Using BaseTextParser for every Text Parser

ToxyFramework/Base/BaseTextParser.cs

@@ -1,4 +1,5 @@
 using System;
+using System.IO;
 
 namespace Toxy.Base
 {
@@ -20,7 +21,6 @@ private protected BaseTextParser(ParserContext context)
 			Context = context;
 		}
 
-#nullable enable
         internal virtual void ValidateContext()
 		{
 		    Utility.ValidateContext(Context);
@@ -35,15 +35,20 @@ internal virtual void ValidateContext()
 		public string Parse()
 		{
 		    ValidateContext();
-			IDisposable? disposable = null;
+			Stream stream = Utility.GetStream(Context);
 			try
 			{
-				return ParseText(out disposable);
+				return ParseText(stream);
 			}
 			finally
 			{
 				// we don't care if it throws just dispose if needed
-				disposable?.Dispose();
+				// Streams passed by the User should not be disposed
+				// this should be the only place to do so
+				if (!Context.IsStreamContext)
+				{
+					stream.Dispose();
+				}
 			}
 		}
 
@@ -53,7 +58,6 @@ public string Parse()
 		/// <param name="disposable">An optional <see cref="IDisposable"/>, which should be disposed at the end of parsing.
 		/// It will be disposed if an <see cref="Exception"/> will be thrown.</param>
 		/// <returns>Returns the extracted Text.</returns>
-		internal abstract string ParseText(out IDisposable? disposable);
-#nullable disable
+		internal abstract string ParseText(Stream stream);
 	}
 }

ToxyFramework/ParserContext.cs

@@ -1,65 +1,63 @@
-using DocumentFormat.OpenXml.InkML;
-using DocumentFormat.OpenXml.Math;
-using FileSignatures;
+using FileSignatures;
 using System.Collections.Generic;
 using System.IO;
 using System.Text;
 
 namespace Toxy
 {
-    public class ParserContext
-    {
-        Stream _filestream;
-        public ParserContext(string path):this(path, null)
-        { 
-
-        }
-        public ParserContext(string path, Encoding encoding)
-        {
-            this.Path = path;
-            this.Properties = new Dictionary<string, string>();
-            this.Encoding = encoding;
-        }
-        public ParserContext(Stream stream):this(null, Encoding.UTF8)
-        {
-            this._filestream = stream;
-        }
-
-        public bool IsStreamContext { get=> this._filestream != null; }
-        public Stream Stream { get => this._filestream; }
-        public string Path { get; set; }
+	public class ParserContext
+	{
+		private Stream _filestream;
+		public bool IsStreamContext { get => _filestream != null; }
+		public Stream Stream { get => _filestream; }
+		public string Path { get; set; }
 
-        public string Extension
-        {
-            get
-            {
-                return Utility.GetFileExtention(this);
-            }
-        }
-        public string MimeType
-        {
-            get
-            {
-                FileFormat? fileformat = null;
-                FileFormatInspector inspector = new FileFormatInspector();
-                if (this.IsStreamContext)
-                {
-                    this.Stream.Position = 0;
-                    fileformat = inspector.DetermineFileFormat(this.Stream);
-                }
-                else
-                {
-                    using var stream = File.OpenRead(this.Path);
-                    fileformat = inspector.DetermineFileFormat(stream);
-                }
-                if (fileformat == null)
-                    throw new InvalidDataException("File format could not be determined for the input stream");
-                return fileformat.MediaType;
+		public string Extension
+		{
+			get
+			{
+				return Utility.GetFileExtention(this);
+			}
+		}
+		public string MimeType
+		{
+			get
+			{
+				FileFormat fileformat = null;
+				FileFormatInspector inspector = new FileFormatInspector();
+				if (IsStreamContext)
+				{
+					Stream.Position = 0;
+					fileformat = inspector.DetermineFileFormat(Stream);
+				}
+				else
+				{
+					using var stream = File.OpenRead(Path);
+					fileformat = inspector.DetermineFileFormat(stream);
+				}
+				if (fileformat == null)
+				{
+					throw new InvalidDataException("File format could not be determined for the input stream");
+				}
+				return fileformat.MediaType;
 
-            }
-        }
-        public Encoding Encoding { get; set; }
-
-        public Dictionary<string, string> Properties { get; set; }
-    }
+			}
+		}
+		public Encoding Encoding { get; set; }
+
+		public Dictionary<string, string> Properties { get; set; }
+
+		public ParserContext(string path) : this(path, null)
+		{ }
+		public ParserContext(string path, Encoding encoding)
+		{
+			Path = path;
+			Properties = new Dictionary<string, string>();
+			Encoding = encoding;
+		}
+		public ParserContext(Stream stream) : this(null, Encoding.UTF8)
+		{
+			_filestream = stream;
+		}
+	}
 }

ToxyFramework/Parsers/EPUB/EPUBTextParser.cs

@@ -1,20 +1,22 @@
-using HtmlAgilityPack;
+using DocumentFormat.OpenXml.InkML;
+using HtmlAgilityPack;
+using System;
+using System.IO;
 using System.Text;
 using System.Text.RegularExpressions;
+using Toxy.Base;
 using VersOne.Epub;
 
 namespace Toxy.Parsers
 {
-	public sealed class EPUBTextParser : ITextParser
+	public sealed class EPUBTextParser : BaseTextParser
 	{
-		public EPUBTextParser(ParserContext context)
-		{
-			Context = context;
-		}
-		public ParserContext Context { get; set; }
-		public string Parse()
+		public EPUBTextParser(ParserContext context) : base(context)
+		{ }
+
+		internal override string ParseText(Stream stream)
 		{
-			EpubBook book = EPUBHelper.GetEpubBook(Context);
+			EpubBook book = EpubReader.ReadBook(stream);
 			StringBuilder result = new StringBuilder();
 			HtmlDocument html = new HtmlDocument();
 			Regex whiteSpaceStart = new Regex("^\\s*", RegexOptions.Multiline);

ToxyFramework/Parsers/Email/EMLTextParser.cs

@@ -18,10 +18,8 @@ public class EMLTextParser : BaseTextParser
 		public EMLTextParser(ParserContext context) : base(context)
 		{ }
 
-		internal override string ParseText(out IDisposable disposable)
+		internal override string ParseText(Stream stream)
 		{
-			Stream stream = Utility.GetStream(Context);
-			disposable = Context.IsStreamContext ? null : stream;
 			using (MimeMessage message = MimeMessage.Load(stream))
 			{
 				return MimeMessageHelper.ConvertToText(message);

ToxyFramework/Parsers/Email/MsgTextParser.cs

@@ -3,6 +3,7 @@
 using System;
 using System.IO;
 using System.Text;
+using Toxy.Base;
 
 namespace Toxy.Parsers
 {
@@ -12,21 +13,15 @@ namespace Toxy.Parsers
     /// <remarks>
     /// http://www.codeproject.com/Articles/19571/MsgReader-DLL
     /// </remarks>
-    public class MsgTextParser : PlainTextParser
+    public class MsgTextParser : BaseTextParser
     {
-        public MsgTextParser(ParserContext context)
-            : base(context)
-        {
-            this.Context = context;
-        }
+        public MsgTextParser(ParserContext context) : base(context)
+        { }
 
-        public override string Parse()
+        internal override string ParseText(Stream stream)
         {
-            Utility.ValidateContext(Context);
-
             StringBuilder result = new StringBuilder();
-            using (var stream = Utility.GetStream(Context))
-            using (var reader = new Storage.Message(stream))
+			using (Storage.Message reader = new Storage.Message(stream, FileAccess.Read, Context.IsStreamContext))
             {
                 if (reader.Sender != null)
                 {
@@ -39,7 +34,7 @@ public override string Parse()
                     StringBuilder recipientTo = new StringBuilder();
                     StringBuilder recipientCc = new StringBuilder();
                     StringBuilder recipientBcc = new StringBuilder();
-                    foreach (var recipient in reader.Recipients)
+                    foreach (Storage.Recipient recipient in reader.Recipients)
                     {
                         string sRecipient = null;
                         if (string.IsNullOrEmpty(recipient.DisplayName))
@@ -79,12 +74,13 @@ public override string Parse()
                         result.Append("[Bcc] ");
                         result.AppendLine(recipientBcc.ToString());
                     }
-
                 }
-                if (!string.IsNullOrEmpty(reader.Subject))
-                    result.AppendFormat("[Subject] {0}{1}", reader.Subject, Environment.NewLine);
+				if (!string.IsNullOrEmpty(reader.Subject))
+				{
+					result.AppendFormat("[Subject] {0}{1}", reader.Subject, Environment.NewLine);
+				}
 
-                result.AppendLine();
+				result.AppendLine();
                 result.AppendLine(reader.BodyText);
             }
             return result.ToString();

ToxyFramework/Parsers/Office/PowerPoint2007TextParser.cs

@@ -0,0 +1,104 @@
+using DocumentFormat.OpenXml;
+using DocumentFormat.OpenXml.Packaging;
+using DocumentFormat.OpenXml.Presentation;
+using PasswordProtectedChecker;
+using System;
+using System.Collections.Generic;
+using System.IO;
+using System.IO.Packaging;
+using System.Linq;
+using System.Text;
+using Toxy.Base;
+
+namespace Toxy.Parsers
+{
+	public class Powerpoint2007TextParser : BaseTextParser
+	{
+		public Powerpoint2007TextParser(ParserContext context) : base(context)
+		{ }
+
+		internal override void ValidateContext()
+		{
+			base.ValidateContext();
+			Utility.ThrowIfProtected(Context);
+		}
+
+		internal override string ParseText(Stream stream)
+		{
+			StringBuilder sb = new StringBuilder();
+			// closes the Stream anyways
+			using (Package pkg = Package.Open(stream, FileMode.Open, FileAccess.Read))
+			{
+				using PresentationDocument ppt = PresentationDocument.Open(pkg);
+				// Get the relationship ID of the first slide.
+				PresentationPart part = ppt.PresentationPart;
+				OpenXmlElementList slideIds = part.Presentation.SlideIdList.ChildElements;
+				for (int index = 0; index < slideIds.Count; index++)
+				{
+					string relId = (slideIds[index] as SlideId).RelationshipId;
+
+					// Get the slide part from the relationship ID.
+					SlidePart slide = (SlidePart)part.GetPartById(relId);
+					string[] texts = GetAllTextInSlide(slide);
+					foreach (string text in texts)
+					{
+						sb.AppendLine(text);
+					}
+				}
+			}
+			return sb.ToString();
+		}
+
+		public static string[] GetAllTextInSlide(SlidePart slidePart)
+		{
+			// Verify that the slide part exists.
+#if NET8_0_OR_GREATER
+			ArgumentNullException.ThrowIfNull(slidePart, nameof(slidePart));
+#else
+			if (slidePart == null)
+			{
+				throw new ArgumentNullException("slidePart");
+			}
+#endif
+
+			// Create a new linked list of strings.
+			LinkedList<string> texts = new LinkedList<string>();
+
+			// If the slide exists...
+			if (slidePart.Slide != null)
+			{
+				// Iterate through all the paragraphs in the slide.
+				foreach (DocumentFormat.OpenXml.Drawing.Paragraph paragraph in
+					slidePart.Slide.Descendants<DocumentFormat.OpenXml.Drawing.Paragraph>())
+				{
+					// Create a new string builder.                    
+					StringBuilder paragraphText = new StringBuilder();
+
+					// Iterate through the lines of the paragraph.
+					foreach (DocumentFormat.OpenXml.Drawing.Text text in
+						paragraph.Descendants<DocumentFormat.OpenXml.Drawing.Text>())
+					{
+						// Append each line to the previous lines.
+						paragraphText.Append(text.Text);
+					}
+
+					if (paragraphText.Length > 0)
+					{
+						// Add each paragraph to the linked list.
+						texts.AddLast(paragraphText.ToString());
+					}
+				}
+			}
+
+			if (texts.Count > 0)
+			{
+				// Return an array of strings.
+				return texts.ToArray();
+			}
+			else
+			{
+				return null;
+			}
+		}
+	}
+}

ToxyFramework/Parsers/Word2003TextParser.cs → ToxyFramework/Parsers/Office/Word2003TextParser.cs

@@ -5,6 +5,7 @@
 using System.IO;
 using System.Text;
 
+
 namespace Toxy.Parsers
 {
     public class Word2003TextParser : ITextParser