Merge pull request #439 from eikek/mail-subject-filter

Mail subject filter

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

@@ -33,7 +33,9 @@ case class ScanMailboxArgs(
     // set a filter for files when importing archives
     fileFilter: Option[Glob],
     // set a list of tags to apply to new item
-    tags: Option[List[String]]
+    tags: Option[List[String]],
+    // a glob filter for the mail subject
+    subjectFilter: Option[Glob]
 )
 
 object ScanMailboxArgs {

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

@@ -142,7 +142,7 @@ object ScanMailboxTask {
         _       <- Kleisli.liftF(ctx.logger.info(s"Processing folder $name"))
         folder  <- requireFolder(a)(name)
         search  <- searchMails(a)(folder)
-        headers <- Kleisli.liftF(filterMessageIds(search.mails))
+        headers <- Kleisli.liftF(filterSubjects(search.mails).flatMap(filterMessageIds))
         _       <- headers.traverse(handleOne(ctx.args, a, upload))
       } yield ScanResult(name, search.mails.size, search.count - search.mails.size)
 
@@ -178,6 +178,26 @@ object ScanMailboxTask {
       } yield mails
     }
 
+    def filterSubjects(headers: Vector[MailHeader]): F[Vector[MailHeader]] =
+      ctx.args.subjectFilter match {
+        case Some(sf) =>
+          def check(mh: MailHeader): F[Option[MailHeader]] =
+            if (sf.matches(mh.subject))
+              ctx.logger.debug(
+                s"Including mail '${mh.subject}', it matches the filter."
+              ) *> Option(mh).pure[F]
+            else
+              ctx.logger.debug(
+                s"Excluding mail '${mh.subject}', it doesn't match the filter."
+              ) *> (None: Option[MailHeader]).pure[F]
+          ctx.logger.info(
+            s"Filtering mails on subject using filter: ${sf.asString}"
+          ) *> headers.traverseFilter(check)
+
+        case None =>
+          ctx.logger.debug("Not matching on subjects. No filter given") *> headers.pure[F]
+      }
+
     def filterMessageIds(headers: Vector[MailHeader]): F[Vector[MailHeader]] =
       NonEmptyList.fromFoldable(headers.flatMap(_.messageId)) match {
         case Some(nl) =>

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

@@ -3503,7 +3503,12 @@ components:
           $ref: "#/components/schemas/StringList"
         fileFilter:
           description: |
-            A glob to filter attachments to import.
+            A glob to filter attachments to import by file name.
+          type: string
+          format: glob
+        subjectFilter:
+          description: |
+            A glob to filter attachments to import by subject.
           type: string
           format: glob
 

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

@@ -115,7 +115,8 @@ object ScanMailboxRoutes {
           settings.direction,
           settings.itemFolder,
           settings.fileFilter,
-          settings.tags.map(_.items)
+          settings.tags.map(_.items),
+          settings.subjectFilter
         )
       )
     )
@@ -137,7 +138,7 @@ object ScanMailboxRoutes {
       task.id,
       task.enabled,
       conn.getOrElse(Ident.unsafe("")),
-      task.args.folders, //folders
+      task.args.folders,
       task.timer,
       task.args.receivedSince.map(_.hours.toInt),
       task.args.targetFolder,
@@ -145,6 +146,7 @@ object ScanMailboxRoutes {
       task.args.direction,
       task.args.itemFolder,
       task.args.tags.map(StringList.apply),
-      task.args.fileFilter
+      task.args.fileFilter,
+      task.args.subjectFilter
     )
 }

modules/webapp/src/main/elm/Comp/ScanMailboxForm.elm

@@ -63,6 +63,7 @@ type alias Model =
     , tagModel : Comp.Dropdown.Model Tag
     , existingTags : List String
     , fileFilter : Maybe String
+    , subjectFilter : Maybe String
     }
 
 
@@ -94,6 +95,7 @@ type Msg
     | GetTagResp (Result Http.Error TagList)
     | TagDropdownMsg (Comp.Dropdown.Msg Tag)
     | SetFileFilter String
+    | SetSubjectFilter String
 
 
 initWith : Flags -> ScanMailboxSettings -> ( Model, Cmd Msg )
@@ -135,6 +137,7 @@ initWith flags s =
             Maybe.map .items s.tags
                 |> Maybe.withDefault []
         , fileFilter = s.fileFilter
+        , subjectFilter = s.subjectFilter
       }
     , Cmd.batch
         [ Api.getImapSettings flags "" ConnResp
@@ -184,6 +187,7 @@ init flags =
       , tagModel = Util.Tag.makeDropdownModel
       , existingTags = []
       , fileFilter = Nothing
+      , subjectFilter = Nothing
       }
     , Cmd.batch
         [ Api.getImapSettings flags "" ConnResp
@@ -228,6 +232,7 @@ makeSettings model =
                 , schedule = Data.CalEvent.makeEvent timer
                 , itemFolder = model.itemFolderId
                 , fileFilter = model.fileFilter
+                , subjectFilter = model.subjectFilter
                 , tags =
                     case Comp.Dropdown.getSelected model.tagModel of
                         [] ->
@@ -586,6 +591,12 @@ update flags msg model =
             , Cmd.none
             )
 
+        SetSubjectFilter str ->
+            ( { model | subjectFilter = Util.Maybe.fromString str }
+            , NoAction
+            , Cmd.none
+            )
+
 
 
 --- View
@@ -653,13 +664,6 @@ view extraClasses settings model =
                 [ text "The folders to go through"
                 ]
             ]
-        , Html.map ReceivedHoursMsg
-            (Comp.IntField.viewWithInfo
-                "Select mails newer than `now - receivedHours`"
-                model.receivedHours
-                "field"
-                model.receivedHoursModel
-            )
         , div [ class "field" ]
             [ label [] [ text "Target folder" ]
             , input
@@ -688,6 +692,74 @@ view extraClasses settings model =
                 , text " is not set."
                 ]
             ]
+        , div [ class "ui dividing header" ]
+            [ text "Filter"
+            ]
+        , Html.map ReceivedHoursMsg
+            (Comp.IntField.viewWithInfo
+                "Select mails newer than `now - receivedHours`"
+                model.receivedHours
+                "field"
+                model.receivedHoursModel
+            )
+        , div
+            [ class "field"
+            ]
+            [ label [] [ text "File Filter" ]
+            , input
+                [ type_ "text"
+                , onInput SetFileFilter
+                , placeholder "File Filter"
+                , model.fileFilter
+                    |> Maybe.withDefault ""
+                    |> value
+                ]
+                []
+            , div [ class "small-info" ]
+                [ text "Specify a file glob to filter attachments. For example, to only extract pdf files: "
+                , code []
+                    [ text "*.pdf"
+                    ]
+                , text ". If you want to include the mail body, allow html files or "
+                , code []
+                    [ text "mail.html"
+                    ]
+                , text ". Globs can be combined via OR, like this: "
+                , code []
+                    [ text "*.pdf|mail.html"
+                    ]
+                , text ". No file filter defaults to "
+                , code []
+                    [ text "*"
+                    ]
+                , text " that includes all"
+                ]
+            ]
+        , div
+            [ class "field"
+            ]
+            [ label [] [ text "Subject Filter" ]
+            , input
+                [ type_ "text"
+                , onInput SetSubjectFilter
+                , placeholder "Subject Filter"
+                , model.subjectFilter
+                    |> Maybe.withDefault ""
+                    |> value
+                ]
+                []
+            , div [ class "small-info" ]
+                [ text "Specify a file glob to filter mails by subject. For example: "
+                , code []
+                    [ text "*Scanned Document*"
+                    ]
+                , text ". No file filter defaults to "
+                , code []
+                    [ text "*"
+                    ]
+                , text " that includes all"
+                ]
+            ]
         , div [ class "ui dividing header" ]
             [ text "Metadata"
             ]
@@ -763,39 +835,6 @@ disappear then.
                 [ text "Choose tags that should be applied to items."
                 ]
             ]
-        , div
-            [ class "field"
-            ]
-            [ label [] [ text "File Filter" ]
-            , input
-                [ type_ "text"
-                , onInput SetFileFilter
-                , placeholder "File Filter"
-                , model.fileFilter
-                    |> Maybe.withDefault ""
-                    |> value
-                ]
-                []
-            , div [ class "small-info" ]
-                [ text "Specify a file glob to filter attachments. For example, to only extract pdf files: "
-                , code []
-                    [ text "*.pdf"
-                    ]
-                , text ". If you want to include the mail body, allow html files or "
-                , code []
-                    [ text "mail.html"
-                    ]
-                , text ". Globs can be combined via OR, like this: "
-                , code []
-                    [ text "*.pdf|mail.html"
-                    ]
-                , text "No file filter defaults to "
-                , code []
-                    [ text "*"
-                    ]
-                , text " that includes all"
-                ]
-            ]
         , div [ class "ui dividing header" ]
             [ text "Schedule"
             ]