docx poi 原理_POI实现DOC/DOCX转HTML

public classDocToHtml {private static final String encoding = "UTF-8";public staticString convert2Html(String wordPath)throwsFileNotFoundException, TransformerException, IOException,

ParserConfigurationException {if( wordPath == null || "".equals(wordPath) ) return "";

File file= newFile(wordPath);if( file.exists() &&file.isFile() )return convert2Html(newFileInputStream(file));else

return "";

}public staticString convert2Html(String wordPath, String context)throwsFileNotFoundException, TransformerException, IOException,

ParserConfigurationException {if( wordPath == null || "".equals(wordPath) ) return "";

File file= newFile(wordPath);if( file.exists() &&file.isFile() )return convert2Html(newFileInputStream(file), context);else

return "";

}public staticString convert2Html(InputStream is)throwsTransformerException, IOException,

ParserConfigurationException {return convert2Html(is, "");

}public static String convert2Html(InputStream is, HttpServletRequest req) throwsTransformerException, IOException, ParserConfigurationException {returnconvert2Html(is, req.getContextPath());

}public static String convert2Html(InputStream is, final String context) throwsIOException, ParserConfigurationException, TransformerException {

HWPFDocument wordDocument= newHWPFDocument(is);

WordToHtmlConverter converter= newWordToHtmlConverter(

DocumentBuilderFactory.newInstance().newDocumentBuilder()

.newDocument());

SimpleDateFormat sdf= new SimpleDateFormat("yyyyMMddHHmmssSSS");final String prefix = sdf.format(newDate());final Map suffixMap = new HashMap();

converter.setPicturesManager(newPicturesManager() {public String savePicture(byte[] content, PictureType pictureType,

String suggestedName,float widthInches, floatheightInches) {

String prefixContext= context.replace("\\", "").replace("/", "");

prefixContext= StringUtils.isNotBlank(prefixContext) ? "/" + prefixContext + "/": prefixContext;

suffixMap.put(new String(content).replace(" ", "").length(), suggestedName);returnprefixContext+ UeConstants.VIEW_IMAGE_PATH + "/" +UeConstants.UEDITOR_PATH+ "/" + UeConstants.UEDITOR_IMAGE_PATH + "/"

+ prefix + "_"

+suggestedName;

}

});

converter.processDocument(wordDocument);

List pics =wordDocument.getPicturesTable().getAllPictures();if (pics != null) {for(Picture pic : pics) {try{

pic.writeImageContent(newFileOutputStream(

UeConstants.IMAGE_PATH+ "/" + prefix + "_" + suffixMap.get(new String(pic.getContent()).replace(" ", "").length())));

}catch(FileNotFoundException e) {

e.printStackTrace();

}

}

}

StringWriter writer= newStringWriter();

Transformer serializer=TransformerFactory.newInstance().newTransformer();

serializer.setOutputProperty(OutputKeys.ENCODING, encoding);

serializer.setOutputProperty(OutputKeys.INDENT,"yes");

serializer.setOutputProperty(OutputKeys.METHOD,"html");

serializer.transform(newDOMSource(converter.getDocument()),newStreamResult(writer) );

writer.close();returnwriter.toString();

}

}