added real auto-ban with social score system (#47)

.env.example

@@ -2,6 +2,8 @@
 BOT_TELEGRAM_TOKEN=SECRET_FROM_TELEGRAM
 BOT_AUTH_TOKEN=JUST_YOUR_SECRET
 BOT_HOOK_ROUTE=/bot
+BOT_USER_ID=123456789
+BOT_USER_NAME=your_bot
 ASPNETCORE_URLS=http://+:88
 DEBUG=true
 LOGS_CHANNEL_ID=-1000000000000
@@ -19,6 +21,9 @@ ML_ENABLED=false
 ML_RETRAIN_INTERVAL_SEC=86400
 ML_SEED=
 ML_SPAM_DELETION_ENABLED=false
+ML_SPAM_AUTOBAN_ENABLED=true
+ML_SPAM_AUTOBAN_SCORE_THRESHOLD=-5.0
+ML_SPAM_AUTOBAN_CHECK_LAST_MSG_COUNT=10
 ML_TRAIN_INTERVAL_DAYS=30
 ML_TRAIN_CRITICAL_MSG_COUNT=5
 ML_TRAINING_SET_FRACTION=0.2

src/VahterBanBot.Tests/ContainerTestBase.fs

@@ -81,6 +81,8 @@ type VahterTestContainers() =
             .WithImage(image)
             .WithNetwork(network)
             .WithPortBinding(80, true)
+            .WithEnvironment("BOT_USER_ID", "1337")
+            .WithEnvironment("BOT_USER_NAME", "test_bot")
             .WithEnvironment("BOT_TELEGRAM_TOKEN", "TELEGRAM_SECRET")
             .WithEnvironment("BOT_AUTH_TOKEN", "OUR_SECRET")
             .WithEnvironment("LOGS_CHANNEL_ID", "-123")
@@ -97,6 +99,9 @@ type VahterTestContainers() =
             .WithEnvironment("ML_SPAM_DELETION_ENABLED", "true")
             .WithEnvironment("ML_SPAM_THRESHOLD", "1.0")
             .WithEnvironment("ML_STOP_WORDS_IN_CHATS", """{"-42":["2"]}""")
+            .WithEnvironment("ML_SPAM_AUTOBAN_ENABLED", "true")
+            .WithEnvironment("ML_SPAM_AUTOBAN_CHECK_LAST_MSG_COUNT", "10")
+            .WithEnvironment("ML_SPAM_AUTOBAN_SCORE_THRESHOLD", "-4.0")
             // .net 8.0 upgrade has a breaking change
             // https://learn.microsoft.com/en-us/dotnet/core/compatibility/containers/8.0/aspnet-port
             // Azure default port for containers is 80, se we need explicitly set it
@@ -203,7 +208,7 @@ type VahterTestContainers() =
         return count > 0
     }
 
-    member _.MessageIsAutoBanned(msg: Message) = task {
+    member _.MessageIsAutoDeleted(msg: Message) = task {
         use conn = new NpgsqlConnection(publicConnectionString)
         //language=postgresql
         let sql = "SELECT COUNT(*) FROM banned_by_bot WHERE banned_in_chat_id = @chatId AND message_id = @messageId"

src/VahterBanBot.Tests/MLBanTests.fs

@@ -16,7 +16,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.True msgBanned
     }
 
@@ -29,7 +29,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.False msgBanned
     }
 
@@ -42,7 +42,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.False msgBanned
     }
 
@@ -54,7 +54,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.False msgBanned
     }
 
@@ -65,7 +65,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.False msgBanned
     }
 
@@ -76,7 +76,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.True msgBanned
     }
 
@@ -88,7 +88,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.True msgBanned
         // assert it is not false-positive
         let! isFalsePositive = fixture.IsMessageFalsePositive msgUpdate.Message
@@ -112,7 +112,7 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! _ = fixture.SendMessage msgUpdate
 
         // assert that the message got auto banned
-        let! msgBanned = fixture.MessageIsAutoBanned msgUpdate.Message
+        let! msgBanned = fixture.MessageIsAutoDeleted msgUpdate.Message
         Assert.True msgBanned
 
         // send a callback to mark it as false-positive
@@ -125,6 +125,104 @@ type MLBanTests(fixture: VahterTestContainers, _unused: MlAwaitFixture) =
         let! isFalsePositive = fixture.IsMessageFalsePositive msgUpdate.Message
         Assert.False isFalsePositive
     }
+
+    [<Fact>]
+    let ``User will be autobanned after consecutive spam`` () = task {
+        // record a message, where 2 is in a training set as spam word
+        // ChatsToMonitor[0] doesn't have stopwords
+        let user = Tg.user()
+        let msgUpdate = Tg.quickMsg(chat = fixture.ChatsToMonitor[0], text = "66666666", from = user)
+
+        // 1 - no ban
+        let! _ = fixture.SendMessage msgUpdate
+        let! msgBanned = fixture.MessageBanned msgUpdate.Message
+        Assert.False msgBanned
+
+        // 2 - no ban
+        let! _ = fixture.SendMessage msgUpdate
+        let! msgBanned = fixture.MessageBanned msgUpdate.Message
+        Assert.False msgBanned
+
+        // 3 - no ban
+        let! _ = fixture.SendMessage msgUpdate
+        let! msgBanned = fixture.MessageBanned msgUpdate.Message
+        Assert.False msgBanned
+
+        // 4 - ban (depends on the ML_SPAM_AUTOBAN_SCORE_THRESHOLD)
+        let! _ = fixture.SendMessage msgUpdate
+        let! msgBanned = fixture.MessageBanned msgUpdate.Message
+        Assert.True msgBanned
+    }
+
+    [<Fact>]
+    let ``User can recover from autoban by sending good messages`` () = task {
+        // record a message, where 2 is in a training set as spam word
+        // ChatsToMonitor[0] doesn't have stopwords
+        let user = Tg.user()
+        let spam = Tg.quickMsg(chat = fixture.ChatsToMonitor[0], text = "66666666", from = user)
+        let notSpam = Tg.quickMsg(chat = fixture.ChatsToMonitor[0], text = "b", from = user)
+
+        // 1 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+
+        // 1.5 - no ban
+        let! _ = fixture.SendMessage notSpam
+        let! msgBanned = fixture.MessageBanned notSpam.Message
+        Assert.False msgBanned
+
+        // 2 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+
+        // 3 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+
+        // 4 - no ban (as user posted 1 good message in beetween)
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+    }
+
+    [<Fact>]
+    let ``User can be saved from auto ban by vahter marking it false-positive`` () = task {
+        // record a message, where 777777777777777777 is in a training set as spam word
+        // ChatsToMonitor[0] doesn't have stopwords
+        let user = Tg.user()
+        let spam = Tg.quickMsg(chat = fixture.ChatsToMonitor[0], text = "777777777777777777", from = user)
+
+        // 1 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        let! msgDeleted = fixture.MessageIsAutoDeleted spam.Message
+        Assert.True msgDeleted
+        Assert.False msgBanned
+
+        // 1.5 - vahter marked as false-positive via button
+        // send a callback to mark it as false-positive
+        let! callbackId = fixture.GetCallbackId spam.Message "NotASpam"
+        let msgCallback = Tg.callback(string callbackId, from = fixture.Vahters[0])
+        let! _ = fixture.SendMessage msgCallback
+
+        // 2 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+
+        // 3 - no ban
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+
+        // 4 - no ban (as vahter marked this as false positive)
+        let! _ = fixture.SendMessage spam
+        let! msgBanned = fixture.MessageBanned spam.Message
+        Assert.False msgBanned
+    }
 
     interface IAssemblyFixture<VahterTestContainers>
     interface IClassFixture<MlAwaitFixture>

src/VahterBanBot/Bot.fs

@@ -447,7 +447,29 @@ let killSpammerAutomated
     logger.LogInformation logMsg
 }
 
+let autoBan
+    (botUser: DbUser)
+    (botClient: ITelegramBotClient)
+    (botConfig: BotConfiguration)
+    (message: Message)
+    (logger: ILogger) = task {
+    use banOnReplyActivity = botActivity.StartActivity("autoBan")
+    %banOnReplyActivity
+        .SetTag("spammerId", message.From.Id)
+        .SetTag("spammerUsername", message.From.Username)
+
+    let! userStats = DB.getUserStatsByLastNMessages botConfig.MlSpamAutobanCheckLastMsgCount message.From.Id
+    let socialScore = userStats.good - userStats.bad
+
+    %banOnReplyActivity.SetTag("socialScore", socialScore)
+
+    if double socialScore <= botConfig.MlSpamAutobanScoreThreshold then
+        // ban user in all monitored chats
+        do! totalBan botClient botConfig message botUser logger
+}
+
 let justMessage
+    (botUser: DbUser)
     (botClient: ITelegramBotClient)
     (botConfig: BotConfiguration)
     (logger: ILogger)
@@ -460,7 +482,16 @@ let justMessage
             .SetTag("fromUserId", message.From.Id)
             .SetTag("fromUsername", message.From.Username)
 
-    if botConfig.MlEnabled && message.Text <> null then
+    // check if user got auto-banned already
+    // that could happen due to the race condition between spammers mass messages
+    // and the bot's processing queue
+    let! isAutoBanned = DB.isBannedByVahter botUser.id message.From.Id 
+    if isAutoBanned then
+        // just delete message and move on
+        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
         use mlActivity = botActivity.StartActivity("mlPrediction")
 
         let shouldBeSkipped =
@@ -487,6 +518,10 @@ let justMessage
                 if prediction.Score >= botConfig.MlSpamThreshold then
                     // delete message
                     do! killSpammerAutomated botClient botConfig message logger botConfig.MlSpamDeletionEnabled prediction.Score
+
+                    if botConfig.MlSpamAutobanEnabled then
+                        // trigger auto-ban check
+                        do! autoBan botUser botClient botConfig message logger
                 elif prediction.Score >= botConfig.MlWarningThreshold then
                     // just warn
                     do! killSpammerAutomated botClient botConfig message logger false prediction.Score
@@ -496,7 +531,7 @@ let justMessage
             | None ->
                 // no prediction (error or not ready yet)
                 ()
-    
+
     do!
         message
         |> DbMessage.newMessage
@@ -570,6 +605,7 @@ let adminCommand
     }
 
 let onMessage
+    (botUser: DbUser)
     (botClient: ITelegramBotClient)
     (botConfig: BotConfiguration)
     (logger: ILogger)
@@ -606,7 +642,7 @@ let onMessage
 
     // if message is not a command from authorized user, just save it ID to DB
     else
-        do! justMessage botClient botConfig logger ml message
+        do! justMessage botUser botClient botConfig logger ml message
 }
 
 let vahterMarkedAsNotSpam
@@ -698,6 +734,7 @@ let onCallback
 }
 
 let onUpdate
+    (botUser: DbUser)
     (botClient: ITelegramBotClient)
     (botConfig: BotConfiguration)
     (logger: ILogger)
@@ -707,5 +744,5 @@ let onUpdate
     if update.CallbackQuery <> null then
         do! onCallback botClient botConfig logger update.CallbackQuery
     else
-        do! onMessage botClient botConfig logger ml update.Message
+        do! onMessage botUser botClient botConfig logger ml update.Message
 }

src/VahterBanBot/DB.fs

@@ -106,16 +106,6 @@ let getUserMessages (userId: int64): Task<DbMessage array> =
         return Array.ofSeq messages
     }
 
-let deleteMsgs (msg: DbMessage[]): Task<int> =
-    task {
-        let msgIds = msg |> Array.map (_.message_id)
-        use conn = new NpgsqlConnection(connString)
-
-        //language=postgresql
-        let sql = "DELETE FROM message WHERE message_id = ANY(@msgIds)"
-        return! conn.ExecuteAsync(sql, {| msgIds = msgIds |})
-    }
-
 let cleanupOldMessages (howOld: TimeSpan): Task<int> =
     task {
         use conn = new NpgsqlConnection(connString)
@@ -315,3 +305,50 @@ let countUniqueUserMsg (userId: int64): Task<int> =
         let! result = conn.QuerySingleAsync<int>(sql, {| userId = userId |})
         return result
     }
+
+let isBannedByVahter (vahterId: int64) (userId: int64): Task<bool> =
+    task {
+        use conn = new NpgsqlConnection(connString)
+
+        //language=postgresql
+        let sql = "SELECT EXISTS(SELECT 1 FROM banned WHERE banned_user_id = @userId AND banned_by = @vahterId)"
+
+        let! result = conn.QuerySingleAsync<bool>(sql, {| userId = userId; vahterId = vahterId |})
+        return result
+    }
+
+let getUserStatsByLastNMessages (n: int) (userId: int64): Task<UserStats> =
+    task {
+        use conn = new NpgsqlConnection(connString)
+
+        //language=postgresql
+        let sql =
+            """
+WITH stats AS (SELECT m.message_id,
+                      m.chat_id,
+                      b.id IS NOT NULL        AS banned,
+                      bbb.id IS NOT NULL      AS banned_by_bot,
+                      fnm.chat_id IS NOT NULL AS false_neg,
+                      fpm.text IS NOT NULL    AS false_pos
+               FROM message m
+                        LEFT JOIN banned b ON m.message_id = b.message_id AND m.chat_id = b.banned_in_chat_id
+                        LEFT JOIN public.banned_by_bot bbb
+                                  ON m.message_id = bbb.message_id AND m.chat_id = bbb.banned_in_chat_id
+                        LEFT JOIN public.false_negative_messages fnm
+                                  ON m.message_id = fnm.message_id AND m.chat_id = fnm.chat_id
+                        LEFT JOIN false_positive_messages fpm ON m.text = fpm.text
+               WHERE m.user_id = @userId
+               ORDER BY m.created_at DESC
+               LIMIT @n),
+     stats_count AS (SELECT message_id,
+                            chat_id,
+                            CASE WHEN false_pos THEN FALSE ELSE banned OR banned_by_bot OR false_neg END AS spam
+                     FROM stats)
+SELECT COUNT(*) FILTER (WHERE NOT spam) AS good,
+       COUNT(*) FILTER (WHERE spam)     AS bad
+FROM stats_count;
+            """
+
+        let! result = conn.QuerySingleAsync<UserStats>(sql, {| userId = userId; n = n |})
+        return result
+    }