nas/MeiamSubtitles
1
0
Fork 0
mirror of https://github.com/91270/MeiamSubtitles.git synced 2026-10-29 03:25:15 +08:00
Code Issues Packages Projects Releases Wiki Activity

修复: 统一迅雷字幕编码为 UTF-8

Browse Source
This commit is contained in:
Meiam
2026-09-17 10:41:37 +08:00
parent f811632e8a
commit c40d83eb8d
7 changed files with 119 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
Emby.MeiamSub.Thunder/Emby.MeiamSub.Thunder.csproj
View File

@@ -27,6 +27,7 @@
<!-- 使用旧版宿主可解析的 SDK，避免插件要求 4.9.1.90 程序集。 -->
<PackageReference Include="MediaBrowser.Common" Version="4.8.10" />
<PackageReference Include="MediaBrowser.Server.Core" Version="4.8.10" />
<PackageReference Include="System.Text.Encoding.CodePages" Version="8.0.0" />
</ItemGroup>

4
Emby.MeiamSub.Thunder/ThunderProvider.cs
View File

@@ -286,14 +286,14 @@ namespace Emby.MeiamSub.Thunder
await response.Content.CopyToAsync(stream, 81920, cancellationToken);
var data = stream.ToArray();
stream.Dispose();
SubtitleContentValidator.Validate(data, format, response.ContentType);
var utf8Data = SubtitleContentValidator.ValidateAndConvertToUtf8(data, format, response.ContentType);
return new SubtitleResponse()
{
Language = downloadSub.Language,
IsForced = downloadSub.IsForced ?? false,
Format = format,
Stream = new MemoryStream(data, writable: false),
Stream = new MemoryStream(utf8Data, writable: false),
};
}
}

1
Jellyfin.MeiamSub.Thunder/Jellyfin.MeiamSub.Thunder.csproj
View File

@@ -14,6 +14,7 @@
<ItemGroup>
<!-- 使用已验证的最低兼容基线，保持同一 DLL 可运行于 Jellyfin 10.10.7、10.11.11 和 12.1。 -->
<PackageReference Include="Jellyfin.Controller" Version="10.10.7" />
<PackageReference Include="System.Text.Encoding.CodePages" Version="8.0.0" />
</ItemGroup>
<ItemGroup>

7
Jellyfin.MeiamSub.Thunder/ThunderProvider.cs
View File

@@ -283,8 +283,11 @@ namespace Jellyfin.MeiamSub.Thunder
var format = NormalizeFormat(downloadSub.Format)
?? throw new InvalidDataException($"Unsupported subtitle format: {downloadSub.Format}");
var data = await response.Content.ReadAsByteArrayAsync(cancellationToken);
SubtitleContentValidator.Validate(data, format, response.Content.Headers.ContentType?.MediaType);
var stream = new MemoryStream(data, writable: false);
var utf8Data = SubtitleContentValidator.ValidateAndConvertToUtf8(
data,
format,
response.Content.Headers.ContentType?.MediaType);
var stream = new MemoryStream(utf8Data, writable: false);
return new SubtitleResponse()
{

59
Shared/SubtitleContentValidator.cs
View File

@@ -7,7 +7,27 @@ namespace MeiamSubtitles.Shared
{
internal static class SubtitleContentValidator
{
private static readonly Encoding StrictUtf8 = new UTF8Encoding(false, true);
private static readonly Encoding StrictUtf16Le = new UnicodeEncoding(false, true, true);
private static readonly Encoding StrictUtf16Be = new UnicodeEncoding(true, true, true);
static SubtitleContentValidator()
{
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
}
public static void Validate(byte[] data, string format, string mediaType)
{
DecodeAndValidate(data, format, mediaType);
}
public static byte[] ValidateAndConvertToUtf8(byte[] data, string format, string mediaType)
{
var text = DecodeAndValidate(data, format, mediaType);
return Encoding.UTF8.GetBytes(text);
}
private static string DecodeAndValidate(byte[] data, string format, string mediaType)
{
if (data == null || data.Length < 8)
{
@@ -38,7 +58,7 @@ namespace MeiamSubtitles.Shared
if (IsValidSubtitle(sample, format))
{
return;
return sample;
}
if (sample.StartsWith("[", StringComparison.Ordinal))
@@ -65,35 +85,52 @@ namespace MeiamSubtitles.Shared
private static IEnumerable<string> DecodeSamples(byte[] data)
{
var length = Math.Min(data.Length, 4096);
// Decode the complete subtitle because the selected text is returned to
// the host after validation and must not be truncated to a probe window.
var length = data.Length;
if (data.Length >= 2)
{
if (data[0] == 0xFF && data[1] == 0xFE)
{
yield return DecodeSample(data, length, Encoding.Unicode);
yield return DecodeSample(data, length, StrictUtf16Le);
yield break;
}
if (data[0] == 0xFE && data[1] == 0xFF)
{
yield return DecodeSample(data, length, Encoding.BigEndianUnicode);
yield return DecodeSample(data, length, StrictUtf16Be);
yield break;
}
}
if (data.Length >= 3 && data[0] == 0xEF && data[1] == 0xBB && data[2] == 0xBF)
{
yield return DecodeSample(data, length, Encoding.UTF8);
yield return DecodeSample(data, length, StrictUtf8);
yield break;
}
// Thunder may return UTF-16 subtitles without a BOM. Try the common
// single-byte/UTF-8 representation and both UTF-16 byte orders, then
// accept only a candidate that contains the expected subtitle syntax.
yield return DecodeSample(data, length, new UTF8Encoding(false, false));
yield return DecodeSample(data, length, Encoding.Unicode);
yield return DecodeSample(data, length, Encoding.BigEndianUnicode);
// Thunder may return UTF-16 or GB18030 subtitles without a BOM. Decode
// strictly so invalid UTF-8 cannot hide the actual legacy encoding.
foreach (var encoding in new[] { StrictUtf8, StrictUtf16Le, StrictUtf16Be, GetGb18030Encoding() })
{
string sample;
try
{
sample = DecodeSample(data, length, encoding);
}
catch (DecoderFallbackException)
{
continue;
}
yield return sample;
}
}
private static Encoding GetGb18030Encoding()
{
return Encoding.GetEncoding(54936, EncoderFallback.ExceptionFallback, DecoderFallback.ExceptionFallback);
}
private static string DecodeSample(byte[] data, int length, Encoding encoding)

1
tests/MeiamSubtitles.Tests/MeiamSubtitles.Tests.csproj
View File

@@ -10,6 +10,7 @@
<ItemGroup>
<PackageReference Include="Microsoft.NET.Test.Sdk" Version="17.14.1" />
<PackageReference Include="System.Text.Encoding.CodePages" Version="8.0.0" />
<PackageReference Include="xunit" Version="2.9.3" />
<PackageReference Include="xunit.runner.visualstudio" Version="3.0.2" />
</ItemGroup>

61
tests/MeiamSubtitles.Tests/SubtitleContentValidatorTests.cs
View File

@@ -32,6 +32,67 @@ public class SubtitleContentValidatorTests
SubtitleContentValidator.Validate(data, "srt", "application/octet-stream");
}
[Fact]
public void ConvertsGb18030SrtToUtf8ForPreview()
{
Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
var content = "1\r\n00:00:01,000 --> 00:00:02,000\r\n中文字幕\r\n";
var data = Encoding.GetEncoding(54936).GetBytes(content);
var normalized = SubtitleContentValidator.ValidateAndConvertToUtf8(
data,
"srt",
"application/octet-stream");
Assert.Equal(content, Encoding.UTF8.GetString(normalized));
}
[Fact]
public void PreservesCompleteUtf8SubtitleWhenNormalizing()
{
var content = "1\r\n00:00:01,000 --> 00:00:20,000\r\n" + new string('中', 3000) + "\r\n";
var normalized = SubtitleContentValidator.ValidateAndConvertToUtf8(
Encoding.UTF8.GetBytes(content),
"srt",
"application/octet-stream");
Assert.Equal(content, Encoding.UTF8.GetString(normalized));
}
[Theory]
[InlineData(false)]
[InlineData(true)]
public void ConvertsUtf16SrtToUtf8ForPreview(bool bigEndian)
{
var encoding = new UnicodeEncoding(bigEndian, true);
var content = "1\r\n00:00:01,000 --> 00:00:02,000\r\n中文字幕\r\n";
var data = encoding.GetPreamble().Concat(encoding.GetBytes(content)).ToArray();
var normalized = SubtitleContentValidator.ValidateAndConvertToUtf8(
data,
"srt",
"application/octet-stream");
Assert.Equal(content, Encoding.UTF8.GetString(normalized));
}
[Theory]
[InlineData(false)]
[InlineData(true)]
public void ConvertsUtf16WithoutBomToUtf8ForPreview(bool bigEndian)
{
var encoding = new UnicodeEncoding(bigEndian, false);
var content = "1\r\n00:00:01,000 --> 00:00:02,000\r\n中文字幕\r\n";
var normalized = SubtitleContentValidator.ValidateAndConvertToUtf8(
encoding.GetBytes(content),
"srt",
"application/octet-stream");
Assert.Equal(content, Encoding.UTF8.GetString(normalized));
}
[Theory]
[InlineData(false, false)]
[InlineData(false, true)]