fix: accept valid Thunder subtitle encodings

Emby.MeiamSub.Thunder/Emby.MeiamSub.Thunder.csproj

@@ -33,6 +33,10 @@
     <Folder Include="Configuration\" />
   </ItemGroup>
 
+  <ItemGroup>
+    <Compile Include="..\Shared\SubtitleContentValidator.cs" Link="Shared\SubtitleContentValidator.cs" />
+  </ItemGroup>
+
   <Target Name="PostBuild" AfterTargets="PostBuildEvent">
     <Copy SourceFiles="$(TargetDir)$(TargetFileName)" DestinationFolder="$(SolutionDir)$(ConfigurationName)\" SkipUnchangedFiles="false" />
   </Target>

Emby.MeiamSub.Thunder/ThunderProvider.cs

@@ -1,4 +1,5 @@
 using Emby.MeiamSub.Thunder.Model;
+using MeiamSubtitles.Shared;
 using MediaBrowser.Common;
 using MediaBrowser.Common.Net;
 using MediaBrowser.Controller.Base;
@@ -285,7 +286,7 @@ namespace Emby.MeiamSub.Thunder
                     await response.Content.CopyToAsync(stream, 81920, cancellationToken);
                     var data = stream.ToArray();
                     stream.Dispose();
-                    ValidateSubtitleData(data, format, response.ContentType);
+                    SubtitleContentValidator.Validate(data, format, response.ContentType);
 
                     return new SubtitleResponse()
                     {
@@ -355,38 +356,6 @@ namespace Emby.MeiamSub.Thunder
             return value == SRT || value == ASS || value == SSA ? value : null;
         }
 
-        private static void ValidateSubtitleData(byte[] data, string format, string mediaType)
-        {
-            if (data == null || data.Length < 8)
-            {
-                throw new InvalidDataException("Subtitle response is empty.");
-            }
-
-            if (data[0] == (byte)'P' && data[1] == (byte)'K' ||
-                data.Length >= 7 && Encoding.ASCII.GetString(data, 0, 7) == "Rar!\u001a\u0007")
-            {
-                throw new InvalidDataException("Compressed subtitle responses are not supported by Thunder.");
-            }
-
-            var sample = Encoding.UTF8.GetString(data, 0, Math.Min(data.Length, 4096)).TrimStart('\uFEFF', ' ', '\r', '\n', '\t');
-            if (sample.StartsWith("<", StringComparison.Ordinal) ||
-                sample.StartsWith("{", StringComparison.Ordinal) ||
-                sample.StartsWith("[", StringComparison.Ordinal) ||
-                mediaType?.IndexOf("html", StringComparison.OrdinalIgnoreCase) >= 0 ||
-                mediaType?.IndexOf("json", StringComparison.OrdinalIgnoreCase) >= 0)
-            {
-                throw new InvalidDataException("Thunder returned an error document instead of a subtitle.");
-            }
-
-            var valid = format == SRT
-                ? sample.IndexOf("-->", StringComparison.Ordinal) >= 0
-                : sample.IndexOf("[Script Info]", StringComparison.OrdinalIgnoreCase) >= 0 || sample.IndexOf("Dialogue:", StringComparison.OrdinalIgnoreCase) >= 0;
-            if (!valid)
-            {
-                throw new InvalidDataException($"Downloaded content is not a valid {format} subtitle.");
-            }
-        }
-
         /// <summary>
         /// 规范化语言代码
         /// </summary>

Jellyfin.MeiamSub.Thunder/Jellyfin.MeiamSub.Thunder.csproj

@@ -21,6 +21,10 @@
     </None>
   </ItemGroup>
 
+  <ItemGroup>
+    <Compile Include="..\Shared\SubtitleContentValidator.cs" Link="Shared\SubtitleContentValidator.cs" />
+  </ItemGroup>
+
   <ItemGroup>
     <EmbeddedResource Include="Configuration\configPage.html" />
   </ItemGroup>

Jellyfin.MeiamSub.Thunder/ThunderProvider.cs

@@ -1,4 +1,5 @@
 using Jellyfin.MeiamSub.Thunder.Model;
+using MeiamSubtitles.Shared;
 using MediaBrowser.Controller.Providers;
 using MediaBrowser.Controller.Subtitles;
 using MediaBrowser.Model.Providers;
@@ -282,7 +283,7 @@ namespace Jellyfin.MeiamSub.Thunder
                     var format = NormalizeFormat(downloadSub.Format)
                         ?? throw new InvalidDataException($"Unsupported subtitle format: {downloadSub.Format}");
                     var data = await response.Content.ReadAsByteArrayAsync(cancellationToken);
-                    ValidateSubtitleData(data, format, response.Content.Headers.ContentType?.MediaType);
+                    SubtitleContentValidator.Validate(data, format, response.Content.Headers.ContentType?.MediaType);
                     var stream = new MemoryStream(data, writable: false);
 
                     return new SubtitleResponse()
@@ -353,38 +354,6 @@ namespace Jellyfin.MeiamSub.Thunder
             return value == SRT || value == ASS || value == SSA ? value : null;
         }
 
-        private static void ValidateSubtitleData(byte[] data, string format, string mediaType)
-        {
-            if (data == null || data.Length < 8)
-            {
-                throw new InvalidDataException("Subtitle response is empty.");
-            }
-
-            if (data[0] == (byte)'P' && data[1] == (byte)'K' ||
-                data.Length >= 7 && Encoding.ASCII.GetString(data, 0, 7) == "Rar!\u001a\u0007")
-            {
-                throw new InvalidDataException("Compressed subtitle responses are not supported by Thunder.");
-            }
-
-            var sample = Encoding.UTF8.GetString(data, 0, Math.Min(data.Length, 4096)).TrimStart('\uFEFF', ' ', '\r', '\n', '\t');
-            if (sample.StartsWith("<", StringComparison.Ordinal) ||
-                sample.StartsWith("{", StringComparison.Ordinal) ||
-                sample.StartsWith("[", StringComparison.Ordinal) ||
-                mediaType?.IndexOf("html", StringComparison.OrdinalIgnoreCase) >= 0 ||
-                mediaType?.IndexOf("json", StringComparison.OrdinalIgnoreCase) >= 0)
-            {
-                throw new InvalidDataException("Thunder returned an error document instead of a subtitle.");
-            }
-
-            var valid = format == SRT
-                ? sample.Contains("-->", StringComparison.Ordinal)
-                : sample.Contains("[Script Info]", StringComparison.OrdinalIgnoreCase) || sample.Contains("Dialogue:", StringComparison.OrdinalIgnoreCase);
-            if (!valid)
-            {
-                throw new InvalidDataException($"Downloaded content is not a valid {format} subtitle.");
-            }
-        }
-
         /// <summary>
         /// 规范化语言代码
         /// </summary>

MeiamSubtitles.sln

@@ -13,6 +13,8 @@ Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "Jellyfin.MeiamSub.Assrt", "
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Emby.MeiamSub.Assrt", "Emby.MeiamSub.Assrt\Emby.MeiamSub.Assrt.csproj", "{D276A8D0-4B17-4EEF-94B6-61D646DE5A22}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "MeiamSubtitles.Tests", "tests\MeiamSubtitles.Tests\MeiamSubtitles.Tests.csproj", "{B8A8BFC1-8A0D-48C5-89AC-4DDDC2DDC3C9}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
@@ -39,6 +41,10 @@ Global
 		{D276A8D0-4B17-4EEF-94B6-61D646DE5A22}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{D276A8D0-4B17-4EEF-94B6-61D646DE5A22}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{D276A8D0-4B17-4EEF-94B6-61D646DE5A22}.Release|Any CPU.Build.0 = Release|Any CPU
+		{B8A8BFC1-8A0D-48C5-89AC-4DDDC2DDC3C9}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{B8A8BFC1-8A0D-48C5-89AC-4DDDC2DDC3C9}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{B8A8BFC1-8A0D-48C5-89AC-4DDDC2DDC3C9}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{B8A8BFC1-8A0D-48C5-89AC-4DDDC2DDC3C9}.Release|Any CPU.Build.0 = Release|Any CPU
 	EndGlobalSection
 	GlobalSection(SolutionProperties) = preSolution
 		HideSolutionNode = FALSE

Shared/SubtitleContentValidator.cs

@@ -0,0 +1,108 @@
+using System;
+using System.IO;
+using System.Text;
+
+namespace MeiamSubtitles.Shared
+{
+    internal static class SubtitleContentValidator
+    {
+        public static void Validate(byte[] data, string format, string mediaType)
+        {
+            if (data == null || data.Length < 8)
+            {
+                throw new InvalidDataException("Subtitle response is empty.");
+            }
+
+            if ((data[0] == (byte)'P' && data[1] == (byte)'K') ||
+                (data.Length >= 7 && Encoding.ASCII.GetString(data, 0, 7) == "Rar!\u001a\u0007"))
+            {
+                throw new InvalidDataException("Compressed subtitle responses are not supported by Thunder.");
+            }
+
+            var sample = DecodeSample(data);
+            if (mediaType?.IndexOf("html", StringComparison.OrdinalIgnoreCase) >= 0 ||
+                mediaType?.IndexOf("json", StringComparison.OrdinalIgnoreCase) >= 0)
+            {
+                throw new InvalidDataException("Thunder returned an error document instead of a subtitle.");
+            }
+
+            var valid = format == "srt"
+                ? sample.IndexOf("-->", StringComparison.Ordinal) >= 0
+                : sample.IndexOf("[Script Info]", StringComparison.OrdinalIgnoreCase) >= 0 ||
+                  sample.IndexOf("Dialogue:", StringComparison.OrdinalIgnoreCase) >= 0;
+
+            if (valid)
+            {
+                return;
+            }
+
+            if (sample.StartsWith("<", StringComparison.Ordinal) ||
+                sample.StartsWith("{", StringComparison.Ordinal) ||
+                sample.StartsWith("[", StringComparison.Ordinal))
+            {
+                throw new InvalidDataException("Thunder returned an error document instead of a subtitle.");
+            }
+
+            throw new InvalidDataException($"Downloaded content is not a valid {format} subtitle.");
+        }
+
+        private static string DecodeSample(byte[] data)
+        {
+            var length = Math.Min(data.Length, 4096);
+            var encoding = DetectEncoding(data);
+            return encoding.GetString(data, 0, length)
+                .TrimStart('\uFEFF', ' ', '\r', '\n', '\t');
+        }
+
+        private static Encoding DetectEncoding(byte[] data)
+        {
+            if (data.Length >= 2)
+            {
+                if (data[0] == 0xFF && data[1] == 0xFE)
+                {
+                    return Encoding.Unicode;
+                }
+
+                if (data[0] == 0xFE && data[1] == 0xFF)
+                {
+                    return Encoding.BigEndianUnicode;
+                }
+            }
+
+            var length = Math.Min(data.Length, 512);
+            var pairs = length / 2;
+            if (pairs >= 4)
+            {
+                // Some Thunder SRT responses are UTF-16 without a BOM. Detect the
+                // characteristic null-byte pattern before falling back to UTF-8.
+                var littleEndianNulls = 0;
+                var bigEndianNulls = 0;
+                for (var i = 0; i < pairs * 2; i += 2)
+                {
+                    if (data[i + 1] == 0)
+                    {
+                        littleEndianNulls++;
+                    }
+
+                    if (data[i] == 0)
+                    {
+                        bigEndianNulls++;
+                    }
+                }
+
+                var threshold = Math.Max(2, pairs / 4);
+                if (littleEndianNulls >= threshold && littleEndianNulls > bigEndianNulls * 2)
+                {
+                    return Encoding.Unicode;
+                }
+
+                if (bigEndianNulls >= threshold && bigEndianNulls > littleEndianNulls * 2)
+                {
+                    return Encoding.BigEndianUnicode;
+                }
+            }
+
+            return new UTF8Encoding(false, false);
+        }
+    }
+}

tests/MeiamSubtitles.Tests/MeiamSubtitles.Tests.csproj

@@ -0,0 +1,21 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <TargetFramework>net8.0</TargetFramework>
+    <IsPackable>false</IsPackable>
+    <IsTestProject>true</IsTestProject>
+    <ImplicitUsings>enable</ImplicitUsings>
+    <Nullable>enable</Nullable>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="17.14.1" />
+    <PackageReference Include="xunit" Version="2.9.3" />
+    <PackageReference Include="xunit.runner.visualstudio" Version="3.0.2" />
+  </ItemGroup>
+
+  <ItemGroup>
+    <Compile Include="..\..\Shared\SubtitleContentValidator.cs" Link="Shared\SubtitleContentValidator.cs" />
+  </ItemGroup>
+
+</Project>

tests/MeiamSubtitles.Tests/SubtitleContentValidatorTests.cs

@@ -0,0 +1,48 @@
+using System.IO;
+using System.Text;
+using MeiamSubtitles.Shared;
+using Xunit;
+
+namespace MeiamSubtitles.Tests;
+
+public class SubtitleContentValidatorTests
+{
+    [Fact]
+    public void AcceptsAssWithScriptInfoHeader()
+    {
+        var data = Encoding.UTF8.GetBytes("[Script Info]\r\nScriptType: v4.00+\r\n[Events]\r\nDialogue: 0,0:00:01.00,0:00:02.00,Default,,0,0,0,,Hello");
+
+        SubtitleContentValidator.Validate(data, "ass", "application/octet-stream");
+    }
+
+    [Fact]
+    public void AcceptsUtf16LittleEndianSrtWithoutBom()
+    {
+        var data = Encoding.Unicode.GetBytes("1\r\n00:00:01,000 --> 00:00:02,000\r\nHello\r\n");
+
+        SubtitleContentValidator.Validate(data, "srt", "application/octet-stream");
+    }
+
+    [Theory]
+    [InlineData("<html><body>temporary error</body></html>", "application/octet-stream")]
+    [InlineData("{\"error\":\"temporary error\"}", "application/json")]
+    [InlineData("[\"temporary error\"]", "application/octet-stream")]
+    public void RejectsErrorDocuments(string content, string mediaType)
+    {
+        var exception = Assert.Throws<InvalidDataException>(() =>
+            SubtitleContentValidator.Validate(Encoding.UTF8.GetBytes(content), "srt", mediaType));
+
+        Assert.Contains("error document", exception.Message, StringComparison.OrdinalIgnoreCase);
+    }
+
+    [Fact]
+    public void RejectsCompressedSubtitleResponse()
+    {
+        var data = new byte[] { (byte)'P', (byte)'K', 3, 4, 5, 6, 7, 8 };
+
+        var exception = Assert.Throws<InvalidDataException>(() =>
+            SubtitleContentValidator.Validate(data, "srt", "application/octet-stream"));
+
+        Assert.Contains("Compressed", exception.Message, StringComparison.Ordinal);
+    }
+}