From 5c9e62613b259a7f1d5027992d9a985fe053f8c0 Mon Sep 17 00:00:00 2001
From: 13xforever <ilya.veselov@gmail.com>
Date: Wed, 28 Jan 2026 21:55:58 +0500
Subject: [PATCH] try to match spam messages between different users to further
 reduce ocr lag

---
 .../EventHandlers/MediaScreenshotMonitor.cs   | 58 +++++++++++++++----
 1 file changed, 47 insertions(+), 11 deletions(-)

diff --git a/CompatBot/EventHandlers/MediaScreenshotMonitor.cs b/CompatBot/EventHandlers/MediaScreenshotMonitor.cs
index 7a6d9e79..22a89702 100644
--- a/CompatBot/EventHandlers/MediaScreenshotMonitor.cs
+++ b/CompatBot/EventHandlers/MediaScreenshotMonitor.cs
@@ -1,11 +1,11 @@
-using System.Collections.Concurrent;
-using CompatApiClient.Utils;
+using CompatApiClient.Utils;
 using CompatBot.Commands;
 using CompatBot.Database;
 using CompatBot.Database.Providers;
 using CompatBot.Ocr;
 using CompatBot.Utils.Extensions;
 using Microsoft.Extensions.Caching.Memory;
+using System.Collections.Concurrent;
 
 namespace CompatBot.EventHandlers;
 
@@ -13,8 +13,8 @@ internal sealed class MediaScreenshotMonitor
 {
     private static readonly SemaphoreSlim WorkSemaphore = new(0);
     private static readonly ConcurrentQueue<(DiscordMessage msg, string imgUrl)> WorkQueue = new();
-    private static readonly MemoryCache RemovedMessages = new(new MemoryCacheOptions() { ExpirationScanFrequency = TimeSpan.FromMinutes(5) });
-    private static readonly TimeSpan CachedTime = TimeSpan.FromMinutes(1);
+    private static readonly MemoryCache RemovedMessages = new(new MemoryCacheOptions() { ExpirationScanFrequency = TimeSpan.FromMinutes(10) });
+    private static readonly TimeSpan CachedTime = TimeSpan.FromMinutes(5);
     public DiscordClient Client { get; internal set; } = null!;
     public static int MaxQueueLength { get; private set; }
 
@@ -82,16 +82,40 @@ internal sealed class MediaScreenshotMonitor
 
             try
             {
-                if (await OcrProvider.GetTextAsync(item.imgUrl, Config.Cts.Token).ConfigureAwait(false) is ({Length: >0} result, var confidence)
-                    && !Config.Cts.Token.IsCancellationRequested)
+                var signature = GetSignature(item.msg);
+                var prefix = $"[{item.msg.Id % 100:00}]";
+                if (RemovedMessages.TryGetValue(signature, out (Piracystring hit, DiscordMessage msg) previousItem))
+                {
+                    Config.Log.Debug($"{prefix} OCR result of message {item.msg.JumpLink} from user {item.msg.Author?.Username} ({item.msg.Author?.Id}):");
+                    var ocrTextBuf = new StringBuilder($"OCR result of message <{item.msg.JumpLink}>:").AppendLine();
+                    FilterAction suppressFlags = 0;
+                    if ("media".Equals(item.msg.Channel?.Name))
+                    {
+                        suppressFlags = FilterAction.SendMessage | FilterAction.ShowExplain;
+                    }
+                    await ContentFilter.PerformFilterActions(
+                        Client,
+                        item.msg,
+                        previousItem.hit,
+                        suppressFlags,
+                        $"Matched to previously removed message from {previousItem.msg.Author?.Mention}: {previousItem.msg.JumpLink}",
+                        "🖼 Screenshot of an undesirable content",
+                        "Screenshot of an undesirable content"
+                    ).ConfigureAwait(false);
+                    if (previousItem.hit.Actions.HasFlag(FilterAction.RemoveContent))
+                        RemovedMessages.Set(item.msg.Id, true, CachedTime);
+                }
+                else if (await OcrProvider.GetTextAsync(item.imgUrl, Config.Cts.Token).ConfigureAwait(false) is ({ Length: > 0 } result, var confidence))
                 {
                     var cnt = true;
-                    var prefix = $"[{item.msg.Id % 100:00}]";
-                    var ocrTextBuf = new StringBuilder($"OCR result of message <{item.msg.JumpLink}> ({confidence*100:0.00}%):").AppendLine();
-                    Config.Log.Debug($"{prefix} OCR result of message {item.msg.JumpLink} from user {item.msg.Author?.Username} ({item.msg.Author?.Id}) ({confidence*100:0.00}%):");
                     var duplicates = new HashSet<string>();
-                    ocrTextBuf.AppendLine(result.Sanitize());
-                    Config.Log.Debug($"{prefix} {result}");
+                    Config.Log.Debug($"""
+                        {prefix} OCR result of message {item.msg.JumpLink} from user {item.msg.Author?.Username} ({item.msg.Author?.Id}) ({confidence * 100:0.00}%):
+                        {result}
+                        """
+                    );
+                    var ocrTextBuf = new StringBuilder($"OCR result of message <{item.msg.JumpLink}> ({confidence * 100:0.00}%):").AppendLine()
+                        .AppendLine(result.Sanitize());
                     if (cnt
                         && confidence > 0.65
                         && await ContentFilter.FindTriggerAsync(FilterContext.Chat, result).ConfigureAwait(false) is Piracystring hit
@@ -112,7 +136,11 @@ internal sealed class MediaScreenshotMonitor
                             "Screenshot of an undesirable content"
                         ).ConfigureAwait(false);
                         if (hit.Actions.HasFlag(FilterAction.RemoveContent))
+                        {
                             RemovedMessages.Set(item.msg.Id, true, CachedTime);
+                            if (signature is { Length: >0})
+                                RemovedMessages.Set(signature, (hit, item.msg), CachedTime);
+                        }
                         cnt &= !hit.Actions.HasFlag(FilterAction.RemoveContent) && !hit.Actions.HasFlag(FilterAction.IssueWarning);
                     }
                     var ocrText = ocrTextBuf.ToString();
@@ -138,4 +166,12 @@ internal sealed class MediaScreenshotMonitor
             }
         } while (!Config.Cts.IsCancellationRequested);
     }
+
+    private static string GetSignature(DiscordMessage msg)
+    {
+        var result = msg.Content ?? "";
+        foreach (var att in msg.Attachments)
+            result += $"📎 {att.FileName} ({att.FileSize})\n";
+        return result.TrimEnd();
+    }
 }
\ No newline at end of file