R: Word to plain text converter

[Paolo Tortora]

Here some piece of java code to extract plain text from Word:


/**************** START SOURCE CODE *******************/
import org.apache.poi.poifs.filesystem.*;
import org.apache.poi.util.LittleEndian;

import java.util.ArrayList;
import java.io.InputStream;

import java.io.IOException;

class WordExtractor {

  public WordExtractor()
  {
  }

    public String extractText(InputStream in) throws IOException
  {
        ArrayList text = new ArrayList();
        POIFSFileSystem fsys = new POIFSFileSystem(in);

        DocumentEntry headerProps =
                (DocumentEntry)fsys.getRoot().getEntry("WordDocument");
        DocumentInputStream din = fsys.createDocumentInputStream("WordDocument");
        byte[] header = new byte[headerProps.getSize()];

        din.read(header);
        din.close();
        // Prende le informazioni dall'header del documento
        int info = LittleEndian.getShort(header, 0xa);

        boolean useTable1 = (info & 0x200) != 0;

        // Prende informazioni dalla piece table
        int complexOffset = LittleEndian.getInt(header, 0x1a2);


        String tableName = null;
        if (useTable1)
        {
          tableName = "1Table";
        }
        else
        {
          tableName = "0Table";
        }

        DocumentEntry table = (DocumentEntry)fsys.getRoot().getEntry(tableName);
        byte[] tableStream = new byte[table.getSize()];

        din = fsys.createDocumentInputStream(tableName);

        din.read(tableStream);
        din.close();

        din = null;
        fsys = null;
        table = null;
        headerProps = null;

        int multiple = findText(tableStream, complexOffset, text);

        StringBuffer sb = new StringBuffer();
        int size = text.size();
        tableStream = null;

        for (int x = 0; x < size; x++)
        {
          WordTextPiece nextPiece = (WordTextPiece)text.get(x);
          int start = nextPiece.getStart();
          int length = nextPiece.getLength();

          boolean unicode = nextPiece.usesUnicode();
          String toStr = null;
          if (unicode)
          {
                toStr = new String(header, start, length * multiple, "UTF-16LE");
          }
          else
          {
                toStr = new String(header, start, length , "ISO-8859-1");
          }
          sb.append(toStr).append(" ");

        }
        return sb.toString();
  }

   private static int findText(byte[] tableStream, int complexOffset,
ArrayList text) throws IOException
  {
        //actual text
        int pos = complexOffset;
        int multiple = 2;
        //skips through the prms before we reach the piece table. These contain
data
        //for actual fast saved files
        while(tableStream[pos] == 1)
        {
                pos++;
                int skip = LittleEndian.getShort(tableStream, pos);
                pos += 2 + skip;
        }
        if(tableStream[pos] != 2)
        {
                throw new IOException("corrupted Word file");
        }
        else
        {
                //parse out the text pieces
                int pieceTableSize = LittleEndian.getInt(tableStream, ++pos);
                pos += 4;
                int pieces = (pieceTableSize - 4) / 12;
                for (int x = 0; x < pieces; x++)
                {
                        int filePos = LittleEndian.getInt(tableStream, pos + ((pieces 
+ 1) * 4) +
(x * 8) + 2);
                        boolean unicode = false;
                        if ((filePos & 0x40000000) == 0)
                        {
                                unicode = true;
                        }
                        else
                        {
                                unicode = false;
                                multiple = 1;
                                filePos &= ~(0x40000000);//gives me FC in doc stream
                                filePos /= 2;
                        }
                        int totLength = LittleEndian.getInt(tableStream, pos + (x + 1) 
* 4) -
                                                        
LittleEndian.getInt(tableStream, pos + (x * 4));

                        WordTextPiece piece = new WordTextPiece(filePos, totLength, 
unicode);
                        text.add(piece);

                }

        }
        return multiple;
  }

}

class WordTextPiece
{
  private int _fcStart;
  private boolean _usesUnicode;
  private int _length;

  public WordTextPiece(int start, int length, boolean unicode)
  {
        _usesUnicode = unicode;
        _length = length;
        _fcStart = start;
  }
   public boolean usesUnicode()
  {
          return _usesUnicode;
  }

  public int getStart()
  {
          return _fcStart;
  }
  public int getLength()
  {
        return _length;
  }

}

/************** END SOURCE CODE **************/


-----Messaggio originale-----
Da: Ryan Ackley [mailto:[EMAIL PROTECTED]
Inviato: Wednesday, January 28, 2004 12:16 AM
A: POI Users List
Oggetto: Re: Word to plain text converter


http://textmining.org

----- Original Message -----
From: "Dimitri Pissarenko" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Tuesday, January 27, 2004 4:54 PM
Subject: Word to plain text converter


Hello!

I want to convert several Microsoft Word files to plain text files, so
that I can search through them with grep (or with analogous search
functions under Windows).

Has someone already written such a converter? Is this tool perhaps
open-source?

TIA

dap

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]




---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]