Add PDF export to stream

cyanfish · Dec 7, 2023 · e431157 · e431157
1 parent 54de82b
commit e431157
Show file tree

Hide file tree

Showing 5 changed files with 83 additions and 6 deletions.
diff --git a/NAPS2.Sdk.Tests/Pdf/PdfExportTests.cs b/NAPS2.Sdk.Tests/Pdf/PdfExportTests.cs
@@ -30,6 +30,23 @@ public async Task ExportJpegImage(StorageConfig storageConfig)
         PdfAsserts.AssertImageFilter(filePath, 0, "DCTDecode");
     }
 
+    [Theory]
+    [ClassData(typeof(StorageAwareTestData))]
+    public async Task ExportJpegImageToStream(StorageConfig storageConfig)
+    {
+        storageConfig.Apply(this);
+
+        var filePath = Path.Combine(FolderPath, "test.pdf");
+        var fileStream = File.OpenWrite(filePath);
+        using var image = ScanningContext.CreateProcessedImage(LoadImage(ImageResources.dog));
+
+        await _exporter.Export(fileStream, new[] { image });
+        fileStream.Close();
+
+        PdfAsserts.AssertImages(filePath, ImageResources.dog);
+        PdfAsserts.AssertImageFilter(filePath, 0, "DCTDecode");
+    }
+
     [Theory]
     [ClassData(typeof(StorageAwareTestData))]
     public async Task ExportGrayJpegImage(StorageConfig storageConfig)

diff --git a/NAPS2.Sdk.Tests/Pdf/PdfImportExportTests.cs b/NAPS2.Sdk.Tests/Pdf/PdfImportExportTests.cs
@@ -36,6 +36,22 @@ public async Task ImportExport(OcrTestConfig config)
         PdfAsserts.AssertImages(_exportPath, PdfResources.word_p1, PdfResources.word_p2);
     }
 
+    [Theory]
+    [ClassData(typeof(OcrTestData))]
+    public async Task ImportExportToStream(OcrTestConfig config)
+    {
+        config.StorageConfig.Apply(this);
+        SetUpFakeOcr();
+
+        var fileStream = File.OpenWrite(_exportPath);
+        var images = await _importer.Import(_importPath).ToListAsync();
+        Assert.Equal(2, images.Count);
+        await _exporter.Export(fileStream, images, ocrParams: config.OcrParams);
+        fileStream.Close();
+
+        PdfAsserts.AssertImages(_exportPath, PdfResources.word_p1, PdfResources.word_p2);
+    }
+
     [Theory]
     [ClassData(typeof(OcrTestData))]
     public async Task ImportInsertExport(OcrTestConfig config)

diff --git a/NAPS2.Sdk/Pdf/IPdfExporter.cs b/NAPS2.Sdk/Pdf/IPdfExporter.cs
@@ -6,4 +6,7 @@ public interface IPdfExporter
 {
     Task<bool> Export(string path, ICollection<ProcessedImage> images, PdfExportParams? exportParams = null,
         OcrParams? ocrParams = null, ProgressHandler progress = default);
+
+    Task<bool> Export(Stream stream, ICollection<ProcessedImage> images, PdfExportParams? exportParams = null,
+        OcrParams? ocrParams = null, ProgressHandler progress = default);
 }
diff --git a/NAPS2.Sdk/Pdf/PdfExporter.cs b/NAPS2.Sdk/Pdf/PdfExporter.cs
@@ -27,7 +27,15 @@ public PdfExporter(ScanningContext scanningContext)
         _logger = scanningContext.Logger;
     }
 
-    public async Task<bool> Export(string path, ICollection<ProcessedImage> images,
+    public Task<bool> Export(string path, ICollection<ProcessedImage> images,
+        PdfExportParams? exportParams = null, OcrParams? ocrParams = null, ProgressHandler progress = default)
+        => Export(new OutputPathOrStream(path, null), images, exportParams, ocrParams, progress);
+
+    public Task<bool> Export(Stream stream, ICollection<ProcessedImage> images,
+        PdfExportParams? exportParams = null, OcrParams? ocrParams = null, ProgressHandler progress = default)
+        => Export(new OutputPathOrStream(null, stream), images, exportParams, ocrParams, progress);
+
+    private async Task<bool> Export(OutputPathOrStream output, ICollection<ProcessedImage> images,
         PdfExportParams? exportParams = null, OcrParams? ocrParams = null, ProgressHandler progress = default)
     {
         return await Task.Run(async () =>
@@ -123,7 +131,7 @@ void IncrementProgress()
                 var stream = FinalizeAndSaveDocument(document, exportParams);
                 if (progress.IsCancellationRequested) return false;
 
-                return MergePassthroughPages(stream, path, pdfPages, exportParams, progress);
+                return MergePassthroughPages(stream, output, pdfPages, exportParams, progress);
             }
             finally
             {
@@ -136,13 +144,12 @@ void IncrementProgress()
         });
     }
 
-    private bool MergePassthroughPages(MemoryStream stream, string path, List<PageExportState> passthroughPages,
+    private bool MergePassthroughPages(MemoryStream stream, OutputPathOrStream output, List<PageExportState> passthroughPages,
         PdfExportParams exportParams, ProgressHandler progress)
     {
         if (!passthroughPages.Any())
         {
-            using var fileStream = new FileStream(path, FileMode.Create);
-            stream.CopyTo(fileStream);
+            output.CopyFromStream(stream);
             return true;
         }
         // TODO: Should we do this (or maybe the whole pdf export/import) in a worker to avoid contention?
@@ -186,7 +193,7 @@ private bool MergePassthroughPages(MemoryStream stream, string path, List<PageEx
                     }
                     if (progress.IsCancellationRequested) return false;
                 }
-                destDoc.Save(path);
+                output.SaveDoc(destDoc);
                 return true;
             }
             finally
@@ -746,4 +753,32 @@ public void Dispose()
         {
         }
     }
+
+    private record OutputPathOrStream(string? Path, Stream? Stream)
+    {
+        public void CopyFromStream(MemoryStream inputStream)
+        {
+            if (Stream != null)
+            {
+                inputStream.CopyTo(Stream);
+            }
+            else
+            {
+                using var fileStream = new FileStream(Path!, FileMode.Create);
+                inputStream.CopyTo(fileStream);
+            }
+        }
+
+        public void SaveDoc(Pdfium.PdfDocument doc)
+        {
+            if (Stream != null)
+            {
+                doc.Save(Stream);
+            }
+            else
+            {
+                doc.Save(Path!);
+            }
+        }
+    }
 }
diff --git a/NAPS2.Sdk/Pdf/PdfiumPdfExporter.cs b/NAPS2.Sdk/Pdf/PdfiumPdfExporter.cs
@@ -17,6 +17,12 @@ public PdfiumPdfExporter(ScanningContext scanningContext)
         _logger = scanningContext.Logger;
     }
 
+    public Task<bool> Export(Stream stream, ICollection<ProcessedImage> images, PdfExportParams? exportParams = null, OcrParams? ocrParams = null,
+        ProgressHandler progress = default)
+    {
+        throw new NotSupportedException();
+    }
+
     public async Task<bool> Export(string path, ICollection<ProcessedImage> images,
         PdfExportParams? exportParams = null, OcrParams? ocrParams = null, ProgressHandler progress = default)
     {