Updated various parts of scanning and ingestion, either for bug fixes, or for enhancements.

JSMR.Infrastructure/Scanning/Extensions/DLSiteWorkExtensions.cs

@@ -0,0 +1,62 @@
+using JSMR.Application.Scanning.Contracts;
+
+namespace JSMR.Infrastructure.Scanning.Extensions;
+
+public static class DLSiteWorkExtensions
+{
+    public static void InferAndUpdateExpectedDates(this DLSiteWork[] works)
+    {
+        // Precompute nearest known effective date on the left and right for each index.
+        var left = new DateOnly?[works.Length];
+        var right = new DateOnly?[works.Length];
+
+        DateOnly? last = null;
+        for (int i = 0; i < works.Length; i++)
+        {
+            var effective = GetEffectiveDate(works[i]);
+            if (effective.HasValue)
+                last = effective;
+
+            left[i] = last;
+        }
+
+        DateOnly? next = null;
+        for (int i = works.Length - 1; i >= 0; i--)
+        {
+            var effective = GetEffectiveDate(works[i]);
+            if (effective.HasValue)
+                next = effective;
+
+            right[i] = next;
+        }
+
+        // Fill only when BOTH sides exist and match.
+        for (int i = 0; i < works.Length; i++)
+        {
+            DLSiteWork work = works[i];
+
+            if (work.SalesDate.HasValue || work.ExpectedDate.HasValue)
+                continue;
+
+            DateOnly? previous = (i > 0) ? left[i - 1] : null;
+            DateOnly? nxt = (i < works.Length - 1) ? right[i + 1] : null;
+
+            if (previous.HasValue && nxt.HasValue && previous.Value == nxt.Value)
+                work.ExpectedDate = previous.Value;
+        }
+    }
+
+    private static DateOnly? GetEffectiveDate(DLSiteWork work)
+    {
+        if (work.ExpectedDate.HasValue)
+            return work.ExpectedDate.Value;
+
+        if (!work.SalesDate.HasValue)
+            return null;
+
+        // Bucket sales day to Early/Middle/Late => 1/11/21
+        var d = work.SalesDate.Value;
+        int day = d.Day >= 21 ? 21 : d.Day >= 11 ? 11 : 1;
+        return new DateOnly(d.Year, d.Month, day);
+    }
+}

JSMR.Infrastructure/Scanning/VoiceWorksScanner.cs

@@ -5,9 +5,9 @@ using JSMR.Application.Scanning.Ports;
 using JSMR.Domain.Enums;
 using JSMR.Domain.ValueObjects;
 using JSMR.Infrastructure.Http;
+using JSMR.Infrastructure.Scanning.Extensions;
 using JSMR.Infrastructure.Scanning.Models;
 using System.Globalization;
-using System.Text.Json;
 using System.Text.RegularExpressions;
 
 namespace JSMR.Infrastructure.Scanning;
@@ -25,7 +25,10 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
         DLSiteHtmlDocument document = await GetDLSiteHtmlCollectionAsync(options, cancellationToken);
         DLSiteHtmlNode[] nodes = document.GetDLSiteNodes();
 
-        return GetDLSiteWorks(nodes, options);
+        DLSiteWork[] works = GetDLSiteWorks(nodes, options);
+        works.InferAndUpdateExpectedDates();
+
+        return works;
     }
 
     private async Task<DLSiteHtmlDocument> GetDLSiteHtmlCollectionAsync(VoiceWorkScanOptions options, CancellationToken cancellationToken)
@@ -53,7 +56,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
         return filterBuilder.BuildSearchQuery(options.PageNumber, options.PageSize);
     }
 
-    private List<DLSiteWork> GetDLSiteWorks(DLSiteHtmlNode[] nodes, VoiceWorkScanOptions options)
+    private DLSiteWork[] GetDLSiteWorks(DLSiteHtmlNode[] nodes, VoiceWorkScanOptions options)
     {
         var works = new List<DLSiteWork>();
 
@@ -67,7 +70,7 @@ public abstract class VoiceWorksScanner(IHtmlLoader htmlLoader) : IVoiceWorksSca
             works.Add(work);
         }
 
-        return works;
+        return [.. works];
     }
 
     private DLSiteWork GetDLSiteWork(DLSiteHtmlNode node)