feat: add learned document template workflow

DocumentAutomation/tests/DocumentAutomation.IntegrationTests/DocumentAutomation.IntegrationTests.csproj

@@ -9,7 +9,7 @@
     <IsTestProject>true</IsTestProject>
   </PropertyGroup>
 
-  <ItemGroup><ProjectReference Include="..\..\src\DocxWorkbench.WordInterop\DocxWorkbench.WordInterop.csproj" /><ProjectReference Include="..\..\src\StandardsWorkbench.Pdf\StandardsWorkbench.Pdf.csproj" /><PackageReference Include="DocumentFormat.OpenXml" Version="3.3.0" /></ItemGroup>
+  <ItemGroup><ProjectReference Include="..\..\src\DocxWorkbench.WordInterop\DocxWorkbench.WordInterop.csproj" /><ProjectReference Include="..\..\src\DocxWorkbench.OpenXml\DocxWorkbench.OpenXml.csproj" /><ProjectReference Include="..\..\src\StandardsWorkbench.Pdf\StandardsWorkbench.Pdf.csproj" /><PackageReference Include="DocumentFormat.OpenXml" Version="3.3.0" /></ItemGroup>
 
   <ItemGroup>
     <PackageReference Include="coverlet.collector" Version="6.0.0" />

DocumentAutomation/tests/DocumentAutomation.IntegrationTests/WordComIntegrationTests.cs

@@ -4,6 +4,7 @@ using DocumentFormat.OpenXml.Packaging;
 using DocumentFormat.OpenXml.Wordprocessing;
 using DocumentAutomation.Shared;
 using DocxWorkbench.WordInterop;
+using DocxWorkbench.OpenXml;
 using Xunit.Sdk;
 
 namespace DocumentAutomation.IntegrationTests;
@@ -47,6 +48,18 @@ public sealed class WordComIntegrationTests
         await RequireWord(); var template = CreateDocx(true); var source = CreateComplexSource(); var output = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".docx"); try { using var service = new WordDocumentService(); var result = await service.BuildDocumentAsync(new WordDocumentBuildRequest { TemplatePath = template, BodyPath = source, OutputDocxPath = output }, null, CancellationToken.None); using var document = WordprocessingDocument.Open(result.OutputDocxPath, false); Assert.Single(document.MainDocumentPart!.Document.Descendants<SectionProperties>()); Assert.Contains("DOCX complex body", document.MainDocumentPart.Document.InnerText); } finally { File.Delete(template); File.Delete(source); File.Delete(output); File.Delete(Path.ChangeExtension(output, ".generation.json")); }
     }
 
+    [Fact]
+    public async Task TemplateLearning_EndToEnd_RealDocToDtemplateToDocxAndPdf()
+    {
+        await RequireWord(); var source = @"C:\Users\codex\Desktop\НХС-33-УК3-ОЗХ-13011-000-НК-ОД.doc"; var content = @"C:\Users\codex\Desktop\НХС-33-УК3-ОЗХ-13011-000-НК-ОД.docx"; if (!File.Exists(source) || !File.Exists(content)) throw SkipException.ForSkip("The supplied real DOC/DOCX fixtures are not available.");
+        var package = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".dtemplate"); var output = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".docx"); var sourceHash = Convert.ToHexString(System.Security.Cryptography.SHA256.HashData(File.ReadAllBytes(source))); var before = ProcessIds();
+        try
+        {
+            using var normalizer = new WordDocumentNormalizer(); var learning = new TemplateLearningService(normalizer, new TemplatePackageStore(), new OpenXmlDocumentInspector()); var learned = await learning.LearnAsync(new TemplateLearningRequest { SourcePath = source, TemplateName = "Real sample", OutputPackagePath = package }, null, CancellationToken.None); Assert.NotEmpty(learned.Slots); Assert.True(File.Exists(package)); using var generator = new TemplateDocumentGenerator(new TemplatePackageStore(), new WordDocumentNormalizer(), new OpenXmlDocumentInspector()); var result = await generator.GenerateAsync(new TemplateGenerationRequest { TemplatePackagePath = package, ContentSourcePath = content, OutputDocxPath = output }, null, CancellationToken.None); Assert.True(File.Exists(result.OutputDocxPath)); Assert.True(File.Exists(result.OutputPdfPath)); Assert.Equal(sourceHash, Convert.ToHexString(System.Security.Cryptography.SHA256.HashData(File.ReadAllBytes(source)))); Assert.Empty(ProcessIds().Except(before));
+        }
+        finally { File.Delete(package); File.Delete(output); File.Delete(Path.ChangeExtension(output, ".pdf")); File.Delete(Path.ChangeExtension(output, ".generation.json")); File.Delete(Path.ChangeExtension(output, ".validation.json")); }
+    }
+
     private static async Task RequireWord() { if (!OperatingSystem.IsWindows() || Type.GetTypeFromProgID("Word.Application", false) is null) throw SkipException.ForSkip("Microsoft Word COM is not installed or registered on this workstation."); await Task.CompletedTask; }
     private static async Task<WordDocumentResult> BuildTxt() { var template = CreateDocx(true); var body = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".txt"); var output = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".docx"); File.WriteAllText(body, "TXT integration body"); try { using var service = new WordDocumentService(); return await service.BuildDocumentAsync(new WordDocumentBuildRequest { TemplatePath = template, BodyPath = body, OutputDocxPath = output }, null, CancellationToken.None); } finally { File.Delete(template); File.Delete(body); } }
     private static HashSet<int> ProcessIds() => Process.GetProcessesByName("WINWORD").Select(x => x.Id).ToHashSet();

DocumentAutomation/tests/DocxWorkbench.Tests/DocxWorkbench.Tests.csproj

@@ -9,7 +9,7 @@
     <IsTestProject>true</IsTestProject>
   </PropertyGroup>
 
-  <ItemGroup><ProjectReference Include="..\..\src\DocxWorkbench.Core\DocxWorkbench.Core.csproj" /><ProjectReference Include="..\..\src\DocxWorkbench.OpenXml\DocxWorkbench.OpenXml.csproj" /><PackageReference Include="DocumentFormat.OpenXml" Version="3.3.0" /></ItemGroup>
+  <ItemGroup><ProjectReference Include="..\..\src\DocxWorkbench.Core\DocxWorkbench.Core.csproj" /><ProjectReference Include="..\..\src\DocxWorkbench.OpenXml\DocxWorkbench.OpenXml.csproj" /><ProjectReference Include="..\..\src\DocxWorkbench.WordInterop\DocxWorkbench.WordInterop.csproj" /><PackageReference Include="DocumentFormat.OpenXml" Version="3.3.0" /></ItemGroup>
 
   <ItemGroup>
     <PackageReference Include="coverlet.collector" Version="6.0.0" />

DocumentAutomation/tests/DocxWorkbench.Tests/TemplateLearningTests.cs

@@ -0,0 +1,25 @@
+using System.IO.Compression;
+using System.Security.Cryptography;
+using DocumentAutomation.Shared;
+using DocxWorkbench.OpenXml;
+using DocxWorkbench.WordInterop;
+
+namespace DocxWorkbench.Tests;
+
+public sealed class TemplateLearningTests
+{
+    [Fact]
+    public async Task NormalizeDocx_UsesSafeWorkingCopy()
+    {
+        var source = CreateDocx("template"); var hash = Hash(source); var working = Path.Combine(Path.GetTempPath(), "learning-test-" + Guid.NewGuid().ToString("N")); using var normalizer = new WordDocumentNormalizer(); try { var result = await normalizer.NormalizeAsync(new DocumentNormalizationRequest { SourcePath = source, WorkingDirectory = working }, null, CancellationToken.None); Assert.False(result.WasConverted); Assert.NotEqual(Path.GetFullPath(source), Path.GetFullPath(result.NormalizedPath)); Assert.Equal(hash, Hash(source)); Assert.Equal(hash, result.SourceSha256); Assert.True(File.Exists(result.NormalizedPath)); } finally { File.Delete(source); if (Directory.Exists(working)) Directory.Delete(working, true); }
+    }
+
+    [Fact]
+    public async Task TemplatePackage_CanBeCreatedAndReopened()
+    {
+        var source = CreateDocx("{{DOC_BODY}}"); var package = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".dtemplate"); using var normalizer = new WordDocumentNormalizer(); try { var service = new TemplateLearningService(normalizer, new TemplatePackageStore()); var learned = await service.LearnAsync(new TemplateLearningRequest { SourcePath = source, TemplateName = "Unit template", OutputPackagePath = package }, null, CancellationToken.None); Assert.NotEmpty(learned.Slots); Assert.Contains(learned.PageArchetypes, x => x.IsFirstPage); using var archive = ZipFile.OpenRead(package); Assert.Contains(archive.Entries, x => x.FullName == "manifest.json"); Assert.Contains(archive.Entries, x => x.FullName == "template.docx"); var opened = new TemplatePackageStore().Open(package); Assert.Equal(learned.TemplateId, opened.Template.TemplateId); Assert.True(File.Exists(opened.TemplateDocxPath)); Directory.Delete(opened.WorkingDirectory, true); } finally { File.Delete(source); File.Delete(package); }
+    }
+
+    private static string CreateDocx(string text) { var path = Path.Combine(Path.GetTempPath(), Guid.NewGuid() + ".docx"); using var document = DocumentFormat.OpenXml.Packaging.WordprocessingDocument.Create(path, DocumentFormat.OpenXml.WordprocessingDocumentType.Document); var main = document.AddMainDocumentPart(); main.Document = new DocumentFormat.OpenXml.Wordprocessing.Document(new DocumentFormat.OpenXml.Wordprocessing.Body(new DocumentFormat.OpenXml.Wordprocessing.Paragraph(new DocumentFormat.OpenXml.Wordprocessing.Run(new DocumentFormat.OpenXml.Wordprocessing.Text(text))))); main.Document.Save(); return path; }
+    private static string Hash(string path) => Convert.ToHexString(SHA256.HashData(File.ReadAllBytes(path)));
+}