+ abstruct class BaseUser to extract fields (Display)?Name from e…

…ntity class `User`

* fix false positive of signature existence due to `WHERE signatureId IN (s1, s2) AND xxHash3 IN (x1, x2)` is not equality with `WHERE (signatureId = x1 AND xxHash3 = x1) OR (signatureId = x2 AND xxHash3 = x2)` @ `ReplySignatureSaver.Save()`
* fix username of historical anonymous user won't get fallbacked to null when it's emtpy string @ `UserParser.Parse()`
* fix outdated comments since a0f48f3 @ `UserSaver.ShouldIgnoreEntityRevision()`
@ fe

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

@@ -0,0 +1,7 @@
+namespace tbm.Crawler.Db;
+
+public abstract class BaseUser : TimestampedEntity
+{
+    public string? Name { get; set; }
+    public string? DisplayName { get; set; }
+}

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

@@ -1,10 +1,8 @@
 namespace tbm.Crawler.Db;
 
-public class User : TimestampedEntity
+public class User : BaseUser
 {
     [Key] public long Uid { get; set; }
-    public string? Name { get; set; }
-    public string? DisplayName { get; set; }
     public required string Portrait { get; set; }
     public uint? PortraitUpdatedAt { get; set; }
     public byte? Gender { get; set; }

c#/crawler/src/Tieba/Crawl/Parser/UserParser.cs

@@ -20,7 +20,7 @@ public void Parse(IEnumerable<TbClient.User> inUsers) =>
                 return new()
                 {
                     Uid = uid,
-                    Name = el.NameShow,
+                    Name = el.NameShow.NullIfEmpty(),
                     Portrait = portrait,
                     PortraitUpdatedAt = portraitUpdatedAt
                 };

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

@@ -17,7 +17,7 @@ public override SaverChangeSet<ReplyPost> Save(CrawlerDbContext db)
             r => new ReplyRevision {TakenAt = r.UpdatedAt ?? r.CreatedAt, Pid = r.Pid},
             LinqKit.PredicateBuilder.New<ReplyPost>(r => Posts.Keys.Contains(r.Pid)));
 
-        db.ReplyContents.AddRange(changeSet.NewlyAdded
+        db.ReplyContents.AddRange(changeSet.NewlyAdded // https://github.com/dotnet/efcore/issues/33945
             .Select(r => new ReplyContent {Pid = r.Pid, ProtoBufBytes = r.Content}));
         PostSaveHandlers += replyContentImageSaver.Save(db, changeSet.NewlyAdded).Invoke;
         PostSaveHandlers += AuthorRevisionSaver.SaveAuthorExpGradeRevisions(db, changeSet.AllAfter).Invoke;

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

@@ -15,7 +15,7 @@ public override SaverChangeSet<SubReplyPost> Save(CrawlerDbContext db)
             sr => new SubReplyRevision {TakenAt = sr.UpdatedAt ?? sr.CreatedAt, Spid = sr.Spid},
             LinqKit.PredicateBuilder.New<SubReplyPost>(sr => Posts.Keys.Contains(sr.Spid)));
 
-        db.SubReplyContents.AddRange(changeSet.NewlyAdded.Select(sr =>
+        db.SubReplyContents.AddRange(changeSet.NewlyAdded.Select(sr => // https://github.com/dotnet/efcore/issues/33945
             new SubReplyContent {Spid = sr.Spid, ProtoBufBytes = sr.Content}));
         PostSaveHandlers += AuthorRevisionSaver.SaveAuthorExpGradeRevisions(db, changeSet.AllAfter).Invoke;
 

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

@@ -40,14 +40,16 @@ public Action Save(CrawlerDbContext db, IEnumerable<ReplyPost> replies)
                 r => r.Signature,
                 SignatureIdAndValueEqualityComparer.Instance);
 
-        var existingSignatures = (
-            from s in db.ReplySignatures.AsTracking()
-            where signatures.Select(s2 => s2.SignatureId).Contains(s.SignatureId)
-
-                  // server side eval doesn't need ByteArrayEqualityComparer
-                  && signatures.Select(s2 => s2.XxHash3).Contains(s.XxHash3)
-            select s
-        ).ToList();
+        var existingSignatures = db.ReplySignatures.AsTracking()
+            .Where(signatures.Aggregate(
+                LinqKit.PredicateBuilder.New<ReplySignature>(),
+                (predicate, newOrExisting) =>
+                    predicate.Or(LinqKit.PredicateBuilder
+                        .New<ReplySignature>(existing =>
+                            existing.SignatureId == newOrExisting.SignatureId)
+                        .And(existing =>
+                            existing.XxHash3 == newOrExisting.XxHash3))))
+            .ToList();
         (from existing in existingSignatures
                 join newInReply in signatures on existing.SignatureId equals newInReply.SignatureId
                 select (existing, newInReply))

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

@@ -72,22 +72,13 @@ public partial class UserSaver
 {
     protected override bool ShouldIgnoreEntityRevision(string propName, PropertyEntry propEntry, EntityEntry entityEntry)
     {
-        // ThreadCrawlFacade.ParseLatestRepliers() will save users with empty string as portrait
-        // they may soon be updated by (sub) reply crawler after it find out the latest reply
+        // ThreadCrawlFacade.ParseLatestRepliers() will save partial filled user of latest repliers for livepost thread
+        // they may later get updated by (sub) reply crawler after it find out the latest reply
         // so we should ignore its revision update for all fields
-        // ignore entire record is not possible via IFieldChangeIgnorance.GlobalFieldChangeIgnorance.Revision()
-        // since it can only determine one field at the time
         if (propName != nameof(User.Portrait) || propEntry.OriginalValue is not "") return false;
-
-        // invokes OriginalValues.ToObject() to get a new instance
-        // since entityInTracking is reference to the changed one
         var user = (User)entityEntry.OriginalValues.ToObject();
-
-        // create another user instance with only fields of latest replier filled
         var latestReplier = User.CreateLatestReplier(user.Uid, user.Name, user.DisplayName);
 
-        // if they are same by fields values, the original one is the latest replier
-        // that previously generated by ParseLatestRepliers()
         return User.EqualityComparer.Instance.Equals(user, latestReplier);
     }