java通过POI技术将HTML文件转成Word

最新推荐文章于 2024-06-24 15:16:35 发布

cjccxy

最新推荐文章于 2024-06-24 15:16:35 发布

阅读量894

点赞数

分类专栏：导出Word技术文章标签： HTML CSS DOC转换 Apache POI 文件操作

本文链接：https://blog.csdn.net/cjccxy/article/details/118414036

版权

导出Word技术专栏收录该内容

2 篇文章 0 订阅

订阅专栏

public void htmlToWord2() throws Exception {
   InputStream bodyIs = new FileInputStream("f:\\1.html");
   InputStream cssIs = new FileInputStream("f:\\1.css");
   String body = this.getContent(bodyIs);
   String css = this.getContent(cssIs);
   //拼一个标准的HTML格式文档
   String content = "<html><head><style>" + css + "</style></head><body>" + body + "</body></html>";
   InputStream is = new ByteArrayInputStream(content.getBytes("GBK"));
   OutputStream os = new FileOutputStream("f:\\1.doc");
   this.inputStreamToWord(is, os);
}

/**
 * 把is写入到对应的word输出流os中
 * 不考虑异常的捕获，直接抛出
 * @param is
 * @param os
 * @throws IOException
 */
private void inputStreamToWord(InputStream is, OutputStream os) throws IOException {
   POIFSFileSystem fs = new POIFSFileSystem();
   //对应于org.apache.poi.hdf.extractor.WordDocument
   fs.createDocument(is, "WordDocument");
   fs.writeFilesystem(os);
   os.close();
   is.close();
}

/**
 * 把输入流里面的内容以UTF-8编码当文本取出。
 * 不考虑异常，直接抛出
 * @param ises
 * @return
 * @throws IOException
 */
private String getContent(InputStream... ises) throws IOException {
   if (ises != null) {
      StringBuilder result = new StringBuilder();
      BufferedReader br;
      String line;
      for (InputStream is : ises) {
         br = new BufferedReader(new InputStreamReader(is, "UTF-8"));
         while ((line=br.readLine()) != null) {
            result.append(line);
         }
      }
      return result.toString();
   }
   return null;
}