public void initial() throws ClassNotFoundException {
crawledSongs = new HashSet<String>();
sDao = new SongInfoDao();
ArrayList<Map<String, String>> songs = sDao.getResult("mid", "qqsonginfoc");
for (int i = 0; i < songs.size(); i++) {
crawledSongs.add(songs.get(i).get("mid"));
}
}
public void crawlSongInfo() throws ClassNotFoundException, URISyntaxException {
sDao = new SongInfoDao();
ArrayList<Map<String, String>> search = sDao.getResult("*", "qqsonginfos");
for (int i = 0; i < search.size(); i++) {
String id = search.get(i).get("id");
String sname = search.get(i).get("sname").trim().replace(" ", "%20");
URI uri = new URI(sname);
sname = uri.toASCIIString();
String singer = search.get(i).get("singer").trim().replace(" ", "%20");
// System.out.println(singer.length());
uri = new URI(singer);
singer = uri.toASCIIString();
String album = search.get(i).get("album").trim().replace(" ", "%20");
System.out.println("album:" + album);
uri = new URI(album);
album = uri.toASCIIString();
// System.out.println("album2:"+album);
System.out.println(id + sname + singer + album);
String url3 =
"http://s.plcloud.music.qq.com/fcgi-bin/smartbox.fcg?o_utf8=1&utf8=1&key="
+ sname
+ "%20"
+ singer
+ "&g_tk=5381&loginUin=0&hostUin=0&format=jsonp&inCharset=GB2312&outCharset=utf-8¬ice=0&platform=yqq&jsonpCallback=MusicJsonCallBack&needNewCode=0";
Document doc3 = null;
doc3 = this.getHtmlContent(url3);
String res = doc3.select("body").text();
int length = res.length();
res = res.substring(18, length - 2);
System.out.println("res:" + res);
JSONObject resObject = JSONObject.fromObject(res);
String tips = resObject.getString("tips");
// String t = "{\"song\":[],\"singer\":[],\"album\":[],\"mv\":[]}";
System.out.println("tips:" + tips);
JSONObject dataObject = JSONObject.fromObject(tips);
String song = dataObject.getString("song");
System.out.println("song:" + song);
// JSONObject songObject = JSONObject.fromObject(song);
// String itemlist = songObject.getString("itemlist");
if (song.length() > 3) {
JSONArray songArray = JSONArray.fromObject(song);
// JSONObject songObject = JSONObject.fromObject(song);
JSONObject midObject = (JSONObject) songArray.get(0);
String mid = midObject.getString("mid");
System.out.println("mid:" + mid);
String url4 =
"http://s.plcloud.music.qq.com/fcgi-bin/fcg_yqq_song_detail_info.fcg?songmid=" + mid;
Document doc4 = null;
doc4 = this.getHtmlContent(url4);
this.getAndSaveSongInfo(doc4, mid, id);
} else {
String url =
"http://soso.music.qq.com/fcgi-bin/multiple_music_search.fcg?mid=1&p=1&catZhida=1&lossless=0&t=100&utf8=1&w="
+ sname
+ "%20"
+ singer
+ "%20"
+ album;
// String url =
// "http://soso.music.qq.com/fcgi-bin/multiple_music_search.fcg?mid=1&p=1&catZhida=1&lossless=0&t=100&utf8=1&w=ϲ����%20BEYOND%20���ܾ���";
System.out.println(url);
Document doc = null;
doc = this.getHtmlContent(url);
String url2 = this.getURL(doc);
if (url2 != null) {
System.out.println("url2:" + url2);
String mid = url2.split("=")[1].split("&")[0];
// System.out.println(mid);
Document doc2 = null;
doc2 = this.getHtmlContent(url2);
this.getAndSaveSongInfo(doc2, mid, id);
}
// else sDao.updateSongInfo("notfound", id);
}
}
}