Add average pretranslation confidence to build execution data

src/Echo/src/EchoEngine/TranslationEngineService.cs

@@ -237,6 +237,7 @@ .. tokens.Select(
 
                     await platform.InsertPretranslationsAsync(
                         engineId,
+                        buildId,
                         pretranslations.ToAsyncEnumerable(),
                         linkedCts.Token
                     );

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

@@ -28,6 +28,7 @@ Task BuildCompletedAsync(
 
     Task InsertInferenceResultsAsync(
         string engineId,
+        string buildId,
         Stream pretranslationsStream,
         CancellationToken cancellationToken = default
     );

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

@@ -70,12 +70,14 @@ public Task UpdateBuildStatusAsync(string buildId, int step, CancellationToken c
 
     public async Task InsertInferenceResultsAsync(
         string engineId,
+        string buildId,
         Stream pretranslationsStream,
         CancellationToken cancellationToken = default
     )
     {
         await _platformService.InsertPretranslationsAsync(
             engineId,
+            buildId,
             ReadPretranslationsAsync(pretranslationsStream, cancellationToken),
             cancellationToken
         );

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

@@ -70,6 +70,7 @@ public Task UpdateBuildStatusAsync(string buildId, int step, CancellationToken c
 
     public async Task InsertInferenceResultsAsync(
         string engineId,
+        string buildId,
         Stream wordAlignmentsStream,
         CancellationToken cancellationToken = default
     )

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

@@ -43,7 +43,12 @@ CancellationToken cancellationToken
             )
         )
         {
-            await PlatformService.InsertInferenceResultsAsync(engineId, wordAlignmentStream, cancellationToken);
+            await PlatformService.InsertInferenceResultsAsync(
+                engineId,
+                buildId,
+                wordAlignmentStream,
+                cancellationToken
+            );
         }
 
         int additionalCorpusSize = await SaveModelAsync(engineId, buildId, cancellationToken);

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

@@ -36,7 +36,12 @@ CancellationToken cancellationToken
             )
         )
         {
-            await PlatformService.InsertInferenceResultsAsync(engineId, pretranslationsStream, cancellationToken);
+            await PlatformService.InsertInferenceResultsAsync(
+                engineId,
+                buildId,
+                pretranslationsStream,
+                cancellationToken
+            );
         }
 
         int additionalCorpusSize = await SaveModelAsync(engineId, buildId, cancellationToken);

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

@@ -24,13 +24,14 @@ await JsonSerializer.SerializeAsync(
                 }
             );
             stream.Seek(0, SeekOrigin.Begin);
-            await env.Service.InsertInferenceResultsAsync("engine1", stream);
+            await env.Service.InsertInferenceResultsAsync("engine1", "build1", stream);
         }
 
         await env
             .PlatformService.Received()
             .InsertPretranslationsAsync(
                 "engine1",
+                "build1",
                 Arg.Any<IAsyncEnumerable<PretranslationContract>>(),
                 Arg.Any<CancellationToken>()
             );
@@ -79,13 +80,14 @@ await JsonSerializer.SerializeAsync(
                 }
             );
             stream.Seek(0, SeekOrigin.Begin);
-            await env.Service.InsertInferenceResultsAsync("engine1", stream);
+            await env.Service.InsertInferenceResultsAsync("engine1", "build1", stream);
         }
 
         await env
             .PlatformService.Received()
             .InsertPretranslationsAsync(
                 "engine1",
+                "build1",
                 Arg.Any<IAsyncEnumerable<PretranslationContract>>(),
                 Arg.Any<CancellationToken>()
             );
@@ -125,6 +127,7 @@ public TestEnvironment()
             PlatformService = Substitute.For<ITranslationPlatformService>();
             PlatformService
                 .InsertPretranslationsAsync(
+                    Arg.Any<string>(),
                     Arg.Any<string>(),
                     Arg.Any<IAsyncEnumerable<PretranslationContract>>(),
                     Arg.Any<CancellationToken>()

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

@@ -25,7 +25,7 @@ await JsonSerializer.SerializeAsync(
                 }
             );
             stream.Seek(0, SeekOrigin.Begin);
-            await env.Service.InsertInferenceResultsAsync("engine1", stream);
+            await env.Service.InsertInferenceResultsAsync("engine1", "build1", stream);
         }
 
         await env
@@ -85,7 +85,7 @@ await JsonSerializer.SerializeAsync(
                 }
             );
             stream.Seek(0, SeekOrigin.Begin);
-            await env.Service.InsertInferenceResultsAsync("engine1", stream);
+            await env.Service.InsertInferenceResultsAsync("engine1", "build1", stream);
         }
 
         await env

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

@@ -11521,6 +11521,9 @@ public partial class ExecutionData
         [Newtonsoft.Json.JsonProperty("resolvedTargetLanguage", Required = Newtonsoft.Json.Required.Default, NullValueHandling = Newtonsoft.Json.NullValueHandling.Ignore)]
         public string? ResolvedTargetLanguage { get; set; } = default!;
 
+        [Newtonsoft.Json.JsonProperty("averagePretranslationConfidence", Required = Newtonsoft.Json.Required.Default, NullValueHandling = Newtonsoft.Json.NullValueHandling.Ignore)]
+        public double? AveragePretranslationConfidence { get; set; } = default!;
+
     }
 
     [System.CodeDom.Compiler.GeneratedCode("NJsonSchema", "14.6.3.0 (NJsonSchema v11.5.2.0 (Newtonsoft.Json v13.0.0.0))")]

src/Serval/src/Serval.Translation.Contracts/ITranslationPlatformService.cs

@@ -26,6 +26,7 @@ Task UpdateBuildStatusAsync(
     Task IncrementEngineCorpusSizeAsync(string engineId, int count = 1, CancellationToken cancellationToken = default);
     Task InsertPretranslationsAsync(
         string engineId,
+        string buildId,
         IAsyncEnumerable<PretranslationContract> pretranslations,
         CancellationToken cancellationToken = default
     );

src/Serval/src/Serval.Translation/Dtos/ExecutionDataDto.cs

@@ -9,4 +9,5 @@ public record ExecutionDataDto
     public string? EngineTargetLanguageTag { get; init; }
     public string? ResolvedSourceLanguage { get; init; }
     public string? ResolvedTargetLanguage { get; init; }
+    public double? AveragePretranslationConfidence { get; init; }
 }

src/Serval/src/Serval.Translation/Models/ExecutionData.cs

@@ -9,4 +9,5 @@ public record ExecutionData
     public string? EngineTargetLanguageTag { get; init; }
     public string? ResolvedSourceLanguage { get; init; }
     public string? ResolvedTargetLanguage { get; init; }
+    public double? AveragePretranslationConfidence { get; init; }
 }

src/Serval/src/Serval.Translation/Services/DtoMapper.cs

@@ -236,6 +236,7 @@ private static ExecutionDataDto Map(ExecutionData source) =>
             EngineTargetLanguageTag = source.EngineTargetLanguageTag,
             ResolvedSourceLanguage = source.ResolvedSourceLanguage,
             ResolvedTargetLanguage = source.ResolvedTargetLanguage,
+            AveragePretranslationConfidence = source.AveragePretranslationConfidence,
         };
 }
 

src/Serval/src/Serval.Translation/Services/PlatformService.cs

@@ -365,6 +365,7 @@ await _engines.UpdateAsync(
 
     public async Task InsertPretranslationsAsync(
         string engineId,
+        string buildId,
         IAsyncEnumerable<PretranslationContract> pretranslations,
         CancellationToken cancellationToken = default
     )
@@ -375,6 +376,8 @@ public async Task InsertPretranslationsAsync(
         int nextModelRevision = engine.ModelRevision + 1;
 
         var batch = new List<Pretranslation>();
+        double confidenceTotal = 0.0;
+        int numPretranslations = 0;
         await foreach (PretranslationContract item in pretranslations.WithCancellation(cancellationToken))
         {
             batch.Add(
@@ -401,6 +404,8 @@ public async Task InsertPretranslationsAsync(
                     Confidence = item.Confidence,
                 }
             );
+            confidenceTotal += item.Confidence ?? 0.0;
+            numPretranslations += 1;
             if (batch.Count == PretranslationInsertBatchSize)
             {
                 await _pretranslations.InsertAllAsync(batch, cancellationToken);
@@ -409,5 +414,11 @@ public async Task InsertPretranslationsAsync(
         }
         if (batch.Count > 0)
             await _pretranslations.InsertAllAsync(batch, CancellationToken.None);
+
+        await _builds.UpdateAsync(
+            b => b.Id == buildId,
+            u => u.Set(b => b.ExecutionData.AveragePretranslationConfidence, confidenceTotal / numPretranslations),
+            cancellationToken: cancellationToken
+        );
     }
 }

src/Serval/test/Serval.E2ETests/ServalApiTests.cs

@@ -292,6 +292,7 @@ public async Task Nmt_Paratext(bool withAdditionalFiles)
         string buildId = await _helperClient.BuildEngineAsync(engineId);
         TranslationBuild build = await _helperClient.TranslationEnginesClient.GetBuildAsync(engineId, buildId);
         Assert.That(build.State, Is.EqualTo(JobState.Completed), JsonSerializer.Serialize(build));
+        Assert.That(build.ExecutionData.AveragePretranslationConfidence, Is.GreaterThan(0.3));
 
         IList<Pretranslation> translations = await _helperClient.TranslationEnginesClient.GetAllPretranslationsAsync(
             engineId,

src/Serval/test/Serval.Translation.Tests/Services/PlatformServiceTests.cs

@@ -39,7 +39,7 @@ await env.Builds.InsertAsync(
         Assert.That(env.Engines.Get("e0").IsBuilding, Is.False);
 
         Assert.That(env.Pretranslations.Count, Is.EqualTo(0));
-        await env.PlatformService.InsertPretranslationsAsync("e0", GetTestPretranslations());
+        await env.PlatformService.InsertPretranslationsAsync("e0", "b0", GetTestPretranslations());
         Assert.That(env.Pretranslations.Count, Is.EqualTo(1));
 
         await env.PlatformService.BuildFaultedAsync("b0", "");
@@ -48,14 +48,18 @@ await env.Builds.InsertAsync(
         Assert.That(env.Engines.Get("e0").IsBuilding, Is.False);
 
         await env.PlatformService.BuildRestartingAsync("b0");
-        await env.PlatformService.InsertPretranslationsAsync("e0", GetTestPretranslations());
+        await env.PlatformService.InsertPretranslationsAsync("e0", "b0", GetTestPretranslations());
         Assert.That(env.Pretranslations.Count, Is.EqualTo(1));
         await env.PlatformService.BuildCompletedAsync("b0", 0, 0.0);
         Assert.That(env.Pretranslations.Count, Is.EqualTo(1));
         await env.PlatformService.BuildStartedAsync("b0");
-        await env.PlatformService.InsertPretranslationsAsync("e0", GetTestPretranslations());
+        await env.PlatformService.InsertPretranslationsAsync("e0", "b0", GetTestPretranslations());
         await env.PlatformService.BuildCompletedAsync("b0", 0, 0.0);
         Assert.That(env.Pretranslations.Count, Is.EqualTo(1));
+        Assert.That(
+            (await env.Builds.GetAsync(b => b.Id == "b0"))?.ExecutionData.AveragePretranslationConfidence,
+            Is.Zero
+        );
     }
 
     [Test]