} // SetupDefaultText
// ----------------------------------------------------------------------
private void ToTextButtonClick( object sender, EventArgs e )
{
try
{
IRtfGroup rtfStructure = RtfParserTool.Parse( ConversionText );
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterTool.Interpret( rtfStructure, textConverter );
this.textBox.Text = textConverter.PlainText;
}
catch ( Exception exception )
{
MessageBox.Show( this, "Error " + exception.Message, this.Text, MessageBoxButtons.OK, MessageBoxIcon.Error );
}
} // ToTextButtonClick
public void ListSupportTest()
{
const RtfParserListenerLogger parserLogger = null;
//parserLogger = new RtfParserListenerLogger();
//string testCaseName = "RtfInterpreterTest_9";
const string testCaseName = "RtfInterpreterTest_10";
IRtfGroup rtfStructure = RtfParserTool.Parse( GetTestResource( testCaseName + ".rtf" ), parserLogger );
const RtfInterpreterListenerLogger interpreterLogger = null;
//interpreterLogger = new RtfInterpreterListenerLogger();
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterTool.Interpret( rtfStructure, textConverter, interpreterLogger );
string plainText = textConverter.PlainText;
AssertEqualLines( "list support:",
new StreamReader( GetTestResource( testCaseName + ".txt" ), RtfSpec.AnsiEncoding ),
new StringReader( plainText ) );
} // ListSupportTest
} // ConversionText
// ----------------------------------------------------------------------
private void ToTextButtonClick( object sender, RoutedEventArgs e )
{
try
{
TextRange conversionText = ConversionText;
using ( MemoryStream stream = new MemoryStream() )
{
conversionText.Save( stream, DataFormats.Rtf );
stream.Seek( 0, SeekOrigin.Begin );
using ( StreamReader reader = new StreamReader( stream ) )
{
IRtfGroup rtfStructure = RtfParserTool.Parse( reader );
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterTool.Interpret( rtfStructure, textConverter );
this.textBox.Text = textConverter.PlainText;
}
}
}
catch ( Exception exception )
{
MessageBox.Show( this, "Error " + exception.Message, this.Title, MessageBoxButton.OK, MessageBoxImage.Error );
}
} // ToTextButtonClick
// ----------------------------------------------------------------------
private string InterpretRtf( IRtfGroup rtfStructure )
{
string text = null;
RtfInterpreterListenerFileLogger interpreterLogger = null;
try
{
// logger
if ( settings.LogInterpreter )
{
string logFileName = settings.BuildDestinationFileName(
settings.LogDirectory,
RtfInterpreterListenerFileLogger.DefaultLogFileExtension );
interpreterLogger = new RtfInterpreterListenerFileLogger( logFileName );
}
// text converter
RtfTextConverter textConverter = null;
if ( settings.SaveText )
{
RtfTextConvertSettings textConvertSettings = new RtfTextConvertSettings();
textConvertSettings.IsShowHiddenText = settings.ShowHiddenText;
textConverter = new RtfTextConverter( textConvertSettings );
}
// image converter
RtfImageConverter imageConverter = null;
if ( settings.SaveImage )
{
RtfVisualImageAdapter imageAdapter = new RtfVisualImageAdapter(
settings.ImageFileNamePattern,
settings.ImageFormat );
RtfImageConvertSettings imageConvertSettings = new RtfImageConvertSettings( imageAdapter );
imageConvertSettings.ImagesPath = settings.DestinationDirectory;
imageConvertSettings.ScaleImage = settings.ScaleImage;
imageConvertSettings.BackgroundColor = settings.ImageBackgroundColor;
if ( settings.ExtendedImageScale )
{
imageConvertSettings.ScaleExtension = 0.5f;
}
imageConverter = new RtfImageConverter( imageConvertSettings );
}
// rtf parser
// interpret the rtf structure using the extractors
RtfInterpreterTool.Interpret( rtfStructure, interpreterLogger, textConverter, imageConverter );
// get the resulting text
if ( textConverter != null )
{
text = textConverter.PlainText;
}
}
catch ( Exception e )
{
if ( interpreterLogger != null )
{
interpreterLogger.Dispose();
}
Console.WriteLine( "error while interpreting rtf: " + e.Message );
ExitCode = ProgramExitCode.InterpretRtf;
return null;
}
return text;
}
// ----------------------------------------------------------------------
protected override void DoTest( string kind, Stream testRes, string testCaseName )
{
const RtfParserListenerLogger parserLogger = null;
//parserLogger = new RtfParserListenerLogger();
const RtfInterpreterListenerLogger interpreterLogger = null;
//interpreterLogger = new RtfInterpreterListenerLogger();
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterListenerDocumentBuilder docBuilder =
new RtfInterpreterListenerDocumentBuilder();
RtfInterpreterTool.Interpret( RtfParserTool.Parse( testRes, parserLogger ),
interpreterLogger, textConverter, docBuilder );
string plainText = textConverter.PlainText;
Assert.IsFalse( string.IsNullOrEmpty( plainText ) );
Assert.IsNotNull( docBuilder.Document );
string testName = testCaseName.Substring( 0, testCaseName.Length - 4 );
//string unicode = "バージョンアップ注文書（銀行振込）";
Encoding enc;
switch ( testName )
{
case "RtfInterpreterTest_12":
case "RtfInterpreterTest_13":
case "RtfInterpreterTest_14":
case "RtfInterpreterTest_16":
case "RtfInterpreterTest_17":
case "RtfInterpreterTest_18":
case "RtfInterpreterTest_19":
case "RtfInterpreterTest_21":
enc = Encoding.Unicode;
break;
default:
enc = RtfSpec.AnsiEncoding;
break;
}
string referenceResName = testName + ".txt";
/*
if ( "RtfInterpreterTest_21".Equals( testName ) )
{
using ( Stream s = new FileStream( @"w:\temp\rtf\RtfInterpreterTest_21.txt", FileMode.Create ) )
{
using ( TextWriter w = new StreamWriter( s, enc ) )
{
w.Write( plainText );
}
}
}
//*/
if ( plainText == null )
{
plainText = string.Empty;
}
AssertEqualLines( "document recognition: " + testCaseName + ":",
new StreamReader( GetTestResource( referenceResName ), enc ),
new StringReader( plainText ) );
}
public void UnicodeSupportTest()
{
const RtfParserListenerLogger parserLogger = null;
//parserLogger = new RtfParserListenerLogger();
IRtfGroup rtfStructure =
RtfParserTool.Parse( GetTestResource( "RtfInterpreterTest_2.rtf" ), parserLogger );
Assert.IsNotNull( rtfStructure );
const RtfInterpreterListenerLogger interpreterLogger = null;
//interpreterLogger = new RtfInterpreterListenerLogger();
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterTool.Interpret( rtfStructure, textConverter, interpreterLogger );
string plainText = textConverter.PlainText;
AssertEqualLines( "unicode support:",
new StreamReader( GetTestResource( "RtfInterpreterTest_2.txt" ), RtfSpec.AnsiEncoding ),
new StringReader( plainText ) );
}
public void ExtractPlainTextTest()
{
const RtfParserListenerLogger parserLogger = null;
//parserLogger = new RtfParserListenerLogger();
IRtfGroup rtfStructure =
RtfParserTool.Parse( GetTestResource( "RtfInterpreterTest_0.rtf" ), parserLogger );
Assert.IsNotNull( rtfStructure );
const RtfInterpreterListenerLogger interpreterLogger = null;
//interpreterLogger = new RtfInterpreterListenerLogger();
RtfTextConverter textConverter = new RtfTextConverter();
RtfInterpreterTool.Interpret( rtfStructure, textConverter, interpreterLogger );
string plainText = textConverter.PlainText;
Assert.AreEqual( "Hello RTF World\r\n", plainText );
}
public void ExtractPlainTextTest()
{
const RtfParserListenerLogger parserLogger = null;
//parserLogger = new RtfParserListenerLogger();
IRtfGroup rtfStructure =
RtfParserTool.Parse( GetTestResource( "RtfInterpreterTest_0.rtf" ), parserLogger );
Assert.IsNotNull( rtfStructure );
const RtfInterpreterListenerLogger interpreterLogger = null;
//interpreterLogger = new RtfInterpreterListenerLogger();
var rtfString =
@"{\rtf1\sste16000\ansi\deflang1033\ftnbj\uc1\deff0 {\fonttbl{\f0 \fnil Verdana;}{\f1 \fnil \fcharset0 Verdana;}} {\colortbl;\red0\green0\blue0;} {\stylesheet{\f0\fs28 Normal;}{\cs1 Default Paragraph Font;}} {\*\revtbl{Unknown;}} \paperw12240\paperh15840\margl1800\margr1800\margt1440\margb1440\headery720\footery720\nogrowautofit\deftab720\formshade\fet4\aendnotes\aftnnrlc\pgbrdrhead\pgbrdrfoot \sectd\pgwsxn12240\pghsxn15840\marglsxn1800\margrsxn1800\margtsxn1440\margbsxn1440\headery720\footery720\sbkpage\pgncont\pgndec \plain\plain\f0\fs24\plain\f0\fs24\plain\f1\fs28\lang1033\hich\f1\dbch\f1\loch\f1\cf2\fs28 test again and again and again }";
var rtfString2 =
@"{\rtf1\sste16000\ansi\deflang1033\ftnbj\uc1\deff0 {\fonttbl{\f0 \f \fcharset0 Arial;}} {\colortbl ;\red255\green255\blue255 ;\red0\green0\blue0 ;\red0\green0\blue0 ;} {\stylesheet{\f0\fs24 Normal;}{\cs1 Default Paragraph Font;}} {\*\revtbl{Unknown;}} \paperw12240\paperh15840\margl1800\margr1800\margt1440\margb1440\headery720\footery720\htmautsp1\nogrowautofit\deftab720\formshade\fet4\aendnotes\aftnnrlc\pgbrdrhead\pgbrdrfoot \sectd\pgwsxn12240\pghsxn15840\marglsxn1800\margrsxn1800\margtsxn1440\margbsxn1440\headery720\footery720\sbkpage\pgncont\pgndec \plain\plain\f0\fs24\ltrpar\ql\plain\f0\fs24\plain\f0\fs24\lang1033\hich\f0\dbch\f0\loch\f0\cf2\fs24\ltrch this is my cover changed in bulk\plain\f0\fs24\par and then in text\par }";
var rtfString3 =
@"{\rtf1\sste16000\ansi\deflang1033\ftnbj\uc1\deff0 {\fonttbl{\f0 \fnil Arial;}{\f1 \fnil \fcharset0 Arial;}} {\colortbl ;\red0\green0\blue0 ;\red255\green255\blue255 ;\red0\green0\blue0 ;\red255\green0\blue0 ;\red0\green0\blue0 ;} {\stylesheet{\f0\fs24 Normal;}{\cs1 Default Paragraph Font;}} {\*\revtbl{Unknown;}} \paperw12240\paperh15840\margl1800\margr1800\margt1440\margb1440\headery720\footery720\nogrowautofit\deftab720\formshade\fet4\aendnotes\aftnnrlc\pgbrdrhead\pgbrdrfoot \sectd\pgwsxn12240\pghsxn15840\marglsxn1800\margrsxn1800\margtsxn1440\margbsxn1440\headery720\footery720\sbkpage\pgncont\pgndec \plain\hich\f0\dbch\f0\loch\f0\cf1\fs24\pard\lang1033\hich\f1\dbch\f1\loch\f1 12:24 v3465 AP-APTN-1630: Turkey Syria Wednesday, 3 October 2012\par STORY:Turkey Syria- Shell fired from Syria kills three in Turkey LENGTH: 00:40\tab FIRST RUN: 1530 RESTRICTIONS: No Access Turkey/ROJ TV TYPE: Natsound SOURCE: Anadolu Agency STORY NUMBER: 861347\par DATELINE: Sanliurfa - 3 Oct 2012 LENGTH: 00:40\tab\par \par \par \cf4\b\protect [Duration: 07:06]\plain\f1\fs24 \par \par \par SHOTLIST:\par 1. Wide of smoke rising, policemen run for cover and surround wounded officer lying on ground 2. Wide of street, smoke seen rising from the distance 3. Tracking view of wounded person in a car which pulls away, policemen seen in background surrounding injured officer lying on the ground 4. Mid plainclothes officers leading crowd and media away from injured officer 5. Wide of street with smoke seen, people running along\par STORYLINE:\par A Turkish official said on Wednesday that a shell fired from Syria had landed on a house in neighbouring Turkey, killing at least three people, including a 6-year-old boy.\par Abdulhakim Ayhan, the mayor of the Turkish town of Akcakale which is along the Syrian border, told Turkey's state-owned Anadolu Agency that the boy and a woman were among the dead in Wednesday's shelling.\par Anadolu Agency reported angry townspeople marched to the mayor's office to protest against the deaths.\par Video from the Anadolu agency purportedly also showed at least one policeman was also injured in the incident.\par The village was also hit by a mortar shell on Friday, though no-one was injured on that occasion.\par The Turkish Foreign Minister Ahmet Davutoglu called an emergency meeting to discuss developments, after receiving information on the incident from Turkish Armed Forces.\par ===========================================================\par Clients are reminded: (i) to check the terms of their licence agreements for use of content outside news programming and that further advice and assistance can be obtained from the AP Archive on: Tel +44 (0) 20 7482 7482 Email: infoaparchive.com (ii) they should check with the applicable collecting society in their Territory regarding the clearance of any sound recording or performance included within the AP Television News service (iii) they have editorial responsibility for the use of all and any content included within the AP Television News service and for libel, privacy, compliance and third party rights applicable to their Territory.\par APTN\par \par \par \tab \par \tab (Copyright 2012 by The Associated Press. All Rights Reserved.)\par \tab AP-NY-10-03-12 1224EDT\par \par \fs18\par \ql\lang1033\hich\f0\dbch\f0\loch\f0\cf1\fs24\plain\f1\fs24\par }";
RtfTextConverter textConverter = new RtfTextConverter();
//RtfInterpreterTool.Interpret( rtfStructure, textConverter, interpreterLogger );
RtfInterpreterTool.Interpret(rtfString3, textConverter, interpreterLogger);
string plainText = textConverter.PlainText;
Assert.AreEqual( "Hello RTF World\r\n", plainText );
}
