, 0, $limit);
+        $objectIdsPriority = [];
+        if ($priorityTagId !== null) {
+            $objectIdsPriority = $this->dbFileMapper->getFileIdsWithTags([$priorityTagId->getId()], $limit);
+            shuffle($objectIdsPriority);
         }
-        $objectIdsPrior = $this->tagMapper->getObjectIdsForTags([$priorTagId->getId()], 'files', $limit, 0);
-        if (count($objectIdsPrior) >= $limit) {
-            return $objectIdsPrior;
+        if (count($objectIdsPriority) < $limit) {
+            $objectIds = $this->dbFileMapper->getFileIdsWithTags($tagIds, $limit - count($objectIdsPriority));
+            shuffle($objectIds);
+            return array_merge($objectIdsPriority, $objectIds);
         }
-        $objectIds = $this->tagMapper->getObjectIdsForTags($tagIds, 'files');
-        shuffle($objectIds);
-        return array_merge($objectIdsPrior, array_slice($objectIds, 0, $limit - count($objectIdsPrior)));
+        return $objectIdsPriority;
     }
 
     /**