diff --git a/src/scratchpad/src/org/apache/poi/hwpf/extractor/WordExtractor.java b/src/scratchpad/src/org/apache/poi/hwpf/extractor/WordExtractor.java
new file mode 100644
index 000000000..639fd36ac
--- /dev/null
+++ b/src/scratchpad/src/org/apache/poi/hwpf/extractor/WordExtractor.java
@@ -0,0 +1,123 @@
+package org.apache.poi.hwpf.extractor;
+
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.UnsupportedEncodingException;
+import java.util.Iterator;
+
+import org.apache.poi.hwpf.HWPFDocument;
+import org.apache.poi.hwpf.model.TextPiece;
+import org.apache.poi.hwpf.usermodel.Paragraph;
+import org.apache.poi.hwpf.usermodel.Range;
+import org.apache.poi.poifs.filesystem.POIFSFileSystem;
+
+/**
+ * Class to extract the text from a Word Document.
+ * 
+ * You should use either getParagraphText() or getText() unless
+ *  you have a strong reason otherwise.
+ *
+ * @author Nick Burch (nick at torchbox dot com)
+ */
+public class WordExtractor {
+	private POIFSFileSystem fs;
+	private HWPFDocument doc;
+	
+	/**
+	 * Create a new Word Extractor
+	 * @param is InputStream containing the word file
+	 */
+	public WordExtractor(InputStream is) throws IOException {
+		this(new POIFSFileSystem(is));
+	}
+
+	/**
+	 * Create a new Word Extractor
+	 * @param fs POIFSFileSystem containing the word file
+	 */
+	public WordExtractor(POIFSFileSystem fs) throws IOException {
+		this.fs = fs;
+		doc = new HWPFDocument(fs);
+	}
+	
+	/**
+	 * Get the text from the word file, as an array with one String
+	 *  per paragraph
+	 */
+	public String[] getParagraphText() {
+		String[] ret;
+		
+		// Extract using the model code
+		try {
+	    	Range r = doc.getRange();
+
+			ret = new String[r.numParagraphs()];
+			for(int i=0; i<ret.length; i++) {
+				Paragraph p = r.getParagraph(i);
+				ret[i] = p.text();
+				
+				// Fix the line ending
+				if(ret[i].endsWith("\r")) {
+					ret[i] = ret[i] + "\n";
+				}
+			}
+		} catch(Exception e) {
+			// Something's up with turning the text pieces into paragraphs
+			// Fall back to ripping out the text pieces
+			ret = new String[1];
+			ret[0] = getTextFromPieces();
+		}
+		
+		return ret;
+	}
+	
+	/**
+	 * Grab the text out of the text pieces. Might also include various
+	 *  bits of crud, but will work in cases where the text piece -> paragraph
+	 *  mapping is broken. Fast too.
+	 */
+	public String getTextFromPieces() {
+    	StringBuffer textBuf = new StringBuffer();
+    	
+    	Iterator textPieces = doc.getTextTable().getTextPieces().iterator();
+    	while (textPieces.hasNext()) {
+    		TextPiece piece = (TextPiece) textPieces.next();
+
+    		String encoding = "Cp1252";
+    		if (piece.usesUnicode()) {
+    			encoding = "UTF-16LE";
+    		}
+    		try {
+    			String text = new String(piece.getRawBytes(), encoding);
+    			textBuf.append(text);
+    		} catch(UnsupportedEncodingException e) {
+    			throw new InternalError("Standard Encoding " + encoding + " not found, JVM broken");
+    		}
+    	}
+    	
+    	String text = textBuf.toString();
+    	
+    	// Fix line endings (Note - won't get all of them
+    	text = text.replaceAll("\r\r\r", "\r\n\r\n\r\n");
+    	text = text.replaceAll("\r\r", "\r\n\r\n");
+    	
+    	if(text.endsWith("\r")) {
+    		text += "\n";
+    	}
+    	
+    	return text;
+	}
+	
+	/*
+	 * Grab the text, based on the paragraphs. Shouldn't include any crud,
+	 *  but slightly slower than getTextFromPieces()
+	 */
+	public String getText() {
+		StringBuffer ret = new StringBuffer();
+		String[] text = getParagraphText();
+		for(int i=0; i<text.length; i++) {
+			ret.append(text[i]);
+		}
+		return ret.toString();
+	}
+}