Restore IETF Tags When a Misnamed Matroska File Remuxes Through FFmpeg

ARCHITECTURE.md

@@ -147,7 +147,7 @@ Critical performance feature - DO NOT break compatibility:
 
 - Only MkvMerge reports IETF tags, and a missing one is derived from the ISO 639-2B tag
 - MkvMerge normalizes to IETF tags when `SetIetfLanguageTags` enabled
-- FfMpeg and HandBrake drop the IETF tags when they write a file. Each of their rewrites restores the tags with MkvPropEdit before the output replaces the original, from the track list read before the rewrite. `Convert.RestoreLanguageTags()` first checks that the output keeps each track's number, type, and ISO 639-2B tag. Otherwise the rewrite fails, and the original stays. A HandBrake output holds the video, then the audio, then the subtitle tracks, so `GetHandBrakeTrackOrder()` numbers copies of the tags that way. The FfMpeg remux fallback restores them only for a Matroska input, since another container carries no IETF tags.
+- FfMpeg and HandBrake drop the IETF tags when they write a file. Each of their rewrites restores the tags with MkvPropEdit before the output replaces the original, from the original input's track list. `Convert.RestoreLanguageTags()` first checks that the output keeps each track's number, type, and ISO 639-2B tag. Otherwise the rewrite fails, and the original stays. A HandBrake output holds the video, then the audio, then the subtitle tracks, so `GetHandBrakeTrackOrder()` numbers copies of the tags that way. The FfMpeg remux fallback restores them only for a Matroska input, since another container carries no IETF tags.
 - FFprobe uses tag metadata which may differ from track metadata
 - Track validation: Checks ISO/IETF consistency, sets error states for mismatches
 
@@ -442,7 +442,9 @@ flowchart TD
   X3 -->|fail| XF
   X3 --> X4["ReMux(): MkvMerge pass over the FfMpeg output"]
   X4 -->|fail| XF
-  X4 --> X9["RestoreLanguageTags() when the input is Matroska"]
+  X4 --> X10["The input's MkvMerge track list, read here when the caller passed none"]
+  X10 -->|"read cancelled"| XF
+  X10 --> X9["RestoreLanguageTags() when the input is Matroska"]
   X9 -->|fail| XF
   X1 --> X5{"The .mkv name is the input file itself?"}
   X9 --> X5

PlexCleaner/Convert.cs

@@ -108,6 +108,21 @@ out string outputName
                 return false;
             }
 
+            // A caller that has not read the props yet passes none, so read them only for this fallback
+            // An input MkvMerge cannot identify keeps the fallback, but a cancel must not skip the restore
+            if (languageProps == null)
+            {
+                if (Tools.MkvMerge.GetMediaProps(inputName, out MediaProps props))
+                {
+                    languageProps = props;
+                }
+                else if (Program.IsCancelledError())
+                {
+                    File.Delete(tempName);
+                    return false;
+                }
+            }
+
             // Only a Matroska input carries IETF language tags to restore
             if (
                 languageProps?.IsContainerMkv() == true

PlexCleanerTests/ReplaceOriginalTests.cs

@@ -538,8 +538,15 @@ bool success
         _ = Directory.GetFiles(_directory, "name.tmp*").Should().BeEmpty();
     }
 
-    [Fact]
-    public void RemuxByExtension_FfMpegFallbackBeforeMediaProps_RemuxesWithoutRestore()
+    [Theory]
+    [InlineData("Matroska", 0, true)]
+    [InlineData("AVI", 0, false)]
+    [InlineData("Matroska", 2, false)]
+    public void RemuxByExtension_FfMpegFallback_RestoresLanguageOnlyForMatroska(
+        string container,
+        int identifyExit,
+        bool restored
+    )
     {
         Assert.SkipWhen(OperatingSystem.IsWindows(), "The fake tools are shell scripts");
         using IDisposable toolErrorScope = ToolErrorScope.BeginScope();
@@ -554,10 +561,24 @@ public void RemuxByExtension_FfMpegFallbackBeforeMediaProps_RemuxesWithoutRestor
         string propEditLog = WriteFakeTools("*.tmp2|*.tmp4", 0);
         FakeTools.WriteSampleToolData(_directory);
 
+        // The input identifies as the given container, the FfMpeg output as the sample Matroska tracks
+        // Only the input's audio track carries en-GB, so the restored tag shows which list was read
+        string identify = FakeTools.SampleIdentifyCase(_directory);
+        string identifyJson = Path.Combine(_directory, "MkvMergeInfoData.json");
+        JsonNode inputInfo = JsonNode.Parse(File.ReadAllText(identifyJson))!;
+        inputInfo["container"]!["type"] = container;
+        JsonNode audio = inputInfo["tracks"]!
+            .AsArray()
+            .First(track => track!["type"]!.GetValue<string>() == "audio")!;
+        audio["properties"]!["language_ietf"] = "en-GB";
+        string inputJson = Path.Combine(_directory, "InputInfoData.json");
+        File.WriteAllText(inputJson, inputInfo.ToJsonString());
+
         // MkvMerge fails to remux the input, then remuxes the FfMpeg output
         FakeTools.WriteScript(
             Tools.MkvMerge.GetToolPath(),
-            FakeTools.SampleIdentifyCase(_directory)
+            $"case \"$*\" in *--identify*name.avi*) cat \"{inputJson}\"; exit {identifyExit} ;; esac\n"
+                + identify
                 + "case \"$*\" in *.tmp4*) ;; *) exit 2 ;; esac\n"
                 + "for a; do case $a in *.tmp4) echo mkvmerge > \"$a\" ;; esac; done\n"
         );
@@ -570,7 +591,16 @@ public void RemuxByExtension_FfMpegFallbackBeforeMediaProps_RemuxesWithoutRestor
         _ = processFile.FileInfo.FullName.Should().Be(mkvFile);
         _ = File.ReadAllText(mkvFile).Should().StartWith("mkvmerge");
         _ = File.Exists(aviFile).Should().BeFalse();
-        _ = File.Exists(propEditLog).Should().BeFalse();
+        _ = File.Exists(propEditLog).Should().Be(restored);
+        if (restored)
+        {
+            string propEditArgs = File.ReadAllText(propEditLog);
+            _ = propEditArgs.Should().Contain("name.tmp2");
+            long number = audio["properties"]!["number"]!.GetValue<long>();
+            _ = $" {propEditArgs.Trim()} "
+                .Should()
+                .Contain($" --edit track:@{number} --set language=en-GB ");
+        }
         _ = Directory.GetFiles(_directory, "name.tmp*").Should().BeEmpty();
     }