Now ML triggers on caption as well (#54)

src/VahterBanBot.Tests/MLBanTests.fs

@@ -223,6 +223,19 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! msgBanned = fixture.MessageBanned spam.Message
         Assert.False msgBanned
     }
+
+    [<Fact>]
+    let ``Message with spam in photo caption also triggers auto-delete`` () = task {
+        // record a message, where 2222222 is in a training set as spam word
+        // but text is in a message.Caption
+        // ChatsToMonitor[0] doesn't have stopwords
+        let msgUpdate = Tg.quickMsg(chat = fixture.ChatsToMonitor[0], text = null, caption = "2222222")
+        let! _ = fixture.SendMessage msgUpdate
+
+        // assert that the message got auto banned
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
+        Assert.True msgBanned
+    }
 
     interface IAssemblyFixture<VahterTestContainers>
     interface IClassFixture<MlAwaitFixture>

src/VahterBanBot.Tests/TgMessageUtils.fs

@@ -32,7 +32,7 @@ type Tg() =
             )
         )
 
-    static member quickMsg (?text: string, ?chat: Chat, ?from: User, ?date: DateTime, ?callback: CallbackQuery) =
+    static member quickMsg (?text: string, ?chat: Chat, ?from: User, ?date: DateTime, ?callback: CallbackQuery, ?caption: string) =
         Update(
             Id = next(),
             Message = 
@@ -42,6 +42,7 @@ type Tg() =
                     Chat = (chat |> Option.defaultValue (Tg.chat())),
                     From = (from |> Option.defaultValue (Tg.user())),
                     Date = (date |> Option.defaultValue DateTime.UtcNow),
+                    Caption = (caption |> Option.defaultValue null),
                     ReplyToMessage = null
                 )
             )

src/VahterBanBot/Bot.fs

@@ -424,7 +424,7 @@ let killSpammerAutomated
             |> DB.banUserByBot
 
     let msgType = if deleteMessage then "Deleted" else "Detected"
-    let logMsg = $"{msgType} spam (score: {score}) in {prependUsername message.Chat.Username} ({message.Chat.Id}) from {prependUsername message.From.Username} ({message.From.Id}) with text:\n{message.Text}"
+    let logMsg = $"{msgType} spam (score: {score}) in {prependUsername message.Chat.Username} ({message.Chat.Id}) from {prependUsername message.From.Username} ({message.From.Id}) with text:\n{message.TextOrCaption}"
 
     let! replyMarkup = task {
         if deleteMessage then
@@ -493,7 +493,7 @@ let justMessage
         do! botClient.DeleteMessageAsync(ChatId(message.Chat.Id), message.MessageId)
             |> safeTaskAwait (fun e -> logger.LogError ($"Failed to delete message {message.MessageId} from chat {message.Chat.Id}", e))
 
-    elif botConfig.MlEnabled && message.Text <> null then
+    elif botConfig.MlEnabled && message.TextOrCaption <> null then
         use mlActivity = botActivity.StartActivity("mlPrediction")
 
         let shouldBeSkipped =
@@ -506,14 +506,14 @@ let justMessage
             match botConfig.MlStopWordsInChats.TryGetValue message.Chat.Id with
             | true, stopWords ->
                 stopWords
-                |> Seq.exists (fun sw -> message.Text.Contains(sw, StringComparison.OrdinalIgnoreCase))
+                |> Seq.exists (fun sw -> message.TextOrCaption.Contains(sw, StringComparison.OrdinalIgnoreCase))
             | _ -> false
         %mlActivity.SetTag("skipPrediction", shouldBeSkipped)
 
         if not shouldBeSkipped then
             let! usrMsgCount = DB.countUniqueUserMsg message.From.Id
 
-            match ml.Predict(message.Text, usrMsgCount)  with
+            match ml.Predict(message.TextOrCaption, usrMsgCount)  with
             | Some prediction ->
                 %mlActivity.SetTag("spamScoreMl", prediction.Score)
 
@@ -667,7 +667,7 @@ let vahterMarkedAsNotSpam
 
     let vahterUsername = vahter.username |> Option.defaultValue null
 
-    let logMsg = $"Vahter {prependUsername vahterUsername} ({vahter.id}) marked message {msgId} in {prependUsername chatName}({chatId}) as false-positive (NOT A SPAM)\n{msg.message.Text}"
+    let logMsg = $"Vahter {prependUsername vahterUsername} ({vahter.id}) marked message {msgId} in {prependUsername chatName}({chatId}) as false-positive (NOT A SPAM)\n{msg.message.TextOrCaption}"
     do! botClient.SendTextMessageAsync(ChatId(botConfig.LogsChannelId), logMsg) |> taskIgnore
     logger.LogInformation logMsg
 }

src/VahterBanBot/Types.fs

@@ -72,7 +72,7 @@ module DbBanned =
         if isNull message.From || isNull message.Chat then
             failwith "Message should have a user and a chat"
         { message_id = Some message.MessageId
-          message_text = message.Text
+          message_text = message.TextOrCaption
           banned_user_id = message.From.Id
           banned_at = DateTime.UtcNow
           banned_in_chat_id = Some message.Chat.Id
@@ -92,7 +92,7 @@ type DbMessage =
           message_id = message.MessageId
           user_id = message.From.Id
           created_at = DateTime.UtcNow
-          text = message.Text
+          text = message.TextOrCaption
           raw_message = JsonConvert.SerializeObject message }
 
 [<CLIMutable>]

src/VahterBanBot/Utils.fs

@@ -56,3 +56,10 @@ type Task<'x> with
     member this.Ignore() = task { let! _ = this in () }
 
 let inline taskIgnore (t: Task<'x>) = t.Ignore()
+
+type Telegram.Bot.Types.Message with
+    member msg.TextOrCaption =
+        if isNull msg.Text then
+            msg.Caption
+        else
+            msg.Text