perf: hash and delete pages in a single scan

komga/src/main/kotlin/org/gotson/komga/application/tasks/Task.kt

@@ -9,6 +9,7 @@ import java.io.Serializable
 const val HIGHEST_PRIORITY = 8
 const val HIGH_PRIORITY = 6
 const val DEFAULT_PRIORITY = 4
+const val LOW_PRIORITY = 2
 const val LOWEST_PRIORITY = 0
 
 sealed class Task(priority: Int = DEFAULT_PRIORITY, val groupId: String? = null) : Serializable {
@@ -25,6 +26,16 @@ sealed class Task(priority: Int = DEFAULT_PRIORITY, val groupId: String? = null)
     override fun toString(): String = "FindBooksToConvert(libraryId='$libraryId', priority='$priority')"
   }
 
+  class FindBooksWithMissingPageHash(val libraryId: String, priority: Int = DEFAULT_PRIORITY) : Task(priority) {
+    override fun uniqueId() = "FIND_BOOKS_WITH_MISSING_PAGE_HASH_$libraryId"
+    override fun toString(): String = "FindBooksWithMissingPageHash(libraryId='$libraryId', priority='$priority')"
+  }
+
+  class FindDuplicatePagesToDelete(val libraryId: String, priority: Int = DEFAULT_PRIORITY) : Task(priority) {
+    override fun uniqueId() = "FIND_DUPLICATE_PAGES_TO_DELETE_$libraryId"
+    override fun toString(): String = "FindDuplicatePagesToDelete(libraryId='$libraryId', priority='$priority')"
+  }
+
   class EmptyTrash(val libraryId: String, priority: Int = DEFAULT_PRIORITY) : Task(priority) {
     override fun uniqueId() = "EMPTY_TRASH_$libraryId"
     override fun toString(): String = "EmptyTrash(libraryId='$libraryId', priority='$priority')"

komga/src/main/kotlin/org/gotson/komga/application/tasks/TaskHandler.kt

@@ -12,6 +12,7 @@ import org.gotson.komga.domain.service.BookMetadataLifecycle
 import org.gotson.komga.domain.service.BookPageEditor
 import org.gotson.komga.domain.service.LibraryContentLifecycle
 import org.gotson.komga.domain.service.LocalArtworkLifecycle
+import org.gotson.komga.domain.service.PageHashLifecycle
 import org.gotson.komga.domain.service.SeriesLifecycle
 import org.gotson.komga.domain.service.SeriesMetadataLifecycle
 import org.gotson.komga.infrastructure.jms.QUEUE_FACTORY
@@ -43,6 +44,7 @@ class TaskHandler(
   private val bookConverter: BookConverter,
   private val bookPageEditor: BookPageEditor,
   private val searchIndexLifecycle: SearchIndexLifecycle,
+  private val pageHashLifecycle: PageHashLifecycle,
   private val meterRegistry: MeterRegistry,
 ) {
 
@@ -57,11 +59,11 @@ class TaskHandler(
             libraryRepository.findByIdOrNull(task.libraryId)?.let { library ->
               libraryContentLifecycle.scanRootFolder(library)
               taskReceiver.analyzeUnknownAndOutdatedBooks(library)
-              taskReceiver.hashBooksWithoutHash(library)
+              taskReceiver.repairExtensions(library, LOW_PRIORITY)
-              taskReceiver.hashBookPagesWithMissingHash(library)
-              taskReceiver.repairExtensions(library, LOWEST_PRIORITY)
               taskReceiver.findBooksToConvert(library, LOWEST_PRIORITY)
-              taskReceiver.removeDuplicatePages(library, LOWEST_PRIORITY)
+              taskReceiver.findBooksWithMissingPageHash(library, LOWEST_PRIORITY)
+              taskReceiver.findDuplicatePagesToDelete(library, LOWEST_PRIORITY)
+              taskReceiver.hashBooksWithoutHash(library)
             } ?: logger.warn { "Cannot execute task $task: Library does not exist" }
 
           is Task.FindBooksToConvert ->
@@ -71,6 +73,20 @@ class TaskHandler(
               }
             } ?: logger.warn { "Cannot execute task $task: Library does not exist" }
 
+          is Task.FindBooksWithMissingPageHash ->
+            libraryRepository.findByIdOrNull(task.libraryId)?.let { library ->
+              pageHashLifecycle.getBookAndSeriesIdsWithMissingPageHash(library).forEach {
+                taskReceiver.hashBookPages(it.first, it.second, task.priority + 1)
+              }
+            } ?: logger.warn { "Cannot execute task $task: Library does not exist" }
+
+          is Task.FindDuplicatePagesToDelete ->
+            libraryRepository.findByIdOrNull(task.libraryId)?.let { library ->
+              pageHashLifecycle.getBookPagesToDeleteAutomatically(library).forEach { (bookId, pages) ->
+                taskReceiver.removeDuplicatePages(bookId, pages, task.priority + 1)
+              }
+            } ?: logger.warn { "Cannot execute task $task: Library does not exist" }
+
           is Task.EmptyTrash ->
             libraryRepository.findByIdOrNull(task.libraryId)?.let { library ->
               libraryContentLifecycle.emptyTrash(library)

komga/src/main/kotlin/org/gotson/komga/application/tasks/TaskReceiver.kt

@@ -11,7 +11,6 @@ import org.gotson.komga.domain.model.Media
 import org.gotson.komga.domain.persistence.BookRepository
 import org.gotson.komga.domain.persistence.LibraryRepository
 import org.gotson.komga.domain.service.BookConverter
-import org.gotson.komga.domain.service.PageHashLifecycle
 import org.gotson.komga.infrastructure.jms.JMS_PROPERTY_TYPE
 import org.gotson.komga.infrastructure.jms.QUEUE_TASKS
 import org.gotson.komga.infrastructure.jms.QUEUE_UNIQUE_ID
@@ -30,7 +29,6 @@ class TaskReceiver(
   private val libraryRepository: LibraryRepository,
   private val bookRepository: BookRepository,
   private val bookConverter: BookConverter,
-  private val pageHashLifecycle: PageHashLifecycle,
 ) {
 
   private val jmsTemplates = (0..9).associateWith {
@@ -71,11 +69,12 @@ class TaskReceiver(
       }
   }
 
-  fun hashBookPagesWithMissingHash(library: Library) {
+  fun findBooksWithMissingPageHash(library: Library, priority: Int = DEFAULT_PRIORITY) {
-    if (library.hashPages)
+    submitTask(Task.FindBooksWithMissingPageHash(library.id, priority))
-      pageHashLifecycle.getBookAndSeriesIdsWithMissingPageHash(library).forEach {
+  }
-        submitTask(Task.HashBookPages(it.first, LOWEST_PRIORITY, it.second))
+
-      }
+  fun hashBookPages(bookId: String, seriesId: String, priority: Int = DEFAULT_PRIORITY) {
+    submitTask(Task.HashBookPages(bookId, priority, seriesId))
   }
 
   fun findBooksToConvert(library: Library, priority: Int = DEFAULT_PRIORITY) {
@@ -93,10 +92,8 @@ class TaskReceiver(
       }
   }
 
-  fun removeDuplicatePages(library: Library, priority: Int = DEFAULT_PRIORITY) {
+  fun findDuplicatePagesToDelete(library: Library, priority: Int = DEFAULT_PRIORITY) {
-    pageHashLifecycle.getBookPagesToDeleteAutomatically(library).forEach { (bookId, pages) ->
+    submitTask(Task.FindDuplicatePagesToDelete(library.id, priority))
-      removeDuplicatePages(bookId, pages, priority)
-    }
   }
 
   fun removeDuplicatePages(bookId: String, pages: Collection<BookPageNumbered>, priority: Int = DEFAULT_PRIORITY) {

komga/src/main/kotlin/org/gotson/komga/domain/service/PageHashLifecycle.kt

@@ -1,5 +1,6 @@
 package org.gotson.komga.domain.service
 
+import mu.KotlinLogging
 import org.gotson.komga.domain.model.BookPageContent
 import org.gotson.komga.domain.model.BookPageNumbered
 import org.gotson.komga.domain.model.Library
@@ -13,6 +14,8 @@ import org.gotson.komga.infrastructure.configuration.KomgaProperties
 import org.springframework.data.domain.Pageable
 import org.springframework.stereotype.Service
 
+private val logger = KotlinLogging.logger {}
+
 @Service
 class PageHashLifecycle(
   private val pageHashRepository: PageHashRepository,
@@ -28,7 +31,13 @@ class PageHashLifecycle(
    * @return a Collection of Pair of BookId/SeriesId
    */
   fun getBookAndSeriesIdsWithMissingPageHash(library: Library): Collection<Pair<String, String>> =
-    mediaRepository.findAllBookAndSeriesIdsByLibraryIdAndMediaTypeAndWithMissingPageHash(library.id, hashableMediaTypes, komgaProperties.pageHashing)
+    if (library.hashPages)
+      mediaRepository.findAllBookAndSeriesIdsByLibraryIdAndMediaTypeAndWithMissingPageHash(library.id, hashableMediaTypes, komgaProperties.pageHashing)
+        .also { logger.info { "Found ${it.size} books with missing page hash" } }
+    else {
+      logger.info { "Page hashing is not enabled, skipping" }
+      emptyList()
+    }
 
   fun getPage(pageHash: PageHash, resizeTo: Int? = null): BookPageContent? {
     val match = pageHashRepository.findMatchesByHash(pageHash, null, Pageable.ofSize(1)).firstOrNull() ?: return null