* opt out entity splitting due to currently it only supporting `one-o…

…ne` relationships instead of desired `one-zeroOrOne`(optional entity), leading to inserting many records with all fields except the primary key with NULL values as placeholder record for `one-one` relation, so without re-introducing something like `IRevision.NullFieldsBitMask`, we can't distinguish between the original literal NULL value and these empty records: dotnet/efcore#27974 dotnet/efcore#29113 @ `TbmDbContext.OnModelCreating()`

+ abstract class `RevisionWithSplitting` as base class of all derived classes of `IRevision`
* insert all entities returned from `RevisionWithSplitting.GetSplitEntities()` into DB @ `CommonInSavers.SavePostsOrUsers()`
* change `abstract class BaseRevision` to `interface IRevision` to comply with single inheritance
@ crawler

crawler/src/Db/Revision/IRevision.cs

@@ -0,0 +1,9 @@
+// ReSharper disable UnusedMember.Global
+namespace tbm.Crawler.Db.Revision
+{
+    public interface IRevision
+    {
+        public uint TakenAt { get; set; }
+        public ushort? NullFieldsBitMask { get; set; }
+    }
+}

crawler/src/Db/Revision/ReplyRevision.cs

@@ -1,14 +1,45 @@
 // ReSharper disable PropertyCanBeMadeInitOnly.Global
 namespace tbm.Crawler.Db.Revision
 {
-    public class ReplyRevision : BaseRevision
+    public class ReplyRevision : ReplyRevision.BaseReplyRevision
     {
-        public ulong Pid { get; set; }
-        public uint Floor { get; set; }
-        public uint SubReplyCount { get; set; }
+        public abstract class BaseReplyRevision : RevisionWithSplitting<BaseReplyRevision>
+        {
+            public ulong Pid { get; set; }
+        }
+        [NotMapped] public uint Floor
+        {
+            get => GetSplitEntityValue<SplitFloor, uint>(r => r.Floor);
+            set => SetSplitEntityValue<SplitFloor, uint>(value, (r, v) => r.Floor = v,
+                () => new() {TakenAt = TakenAt, Pid = Pid, Floor = value});
+        }
+        [NotMapped] public uint SubReplyCount
+        {
+            get => GetSplitEntityValue<SplitSubReplyCount, uint>(r => r.SubReplyCount);
+            set => SetSplitEntityValue<SplitSubReplyCount, uint>(value, (r, v) => r.SubReplyCount = v,
+                () => new() {TakenAt = TakenAt, Pid = Pid, SubReplyCount = value});
+        }
         public ushort? IsFold { get; set; }
-        public int AgreeCount { get; set; }
+        [NotMapped] public int AgreeCount
+        {
+            get => GetSplitEntityValue<SplitAgreeCount, int>(r => r.AgreeCount);
+            set => SetSplitEntityValue<SplitAgreeCount, int>(value, (r, v) => r.AgreeCount = v,
+                () => new() {TakenAt = TakenAt, Pid = Pid, AgreeCount = value});
+        }
         public int? DisagreeCount { get; set; }
         public byte[]? Geolocation { get; set; }
+
+        public class SplitFloor : BaseReplyRevision
+        {
+            public uint Floor { get; set; }
+        }
+        public class SplitSubReplyCount : BaseReplyRevision
+        {
+            public uint SubReplyCount { get; set; }
+        }
+        public class SplitAgreeCount : BaseReplyRevision
+        {
+            public int AgreeCount { get; set; }
+        }
     }
 }

crawler/src/Db/Revision/RevisionWithSplitting.cs

@@ -0,0 +1,27 @@
+namespace tbm.Crawler.Db.Revision
+{
+    public abstract class RevisionWithSplitting<TSplitEntities> : IRevision
+    {
+        public uint TakenAt { get; set; }
+        public ushort? NullFieldsBitMask { get; set; }
+
+        private Dictionary<Type, TSplitEntities> SplitEntities { get; } = new();
+        public IEnumerable<TSplitEntities> GetSplitEntities() => SplitEntities.Values;
+
+        protected TValue? GetSplitEntityValue<TSplitEntity, TValue>(Func<TSplitEntity, TValue?> valueSelector)
+            where TSplitEntity : class, TSplitEntities =>
+            SplitEntities.ContainsKey(typeof(TSplitEntity))
+                ? valueSelector((TSplitEntity)SplitEntities[typeof(TSplitEntity)]!)
+                : default;
+
+        protected void SetSplitEntityValue<TSplitEntity, TValue>(TValue? value,
+            Action<TSplitEntity, TValue?> valueSetter, Func<TSplitEntity> entityFactory)
+            where TSplitEntity : class, TSplitEntities
+        {
+            if (SplitEntities.ContainsKey(typeof(TSplitEntity)))
+                valueSetter((TSplitEntity)SplitEntities[typeof(TSplitEntity)]!, value);
+            else
+                SplitEntities[typeof(TSplitEntity)] = entityFactory();
+        }
+    }
+}

crawler/src/Db/Revision/SubReplyRevision.cs

@@ -1,10 +1,32 @@
 // ReSharper disable PropertyCanBeMadeInitOnly.Global
 namespace tbm.Crawler.Db.Revision
 {
-    public class SubReplyRevision : BaseRevision
+    public class SubReplyRevision : SubReplyRevision.BaseSubReplyRevision
     {
-        public ulong Spid { get; set; }
-        public int AgreeCount { get; set; }
-        public int DisagreeCount { get; set; }
+        public abstract class BaseSubReplyRevision : RevisionWithSplitting<BaseSubReplyRevision>
+        {
+            public ulong Spid { get; set; }
+        }
+        [NotMapped] public int AgreeCount
+        {
+            get => GetSplitEntityValue<SplitAgreeCount, int>(r => r.AgreeCount);
+            set => SetSplitEntityValue<SplitAgreeCount, int>(value, (r, v) => r.AgreeCount = v,
+                () => new() {TakenAt = TakenAt, Spid = Spid, AgreeCount = value});
+        }
+        [NotMapped] public int DisagreeCount
+        {
+            get => GetSplitEntityValue<SplitDisagreeCount, int>(r => r.DisagreeCount);
+            set => SetSplitEntityValue<SplitDisagreeCount, int>(value, (r, v) => r.DisagreeCount = v,
+                () => new() {TakenAt = TakenAt, Spid = Spid, DisagreeCount = value});
+        }
+
+        public class SplitAgreeCount : BaseSubReplyRevision
+        {
+            public int AgreeCount { get; set; }
+        }
+        public class SplitDisagreeCount : BaseSubReplyRevision
+        {
+            public int DisagreeCount { get; set; }
+        }
     }
 }

crawler/src/Db/Revision/ThreadRevision.cs

@@ -1,20 +1,33 @@
 // ReSharper disable PropertyCanBeMadeInitOnly.Global
 namespace tbm.Crawler.Db.Revision
 {
-    public class ThreadRevision : BaseRevision
+    public class ThreadRevision : ThreadRevision.BaseThreadRevision
     {
-        public ulong Tid { get; set; }
+        public abstract class BaseThreadRevision : RevisionWithSplitting<BaseThreadRevision>
+        {
+            public ulong Tid { get; set; }
+        }
         public ulong? ThreadType { get; set; }
         public string? StickyType { get; set; }
         public string? TopicType { get; set; }
         public ushort? IsGood { get; set; }
         public uint? LatestReplyPostedAt { get; set; }
         public long? LatestReplierUid { get; set; }
         public uint? ReplyCount { get; set; }
-        public uint ViewCount { get; set; }
+        [NotMapped] public uint ViewCount
+        {
+            get => GetSplitEntityValue<SplitViewCount, uint>(r => r.ViewCount);
+            set => SetSplitEntityValue<SplitViewCount, uint>(value, (r, v) => r.ViewCount = v,
+                () => new() {TakenAt = TakenAt, Tid = Tid, ViewCount = value});
+        }
         public uint? ShareCount { get; set; }
         public int? AgreeCount { get; set; }
         public int? DisagreeCount { get; set; }
         public byte[]? Geolocation { get; set; }
+
+        public class SplitViewCount : BaseThreadRevision
+        {
+            public uint ViewCount { get; set; }
+        }
     }
 }

crawler/src/Db/Revision/UserRevision.cs

@@ -2,16 +2,47 @@
 // ReSharper disable PropertyCanBeMadeInitOnly.Global
 namespace tbm.Crawler.Db.Revision
 {
-    public class UserRevision : BaseRevision
+    public class UserRevision : UserRevision.BaseUserRevision
     {
-        public long Uid { get; set; }
-        public string TriggeredBy { get; set; } = "";
+        public abstract class BaseUserRevision : RevisionWithSplitting<BaseUserRevision>
+        {
+            public long Uid { get; set; }
+            public string TriggeredBy { get; set; } = "";
+        }
         public string? Name { get; set; }
-        public string? DisplayName { get; set; }
+        [NotMapped] public string? DisplayName
+        {
+            get => GetSplitEntityValue<SplitDisplayName, string?>(r => r.DisplayName);
+            set => SetSplitEntityValue<SplitDisplayName, string?>(value, (r, v) => r.DisplayName = v,
+                () => new() {TakenAt = TakenAt, Uid = Uid, TriggeredBy = TriggeredBy, DisplayName = value});
+        }
         public string? Portrait { get; set; }
-        public uint? PortraitUpdatedAt { get; set; }
+        [NotMapped] public uint? PortraitUpdatedAt
+        {
+            get => GetSplitEntityValue<SplitPortraitUpdatedAt, uint?>(r => r.PortraitUpdatedAt);
+            set => SetSplitEntityValue<SplitPortraitUpdatedAt, uint?>(value, (r, v) => r.PortraitUpdatedAt = v,
+                () => new() {TakenAt = TakenAt, Uid = Uid, TriggeredBy = TriggeredBy, PortraitUpdatedAt = value});
+        }
         public ushort? Gender { get; set; }
         public byte[]? Icon { get; set; }
-        public string? IpGeolocation { get; set; }
+        [NotMapped] public string? IpGeolocation
+        {
+            get => GetSplitEntityValue<SplitIpGeolocation, string?>(r => r.IpGeolocation);
+            set => SetSplitEntityValue<SplitIpGeolocation, string?>(value, (r, v) => r.IpGeolocation = v,
+                () => new() {TakenAt = TakenAt, Uid = Uid, TriggeredBy = TriggeredBy, IpGeolocation = value});
+        }
+
+        public class SplitDisplayName : BaseUserRevision
+        {
+            public string? DisplayName { get; set; }
+        }
+        public class SplitPortraitUpdatedAt : BaseUserRevision
+        {
+            public uint? PortraitUpdatedAt { get; set; }
+        }
+        public class SplitIpGeolocation : BaseUserRevision
+        {
+            public string? IpGeolocation { get; set; }
+        }
     }
 }

crawler/src/Db/TbmDbContext.cs

@@ -40,35 +40,19 @@ protected override void OnModelCreating(ModelBuilder b)
             b.Entity<ReplyContent>().ToTable($"tbmc_f{Fid}_reply_content");
             b.Entity<SubReplyPost>().ToTable($"tbmc_f{Fid}_subReply");
             b.Entity<SubReplyContent>().ToTable($"tbmc_f{Fid}_subReply_content");
-            b.Entity<ThreadRevision>()
-                .SplitToTable("tbmc_revision_thread_viewCount", tb => tb.Property(e => e.ViewCount))
-                .ToTable("tbmc_revision_thread").HasKey(e => new {e.Tid, e.TakenAt});
-            b.Entity<ReplyRevision>()
-                .SplitToTable("tbmc_revision_reply_agreeCount", tb => tb.Property(e => e.AgreeCount))
-                .SplitToTable("tbmc_revision_reply_subReplyCount", tb => tb.Property(e => e.SubReplyCount))
-                .SplitToTable("tbmc_revision_reply_floor", tb => tb.Property(e => e.Floor))
-                .ToTable("tbmc_revision_reply").HasKey(e => new {e.Pid, e.TakenAt});
-            b.Entity<SubReplyRevision>()
-                .SplitToTable("tbmc_revision_subReply_agreeCount", tb => tb.Property(e => e.AgreeCount))
-                .SplitToTable("tbmc_revision_subReply_disagreeCount", tb => tb.Property(e => e.DisagreeCount))
-                .ToTable("tbmc_revision_subReply").HasKey(e => new {e.Spid, e.TakenAt});
-            b.Entity<UserRevision>()
-                .SplitToTable("tbmc_revision_user_ipGeolocation", tb =>
-                {
-                    tb.Property(e => e.TriggeredBy);
-                    tb.Property(e => e.IpGeolocation);
-                })
-                .SplitToTable("tbmc_revision_user_portraitUpdatedAt", tb =>
-                {
-                    tb.Property(e => e.TriggeredBy);
-                    tb.Property(e => e.PortraitUpdatedAt);
-                })
-                .SplitToTable("tbmc_revision_user_displayName", tb =>
-                {
-                    tb.Property(e => e.TriggeredBy);
-                    tb.Property(e => e.DisplayName);
-                })
-                .ToTable("tbmc_revision_user").HasKey(e => new {e.Uid, e.TakenAt});
+            b.Entity<ThreadRevision>().ToTable("tbmc_revision_thread").HasKey(e => new {e.Tid, e.TakenAt});
+            b.Entity<ThreadRevision.SplitViewCount>().ToTable("tbmc_revision_thread_viewCount").HasKey(e => new {e.Tid, e.TakenAt});
+            b.Entity<ReplyRevision>().ToTable("tbmc_revision_reply").HasKey(e => new {e.Pid, e.TakenAt});
+            b.Entity<ReplyRevision.SplitAgreeCount>().ToTable("tbmc_revision_reply_agreeCount").HasKey(e => new {e.Pid, e.TakenAt});
+            b.Entity<ReplyRevision.SplitSubReplyCount>().ToTable("tbmc_revision_reply_subReplyCount").HasKey(e => new {e.Pid, e.TakenAt});
+            b.Entity<ReplyRevision.SplitFloor>().ToTable("tbmc_revision_reply_floor").HasKey(e => new {e.Pid, e.TakenAt});
+            b.Entity<SubReplyRevision>().ToTable("tbmc_revision_subReply").HasKey(e => new {e.Spid, e.TakenAt});
+            b.Entity<SubReplyRevision.SplitAgreeCount>().ToTable("tbmc_revision_subReply_agreeCount").HasKey(e => new {e.Spid, e.TakenAt});
+            b.Entity<SubReplyRevision.SplitDisagreeCount>().ToTable("tbmc_revision_subReply_disagreeCount").HasKey(e => new {e.Spid, e.TakenAt});
+            b.Entity<UserRevision>().ToTable("tbmc_revision_user").HasKey(e => new {e.Uid, e.TakenAt});
+            b.Entity<UserRevision.SplitIpGeolocation>().ToTable("tbmc_revision_user_ipGeolocation").HasKey(e => new {e.Uid, e.TakenAt});
+            b.Entity<UserRevision.SplitPortraitUpdatedAt>().ToTable("tbmc_revision_user_portraitUpdatedAt").HasKey(e => new {e.Uid, e.TakenAt});
+            b.Entity<UserRevision.SplitDisplayName>().ToTable("tbmc_revision_user_displayName").HasKey(e => new {e.Uid, e.TakenAt});
             b.Entity<AuthorExpGradeRevision>().ToTable("tbmc_revision_authorExpGrade").HasKey(e => new {e.Fid, e.Uid, e.DiscoveredAt});
             b.Entity<ForumModeratorRevision>().ToTable("tbmc_revision_forumModerator");
             b.Entity<Forum>().ToTable("tbm_forum");

crawler/src/Tieba/Crawl/Saver/BaseSaver.cs

@@ -33,7 +33,7 @@ protected SaverChangeSet<TPost> SavePosts<TRevision>(TbmDbContext db,
             ExpressionStarter<TPost> existingPostPredicate,
             Func<IEnumerable<TRevision>, Expression<Func<TRevision, bool>>> existingRevisionPredicate,
             Expression<Func<TRevision, TRevision>> revisionKeySelector)
-            where TRevision : BaseRevision, new()
+            where TRevision : class, IRevision, new()
         {
             var dbSet = db.Set<TPost>().TagWith("ForUpdate");
             if (dbSet == null) throw new ArgumentException(

crawler/src/Tieba/Crawl/Saver/CommonInSavers.cs

@@ -19,9 +19,10 @@ protected void SavePostsOrUsers<TPostOrUser, TRevision>(
             Func<TRevision, long> revisionPostOrUserIdSelector,
             Func<IEnumerable<TRevision>, Expression<Func<TRevision, bool>>> existingRevisionPredicate,
             Expression<Func<TRevision, TRevision>> revisionKeySelector)
-            where TPostOrUser : class where TRevision : BaseRevision, new()
+            where TPostOrUser : class where TRevision : class, IRevision, new()
         {
             db.Set<TPostOrUser>().AddRange(existingOrNewLookup[false]); // newly added
+            db.TimestampingEntities();
             var newRevisions = existingOrNewLookup[true].Select(newPostOrUser =>
             {
                 var postOrUserInTracking = existingSelector(newPostOrUser);
@@ -101,7 +102,11 @@ or nameof(ITimestampingEntity.CreatedAt)
                 if (revision != null) revision.NullFieldsBitMask = (ushort?)revisionNullFieldsBitMask;
                 return revision;
             }).OfType<TRevision>().ToList();
-            db.TimestampingEntities();
+
+            db.AddRange(newRevisions.OfType<ThreadRevision>().Select(r => r.GetSplitEntities()));
+            db.AddRange(newRevisions.OfType<ReplyRevision>().Select(r => r.GetSplitEntities()));
+            db.AddRange(newRevisions.OfType<SubReplyRevision>().Select(r => r.GetSplitEntities()));
+            db.AddRange(newRevisions.OfType<UserRevision>().Select(r => r.GetSplitEntities()));
 
             if (!newRevisions.Any()) return; // quick exit to prevent execute sql with WHERE FALSE clause
             var existingRevisions = db.Set<TRevision>()