* partial revert f2388a7 @ ThreadCrawlFacade.cs & `ThreadLatestReplie…

…rSaver.SaveFromThread()`

* fix `EntityEntry.CurrentValues.SetValues()` won't update navigation prop @ `SaverWithRevision.SaveEntitiesWithRevision()`
* add primary ctor param `existingAfterInTracking` to entities `Existing[].After` & `AllAfter` is the one being tracked by DbContext instead of the one from `PostSaver.Posts` that produced by `PostParser.Convert()` @ SaverChangeSet.cs
* remove part of `user_` in the name of table for entity classes `LatestReplier(Revision)?`
@ c#/crawler

c#/crawler/src/Db/CrawlerDbContext.cs

@@ -69,11 +69,11 @@ protected override void OnModelCreating(ModelBuilder b)
         base.OnModelCreating(b);
         OnModelCreatingWithFid(b, Fid);
         b.Entity<User>().ToTable("tbmc_user");
-        b.Entity<LatestReplier>().ToTable("tbmc_user_latestReplier");
+        b.Entity<LatestReplier>().ToTable("tbmc_latestReplier");
         b.Entity<LatestReplier>().Property(e => e.DisplayName).HasConversion<byte[]>();
         b.Entity<LatestReplier>().HasOne<ThreadPost>().WithOne(e => e.LatestReplier)
             .HasForeignKey<ThreadPost>(e => e.LatestReplierId);
-        b.Entity<LatestReplierRevision>().ToTable("tbmcr_user_latestReplier").HasKey(e => new {e.TakenAt, e.Uid});
+        b.Entity<LatestReplierRevision>().ToTable("tbmcr_latestReplier").HasKey(e => new {e.TakenAt, e.Uid});
         b.Entity<LatestReplierRevision>().Property(e => e.DisplayName).HasConversion<byte[]>();
         b.Entity<ThreadPost>().ToTable($"tbmc_f{Fid}_thread");
         b.Entity<ThreadMissingFirstReply>().ToTable("tbmc_thread_missingFirstReply");

c#/crawler/src/Tieba/Crawl/Facade/ThreadCrawlFacade.cs

@@ -14,6 +14,8 @@ public class ThreadCrawlFacade(
         postParser, postSaverFactory.Invoke,
         userParserFactory.Invoke, userSaverFactory.Invoke)
 {
+    private readonly Dictionary<ThreadLatestReplierSaver.UniqueLatestReplier, LatestReplier?> _latestRepliersKeyByUnique = [];
+
     public delegate ThreadCrawlFacade New(Fid fid, string forumName);
 
     protected override void OnPostParse(
@@ -42,17 +44,21 @@ join parsed in Posts.Values on (Tid)inResponse.Tid equals parsed.Tid
             { // replace with more detailed location.name in the 6.0.2 response
                 t.parsed.Geolocation = Helper.SerializedProtoBufOrNullIfEmpty(t.inResponse.Location);
             }
-            var lastReplyer = t.inResponse.LastReplyer;
-            var name = lastReplyer?.Name.NullIfEmpty();
-            var nameShow = lastReplyer?.NameShow.NullIfEmpty();
+            var name = t.inResponse.LastReplyer.Name.NullIfEmpty();
+            var nameShow = t.inResponse.LastReplyer.NameShow.NullIfEmpty();
 
             // LastReplyer will be null when LivePostType != "", but LastTimeInt will have expected timestamp value
-            t.parsed.LatestReplier = lastReplyer == null ? null : new LatestReplier
+            var latestReplierEntity = t.inResponse.LastReplyer == null ? null : new LatestReplier
             {
                 Name = name,
 #pragma warning disable S3358 // Ternary operators should not be nested
                 DisplayName = name == nameShow ? null : nameShow
 #pragma warning restore S3358 // Ternary operators should not be nested
             };
+            var uniqueLatestReplier = ThreadLatestReplierSaver.UniqueLatestReplier.FromLatestReplier(latestReplierEntity);
+
+            var isExists = _latestRepliersKeyByUnique.TryGetValue(uniqueLatestReplier, out var existingLatestReplier);
+            if (!isExists) _latestRepliersKeyByUnique[uniqueLatestReplier] = latestReplierEntity;
+            t.parsed.LatestReplier = isExists ? existingLatestReplier : latestReplierEntity;
         });
 }

c#/crawler/src/Tieba/Crawl/Saver/Post/PostSaver.cs

@@ -33,12 +33,12 @@ protected SaverChangeSet<TPost> Save<TRevision>(
         var existingPostsKeyById = db.Set<TPost>().AsTracking()
             .Where(existingPostPredicate).ToDictionary(postIdSelector);
 
-        // deep copy before entities get mutated by SaverWithRevision.SaveEntitiesWithRevision()
-        var existingBeforeMerge = existingPostsKeyById.Select(pair => (TPost)pair.Value.Clone()).ToList();
+        // clone before entities get mutated by SaverWithRevision.SaveEntitiesWithRevision()
+        var existingPostsBeforeMerge = existingPostsKeyById.Select(pair => (TPost)pair.Value.Clone()).ToList();
 
         SaveEntitiesWithRevision(db, revisionFactory,
             Posts.Values.ToLookup(p => existingPostsKeyById.ContainsKey(postIdSelector(p))),
             p => existingPostsKeyById[postIdSelector(p)]);
-        return new(existingBeforeMerge, Posts.Values, postIdSelector);
+        return new(postIdSelector, existingPostsBeforeMerge, Posts.Values, existingPostsKeyById.Values);
     }
 }

c#/crawler/src/Tieba/Crawl/Saver/Post/ThreadSaver.cs

@@ -72,7 +72,7 @@ protected override bool FieldUpdateIgnorance
         nameof(ThreadPost.Title)
             when newValue is ""
 
-            // prevent repeatedly update with different title
+            // prevent update repeatedly with different title
             // due to the thread is a multi forum topic thread
             // thus its title can be varied within the forum and within the thread
             || (newValue is not "" && oldValue is not "") => true,

c#/crawler/src/Tieba/Crawl/Saver/Related/ThreadLatestReplierSaver.cs

@@ -1,5 +1,3 @@
-using Microsoft.EntityFrameworkCore.ChangeTracking;
-
 namespace tbm.Crawler.Tieba.Crawl.Saver.Related;
 
 public class ThreadLatestReplierSaver(
@@ -12,27 +10,10 @@ public class ThreadLatestReplierSaver(
 
     public Action SaveFromThread(CrawlerDbContext db, IReadOnlyCollection<ThreadPost> threads)
     {
-        static void DetachThenReplace(
-            EntityEntry<LatestReplier> entityEntry,
-            ThreadPost thread,
-            LatestReplier newLatestReplier)
-        {
-            entityEntry.State = EntityState.Detached;
-            thread.LatestReplier = newLatestReplier;
-        }
-
-        var threadsGroupByUniqueLatestReplier = threads
+        var uniqueLatestRepliers = threads
             .Where(th => th.LatestReplier != null)
-            .GroupBy(UniqueLatestReplier.FromThread).ToList();
-        threadsGroupByUniqueLatestReplier.ForEach(g =>
-            (from thread in g.Skip(1)
-                join entityEntry in db.ChangeTracker.Entries<LatestReplier>()
-                    on thread.LatestReplier equals entityEntry.Entity
-                select (thread, entityEntry))
-            .ForEach(t => DetachThenReplace(t.entityEntry, t.thread, g.First().LatestReplier!)));
-
-        var uniqueLatestRepliers = threadsGroupByUniqueLatestReplier.Select(g => g.Key).ToList();
-        var existingLatestRepliers = db.LatestRepliers.AsNoTracking().FilterByItems(
+            .Select(UniqueLatestReplier.FromThread).ToList();
+        var existingLatestRepliers = db.LatestRepliers.AsTracking().FilterByItems(
                 uniqueLatestRepliers, (latestReplier, uniqueLatestReplier) =>
                     latestReplier.Name == uniqueLatestReplier.Name
                     && latestReplier.DisplayName == uniqueLatestReplier.DisplayName)
@@ -43,7 +24,11 @@ on UniqueLatestReplier.FromLatestReplier(existing) equals UniqueLatestReplier.Fr
                 join entityEntry in db.ChangeTracker.Entries<LatestReplier>()
                     on thread.LatestReplier equals entityEntry.Entity // Object.ReferenceEquals()
                 select (existing, thread, entityEntry))
-            .ForEach(t => DetachThenReplace(t.entityEntry, t.thread, t.existing));
+            .ForEach(t =>
+            {
+                t.entityEntry.State = EntityState.Detached;
+                t.thread.LatestReplier = t.existing;
+            });
 
         _ = _saverLocks.Value.Acquire(uniqueLatestRepliers
             .Except(existingLatestRepliers.Select(UniqueLatestReplier.FromLatestReplier))

c#/crawler/src/Tieba/Crawl/Saver/SaverChangeSet.cs

@@ -1,14 +1,15 @@
 namespace tbm.Crawler.Tieba.Crawl.Saver;
 
 public class SaverChangeSet<TPost>(
+    Func<TPost, PostId> postIdSelector,
     IReadOnlyCollection<TPost> existingBefore,
     ICollection<TPost> existingAfterAndNewlyAdded,
-    Func<TPost, PostId> postIdSelector)
+    IReadOnlyCollection<TPost> existingAfterInTracking)
     where TPost : BasePost
 {
     public IReadOnlyCollection<(TPost Before, TPost After)> Existing { get; } = existingBefore
         .OrderBy(postIdSelector)
-        .EquiZip(existingAfterAndNewlyAdded
+        .EquiZip(existingAfterInTracking
                 .IntersectBy(existingBefore.Select(postIdSelector), postIdSelector)
                 .OrderBy(postIdSelector),
             (before, after) => (before, after))
@@ -18,6 +19,11 @@ public class SaverChangeSet<TPost>(
         .ExceptBy(existingBefore.Select(postIdSelector), postIdSelector)
         .ToList().AsReadOnly();
 
-    public IReadOnlyCollection<TPost> AllAfter { get; } = existingAfterAndNewlyAdded
-        .ToList().AsReadOnly();
+    // https://stackoverflow.com/questions/3404975/left-outer-join-in-linq/23558389#23558389
+    public IReadOnlyCollection<TPost> AllAfter { get; } = (
+        from nonTracked in existingAfterAndNewlyAdded
+        join inTracking in existingAfterInTracking
+            on postIdSelector(nonTracked) equals postIdSelector(inTracking) into inTrackings
+        from inTracking in inTrackings.DefaultIfEmpty()
+        select inTracking ?? nonTracked).ToList().AsReadOnly();
 }

c#/crawler/src/Tieba/Crawl/Saver/SaverWithRevision.cs

@@ -83,6 +83,13 @@ protected void SaveEntitiesWithRevision<TEntity, TRevision>(
             entityEntry.CurrentValues.SetValues(newEntity); // mutate existingEntity that referenced by entry
             entityEntry.Property(e => e.Version).IsModified = false; // newEntity.Version will always be default 0
 
+            // https://stackoverflow.com/questions/66206459/update-navigation-property-with-entity-currentvalues-setvalues/66491805#66491805
+            (from newNavigation in db.Entry(newEntity).Navigations
+                    join existingNavigation in entityEntry.Navigations
+                        on newNavigation.Metadata.Name equals existingNavigation.Metadata.Name
+                    select (newNavigation, existingNavigation))
+                .ForEach(t => t.existingNavigation.CurrentValue = t.newNavigation.CurrentValue);
+
             bool IsTimestampingFieldName(string name) => name is nameof(BasePost.LastSeenAt)
                 or nameof(TimestampedEntity.CreatedAt) or nameof(TimestampedEntity.UpdatedAt);