From 2ae2d92e261fb4c500d5b8b46766b0972b9badf9 Mon Sep 17 00:00:00 2001
From: Sergey Gut <kruzefag@gmail.com>
Date: Sun, 18 Feb 2024 18:50:01 +0300
Subject: [PATCH] increase speed of mysql when searching posts

---
 Core/Service/Hash.cs                          | 21 +++++++++++++++++++
 Core/Startup.cs                               |  2 ++
 modules/RichSiteSummary/Data/Post.cs          |  9 +++++++-
 .../RichSiteSummaryRepositoryExtension.cs     |  6 +++++-
 modules/RichSiteSummary/Service/RssFetch.cs   |  8 +++++--
 5 files changed, 42 insertions(+), 4 deletions(-)
 create mode 100644 Core/Service/Hash.cs

diff --git a/Core/Service/Hash.cs b/Core/Service/Hash.cs
new file mode 100644
index 0000000..2e55136
--- /dev/null
+++ b/Core/Service/Hash.cs
@@ -0,0 +1,21 @@
+using System.Security.Cryptography;
+using System.Text;
+
+namespace Kruzya.TelegramBot.Core.Service
+{
+    public class Hash
+    {
+        public string GenerateHash(string input)
+        {
+            var output = string.Empty;
+            var hashedBytes = SHA256.HashData(Encoding.UTF8.GetBytes(input));
+
+            foreach (var b in hashedBytes)
+            {
+                output += string.Format("{0,2:x2}", b);
+            }
+
+            return output;
+        }
+    }
+}
diff --git a/Core/Startup.cs b/Core/Startup.cs
index 98414cc..b81117e 100644
--- a/Core/Startup.cs
+++ b/Core/Startup.cs
@@ -59,6 +59,8 @@ public class Startup
         services.AddScoped<IOptionProvider, DbOptionProvider>();
         services.AddHttpClient();
 
+        services.AddSingleton<Hash>();
+
         // Trigger module handlers.
         _core.Modules.ConfigureServices(services);
     }
diff --git a/modules/RichSiteSummary/Data/Post.cs b/modules/RichSiteSummary/Data/Post.cs
index 8657e96..245a5a2 100644
--- a/modules/RichSiteSummary/Data/Post.cs
+++ b/modules/RichSiteSummary/Data/Post.cs
@@ -68,7 +68,7 @@ public class Post : RepresentableEntity
     /// The URI where this post is located in web.
     /// </summary>
     [Required]
-    [MaxLength(256)]
+    [MaxLength(1024)]
     public string Url { get; set; }
         
     /// <summary>
@@ -83,6 +83,13 @@ public class Post : RepresentableEntity
     [Required]
     public DateTime ReceivedAt { get; set; }
 
+    /// <summary>
+    /// 
+    /// </summary>
+    [Required]
+    [MaxLength(64)]
+    public string UrlHash { get; set; }
+
     public Post()
     {
         PostId = new Guid();
diff --git a/modules/RichSiteSummary/RichSiteSummaryRepositoryExtension.cs b/modules/RichSiteSummary/RichSiteSummaryRepositoryExtension.cs
index 95cf763..d4c44b4 100644
--- a/modules/RichSiteSummary/RichSiteSummaryRepositoryExtension.cs
+++ b/modules/RichSiteSummary/RichSiteSummaryRepositoryExtension.cs
@@ -61,11 +61,15 @@ internal static class RichSiteSummaryRepositoryExtension
         await repository.ForFetching(period, 25);
 
     #endregion
-        
+
     #region Post
 
+    [Obsolete("Use ByFeedAndUrlHash()")]
     public static async Task<Post> ByFeedAndUrl(this DbSet<Post> repository, string url, Feed feed = null) =>
         await repository.FirstOrDefaultAsync(post => post.Url == url && (feed == null || post.Feed == feed));
 
+    public static async Task<Post> ByFeedAndUrlHash(this DbSet<Post> repository, string urlHash, Feed feed = null) =>
+        await repository.FirstOrDefaultAsync(post => post.UrlHash == urlHash && (feed == null || post.Feed == feed));
+
     #endregion
 }
\ No newline at end of file
diff --git a/modules/RichSiteSummary/Service/RssFetch.cs b/modules/RichSiteSummary/Service/RssFetch.cs
index 7e3118a..d424462 100644
--- a/modules/RichSiteSummary/Service/RssFetch.cs
+++ b/modules/RichSiteSummary/Service/RssFetch.cs
@@ -19,6 +19,7 @@ using Telegram.Bot.Types.Enums;
 using Feed = Kruzya.TelegramBot.RichSiteSummary.Data.Feed;
 
 using CodeHollow.FeedReader.Feeds;
+using Kruzya.TelegramBot.Core.Service;
 
 namespace Kruzya.TelegramBot.RichSiteSummary.Service;
 
@@ -35,14 +36,16 @@ public class RssFetch : IHostedService, IDisposable
     private readonly ILogger _logger;
     private readonly IServiceScopeFactory _scopeFactory;
     private readonly ConcurrentQueue<UserMessage> _queue;
+    private readonly Hash _hash;
 
     private static TimeSpan TimerPeriod => TimeSpan.FromSeconds(45);
 
-    public RssFetch(ILogger<RssFetch> logger, ConcurrentQueue<UserMessage> queue, IServiceScopeFactory scopeFactory)
+    public RssFetch(ILogger<RssFetch> logger, ConcurrentQueue<UserMessage> queue, IServiceScopeFactory scopeFactory, Hash hash)
     {
         _scopeFactory = scopeFactory;
         _logger = logger;
         _queue = queue;
+        _hash = hash;
     }
 
     #region IHostedService
@@ -132,7 +135,7 @@ public class RssFetch : IHostedService, IDisposable
         var postsImages = new Dictionary<Post, string>();
         foreach (var post in parsedFeed.Items)
         {
-            feedPost = await dbContext.Posts.ByFeedAndUrl(post.Link, feed);
+            feedPost = await dbContext.Posts.ByFeedAndUrlHash(_hash.GenerateHash(post.Link), feed);
             if (feedPost != null)
             {
                 // skip. This post already exists.
@@ -144,6 +147,7 @@ public class RssFetch : IHostedService, IDisposable
             feedPost.Title = post.Title;
             feedPost.Url = post.Link;
             feedPost.PostedAt = post.PublishingDate.GetValueOrDefault(DateTime.Now);
+            feedPost.UrlHash = _hash.GenerateHash(feedPost.Url);
 
             postsImages.Add(feedPost, await FetchImageUrl(post));
             newPosts.Add(feedPost);