Block invalid language tags

Author: pmachapman

src/Machine/src/Serval.Machine.Shared/Services/LanguageTagService.cs

@@ -1,7 +1,10 @@
 namespace Serval.Machine.Shared.Services;
 
-public class LanguageTagService : ILanguageTagService
+public partial class LanguageTagService : ILanguageTagService
 {
+    [GeneratedRegex("^[A-Za-z0-9_-]+$")]
+    private static partial Regex BroadlyPermissibleLanguageTagPattern();
+
     private readonly HashSet<string> _flores200Languages = [];
     private readonly HashSet<string> _flores200Scripts = [];
     private readonly LanguageTagParser _parser = new();
@@ -40,9 +43,16 @@ private void InitializeFlores200Languages()
     public Flores200Support ConvertToFlores200Code(string languageTag, out string flores200Code)
     {
         if (_parser.TryParse(languageTag, out string? languageCode, out string? scriptCode))
+        {
             flores200Code = $"{languageCode}_{scriptCode}";
+        }
         else
+        {
+            if (!BroadlyPermissibleLanguageTagPattern().IsMatch(languageTag))
+                throw new InvalidOperationException($"Invalid language tag: {languageTag}");
             flores200Code = languageTag;
+        }
+
         if (_flores200Scripts.Contains(scriptCode ?? ""))
         {
             if (_flores200Languages.Contains(flores200Code))

src/Machine/test/Serval.Machine.Shared.Tests/Services/LanguageTagServiceTests.cs

@@ -10,7 +10,6 @@ public void OneTimeSetUp()
             Sldr.Initialize();
     }
 
-    [Test]
     [TestCase("es", "spa_Latn", Description = "Iso639_1Code")]
     [TestCase("hne", "hne_Deva", Description = "Iso639_3Code")]
     [TestCase("ks-Arab", "kas_Arab", Description = "ScriptCode")]
@@ -27,13 +26,19 @@ public void OneTimeSetUp()
     [TestCase("eng-Latn", "eng_Latn", Description = "DashToUnderscore")]
     [TestCase("kor", "kor_Hang", Description = "KoreanScript")]
     [TestCase("kor_Kore", "kor_Hang", Description = "KoreanScriptCorrection")]
+    [TestCase("unknown", "unknown", Description = "UnknownLanguage")]
     public void ConvertToFlores200CodeTest(string language, string internalCodeTruth)
     {
         new LanguageTagService().ConvertToFlores200Code(language, out string internalCode);
         Assert.That(internalCode, Is.EqualTo(internalCodeTruth));
     }
 
     [Test]
+    public void ConvertToFlores200CodeTest_InvalidLanguage() =>
+        Assert.Throws<InvalidOperationException>(() =>
+            new LanguageTagService().ConvertToFlores200Code("'Invalid Language'", out string _)
+        );
+
     [TestCase("en", "eng_Latn", Flores200Support.LanguageAndScript)]
     [TestCase("ms", "zsm_Latn", Flores200Support.LanguageAndScript)]
     [TestCase("cmn", "zho_Hans", Flores200Support.LanguageAndScript)]

src/Serval/src/Serval.Client/Client.g.cs

@@ -2186,7 +2186,7 @@ public partial interface ITranslationEngineTypesClient
 
         /// <param name="cancellationToken">A cancellation token that can be used by other objects or threads to receive notice of cancellation.</param>
         /// <summary>
-        /// Get information regarding a language for a given engine type
+        /// Get information regarding a language for a given engine type.
         /// </summary>
         /// <remarks>
         /// This endpoint exists primarily to support `nmt` model-training since `echo` and `smt-transfer` engines support all languages equally. Given a language tag, it provides the ISO 639-3 code that the tag maps to internally
@@ -2199,9 +2199,9 @@ public partial interface ITranslationEngineTypesClient
         /// <br/>* **`isNative`**: Whether the base translation model supports this language without fine-tuning.
         /// <br/>* **`internalCode`**: The translation model's internal language code. See more details about how the language tag is mapped to an internal code [here](https://github.com/sillsdev/serval/wiki/FLORES%E2%80%90200-Language-Code-Resolution-for-NMT-Engine).
         /// </remarks>
-        /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer</param>
+        /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer.</param>
         /// <param name="language">The language to retrieve information on.</param>
-        /// <returns>Language information for the specified engine type</returns>
+        /// <returns>Language information for the specified engine type.</returns>
         /// <exception cref="ServalApiException">A server side error occurred.</exception>
         System.Threading.Tasks.Task<LanguageInfo> GetLanguageInfoAsync(string engineType, string language, System.Threading.CancellationToken cancellationToken = default(System.Threading.CancellationToken));
 
@@ -2266,7 +2266,7 @@ public string BaseUrl
 
         /// <param name="cancellationToken">A cancellation token that can be used by other objects or threads to receive notice of cancellation.</param>
         /// <summary>
-        /// Get information regarding a language for a given engine type
+        /// Get information regarding a language for a given engine type.
         /// </summary>
         /// <remarks>
         /// This endpoint exists primarily to support `nmt` model-training since `echo` and `smt-transfer` engines support all languages equally. Given a language tag, it provides the ISO 639-3 code that the tag maps to internally
@@ -2279,9 +2279,9 @@ public string BaseUrl
         /// <br/>* **`isNative`**: Whether the base translation model supports this language without fine-tuning.
         /// <br/>* **`internalCode`**: The translation model's internal language code. See more details about how the language tag is mapped to an internal code [here](https://github.com/sillsdev/serval/wiki/FLORES%E2%80%90200-Language-Code-Resolution-for-NMT-Engine).
         /// </remarks>
-        /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer</param>
+        /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer.</param>
         /// <param name="language">The language to retrieve information on.</param>
-        /// <returns>Language information for the specified engine type</returns>
+        /// <returns>Language information for the specified engine type.</returns>
         /// <exception cref="ServalApiException">A server side error occurred.</exception>
         public virtual async System.Threading.Tasks.Task<LanguageInfo> GetLanguageInfoAsync(string engineType, string language, System.Threading.CancellationToken cancellationToken = default(System.Threading.CancellationToken))
         {
@@ -2341,22 +2341,34 @@ public string BaseUrl
                             return objectResponse_.Object;
                         }
                         else
+                        if (status_ == 400)
+                        {
+                            string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);
+                            throw new ServalApiException("The language is not valid.", status_, responseText_, headers_, null);
+                        }
+                        else
                         if (status_ == 401)
                         {
                             string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);
-                            throw new ServalApiException("The client is not authenticated", status_, responseText_, headers_, null);
+                            throw new ServalApiException("The client is not authenticated.", status_, responseText_, headers_, null);
                         }
                         else
                         if (status_ == 403)
                         {
                             string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);
-                            throw new ServalApiException("The authenticated client cannot perform the operation", status_, responseText_, headers_, null);
+                            throw new ServalApiException("The authenticated client cannot perform the operation.", status_, responseText_, headers_, null);
+                        }
+                        else
+                        if (status_ == 404)
+                        {
+                            string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);
+                            throw new ServalApiException("The engine type is not valid.", status_, responseText_, headers_, null);
                         }
                         else
                         if (status_ == 405)
                         {
                             string responseText_ = ( response_.Content == null ) ? string.Empty : await response_.Content.ReadAsStringAsync().ConfigureAwait(false);
-                            throw new ServalApiException("The method is not supported", status_, responseText_, headers_, null);
+                            throw new ServalApiException("The method is not supported.", status_, responseText_, headers_, null);
                         }
                         else
                         {

src/Serval/src/Serval.Translation/Features/EngineTypes/GetLanguageInfo.cs

@@ -38,7 +38,7 @@ public async Task<GetLanguageInfoResponse> HandleAsync(GetLanguageInfo request,
 public partial class TranslationEngineTypesController
 {
     /// <summary>
-    /// Get information regarding a language for a given engine type
+    /// Get information regarding a language for a given engine type.
     /// </summary>
     /// <remarks>
     /// This endpoint exists primarily to support `nmt` model-training since `echo` and `smt-transfer` engines support all languages equally. Given a language tag, it provides the ISO 639-3 code that the tag maps to internally
@@ -51,18 +51,22 @@ public partial class TranslationEngineTypesController
     /// * **`isNative`**: Whether the base translation model supports this language without fine-tuning.
     /// * **`internalCode`**: The translation model's internal language code. See more details about how the language tag is mapped to an internal code [here](https://github.com/sillsdev/serval/wiki/FLORES%E2%80%90200-Language-Code-Resolution-for-NMT-Engine).
     /// </remarks>
-    /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer</param>
+    /// <param name="engineType">A valid engine type: nmt, echo, or smt-transfer.</param>
     /// <param name="language">The language to retrieve information on.</param>
     /// <param name="cancellationToken"></param>
-    /// <response code="200">Language information for the specified engine type</response>
-    /// <response code="401">The client is not authenticated</response>
-    /// <response code="403">The authenticated client cannot perform the operation</response>
-    /// <response code="405">The method is not supported</response>
+    /// <response code="200">Language information for the specified engine type.</response>
+    /// <response code="400">The language is not valid.</response>
+    /// <response code="401">The client is not authenticated.</response>
+    /// <response code="403">The authenticated client cannot perform the operation.</response>
+    /// <response code="404">The engine type is not valid.</response>
+    /// <response code="405">The method is not supported.</response>
     [Authorize(Scopes.ReadTranslationEngines)]
     [HttpGet("{engineType}/languages/{language}")]
     [ProducesResponseType(StatusCodes.Status200OK)]
+    [ProducesResponseType(typeof(void), StatusCodes.Status400BadRequest)]
     [ProducesResponseType(typeof(void), StatusCodes.Status401Unauthorized)]
     [ProducesResponseType(typeof(void), StatusCodes.Status403Forbidden)]
+    [ProducesResponseType(typeof(void), StatusCodes.Status404NotFound)]
     [ProducesResponseType(typeof(void), StatusCodes.Status405MethodNotAllowed)]
     public async Task<ActionResult<LanguageInfoDto>> GetLanguageInfoAsync(
         [NotNull] string engineType,

src/Serval/test/Serval.ApiServer.IntegrationTests/TranslationEngineTests.cs

@@ -2414,24 +2414,31 @@ public void GetQueueAsync_NotAuthorized()
     public async Task GetLanguageInfoAsync(string engineType)
     {
         TranslationEngineTypesClient client = _env.CreateTranslationEngineTypesClient();
-        Client.LanguageInfo languageInfo = await client.GetLanguageInfoAsync(engineType, "Alphabet");
-        Assert.Multiple(() =>
+        LanguageInfo languageInfo = await client.GetLanguageInfoAsync(engineType, "Alphabet");
+        using (Assert.EnterMultipleScope())
         {
             Assert.That(languageInfo.InternalCode, Is.EqualTo("abc_123"));
-            Assert.That(languageInfo.IsNative, Is.EqualTo(true));
-        });
+            Assert.That(languageInfo.IsNative, Is.True);
+        }
     }
 
-    [Test]
-    public void GetLanguageInfo_Error()
+    [TestCase(new[] { Scopes.ReadTranslationEngines }, "Nmt", "invalid_language", StatusCodes.Status400BadRequest)]
+    [TestCase(new[] { Scopes.ReadFiles }, "Nmt", "abc", StatusCodes.Status403Forbidden)]
+    [TestCase(new[] { Scopes.ReadTranslationEngines }, "invalid_engine", "en", StatusCodes.Status404NotFound)]
+    public void GetLanguageInfo_Error(
+        IEnumerable<string> scope,
+        string engineType,
+        string language,
+        int expectedStatusCode
+    )
     {
-        TranslationEngineTypesClient client = _env.CreateTranslationEngineTypesClient([Scopes.ReadFiles]);
+        TranslationEngineTypesClient client = _env.CreateTranslationEngineTypesClient(scope);
         ServalApiException? ex = Assert.ThrowsAsync<ServalApiException>(async () =>
         {
-            Client.LanguageInfo languageInfo = await client.GetLanguageInfoAsync("Nmt", "abc");
+            _ = await client.GetLanguageInfoAsync(engineType, language);
         });
         Assert.That(ex, Is.Not.Null);
-        Assert.That(ex.StatusCode, Is.EqualTo(403));
+        Assert.That(ex?.StatusCode, Is.EqualTo(expectedStatusCode));
     }
 
     [Test]
@@ -2747,6 +2754,9 @@ public TestEnvironment()
             NmtService
                 .GetLanguageInfoAsync(Arg.Any<string>(), Arg.Any<CancellationToken>())
                 .Returns(Task.FromResult(languageInfo));
+            NmtService
+                .GetLanguageInfoAsync(Arg.Is("invalid_language"), Arg.Any<CancellationToken>())
+                .Returns(Task.FromException<LanguageInfoContract>(new InvalidOperationException()));
 
             SmtService = Substitute.For<ITranslationEngineService>();
             SmtService