XML Parser: Visitor and PrettyPrinter

git-svn-id: svn://svn.sharpdevelop.net/sharpdevelop/trunk@4605 1ccf3a8d-04fe-1044-b7c0-cef0b8235c61

samples/XmlDOM/Window1.xaml.cs

@@ -28,6 +28,8 @@ namespace XmlDOM
 	{
 		XmlParser parser;
 		
+		bool textDirty = true;
+		
 		public Window1()
 		{
 			InitializeComponent();
@@ -35,10 +37,11 @@ namespace XmlDOM
 		
 		protected override void OnInitialized(EventArgs e)
 		{
+			editor.Document.Changed += delegate { textDirty = true; };
 			parser = new XmlParser(editor.Document);
 
 			DispatcherTimer timer = new DispatcherTimer();
-			timer.Interval = TimeSpan.FromSeconds(0.25);
+			timer.Interval = TimeSpan.FromSeconds(0.5);
 			timer.Tick += delegate { Button_Click(null, null); };
 			timer.Start();
 			
@@ -47,10 +50,18 @@ namespace XmlDOM
 		
 		void Button_Click(object sender, RoutedEventArgs e)
 		{
+			if (!textDirty) return;
 			RawDocument doc = parser.Parse();
 			if (treeView.Items.Count == 0) {
 				treeView.Items.Add(doc);
 			}
+			PrettyPrintXmlVisitor visitor = new PrettyPrintXmlVisitor();
+			visitor.VisitDocument(doc);
+			string prettyPrintedText = visitor.Output;
+			if (prettyPrintedText != editor.Document.Text) {
+				MessageBox.Show("Original and pretty printer version of XML differ");
+			}
+			textDirty = false;
 		}
 		
 		void BindObject(object sender, EventArgs e)

src/Libraries/AvalonEdit/ICSharpCode.AvalonEdit/ICSharpCode.AvalonEdit.csproj

@@ -310,6 +310,7 @@
     <CodeAnalysisDictionary Include="Properties\CodeAnalysisDictionary.xml" />
     <Compile Include="XmlParser\Collections.cs" />
     <Compile Include="XmlParser\RawObjects.cs" />
+    <Compile Include="XmlParser\Visitors.cs" />
     <Compile Include="XmlParser\XmlParser.cs" />
     <Resource Include="themes\RightArrow.cur" />
     <EmbeddedResource Include="Highlighting\Resources\ASPX.xshd" />

src/Libraries/AvalonEdit/ICSharpCode.AvalonEdit/XmlParser/RawObjects.cs

@@ -79,6 +79,8 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 			return new RawObject[] { this };
 		}
 		
+		public abstract void AcceptVisitor(IXmlVisitor visitor);
+		
 		public virtual void UpdateDataFrom(RawObject source)
 		{
 			this.ReadCallID = source.ReadCallID;
@@ -324,6 +326,11 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 			if (ObjectDettached != null) ObjectDettached(this, new RawObjectEventArgs() { Object = obj } );
 		}
 		
+		public override void AcceptVisitor(IXmlVisitor visitor)
+		{
+			visitor.VisitDocument(this);
+		}
+		
 		XDocument xDoc;
 		
 		public XDocument GetXDocument()
@@ -378,6 +385,11 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		public bool IsDocumentType          { get { return DTDNames.Contains(OpeningBracket); } }
 		public bool IsUnknownBang           { get { return OpeningBracket == "<!"; } }
 		
+		public override void AcceptVisitor(IXmlVisitor visitor)
+		{
+			visitor.VisitTag(this);
+		}
+		
 		public override void UpdateDataFrom(RawObject source)
 		{
 			if (this.ReadCallID == source.ReadCallID) return;
@@ -424,6 +436,11 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 			}
 		}
 		
+		public override void AcceptVisitor(IXmlVisitor visitor)
+		{
+			visitor.VisitElement(this);
+		}
+		
 		XElement xElem;
 		
 		public XElement GetXElement()
@@ -500,6 +517,11 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		
 		// TODO: Provide method to dereference Value - &
 		
+		public override void AcceptVisitor(IXmlVisitor visitor)
+		{
+			visitor.VisitAttribute(this);
+		}
+		
 		public override void UpdateDataFrom(RawObject source)
 		{
 			if (this.ReadCallID == source.ReadCallID) return;
@@ -584,6 +606,11 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		public RawTextType Type { get; set; }
 		public string Value { get; set; }
 		
+		public override void AcceptVisitor(IXmlVisitor visitor)
+		{
+			visitor.VisitText(this);
+		}
+		
 		public override void UpdateDataFrom(RawObject source)
 		{
 			if (this.ReadCallID == source.ReadCallID) return;

src/Libraries/AvalonEdit/ICSharpCode.AvalonEdit/XmlParser/Visitors.cs

@@ -0,0 +1,121 @@
+// <file>
+//     <copyright see="prj:///doc/copyright.txt"/>
+//     <license see="prj:///doc/license.txt"/>
+//     <owner name="David Srbecký" email="dsrbecky@gmail.com"/>
+//     <version>$Revision$</version>
+// </file>
+using System;
+using System.Text;
+
+namespace ICSharpCode.AvalonEdit.XmlParser
+{
+	/// <summary>
+	/// Visitor for the XML tree
+	/// </summary>
+	public interface IXmlVisitor
+	{
+		/// <summary> Visit RawDocument </summary>
+		void VisitDocument(RawDocument document);
+		
+		/// <summary> Visit RawElement </summary>
+		void VisitElement(RawElement element);
+		
+		/// <summary> Visit RawTag </summary>
+		void VisitTag(RawTag tag);
+		
+		/// <summary> Visit RawAttribute </summary>
+		void VisitAttribute(RawAttribute attribute);
+		
+		/// <summary> Visit RawText </summary>
+		void VisitText(RawText text);
+	}
+	
+	/// <summary>
+	/// Derive from this class to create visitor for the XML tree
+	/// </summary>
+	public abstract class AbstractXmlVisitor : IXmlVisitor
+	{
+		/// <summary> Visit RawDocument </summary>
+		public virtual void VisitDocument(RawDocument document)
+		{
+			foreach(RawObject child in document.Children) child.AcceptVisitor(this);
+		}
+		
+		/// <summary> Visit RawElement </summary>
+		public virtual void VisitElement(RawElement element)
+		{
+			foreach(RawObject child in element.Children) child.AcceptVisitor(this);
+		}
+		
+		/// <summary> Visit RawTag </summary>
+		public virtual void VisitTag(RawTag tag)
+		{
+			foreach(RawObject child in tag.Children) child.AcceptVisitor(this);
+		}
+		
+		/// <summary> Visit RawAttribute </summary>
+		public virtual void VisitAttribute(RawAttribute attribute)
+		{
+			
+		}
+		
+		/// <summary> Visit RawText </summary>
+		public virtual void VisitText(RawText text)
+		{
+			
+		}
+	}
+	
+	/// <summary>
+	/// Converts the XML tree back to text.
+	/// The text should exactly match the original.
+	/// </summary>
+	public class PrettyPrintXmlVisitor: AbstractXmlVisitor
+	{
+		StringBuilder sb = new StringBuilder();
+		
+		/// <summary>
+		/// Gets the pretty printed text
+		/// </summary>
+		public string Output {
+			get {
+				return sb.ToString();
+			}
+		}
+		
+		/// <summary> Visit RawDocument </summary>
+		public override void VisitDocument(RawDocument document)
+		{
+			base.VisitDocument(document);
+		}
+		
+		/// <summary> Visit RawElement </summary>
+		public override void VisitElement(RawElement element)
+		{
+			base.VisitElement(element);
+		}
+		
+		/// <summary> Visit RawTag </summary>
+		public override void VisitTag(RawTag tag)
+		{
+			sb.Append(tag.OpeningBracket);
+			sb.Append(tag.Name);
+			base.VisitTag(tag);
+			sb.Append(tag.ClosingBracket);
+		}
+		
+		/// <summary> Visit RawAttribute </summary>
+		public override void VisitAttribute(RawAttribute attribute)
+		{
+			sb.Append(attribute.Name);
+			sb.Append(attribute.EqualsSign);
+			sb.Append(attribute.Value);
+		}
+		
+		/// <summary> Visit RawText </summary>
+		public override void VisitText(RawText text)
+		{
+			sb.Append(text.Value);
+		}
+	}
+}

src/Libraries/AvalonEdit/ICSharpCode.AvalonEdit/XmlParser/XmlParser.cs

@@ -88,7 +88,13 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 	/// </remarks>
 	public class XmlParser
 	{
-		RawDocument userDocument = new RawDocument();
+		// TODO: Error reporting
+		// TODO: Trace touched memory
+		// TODO: Simple tag matching heuristic
+		// TODO: Simple attribute value closing heurisitc
+		// TODO: Backtracking for unclosed long Text sections
+		
+		RawDocument userDocument;
 		XDocument userLinqDocument;
 		TextDocument textDocument;
 		TextSegmentCollection<RawObject> parsedItems = new TextSegmentCollection<RawObject>();
@@ -97,9 +103,18 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		/// <summary>
 		/// Create new parser, but do not parse the text yet.
 		/// </summary>
-		public XmlParser(TextDocument textDocument)
+		public XmlParser(string input)
 		{
+			this.input = input;
+			this.userDocument = new RawDocument();
 			this.userLinqDocument = userDocument.GetXDocument();
+		}
+		
+		/// <summary>
+		/// Create new parser, but do not parse the text yet.
+		/// </summary>
+		public XmlParser(TextDocument textDocument): this(textDocument.Text)
+		{
 			this.textDocument = textDocument;
 			this.textDocument.Changed += delegate(object sender, DocumentChangeEventArgs e) {
 				changesSinceLastParse.Add(e);
@@ -111,18 +126,18 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		/// </summary>
 		public RawDocument Parse()
 		{
-			input = textDocument.Text;
-			readingEnd = input.Length;
-			currentLocation = 0;
+			// Update source text
+			if (textDocument != null) {
+				input = textDocument.Text;
+			}
 			
+			// Use chages to invalidate cache
 			foreach(DocumentChangeEventArgs change in changesSinceLastParse) {
 				// Update offsets of all items
 				parsedItems.UpdateOffsets(change);
 				// Remove any items affected by the change
-				int start = change.Offset - 2;
-				int end = change.Offset + change.InsertionLength + 2;
-				start = Math.Max(Math.Min(start, textDocument.TextLength - 1), 0);
-				end = Math.Max(Math.Min(end, textDocument.TextLength - 1), 0);
+				int start = Math.Max(change.Offset - 2, 0);
+				int end = Math.Max(change.Offset + change.InsertionLength + 2, 0);
 				foreach(RawObject obj in parsedItems.FindOverlappingSegments(start, end - start)) {
 					parsedItems.Remove(obj);
 					Log("Removed cached item {0}", obj);
@@ -130,6 +145,9 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 			}
 			changesSinceLastParse.Clear();
 			
+			currentLocation = 0;
+			readingEnd = input.Length;
+			
 			RawDocument parsedDocument = ReadDocument();
 			// Just in case parse method was called redundantly
 			if (parsedDocument.ReadCallID != userDocument.ReadCallID) {
@@ -357,8 +375,6 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 		/// </summary>
 		bool TryReadName(out string res)
 		{
-			AssertHasMoreData();
-			
 			int start = currentLocation;
 			TryMoveToAnyOf(WhiteSpaceAndReservedChars.ToArray());
 			if (start == currentLocation) {
@@ -433,8 +449,8 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 						break;
 					} else if (TryPeek('<')) {
 						RawObject content = ReadElementOrTag();
-						if (content is RawTag && ((RawTag)content).IsEndTag) break;
 						element.AddChild(content);
+						if (content is RawTag && ((RawTag)content).IsEndTag) break;
 					} else {
 						element.AddChildren(ReadText(RawTextType.CharacterData));
 					}
@@ -479,7 +495,10 @@ namespace ICSharpCode.AvalonEdit.XmlParser
 				while(true) {					
 					// Chech for all forbiden 'name' charcters first - see ReadName
 					if (IsEndOfFile()) break;
-					if (TryPeekWhiteSpace()) tag.AddChildren(ReadText(RawTextType.WhiteSpace));
+					if (TryPeekWhiteSpace()) {
+						tag.AddChildren(ReadText(RawTextType.WhiteSpace));
+						continue;  // End of file might be next
+					}
 					if (TryPeek('<')) break;
 					if (TryPeek('>') || TryPeek('/') || TryPeek('?')) break;  // End tag