https://www.heckjj.com/index.php zh-cn https://www.heckjj.com/post// Heck <@hecks.tk> Thu, 17 Oct 2019 02:17:21 +0000 https://www.heckjj.com/post// private  void downloadPage(NewsItem newsItem,OutputStream output) throws MalformedURLException, IOException {         File baseDir =new File(basePath);         if(!baseDir.exists()) {             baseDir.mkdirs();         }         String portUrl=newsItem.getUrl();          Whitelist whiteList=new Whitelist();          whiteList.addTags("img","p");          whiteList.addAttributes("img","src");              Element contentBody=Jsoup.parse(new URL(portUrl).openStream(),"utf-8",portUrl).body().getElementById("ContentBody");          contentBody.select(".abstract").remove();          contentBody.select(".b-review").remove();          contentBody.select(".res-edit").remove();          Elements images=contentBody.select("img");                  File dataDir=new File(baseDir,DigestUtils.md5Hex(portUrl));          if(!dataDir.exists()) {              dataDir.mkdirs();          }          ZipOutputStream zip=new ZipOutputStream(output);          ZipEntry entry=null;          if(images.size()>0) {                    for(Element image:images) {                  String imgSrc=images.attr("abs:src");                                    image.attr("src", downloadImage(zip,imgSrc));              }          }          String text=Jsoup.clean(contentBody.html(), "",                  whiteList,                  new Document.OutputSettings().prettyPrint(true)          );          System.out.println(text);                   entry=new ZipEntry("index.html");          zip.putNextEntry(entry);          zip.write(text.getBytes());          zip.closeEntry();          entry=new ZipEntry("config.json");          zip.putNextEntry(entry);                  zip.write(new Gson().toJson(newsItem).getBytes());              zip.close();          zip.close();              }

    private String downloadImage(ZipOutputStream zip,String imgSrc) throws IOException {         String result="";         Connection conn=HttpConnection.connect(new URL(imgSrc)).ignoreContentType(true);         Response response=conn.execute();         String contentType=response.header("Content-Type");         if(contentType.contains("image/")) {             String fileExt=contentType.split("/")[1];             result="images/"+DigestUtils.md5Hex(imgSrc)+"."+fileExt;             ZipEntry entry=new ZipEntry(result);             zip.putNextEntry(entry);             zip.write(response.bodyAsBytes());             zip.closeEntry();         }         return result;     } ]]> https://www.heckjj.com/post//#blogcomment <user@domain.com> Thu, 01 Jan 1970 00:00:00 +0000 https://www.heckjj.com/post//#blogcomment