java抓取网页数据

1. try{
2. URL url = new URL("http://www.51leba.com");
3. URLConnection conn = url.openConnection();
5. BufferedReader is = new BufferedReader(new InputStreamReader(conn.getInputStream()));
6. StringBuffer buffer = new StringBuffer();
7. String str;
8. while((str = is.readLine()) != null){
9. buffer.append(str);
10. buffer.append("/n");
12. }
13. str = buffer.toString().replaceAll("<script(.|/n)+?</script>", "").replaceAll("<(.|/n)+?>", "").replaceAll("&nbsp;", " ");
14. String[] s = str.split("/n");
15. buffer = new StringBuffer();
16. for(int i=0;i<s.length;i++){
17. if(s[i].trim().equals("") ){
18. continue;
19. }else{
20. buffer.append(s[i]);
21. buffer.append("/n");
22. }
23. }
24. System.out.println(buffer.toString());
26. is.close();
28. }catch (Exception e) {
29. e.printStackTrace();
30. }