public static String GetText(HWPFDocumentCore wordDocument)
{
WordToTextConverter wordToTextConverter = new WordToTextConverter(new XmlDocument());
wordToTextConverter.ProcessDocument(wordDocument);
return(wordToTextConverter.GetText());
}
public static XmlDocument Process(string docFile)
{
HWPFDocumentCore wordDocument = AbstractWordUtils
.LoadDoc(docFile);
WordToTextConverter wordToTextConverter = new WordToTextConverter(new XmlDocument());
wordToTextConverter.ProcessDocument(wordDocument);
return wordToTextConverter.Document;
}
public static XmlDocument Process(string docFile)
{
HWPFDocumentCore wordDocument = AbstractWordUtils
.LoadDoc(docFile);
WordToTextConverter wordToTextConverter = new WordToTextConverter(new XmlDocument());
wordToTextConverter.ProcessDocument(wordDocument);
return(wordToTextConverter.Document);
}
protected override bool ProcessOle2(HWPFDocument wordDocument, XmlElement block, Entry entry)
{
if (!(entry is DirectoryNode))
{
return(false);
}
DirectoryNode directoryNode = (DirectoryNode)entry;
/*
* even if there is no ExtractorFactory in classpath, still support
* included Word's objects
*/
//TODO: Not completed
if (directoryNode.HasEntry("WordDocument"))
{
String text = WordToTextConverter.GetText((DirectoryNode)entry);
block.AppendChild(textDocumentFacade
.CreateText(UNICODECHAR_ZERO_WIDTH_SPACE + text
+ UNICODECHAR_ZERO_WIDTH_SPACE));
return(true);
}
Object extractor;
/*try
* {
* Class<?> cls = Class
* .ForName( "org.apache.poi.extractor.ExtractorFactory" );
* Method createExtractor = cls.GetMethod( "createExtractor",
* DirectoryNode.class );
* extractor = createExtractor.Invoke( null, directoryNode );
* }
* catch ( Error exc )
* {
* // no extractor in classpath
* logger.Log( POILogger.WARN, "There is an OLE object entry '",
* entry.GetName(),
* "', but there is no text extractor for this object type ",
* "or text extractor factory is not available: ", "" + exc );
* return false;
* }
*
* try
* {
* Method getText = extractor.GetClass().GetMethod( "getText" );
* String text = (String) getText.Invoke( extractor );
*
* block.AppendChild( textDocumentFacade
* .CreateText( UNICODECHAR_ZERO_WIDTH_SPACE + text
+ UNICODECHAR_ZERO_WIDTH_SPACE ) );
+ return true;
+ }
+ catch ( Exception exc )
+ {
+ logger.Log( POILogger.ERROR,
+ "Unable to extract text from OLE entry '", entry.GetName(),
+ "': ", exc, exc );
+ return false;
+ }
* */
return(false);
}
public static String GetText(HWPFDocumentCore wordDocument)
{
WordToTextConverter wordToTextConverter = new WordToTextConverter(new XmlDocument());
wordToTextConverter.ProcessDocument(wordDocument);
return wordToTextConverter.GetText();
}
