Prepend a duplicate check when uploading files

modules/backend/src/main/scala/docspell/backend/ops/OUpload.scala

@@ -59,7 +59,8 @@ object OUpload {
       direction: Option[Direction],
       sourceAbbrev: String,
       folderId: Option[Ident],
-      validFileTypes: Seq[MimeType]
+      validFileTypes: Seq[MimeType],
+      skipDuplicates: Boolean
   )
 
   case class UploadData[F[_]](
@@ -125,7 +126,8 @@ object OUpload {
             data.meta.direction,
             data.meta.sourceAbbrev,
             data.meta.folderId,
-            data.meta.validFileTypes
+            data.meta.validFileTypes,
+            data.meta.skipDuplicates
           )
           args =
             if (data.multiple) files.map(f => ProcessItemArgs(meta, List(f)))

modules/common/src/main/scala/docspell/common/ProcessItemArgs.scala

@@ -37,7 +37,8 @@ object ProcessItemArgs {
       direction: Option[Direction],
       sourceAbbrev: String,
       folderId: Option[Ident],
-      validFileTypes: Seq[MimeType]
+      validFileTypes: Seq[MimeType],
+      skipDuplicate: Boolean
   )
 
   object ProcessMeta {

modules/joex/src/main/scala/docspell/joex/process/DuplicateCheck.scala

@@ -0,0 +1,62 @@
+package docspell.joex.process
+
+import cats.effect._
+import cats.implicits._
+
+import docspell.common._
+import docspell.joex.scheduler.{Context, Task}
+import docspell.store.queries.QItem
+import docspell.store.records.RFileMeta
+
+import bitpeace.FileMeta
+import doobie._
+
+object DuplicateCheck {
+  type Args = ProcessItemArgs
+
+  def apply[F[_]: Sync]: Task[F, Args, Args] =
+    Task { ctx =>
+      if (ctx.args.meta.skipDuplicate)
+        ctx.logger.debug("Checking for duplicate files") *> removeDuplicates(ctx)
+      else ctx.logger.debug("Not checking for duplicates") *> ctx.args.pure[F]
+    }
+
+  def removeDuplicates[F[_]: Sync](ctx: Context[F, Args]): F[ProcessItemArgs] =
+    for {
+      fileMetas <- findDuplicates(ctx)
+      _         <- fileMetas.traverse(deleteDuplicate(ctx))
+      ids = fileMetas.filter(_.exists).map(_.fm.id).toSet
+    } yield ctx.args.copy(files =
+      ctx.args.files.filterNot(f => ids.contains(f.fileMetaId.id))
+    )
+
+  private def deleteDuplicate[F[_]: Sync](
+      ctx: Context[F, Args]
+  )(fd: FileMetaDupes): F[Unit] = {
+    val fname = ctx.args.files.find(_.fileMetaId.id == fd.fm.id).flatMap(_.name)
+    if (fd.exists)
+      ctx.logger
+        .info(s"Deleting duplicate file ${fname}!") *> ctx.store.bitpeace
+        .delete(fd.fm.id)
+        .compile
+        .drain
+    else ().pure[F]
+  }
+
+  private def findDuplicates[F[_]: Sync](
+      ctx: Context[F, Args]
+  ): F[Vector[FileMetaDupes]] =
+    ctx.store.transact(for {
+      fileMetas <- RFileMeta.findByIds(ctx.args.files.map(_.fileMetaId))
+      dupes     <- fileMetas.traverse(checkDuplicate(ctx))
+    } yield dupes)
+
+  private def checkDuplicate[F[_]](
+      ctx: Context[F, Args]
+  )(fm: FileMeta): ConnectionIO[FileMetaDupes] =
+    QItem
+      .findByChecksum(fm.checksum, ctx.args.meta.collective)
+      .map(v => FileMetaDupes(fm, v.nonEmpty))
+
+  case class FileMetaDupes(fm: FileMeta, exists: Boolean)
+}

modules/joex/src/main/scala/docspell/joex/process/ItemHandler.scala

@@ -35,10 +35,18 @@ object ItemHandler {
       analyser: TextAnalyser[F],
       regexNer: RegexNerFile[F]
   ): Task[F, Args, Unit] =
-    CreateItem[F]
-      .flatMap(itemStateTask(ItemState.Processing))
-      .flatMap(safeProcess[F](cfg, itemOps, fts, analyser, regexNer))
-      .map(_ => ())
+    DuplicateCheck[F]
+      .flatMap(args =>
+        if (args.files.isEmpty) logNoFiles
+        else {
+          val create: Task[F, Args, ItemData] =
+            CreateItem[F].contramap(_ => args.pure[F])
+          create
+            .flatMap(itemStateTask(ItemState.Processing))
+            .flatMap(safeProcess[F](cfg, itemOps, fts, analyser, regexNer))
+            .map(_ => ())
+        }
+      )
 
   def itemStateTask[F[_]: Sync, A](
       state: ItemState
@@ -121,4 +129,10 @@ object ItemHandler {
 
   private def logWarn[F[_]](msg: => String): Task[F, Args, Unit] =
     Task(_.logger.warn(msg))
+
+  private def logNoFiles[F[_]]: Task[F, Args, Unit] =
+    logWarn(
+      "No files to process! Either no files were given or duplicate check removed all."
+    )
+
 }

modules/joex/src/main/scala/docspell/joex/process/ReProcessItem.scala

@@ -90,7 +90,8 @@ object ReProcessItem {
               None, //direction
               "",   //source-id
               None, //folder
-              Seq.empty
+              Seq.empty,
+              false
             ),
             Nil
           ).pure[F]

modules/joex/src/main/scala/docspell/joex/scanmailbox/ScanMailboxTask.scala

@@ -254,7 +254,8 @@ object ScanMailboxTask {
           Some(dir),
           s"mailbox-${ctx.args.account.user.id}",
           args.itemFolder,
-          Seq.empty
+          Seq.empty,
+          true
         )
         data = OUpload.UploadData(
           multiple = false,

modules/restapi/src/main/resources/docspell-openapi.yml

@@ -3660,6 +3660,7 @@ components:
           description: DateTime
           type: integer
           format: date-time
+
     ItemUploadMeta:
       description: |
         Meta information for an item upload. The user can specify some
@@ -3674,6 +3675,7 @@ components:
 
         A folderId can be given, the item is placed into this folder
         after creation.
+
       required:
         - multiple
       properties:
@@ -3686,6 +3688,10 @@ components:
         folder:
           type: string
           format: ident
+        skipDuplicates:
+          type: boolean
+          default: false
+
     Collective:
       description: |
         Information about a collective.

modules/restserver/src/main/scala/docspell/restserver/conv/Conversions.scala

@@ -294,10 +294,21 @@ trait Conversions {
       .map(p => parseMeta(p.body))
       .map(fm =>
         fm.map(m =>
-          (m.multiple, UploadMeta(m.direction, "webapp", m.folder, validFileTypes))
+          (
+            m.multiple,
+            UploadMeta(
+              m.direction,
+              "webapp",
+              m.folder,
+              validFileTypes,
+              m.skipDuplicates.getOrElse(false)
+            )
+          )
         )
       )
-      .getOrElse((true, UploadMeta(None, "webapp", None, validFileTypes)).pure[F])
+      .getOrElse(
+        (true, UploadMeta(None, "webapp", None, validFileTypes, false)).pure[F]
+      )
 
     val files = mp.parts
       .filter(p => p.name.forall(s => !s.equalsIgnoreCase("meta")))

modules/store/src/main/scala/docspell/store/records/RFileMeta.scala

@@ -1,5 +1,8 @@
 package docspell.store.records
 
+import cats.data.NonEmptyList
+import cats.implicits._
+
 import docspell.common._
 import docspell.store.impl.Implicits._
 import docspell.store.impl._
@@ -33,6 +36,17 @@ object RFileMeta {
     selectSimple(Columns.all, table, Columns.id.is(fid)).query[FileMeta].option
   }
 
+  def findByIds(ids: List[Ident]): ConnectionIO[Vector[FileMeta]] = {
+    import bitpeace.sql._
+
+    NonEmptyList.fromList(ids) match {
+      case Some(nel) =>
+        selectSimple(Columns.all, table, Columns.id.isIn(nel)).query[FileMeta].to[Vector]
+      case None =>
+        Vector.empty[FileMeta].pure[ConnectionIO]
+    }
+  }
+
   def findMime(fid: Ident): ConnectionIO[Option[MimeType]] = {
     import bitpeace.sql._