src/NLog/Targets/FileArchiveHandlers/BaseFileArchiveHandler.cs (1)

317-321: Optionally trim to the first digit within the chosen segment (handles multiple leading separators)

Today only one leading non-digit is trimmed. If the segment starts with multiple separators (e.g., "__-001"), consider trimming all leading non-digits so the wildcard anchors exactly at the numeric portion.

Apply this diff inside the same block to advance to the first digit:

-                if (lastStart > 0 && lastLength > 1 && !char.IsDigit(filename[lastStart + 1]))
-                {
-                    lastStart += 1;
-                    lastLength -= 1;
-                }
+                if (lastStart > 0 && lastLength > 0)
+                {
+                    int advance = 0;
+                    while (advance < lastLength && !char.IsDigit(filename[lastStart + advance]))
+                        advance++;
+                    if (advance > 0 && advance < lastLength)
+                    {
+                        lastStart += advance;
+                        lastLength -= advance;
+                    }
+                }

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.

src/NLog/Targets/FileArchiveHandlers/BaseFileArchiveHandler.cs (2)

284-296: Include preceding non-letter chars in the same run (optional).

Currently the run starts at the first digit, excluding leading separators like '_' that belong to the same contiguous non-letter run. Broadening the run improves robustness without changing behavior.

Apply:

-            bool hasDigit = false;
+            bool hasDigit = false;
             for (int i = 0; i < filename.Length; ++i)
             {
                 if (!char.IsLetter(filename[i]))
                 {
-                    hasDigit = hasDigit || char.IsDigit(filename[i]);
-                    if (hasDigit)
-                    {
-                        if (currentLength == 0)
-                            currentStart = i;
-                        ++currentLength;
-                    }
+                    if (currentLength == 0)
+                        currentStart = i;
+                    ++currentLength;
+                    if (char.IsDigit(filename[i]))
+                        hasDigit = true;
                 }
                 else
                 {
                     if (currentLength != 0)
                     {
-                        if (lastLength <= currentLength)
+                        if (hasDigit && lastLength <= currentLength)
                         {
                             lastStart = currentStart;
                             lastLength = currentLength;
                         }
                         currentLength = 0;
                     }
-                    hasDigit = false;
+                    hasDigit = false;
                 }
             }
 
-            if (lastLength < currentLength)
+            if (hasDigit && lastLength < currentLength)
             {
                 lastStart = currentStart;
                 lastLength = currentLength;
             }

Also applies to: 308-316

---

52-60: Plumb “contains sequence” to sorting (optional).

When the wildcard stems from a numeric segment, CleanupFiles can sort by sequence for more deterministic trimming. Consider returning a flag from GetDeleteOldFileNameWildcard and passing it to wildCardContainsSeqNo.

Proposed minimal internal change:

-        protected bool DeleteOldFilesBeforeArchive(string filePath, bool initialFileOpen, string? excludeFileName = null)
+        protected bool DeleteOldFilesBeforeArchive(string filePath, bool initialFileOpen, string? excludeFileName = null)
         {
             // Get all files matching the filename, order by timestamp, and when same timestamp then order by filename
             //  - First start with removing the oldest files
             string fileDirectory = Path.GetDirectoryName(filePath);
             // Replace all non-letter with '*' replace all '**' with single '*'
-            string fileWildcard = GetDeleteOldFileNameWildcard(filePath);
-            return DeleteOldFilesBeforeArchive(fileDirectory, fileWildcard, initialFileOpen, excludeFileName);
+            bool wildcardHasDigits;
+            string fileWildcard = GetDeleteOldFileNameWildcard(filePath, out wildcardHasDigits);
+            return DeleteOldFilesBeforeArchive(fileDirectory, fileWildcard, initialFileOpen, excludeFileName, wildcardHasDigits);
         }

-        private static string GetDeleteOldFileNameWildcard(string filepath)
+        private static string GetDeleteOldFileNameWildcard(string filepath, out bool containsDigitsSegment)
         {
             var filename = Path.GetFileNameWithoutExtension(filepath) ?? string.Empty;
             var fileext = Path.GetExtension(filepath) ?? string.Empty;
             if (string.IsNullOrEmpty(filename) && string.IsNullOrEmpty(fileext))
-                return string.Empty;
+            {
+                containsDigitsSegment = false;
+                return string.Empty;
+            }
 ...
-            if (lastLength > 0)
+            containsDigitsSegment = lastLength > 0;
+            if (lastLength > 0)
                 ...
             return string.Concat(filename, "*", fileext);
         }

Please run the affected unit tests that rely on DeleteOldFilesBeforeArchive ordering to confirm no regressions.

tests/NLog.UnitTests/Targets/FileTargetTests.cs (1)

3247-3296: Add a regression test for multi-dot filenames with pre-extension numeric suffix (optional).

To lock the fix for issue #5976, add an explicit test with "nlog-all.ecs.json" and ArchiveSuffixFormat="_{0:000}".

Suggested addition:

@@
         public void FileTarget_SameDirectory_MaxArchiveFiles(int maxArchiveFiles)
         {
@@
         }
+
+        [Fact]
+        public void MaxArchiveFiles_MultiDotFile_PreExtSuffix()
+        {
+            var tempDir = ArchiveFileNameHelper.GenerateTempPath();
+            var baseName = Path.Combine(tempDir, "nlog-all.ecs.json");
+            try
+            {
+                var fileTarget = new FileTarget
+                {
+                    FileName = baseName,
+                    ArchiveAboveSize = 100,
+                    LineEnding = LineEndingMode.LF,
+                    Layout = "${message}",
+                    MaxArchiveFiles = 3,
+                    Encoding = Encoding.ASCII,
+                    ArchiveSuffixFormat = "_{0:000}",
+                };
+
+                LogManager.Setup().LoadConfiguration(c => c.ForLogger().WriteTo(fileTarget));
+                Generate100BytesLog('a'); // -> _000
+                Generate100BytesLog('b'); // -> _001
+                Generate100BytesLog('c'); // -> _002
+                Generate100BytesLog('d'); // -> _003 (current becomes ddd)
+                LogManager.Configuration = null;
+
+                var times = 25;
+                AssertFileContents(Path.Combine(tempDir, "nlog-all.ecs_003.json"), StringRepeat(times, "ddd\n"), Encoding.ASCII);
+                Assert.True(File.Exists(Path.Combine(tempDir, "nlog-all.ecs_001.json")));
+                Assert.True(File.Exists(Path.Combine(tempDir, "nlog-all.ecs_002.json")));
+                Assert.False(File.Exists(Path.Combine(tempDir, "nlog-all.ecs_000.json"))); // trimmed
+                Assert.False(File.Exists(Path.Combine(tempDir, "nlog-all.ecs_004.json"))); // not created
+            }
+            finally
+            {
+                if (Directory.Exists(tempDir))
+                    Directory.Delete(tempDir, true);
+            }
+        }

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.