java读取word文档

需要引用apache的poi开源项目包；下载地址：http://poi.apache.org/download.html

引入6个jar包：poi-3.8-20120326.jar；poi-ooxml-3.8-20120326.jar；poi-ooxml-schemas-3.8-20120326.jar；poi-scratchpad-3.8-20120326.jar；xmlbeans-2.3.0.jar；dom4j-1.6.1.jar。

    public static String readWord(String filePath){
        String text = "";
        File file = new File(filePath);
        //2003
        if(file.getName().endsWith(".doc")){
            try {
                FileInputStream stream = new FileInputStream(file);
                WordExtractor word = new WordExtractor(stream);
                text = word.getText();
                //去掉word文档中的多个换行
                text = text.replaceAll("(\\r\\n){2,}", "\r\n");
                text = text.replaceAll("(\\n){2,}", "\n");
                stream.close();
            } catch (Exception e) {
                e.printStackTrace();
            }

        }else if(file.getName().endsWith(".docx")){       //2007
            try {
                OPCPackage oPCPackage = POIXMLDocument.openPackage(filePath);
                XWPFDocument xwpf = new XWPFDocument(oPCPackage);
                POIXMLTextExtractor ex = new XWPFWordExtractor(xwpf);
                text = ex.getText();
                //去掉word文档中的多个换行
                text = text.replaceAll("(\\r\\n){2,}", "\r\n");
                text = text.replaceAll("(\\n){2,}", "\n");
                System.out.println("ok");
            } catch (Exception e) {
                e.printStackTrace();
            }
        }
        return text;
    }