Add attachments-only filter to uploads

When uploading a file which is an e-mail, this option allows to skip
the mail body when the file is being processed.

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

@@ -73,7 +73,8 @@ object OUpload {
       skipDuplicates: Boolean,
       fileFilter: Glob,
       tags: List[String],
-      language: Option[Language]
+      language: Option[Language],
+      attachmentsOnly: Option[Boolean]
   )
 
   case class UploadData[F[_]](
@@ -150,7 +151,8 @@ object OUpload {
             data.meta.skipDuplicates,
             data.meta.fileFilter.some,
             data.meta.tags.some,
-            false
+            false,
+            data.meta.attachmentsOnly
           )
           args =
             if (data.multiple) files.map(f => ProcessItemArgs(meta, List(f)))

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

@@ -51,7 +51,8 @@ object ProcessItemArgs {
       skipDuplicate: Boolean,
       fileFilter: Option[Glob],
       tags: Option[List[String]],
-      reprocess: Boolean
+      reprocess: Boolean,
+      attachmentsOnly: Option[Boolean]
   )
 
   object ProcessMeta {

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

@@ -44,7 +44,9 @@ case class ScanMailboxArgs(
     // the language for extraction and analysis
     language: Option[Language],
     // apply additional filter to all mails or only imported
-    postHandleAll: Option[Boolean]
+    postHandleAll: Option[Boolean],
+    // Exclude the mail body when importing
+    attachmentsOnly: Option[Boolean]
 )
 
 object ScanMailboxArgs {

modules/joex/src/main/scala/docspell/joex/mail/ReadMail.scala

@@ -23,9 +23,12 @@ object ReadMail {
 
   def readBytesP[F[_]: Async](
       logger: Logger[F],
-      glob: Glob
+      glob: Glob,
+      attachmentsOnly: Boolean
   ): Pipe[F, Byte, Binary[F]] =
-    _.through(bytesToMail(logger)).flatMap(mailToEntries[F](logger, glob))
+    _.through(bytesToMail(logger)).flatMap(
+      mailToEntries[F](logger, glob, attachmentsOnly)
+    )
 
   def bytesToMail[F[_]: Sync](logger: Logger[F]): Pipe[F, Byte, Mail[F]] =
     s =>
@@ -34,10 +37,30 @@ object ReadMail {
 
   def mailToEntries[F[_]: Async](
       logger: Logger[F],
-      glob: Glob
+      glob: Glob,
+      attachmentsOnly: Boolean
+  )(mail: Mail[F]): Stream[F, Binary[F]] =
+    Stream.eval(
+      logger.debug(
+        s"E-mail has ${mail.attachments.size} attachments and ${bodyType(mail.body)}"
+      )
+    ) >>
+      (makeBodyEntry(logger, glob, attachmentsOnly)(mail) ++
+        Stream
+          .eval(TnefExtract.replace(mail))
+          .flatMap(m => Stream.emits(m.attachments.all))
+          .filter(a => a.filename.exists(glob.matches(caseSensitive = false)))
+          .map(a =>
+            Binary(a.filename.getOrElse("noname"), a.mimeType.toLocal, a.content)
+          ))
+
+  private def makeBodyEntry[F[_]: Async](
+      logger: Logger[F],
+      glob: Glob,
+      attachmentsOnly: Boolean
   )(mail: Mail[F]): Stream[F, Binary[F]] = {
     val bodyEntry: F[Option[Binary[F]]] =
-      if (mail.body.isEmpty) (None: Option[Binary[F]]).pure[F]
+      if (mail.body.isEmpty || attachmentsOnly) (None: Option[Binary[F]]).pure[F]
       else {
         val markdownCfg = MarkdownConfig.defaultConfig
         HtmlBodyView(
@@ -49,22 +72,14 @@ object ReadMail {
         ).map(makeHtmlBinary[F] _).map(b => Some(b))
       }
 
-    Stream.eval(
-      logger.debug(
-        s"E-mail has ${mail.attachments.size} attachments and ${bodyType(mail.body)}"
-      )
-    ) >>
-      (Stream
-        .eval(bodyEntry)
-        .flatMap(e => Stream.emits(e.toSeq))
-        .filter(a => glob.matches(caseSensitive = false)(a.name)) ++
+    for {
+      _ <- Stream.eval(logger.debug(s"Import attachments only: $attachmentsOnly"))
+      bin <-
         Stream
-          .eval(TnefExtract.replace(mail))
-          .flatMap(m => Stream.emits(m.attachments.all))
-          .filter(a => a.filename.exists(glob.matches(caseSensitive = false)))
-          .map(a =>
-            Binary(a.filename.getOrElse("noname"), a.mimeType.toLocal, a.content)
-          ))
+          .eval(bodyEntry)
+          .flatMap(e => Stream.emits(e.toSeq))
+          .filter(a => glob.matches(caseSensitive = false)(a.name))
+    } yield bin
   }
 
   private def makeHtmlBinary[F[_]](cnt: BodyContent): Binary[F] =

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

@@ -161,7 +161,8 @@ object ExtractArchive {
       .unNoneTerminate
       .through(ctx.store.bitpeace.fetchData2(RangeDef.all))
 
-    val glob = ctx.args.meta.fileFilter.getOrElse(Glob.all)
+    val glob       = ctx.args.meta.fileFilter.getOrElse(Glob.all)
+    val attachOnly = ctx.args.meta.attachmentsOnly.getOrElse(false)
     ctx.logger.debug(s"Filtering email attachments with '${glob.asString}'") *>
       email
         .through(ReadMail.bytesToMail[F](ctx.logger))
@@ -174,7 +175,7 @@ object ExtractArchive {
             } yield s
 
           ReadMail
-            .mailToEntries(ctx.logger, glob)(mail)
+            .mailToEntries(ctx.logger, glob, attachOnly)(mail)
             .zipWithIndex
             .flatMap(handleEntry(ctx, ra, pos, archive, mId)) ++ Stream.eval(givenMeta)
         }

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

@@ -114,7 +114,8 @@ object ReProcessItem {
               false,
               None,
               None,
-              true
+              true,
+              None // attachOnly (not used when reprocessing attachments)
             ),
             Nil
           ).pure[F]

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

@@ -300,7 +300,8 @@ object ScanMailboxTask {
           true,
           args.fileFilter.getOrElse(Glob.all),
           args.tags.getOrElse(Nil),
-          args.language
+          args.language,
+          args.attachmentsOnly
         )
         data = OUpload.UploadData(
           multiple = false,

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

@@ -4336,6 +4336,10 @@ components:
           format: language
         postHandleAll:
           type: boolean
+        attachmentsOnly:
+          type: boolean
+          description: |
+            Import only the attachments e-mails and discard the body
 
     ImapSettingsList:
       description: |
@@ -5282,6 +5286,14 @@ components:
           description: |
             The `language` of the document may be specified, otherwise
             the one from settings is used.
+        attachmentsOnly:
+          type: boolean
+          default: false
+          description: |
+            Only applies to e-mail files. If `true` then only
+            attachments of the e-mail are imported and the e-mail body
+            is discarded. E-mails that don't have any attachments are
+            skipped.
 
     Collective:
       description: |

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

@@ -337,7 +337,8 @@ trait Conversions {
               m.skipDuplicates.getOrElse(false),
               m.fileFilter.getOrElse(Glob.all),
               m.tags.map(_.items).getOrElse(Nil),
-              m.language
+              m.language,
+              m.attachmentsOnly
             )
           )
         )
@@ -345,7 +346,17 @@ trait Conversions {
       .getOrElse(
         (
           true,
-          UploadMeta(None, sourceName, None, validFileTypes, false, Glob.all, Nil, None)
+          UploadMeta(
+            None,
+            sourceName,
+            None,
+            validFileTypes,
+            false,
+            Glob.all,
+            Nil,
+            None,
+            None
+          )
         )
           .pure[F]
       )

modules/restserver/src/main/scala/docspell/restserver/routes/ScanMailboxRoutes.scala

@@ -125,7 +125,8 @@ object ScanMailboxRoutes {
           settings.tags.map(_.items),
           settings.subjectFilter,
           settings.language,
-          settings.postHandleAll
+          settings.postHandleAll,
+          settings.attachmentsOnly
         )
       )
     )
@@ -159,6 +160,7 @@ object ScanMailboxRoutes {
       task.args.fileFilter,
       task.args.subjectFilter,
       task.args.language,
-      task.args.postHandleAll
+      task.args.postHandleAll,
+      task.args.attachmentsOnly
     )
 }

website/site/content/docs/api/upload.md

@@ -52,6 +52,7 @@ specified via a JSON structure in a part with name `meta`:
 , tags: Maybe StringList
 , fileFilter: Maybe String
 , language: Maybe String
+, attachmentsOnly: Maybe Bool
 }
 ```
 
@@ -90,6 +91,10 @@ specified via a JSON structure in a part with name `meta`:
 - The `language` is used for processing the document(s) contained in
   the request. If not specified the collective's default language is
   used.
+- The `attachmentsOnly` property only applies to e-mail files (usually
+  `*.eml`). If this is `true`, then the e-mail body is discarded and
+  only the attachments are imported. An e-mail without any attachments
+  is therefore skipped.
 
 
 # Endpoints