Add Paratext metadata to project data files

src/Serval/src/Serval.DataFiles/Features/DataFiles/CreateDataFile.cs

@@ -27,8 +27,15 @@ public async Task<CreateDataFileResponse> HandleAsync(CreateDataFile request, Ca
         };
         try
         {
-            await using Stream fileStream = fileSystem.OpenWrite(path);
-            await request.FileStream.CopyToAsync(fileStream, cancellationToken);
+            using (Stream fileStream = fileSystem.OpenWrite(path))
+            {
+                await request.FileStream.CopyToAsync(fileStream, cancellationToken);
+            }
+            if (dataFile.Format == FileFormat.Paratext)
+            {
+                ParatextMetadata metadata = await ParatextProjectDataParser.ParseParatextMetadataAsync(path);
+                dataFile = dataFile with { FileMetadata = metadata };
+            }
             await dataFiles.InsertAsync(dataFile, cancellationToken);
         }
         catch

src/Serval/src/Serval.DataFiles/Features/DataFiles/UpdateDataFile.cs

@@ -36,6 +36,16 @@ await dataAccessContext.WithTransactionAsync(
                     );
                     if (originalDataFile is null)
                         throw new EntityNotFoundException($"Could not find the DataFile '{request.FileId}'.");
+                    if (originalDataFile.Format == FileFormat.Paratext)
+                    {
+                        ParatextMetadata metadata = await ParatextProjectDataParser.ParseParatextMetadataAsync(path);
+                        await dataFiles.UpdateAsync(
+                            request.FileId,
+                            u => u.Set(f => f.FileMetadata, metadata),
+                            cancellationToken: ct
+                        );
+                    }
+
                     await deletedFiles.InsertAsync(
                         new DeletedFile { Filename = originalDataFile.Filename, DeletedAt = DateTime.UtcNow },
                         ct

src/Serval/src/Serval.DataFiles/Models/DataFile.cs

@@ -8,4 +8,5 @@ public record DataFile : IOwnedEntity
     public required string Name { get; init; }
     public string Filename { get; init; } = "";
     public required FileFormat Format { get; init; }
+    public ParatextMetadata? FileMetadata { get; init; }
 }

src/Serval/src/Serval.DataFiles/Models/ParatextMetadata.cs

@@ -0,0 +1,12 @@
+namespace Serval.DataFiles.Models;
+
+public record ParatextMetadata
+{
+    public required string ProjectGuid { get; init; }
+    public required string Name { get; init; }
+    public required string FullName { get; init; }
+    public required string Versification { get; init; }
+    public required string TranslationType { get; init; }
+    public string? LanguageCode { get; init; }
+    public string? Visibility { get; init; }
+}

src/Serval/src/Serval.DataFiles/Services/ParatextProjectDataParser.cs

@@ -0,0 +1,32 @@
+namespace Serval.DataFiles.Services;
+
+public class ParatextProjectDataParser
+{
+    public static async Task<ParatextMetadata> ParseParatextMetadataAsync(string path)
+    {
+        using ZipContainer zipContainer = new(path);
+        try
+        {
+            ParatextProjectSettings projectSettings = new Shared.Services.ZipParatextProjectSettingsParser(
+                zipContainer
+            ).Parse();
+            return new ParatextMetadata
+            {
+                ProjectGuid = projectSettings.Guid,
+                Name = projectSettings.Name,
+                FullName = projectSettings.FullName,
+                Versification = projectSettings.Versification.Name,
+                TranslationType = projectSettings.TranslationType,
+                LanguageCode = projectSettings.LanguageCode,
+                Visibility = projectSettings.Visibility,
+            };
+        }
+        catch (Exception e) when (e is not OperationCanceledException)
+        {
+            throw new InvalidOperationException(
+                "Unable to parse the Paratext project settings for the uploaded data file.",
+                e
+            );
+        }
+    }
+}

src/Serval/src/Serval.DataFiles/Usings.cs

@@ -26,3 +26,4 @@
 global using Serval.Shared.Services;
 global using Serval.Shared.Utils;
 global using SIL.DataAccess;
+global using SIL.Machine.Corpora;

src/Serval/test/Serval.DataFiles.Tests/Features/DataFiles/DataFilesHandlersTests.cs

@@ -90,6 +90,35 @@ public async Task CreateDataFile_Error()
         env.FileSystem.Received().DeleteFile(Arg.Any<string>());
     }
 
+    [Test]
+    public async Task CreateDataFile_Paratext()
+    {
+        var env = new TestEnvironment();
+        env.FileSystem.OpenWrite(Arg.Any<string>())
+            .Returns(callInfo => new FileStream(callInfo.Arg<string>(), FileMode.Create, FileAccess.Write));
+        string paratextZipPath = ZipParatextProject();
+        CreateDataFileHandler handler = new(env.DataFiles, env.IdGenerator, env.Options, env.FileSystem, env.Mapper);
+        using FileStream stream = File.OpenRead(paratextZipPath);
+        CreateDataFileResponse response = await handler.HandleAsync(
+            new(Owner, "file1", "file1.txt", FileFormat.Paratext, stream),
+            CancellationToken.None
+        );
+        DataFile? dataFile = await env.DataFiles.GetAsync(response.DataFile.Id, CancellationToken.None);
+        Assert.That(dataFile, Is.Not.Null);
+        Assert.That(dataFile.FileMetadata, Is.Not.Null);
+        ParatextMetadata metadata = dataFile.FileMetadata;
+        using (Assert.EnterMultipleScope())
+        {
+            Assert.That(metadata.ProjectGuid, Is.EqualTo("a7e0b3ce0200736062f9f810a444dbfbe64aca35"));
+            Assert.That(metadata.Name, Is.EqualTo("Te1"));
+            Assert.That(metadata.FullName, Is.EqualTo("Test1"));
+            Assert.That(metadata.TranslationType, Is.EqualTo("Standard"));
+            Assert.That(metadata.Versification, Does.StartWith("English"));
+            Assert.That(metadata.LanguageCode, Is.EqualTo("en"));
+            Assert.That(metadata.Visibility, Is.EqualTo("Public"));
+        }
+    }
+
     [Test]
     public async Task DownloadDataFile_FileExists()
     {
@@ -273,4 +302,13 @@ public async Task<DataFile> CreateDataFileAsync(string id = "df00000000000000000
             return file;
         }
     }
+
+    private static string ZipParatextProject()
+    {
+        string path = Path.Combine(Path.GetTempPath(), "pt-project.zip");
+        if (File.Exists(path))
+            File.Delete(path);
+        ZipFile.CreateFromDirectory(Path.Combine("..", "..", "..", "data", "pt-project"), path);
+        return path;
+    }
 }

src/Serval/test/Serval.DataFiles.Tests/Usings.cs

@@ -1,3 +1,4 @@
+global using System.IO.Compression;
 global using System.Text;
 global using Microsoft.Extensions.DependencyInjection;
 global using Microsoft.Extensions.Logging;

src/Serval/test/Serval.DataFiles.Tests/data/pt-project/41MATTe1.SFM

@@ -0,0 +1,6 @@
+\id MAT - SRC
+\c 1
+\v 1 SRC - Chapter one, verse one.
+\p new paragraph
+\v 2
+\v 3 SRC - Chapter one, verse three.

src/Serval/test/Serval.DataFiles.Tests/data/pt-project/Settings.xml

@@ -0,0 +1,34 @@
+<ScriptureText>
+  <StyleSheet>usfm.sty</StyleSheet>
+  <Versification>4</Versification>
+  <LanguageIsoCode>en:::</LanguageIsoCode>
+  <Language>English</Language>
+  <MinParatextVersion>8.0.100.76</MinParatextVersion>
+  <FullName>Test1</FullName>
+  <Encoding>65001</Encoding>
+  <Editable>T</Editable>
+  <Copyright />
+  <NormalizationForm>NFC</NormalizationForm>
+  <Name>Te1</Name>
+  <Guid>a7e0b3ce0200736062f9f810a444dbfbe64aca35</Guid>
+  <DefaultFont>Charis SIL</DefaultFont>
+  <DefaultFontSize>12</DefaultFontSize>
+  <FontFeatures />
+  <HtmlLanguage />
+  <AssociatedLexicalProject />
+  <FileNameBookNameForm>41MAT</FileNameBookNameForm>
+  <FileNamePrePart />
+  <FileNamePostPart>Tes.SFM</FileNamePostPart>
+  <BiblicalTermsListSetting>Major::BiblicalTerms.xml</BiblicalTermsListSetting>
+  <MatchBasedOnStems>F</MatchBasedOnStems>
+  <AllowReadAccess>F</AllowReadAccess>
+  <AllowSharingWithSLDR>F</AllowSharingWithSLDR>
+  <Visibility>Public</Visibility>
+  <TranslationInfo>Standard::</TranslationInfo>
+  <EncodingConverter />
+  <UsfmVersion>3</UsfmVersion>
+  <ParallelPassagesBooks>000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000</ParallelPassagesBooks>
+  <BooksPresent>000000000000000000000000000000000000001000000000000000000000000000000000000000000000000000000000000000000000000000000000000</BooksPresent>
+  <BibleModuleAssociations />
+  <Naming PrePart="" PostPart="Te1.SFM" BookNameForm="41MAT" />
+</ScriptureText>

src/Serval/test/Serval.DataFiles.Tests/data/pt-project/custom.vrs

@@ -0,0 +1,31 @@
+# custom.vrs 
+
+LEV 14:56
+ROM 14:26
+REV 12:17
+TOB 5:22
+TOB 10:12
+SIR 23:28
+ESG 1:22
+ESG 3:15
+ESG 5:14
+ESG 8:17
+ESG 10:14
+SIR 33:33
+SIR 41:24
+BAR 1:22
+4MA 7:25
+4MA 12:20
+
+# deliberately missing verses
+-ROM 16:26
+-ROM 16:27
+-3JN 1:15
+-S3Y 1:49
+-ESG 4:6
+-ESG 9:5
+-ESG 9:30
+
+LEV 14:55 = LEV 14:55
+LEV 14:55 = LEV 14:56
+LEV 14:56 = LEV 14:57