Source code for epub2txt.cs

```
using System;
```
```
using System.Collections.Generic;
```
```
using System.IO;
```
```
using System.IO.Compression;
```
```
using System.Text;
```
```
using System.Text.RegularExpressions;
```
```
using System.Xml.Linq;
```
```
 
```
```
 
```
```
namespace RobvanderWoude
```
```
{
```
```
	internal class Epub2Txt
```
```
	{
```
```
		static string progver = "1.01";
```
```
 
```
```
 
```
```
		static int Main( string[] args )
```
```
		{
```
```
			string epub = string.Empty;
```
```
			Encoding encoding = null;
```
```
 
```
```
 
```
```
			#region Parse Command Line
```
```
 
```

			if ( args.Length == 0 || args.Length > 2 )

```
			{
```
```
				return ShowHelp( );
```
```
			}
```
```
 
```
```
			foreach ( string arg in args )
```
```
			{
```
```
				if ( arg[0] == '/' )
```
```
				{
```
```
					if ( arg == "/?" )
```
```
					{
```
```
						return ShowHelp( );
```
```
					}
```
```
					else if ( arg.ToUpper( ) == "/E" )
```
```
					{
```
```
						return ListEncodings( );
```
```
					}
```
```
					else
```
```
					{
```

						return ShowHelp( "Invalid command line switch {0}", arg );

```
					}
```
```
				}
```
```
				else
```
```
				{
```

					if ( string.IsNullOrWhiteSpace( epub ) )

```
					{
```
```
						epub = arg;
```
```
						if ( !File.Exists( epub ) )
```
```
						{
```

							return ShowHelp( "File \"{0}\" not found", epub );

```
						}
```

						if ( Path.GetExtension( epub ).ToLower( ) != ".epub" )

```
						{
```

							return ShowHelp( "This program can handle .EPUB files only" );

```
						}
```
```
					}
```
```
					else if ( encoding == null )
```
```
					{
```
```
						encoding = GetEncoding( arg );
```
```
						if ( encoding == null )
```
```
						{
```

							return ShowHelp( "Invalid encoding \"{0}\"", args[1] );

```
						}
```
```
					}
```
```
					else
```
```
					{
```

						return ShowHelp( "Too many command line arguments" );

```
					}
```
```
				}
```
```
			}
```
```
 
```

			if ( string.IsNullOrWhiteSpace( epub ) )

```
			{
```
```
				return ShowHelp( );
```
```
			}
```
```
 
```
```
			#endregion Parse Command Line
```
```
 
```
```
 
```
```
			#region Extract Text
```
```
 
```
```
			string content = string.Empty;
```
```
			string doctitle = string.Empty;
```

			SortedDictionary<int, string> textcontent = new SortedDictionary<int, string>( );

```
 
```

			// Open document as ZIP file and extract the XML file containing the text content

			using ( ZipArchive archive = ZipFile.OpenRead( epub ) )

```
			{
```

				foreach ( ZipArchiveEntry entry in archive.Entries )

```
				{
```

					if ( entry.Name.ToLower( ) == "toc.ncx" )

```
					{
```

						string toc = Path.GetTempFileName( );

```
						entry.ExtractToFile( toc, true );
```
```
 
```

						XDocument xml = XDocument.Load( toc );

```
						if ( encoding == null )
```
```
						{
```

							encoding = GetEncoding( xml.Declaration.Encoding );

```
						}
```

						foreach ( XElement el in xml.Elements( ).Elements( ) )

```
						{
```

							if ( el.Name.LocalName == "docTitle" )

```
							{
```
```
								doctitle = el.Value;
```
```
							}
```
```
						}
```

						foreach ( XElement el in xml.Elements( ).Elements( ).Elements( ) )

```
						{
```

							if ( el.Name.LocalName == "navPoint" )

```
							{
```
```
								string header = el.Value;
```

								int playorder = int.Parse( el.Attribute( "playOrder" ).Value );

								string chaptersrc = string.Empty;

								foreach ( XElement el2 in el.Elements( ) )

```
								{
```

									if ( el2.Name.LocalName == "content" )

```
									{
```

										chaptersrc = el2.Attribute( "src" ).Value;

										if ( chaptersrc.Contains( "#" ) )

```
										{
```

											chaptersrc = chaptersrc.Split( "#".ToCharArray( ) )[0];

```
										}
```
```
									}
```
```
								}
```

								string chapter = Path.GetTempFileName( );

```
								string text = string.Empty;
```

								foreach ( ZipArchiveEntry entry2 in archive.Entries )

```
								{
```

									if ( entry2.Name == chaptersrc )

```
									{
```

										entry2.ExtractToFile( chapter, true );

										text = File.ReadAllText( chapter );

```
										File.Delete( chapter );
```

										text = Regex.Replace( text, "<title>[^<]*</title>", string.Empty, RegexOptions.IgnoreCase );

										text = Regex.Replace( text, "<br( /)?>", Environment.NewLine, RegexOptions.IgnoreCase );

										text = Regex.Replace( text, "(</p>|</h\\d+>)", Environment.NewLine + Environment.NewLine, RegexOptions.IgnoreCase );

										text = Regex.Replace( text, "<[^>]+>", string.Empty );

```
									}
```
```
								}
```

								textcontent.Add( playorder, text );

```
							}
```
```
						}
```
```
						File.Delete( toc );
```
```
					}
```
```
				}
```
```
			}
```
```
 
```
```
			#endregion Extract Text
```
```
 
```
```
 
```
```
			int textlength = 0;
```

			Encoding oldencoding = Console.OutputEncoding;

```
			Console.OutputEncoding = encoding;
```

			foreach ( int key in textcontent.Keys )

```
			{
```

				Console.WriteLine( textcontent[key] );

				textlength += textcontent[key].Length;

```
			}
```

			Console.OutputEncoding = oldencoding;

```
 
```
```
			if ( textlength > 1000 )
```
```
			{
```
```
				return 0;
```
```
			}
```
```
			else
```
```
			{
```
```
				return 1;
```
```
			}
```
```
		}
```
```
 
```
```
 
```

		static Encoding GetEncoding( string myencoding )

```
		{
```

			if ( string.IsNullOrEmpty( myencoding ) )

```
			{
```
```
				return null;
```
```
			}
```
```
			// Get a list of available encodings
```

			EncodingInfo[] encodings = Encoding.GetEncodings( );

			// Try correctly spelled encodings first

			foreach ( EncodingInfo encoding in encodings )

```
			{
```

				if ( encoding.Name.ToLower( ) == myencoding.ToLower( ) )

```
				{
```

					return Encoding.GetEncoding( encoding.CodePage );

```
				}
```
```
			}
```

			// No direct match found, try again, ignoring dashes

			foreach ( EncodingInfo encoding in encodings )

```
			{
```

				if ( encoding.Name.Replace( "-", "" ).ToLower( ) == myencoding.Replace( "-", "" ).ToLower( ) )

```
				{
```

					return Encoding.GetEncoding( encoding.CodePage );

```
				}
```
```
			}
```
```
			// Still no match, try codepages
```

			foreach ( EncodingInfo encoding in encodings )

```
			{
```

				if ( encoding.CodePage.ToString( ) == myencoding )

```
				{
```

					return Encoding.GetEncoding( encoding.CodePage );

```
				}
```
```
			}
```
```
			// Still no match, giving up
```
```
			return null;
```
```
		}
```
```
 
```
```
 
```
```
		static int ListEncodings( )
```
```
		{
```
```
			try
```
```
			{
```
```
				Console.Clear( );
```
```
			}
```
```
			catch
```
```
			{
```

				// Console.Clear( ) throws an IO exception if the output is redirected

```
			}
```
```
			int columnwidth = 8;
```

			EncodingInfo[] allencodings = Encoding.GetEncodings( );

			List<string> allencodingnames = new List<string>( );

			foreach ( EncodingInfo enc in allencodings )

```
			{
```
```
				allencodingnames.Add( enc.Name );
```
```
			}
```
```
			allencodingnames.Sort( );
```

			foreach ( string enc in allencodingnames )

```
			{
```

				columnwidth = Math.Max( columnwidth, enc.Length );

```
			}
```

			Console.WriteLine( "{0,-" + columnwidth + "}   {1}", "Encoding", "CodePage" );

			Console.WriteLine( "{0,-" + columnwidth + "}   {1}", "========", "========" );

			foreach ( string enc in allencodingnames )

```
			{
```

				Console.WriteLine( "{0,-" + columnwidth + "}   {1}", enc, GetEncoding( enc ).CodePage );

```
			}
```
```
			return 0;
```
```
		}
```
```
 
```
```
 
```

		static int ShowHelp( params string[] errmsg )

```
		{
```
```
			#region Help Text
```
```
 
```
```
			/*
```
```
			Epub2Txt,  Version 1.01
```

			Extract plain text from an EPUB file and send it to the screen

```
 
```

			Usage:   Epub2Txt    "epubfile"  [ encoding ]

```
 
```
```
			or:      Epub2Txt    /E
```
```
 
```

			Where:   epubfile    is the path of the EPUB file to be read

			                     (no wildcards allowed, only .epub extension)

			         encoding    force use of alternative encoding for plain

			                     text, e.g. UTF-8 to preserve accented characters

			                     or IBM437 to convert unicode quotes to ASCII

			                     (default: encoding of EPUB file)

			         /E          list all available encodings

```
 
```

			Notes:   If the specified encoding does not match any available encoding

			         name, the program will try again, ignoring dashes; if that does

			         not provide a match, the program will try matching the specified

			         encoding with the available encodings' codepages.

			         This program requires .NET 4.5.

			         Return code ("errorlevel") 0 means no errors were encountered and

			         extracted text exceeds 1KB; otherwise the return code will be 1.

```
 
```
```
			Written by Rob van der Woude
```
```
			https://www.robvanderwoude.com
```
```
			*/
```
```
 
```
```
			#endregion Help Text
```
```
 
```
```
 
```
```
			#region Error Message
```
```
 
```
```
			if ( errmsg.Length > 0 )
```
```
			{
```

				List<string> errargs = new List<string>( errmsg );

```
				errargs.RemoveAt( 0 );
```
```
				Console.Error.WriteLine( );
```

				Console.ForegroundColor = ConsoleColor.Red;

```
				Console.Error.Write( "ERROR:\t" );
```

				Console.ForegroundColor = ConsoleColor.White;

				Console.Error.WriteLine( errmsg[0], errargs.ToArray( ) );

```
				Console.ResetColor( );
```
```
			}
```
```
 
```
```
			#endregion Error Message
```
```
 
```
```
 
```
```
			#region Display Help Text
```
```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```

			Console.Error.WriteLine( "Epub2Txt,  Version {0}", progver );

```
 
```

			Console.Error.WriteLine( "Extract plain text from an EPUB file and send it to the screen" );

```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```
```
			Console.Error.Write( "Usage:   " );
```

			Console.ForegroundColor = ConsoleColor.White;

			Console.Error.WriteLine( "Epub2Txt    \"epubfile\"  [ encoding ]" );

```
			Console.ResetColor( );
```
```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```
```
			Console.Error.Write( "or:      " );
```

			Console.ForegroundColor = ConsoleColor.White;

			Console.Error.WriteLine( "Epub2Txt    /E" );

```
			Console.ResetColor( );
```
```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```
```
			Console.Error.Write( "Where:   " );
```

			Console.ForegroundColor = ConsoleColor.White;

```
			Console.Error.Write( "epubfile" );
```
```
			Console.ResetColor( );
```

			Console.Error.WriteLine( "    is the path of the EPUB file to be read" );

```
 
```

			Console.Error.WriteLine( "                     (no wildcards allowed, only .epub extension)" );

```
 
```

			Console.ForegroundColor = ConsoleColor.White;

			Console.Error.Write( "         encoding" );

```
			Console.ResetColor( );
```

			Console.Error.WriteLine( "    force use of alternative encoding for plain" );

```
 
```

			Console.Error.Write( "                     text, e.g. " );

			Console.ForegroundColor = ConsoleColor.White;

```
			Console.Error.Write( "UTF-8" );
```
```
			Console.ResetColor( );
```

			Console.Error.WriteLine( " to preserve accented characters" );

```
 
```

			Console.Error.Write( "                     or " );

			Console.ForegroundColor = ConsoleColor.White;

```
			Console.Error.Write( "IBM437" );
```
```
			Console.ResetColor( );
```

			Console.Error.WriteLine( " to convert unicode quotes to ASCII" );

```
 
```

			Console.Error.WriteLine( "                     (default: encoding of EPUB file)" );

```
 
```

			Console.ForegroundColor = ConsoleColor.White;

			Console.Error.Write( "         /E" );

```
			Console.ResetColor( );
```

			Console.Error.WriteLine( "          list all available encodings" );

```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```

			Console.Error.WriteLine( "Notes:   If the specified encoding does not match any available encoding" );

```
 
```

			Console.Error.WriteLine( "         name, the program will try again, ignoring dashes; if that does" );

```
 
```

			Console.Error.WriteLine( "         not provide a match, the program will try matching the specified" );

```
 
```

			Console.Error.WriteLine( "         encoding with the available encodings' codepages." );

```
 
```

			Console.Error.WriteLine( "         Return code (\"errorlevel\") 0 means no errors were encounterd" );

```
 
```

			Console.Error.WriteLine( "         and some text was extracted from the file; otherwise the" );

```
 
```

			Console.Error.WriteLine( "         return code will be 1." );

```
 
```
```
			Console.Error.WriteLine( );
```
```
 
```

			Console.Error.WriteLine( "Written by Rob van der Woude" );

```
 
```

			Console.Error.WriteLine( "https://www.robvanderwoude.com" );

```
 
```
```
			#endregion Display Help Text
```
```
 
```
```
 
```
```
			return 1;
```
```
		}
```
```
	}
```
```
}
```