Refactor to reuse more I/O code

Author: manfromarce

README.md

@@ -5,7 +5,7 @@ DocSharp is a pure C# library to convert between document formats without Office
 The following packages are currently available:
 
 - DocSharp.Binary: convert Office 97-2003 binary documents (doc, xls, ppt) to OpenXML documents (docx, xlsx, pptx). This is a fork of the abandoned [b2xtranslator project](https://github.com/EvolutionJobs/b2xtranslator) which provides critical fixes. 
-- DocSharp.Docx: convert DOCX to RTF, HTML, Markdown and plain text (.txt). Possible applications include generating Open XML documents in C# and exporting for other editors, or loading Microsoft Word documents in a RichTextBox / RichEditBox control.
+- DocSharp.Docx: convert DOCX to RTF, HTML, Markdown and plain text (.txt). Possible applications include generating Open XML documents in C# and exporting for other editors/services, or loading Microsoft Word documents in a RichTextBox / RichEditBox control.
 - DocSharp.Markdown: convert Markdown to DOCX or RTF using custom Markdig renderers.
 
 Packages can be installed via NuGet:  

src/DocSharp.Common/DocumentConverterBase.cs

@@ -0,0 +1,82 @@
+using System.IO;
+using System.Text;
+
+namespace DocSharp;
+
+public abstract class DocumentConverterBase<TOutput> where TOutput : class
+{
+    public abstract void Convert(Stream inputStream, Stream outputStream);
+
+    public virtual void Convert(Stream inputStream, string outputFilePath)
+    {
+        using (var outputStream = File.OpenWrite(outputFilePath))
+            Convert(inputStream, outputStream);
+    }
+
+    public virtual void Convert(string inputFilePath, Stream outputStream)
+    {
+        using (var inputStream = File.OpenRead(inputFilePath))
+            Convert(inputStream, outputStream);
+    }
+
+    public virtual void Convert(string inputFilePath, string outputFilePath)
+    {
+        using (var inputStream = File.OpenRead(inputFilePath))
+            using (var outputStream = File.OpenWrite(outputFilePath))
+                Convert(inputStream, outputStream);
+    }
+}
+
+public abstract class BinaryDocumentConverterBase<TOutput> : DocumentConverterBase<TOutput> where TOutput : class
+{
+    public void Convert(byte[] inputBytes, Stream outputStream)
+    {
+        using (var memoryStream = new MemoryStream(inputBytes))
+            Convert(memoryStream, outputStream);
+    }
+
+    public void Convert(byte[] inputBytes, string outputFilePath)
+    {
+        using (var memoryStream = new MemoryStream(inputBytes))
+            Convert(memoryStream, outputFilePath);
+    }    
+}
+
+public abstract class TextDocumentConverterBase<TOutput> : DocumentConverterBase<TOutput> where TOutput : class
+{
+    public abstract void Convert(TextReader reader, Stream outputStream);
+
+    public void Convert(TextReader reader, string outputFilePath)
+    {
+        using (var outputStream = File.OpenWrite(outputFilePath))
+            Convert(reader, outputStream);
+    }
+
+    public override void Convert(Stream inputStream, Stream outputStream)
+    {
+        using (var reader = new StreamReader(inputStream, Encoding.UTF8, detectEncodingFromByteOrderMarks: true, bufferSize: 1024, leaveOpen: true))
+            Convert(reader, outputStream);
+    }
+
+    public void ConvertString(string inputContent, Stream outputStream, Encoding? encoding = null)
+    {
+        encoding ??= Encoding.UTF8;
+        using (var memoryStream = new MemoryStream())
+        {
+            using (var writer = new StreamWriter(memoryStream, encoding, 1024, leaveOpen: true))
+            {
+                writer.Write(inputContent);
+                writer.Flush();
+                memoryStream.Position = 0;
+                Convert(memoryStream, outputStream);
+            }
+        }
+    }
+
+    public void ConvertString(string inputContent, string outputFilePath, Encoding? encoding = null)
+    {
+        using (var outputStream = File.OpenWrite(outputFilePath))
+            ConvertString(inputContent, outputStream, encoding);
+    }
+}
+

src/DocSharp.Docx/DocSharp.Docx.csproj

@@ -17,6 +17,12 @@
     <PackageTags>docx html rtf markdown txt convert converter openxml office word</PackageTags>
   </PropertyGroup>
 
+  <ItemGroup>
+    <AssemblyAttribute Include="System.Runtime.CompilerServices.InternalsVisibleToAttribute">
+    <_Parameter1>DocSharp.Renderer</_Parameter1>
+    </AssemblyAttribute>
+  </ItemGroup>
+
   <ItemGroup>
     <None Remove="Resources\OMML2MML.XSL" />
   </ItemGroup>

src/DocSharp.Docx/DocxConverterBase.cs

@@ -13,7 +13,11 @@
 
 namespace DocSharp.Docx;
 
-public abstract class DocxConverterBase<TOutput>
+/// <summary>
+/// Base class for DOCX converters.
+/// </summary>
+/// <typeparam name="TOutput"></typeparam>
+public abstract class DocxConverterBase<TOutput> : BinaryDocumentConverterBase<TOutput> where TOutput : class
 {
     /// <summary>
     /// Get or set the base file path for processing external sub-documents (if any).
@@ -28,10 +32,44 @@ static DocxConverterBase()
     }
 #endif
 
-    internal List<(List<OpenXmlElement> content, SectionProperties properties)> Sections;
+    internal List<(List<OpenXmlElement> content, SectionProperties properties)> Sections = new();
     internal bool TitlePage = false;
     internal bool FacingPages = false;
 
+    /// <summary>
+    /// Convert a DOCX file to the output format.
+    /// </summary>
+    /// <param name="inputDocument">The WordprocessingDocument to use.</param>
+    /// <param name="outputStream">The output stream.</param>
+    public abstract void Convert(WordprocessingDocument inputDocument, Stream outputStream);
+    // This is the main function that must be implemented by specific converters.
+
+    /// <summary>
+    /// Convert a DOCX file to the output format.
+    /// </summary>
+    /// <param name="inputStream">The input DOCX stream to use.</param>
+    /// <param name="outputStream">The output stream.</param>
+    public override void Convert(Stream inputStream, Stream outputStream)
+    {
+        using (var wordDocument = WordprocessingDocument.Open(inputStream, false))
+        {
+            Convert(wordDocument, outputStream);
+        }
+    }
+
+    /// <summary>
+    /// Convert a DOCX file to the output format.
+    /// </summary>
+    /// <param name="inputDocument">The WordprocessingDocument to use.</param>
+    /// <param name="outputFilePath">The output file path.</param>
+    public virtual void Convert(WordprocessingDocument inputDocument, string outputFilePath)
+    {
+        using (var fs = new FileStream(outputFilePath, FileMode.Create, FileAccess.Write))
+        {
+            Convert(inputDocument, fs);
+        }
+    }
+
     internal virtual void ProcessDocument(Document document, TOutput sb)
     {
         if (document.DocumentBackground is DocumentBackground bg)

src/DocSharp.Docx/DocxExtensions.cs

@@ -9,6 +9,9 @@
 
 namespace DocSharp.Docx;
 
+/// <summary>
+/// Provides public extension methods for WordprocessingDocument.
+/// </summary>
 public static class DocxExtensions
 {
     /// <summary>

src/DocSharp.Docx/DocxToHtml/DocxToHtmlConverter.Borders.cs

@@ -10,7 +10,7 @@
 
 namespace DocSharp.Docx;
 
-public partial class DocxToHtmlConverter : DocxToTextWriterBase<HtmlTextWriter>
+public partial class DocxToHtmlConverter : DocxToXmlWriterBase<HtmlTextWriter>
 {
     internal void ProcessBorder(BorderType? border, string? cssAttribute, ref List<string> styles)
     {

src/DocSharp.Docx/DocxToHtml/DocxToHtmlConverter.DrawingML.cs

@@ -18,7 +18,7 @@
 
 namespace DocSharp.Docx;
 
-public partial class DocxToHtmlConverter : DocxToTextWriterBase<HtmlTextWriter>
+public partial class DocxToHtmlConverter : DocxToXmlWriterBase<HtmlTextWriter>
 {
     internal override void ProcessDrawing(Drawing drawing, HtmlTextWriter sb)
     {

src/DocSharp.Docx/DocxToHtml/DocxToHtmlConverter.FootnoteEndnote.cs

@@ -12,7 +12,7 @@
 
 namespace DocSharp.Docx;
 
-public partial class DocxToHtmlConverter : DocxToTextWriterBase<HtmlTextWriter>
+public partial class DocxToHtmlConverter : DocxToXmlWriterBase<HtmlTextWriter>
 {
     // Note: FootnoteReference and EndnoteReference are found inside runs in the document body,
     // while FootnoteReferenceMark and EndnoteReferenceMark are in runs in the footnote/endnote part itself. 

src/DocSharp.Docx/DocxToHtml/DocxToHtmlConverter.Image.cs

@@ -22,7 +22,7 @@
 
 namespace DocSharp.Docx;
 
-public partial class DocxToHtmlConverter : DocxToTextWriterBase<HtmlTextWriter>
+public partial class DocxToHtmlConverter : DocxToXmlWriterBase<HtmlTextWriter>
 {
     internal void ProcessImagePart(OpenXmlPart? rootPart, string relId, double width, double height, HtmlTextWriter sb)
     {

src/DocSharp.Docx/DocxToHtml/DocxToHtmlConverter.List.cs

@@ -9,7 +9,7 @@
 
 namespace DocSharp.Docx;
 
-public partial class DocxToHtmlConverter : DocxToTextWriterBase<HtmlTextWriter>
+public partial class DocxToHtmlConverter : DocxToXmlWriterBase<HtmlTextWriter>
 {
     private readonly Dictionary<int, (int numId, int abstractNumId, int counter)> _listLevelCounters = new();