| 1 |
package dk.thoerup.spejdernetscraper;
|
| 2 |
|
| 3 |
import java.util.Map;
|
| 4 |
import java.util.TreeMap;
|
| 5 |
|
| 6 |
import org.apache.commons.lang3.exception.ExceptionUtils;
|
| 7 |
import org.jsoup.Connection.Response;
|
| 8 |
import org.jsoup.Jsoup;
|
| 9 |
import org.jsoup.nodes.Document;
|
| 10 |
import org.jsoup.nodes.FormElement;
|
| 11 |
|
| 12 |
import dk.thoerup.genericjavautils.HttpUtil;
|
| 13 |
|
| 14 |
public class ScraperWorker implements Runnable {
|
| 15 |
|
| 16 |
@Override
|
| 17 |
public void run() {
|
| 18 |
|
| 19 |
ResultBuffer rb = ResultBuffer.getResultBuffer();
|
| 20 |
try {
|
| 21 |
|
| 22 |
rb.clear();
|
| 23 |
rb.addString("Starting");
|
| 24 |
|
| 25 |
|
| 26 |
Response res = Jsoup.connect("http://medlemssystem.spejdernet.dk/login.aspx?ReturnUrl=%2f").execute();
|
| 27 |
Map<String, String> cookies = res.cookies();
|
| 28 |
|
| 29 |
Document loginPage = res.parse();
|
| 30 |
|
| 31 |
rb.addString("Got login page");
|
| 32 |
|
| 33 |
|
| 34 |
|
| 35 |
FormElement loginForm = (FormElement) loginPage.getElementById("aspnetForm");
|
| 36 |
loginForm.getElementById("ctl00_main_loginForm__txtUserName").attr("value","torbenhoerupnielsen");
|
| 37 |
loginForm.getElementById("ctl00_main_loginForm__txtPassword").attr("value","Pwspejder2013");
|
| 38 |
System.out.println( loginForm.formData() );
|
| 39 |
|
| 40 |
res = loginForm.submit().execute();
|
| 41 |
//System.out.println( res.body() ); //DEBUG
|
| 42 |
|
| 43 |
rb.addString("Login OK");
|
| 44 |
//This will get you cookies
|
| 45 |
cookies.putAll( res.cookies() );
|
| 46 |
//printMap(cookies);
|
| 47 |
|
| 48 |
// /////////////////////////////////
|
| 49 |
// Data.csv
|
| 50 |
|
| 51 |
String dataCsvUrl = "http://medlemssystem.spejdernet.dk/csvx.ashx?q=JQBOZXRtZXN0ZXIuS0ZVTS5NZW1iZXIsIE5ldG1lc3Rlci5LRlVNPgB4LlVuaXQuR3JvdXAuSWQgPT0gImd1aWQ6NWMwY2Y2MzgtZGM4Ny00MTNmLTkxYjYtOWM3MzAwYzcwMTM3Ig==&type=Member&view=csv";
|
| 52 |
//And this is the easieste way I've found to remain in session
|
| 53 |
Response data = Jsoup.connect(dataCsvUrl).cookies(cookies).timeout(10000).execute();
|
| 54 |
if ( ! data.contentType().equalsIgnoreCase("text/x-csv; charset=iso-8859-1") ) {
|
| 55 |
|
| 56 |
rb.addString("Data.csv - has the wrong content type: " + data.contentType());
|
| 57 |
return;
|
| 58 |
}
|
| 59 |
String dataCsv = data.body();
|
| 60 |
rb.addString("Got data.csv");
|
| 61 |
|
| 62 |
// /////////////////////////////////
|
| 63 |
// Roller.csv
|
| 64 |
/*
|
| 65 |
String rollerCsvUrl = "http://medlemssystem.spejdernet.dk/DataExport.aspx?Id=5c0cf638-dc87-413f-91b6-9c7300c70137&listid=e8c5ae9d-5ea7-4a00-bea4-a0ce00ea891e&execute=true";
|
| 66 |
Response roller = Jsoup.connect(rollerCsvUrl).cookies(cookies).timeout(10000).execute();
|
| 67 |
if ( ! roller.contentType().equalsIgnoreCase("text/x-csv; charset=iso-8859-1") ) {
|
| 68 |
rb.addString("Roller.csv - has the wrong content type: " + roller.contentType());
|
| 69 |
return;
|
| 70 |
}
|
| 71 |
String rollerCsv = roller.body();
|
| 72 |
rb.addString("Got roller.csv");*/
|
| 73 |
String rollerCsv = ""; //Temporary fix for PrematureEof exception thingie
|
| 74 |
|
| 75 |
Map<String,String> postData = new TreeMap<String,String>();
|
| 76 |
postData.put("roller", rollerCsv);
|
| 77 |
postData.put("data", dataCsv);
|
| 78 |
|
| 79 |
String params = HttpUtil.encodeParams( postData );
|
| 80 |
byte resp[] = HttpUtil.postContent("http://horsensspejder.t-hoerup.dk/data/postdata.php", params, 3000);
|
| 81 |
|
| 82 |
String response = new String(resp);
|
| 83 |
rb.addString("POST Response: " + response);
|
| 84 |
|
| 85 |
rb.addString("Done !");
|
| 86 |
|
| 87 |
} catch (Exception e) {
|
| 88 |
rb.addString("Error occurred: " + e.toString() );
|
| 89 |
rb.addString( ExceptionUtils.getStackTrace(e) );
|
| 90 |
}
|
| 91 |
}
|
| 92 |
|
| 93 |
static void printMap(Map<String,String> map) {
|
| 94 |
for(String key : map.keySet()) {
|
| 95 |
System.out.println( key + ": " +map.get(key) );
|
| 96 |
}
|
| 97 |
|
| 98 |
}
|
| 99 |
|
| 100 |
public static void main(String[] args) throws Exception {
|
| 101 |
new ScraperWorker().run();
|
| 102 |
|
| 103 |
System.out.println( ResultBuffer.getResultBuffer().getString() );
|
| 104 |
}
|
| 105 |
|
| 106 |
|
| 107 |
/*
|
| 108 |
*
|
| 109 |
* Just fore reference, here's the old login method
|
| 110 |
res = Jsoup
|
| 111 |
.connect("http://medlemssystem.spejdernet.dk/Login.aspx?changeuser=1")
|
| 112 |
.data("ctl00$main$loginForm$_txtUserName", "<enter-username-here>",
|
| 113 |
"ctl00$main$loginForm$_txtPassword", "<enter-password-here>",
|
| 114 |
"ctl00$main$loginForm$_buttonLogin", "Login",
|
| 115 |
"__EVENTTARGET", "",
|
| 116 |
"__EVENTARGUMENT", "",
|
| 117 |
"__VIEWSTATE", viewState.attr("value"),
|
| 118 |
"ctl00$ctl04$hiddenTab", ""
|
| 119 |
)
|
| 120 |
.method(Method.POST)
|
| 121 |
.cookies(cookies)
|
| 122 |
.execute();*/
|
| 123 |
|
| 124 |
}
|
Parent Directory
| 
Revision Log