HTML und Regular Expressions

1package app.mensa;
2
3import android.app.Activity;
4import android.view.*;
5import android.view.View.OnClickListener;
6import android.os.Bundle;
7import android.widget.Button;
8import android.widget.TextView;
9import java.net.*;
10import java.util.LinkedList;
11import java.util.ListIterator;
12import java.io.*;
13
14public class Mensa extends Activity implements OnClickListener{
15 private TextView twTest;
16 private TextView twCount;
17 private Button filterBtn;
18 private LinkedList<String> readHTML;
19 /** Called when the activity is first created. */
20 @Override
21 public void onCreate(Bundle savedInstanceState) {
22 super.onCreate(savedInstanceState);
23 setContentView(R.layout.main);
24 twTest = (TextView) findViewById(R.id.twTest);
25 twCount = (TextView) findViewById(R.id.TextView01);
26 filterBtn = (Button) findViewById(R.id.filterBtn);
27 filterBtn.setOnClickListener(this);
28 readHTML = new LinkedList<String>();
29 readMensaplan();
30 }
31
32 public void onClick(View v) {
33 switch (v.getId()) {
34 case R.id.filterBtn:
35 filterHTML();
36 }
37 }
38
39 public void filterHTML() {
40 ListIterator<String> it = readHTML.listIterator();
41 while(it.hasNext()) {
42 String s = (String) it.next();
43 if(!s.matches(".*(<td|<table|<tr|<th).*"))
44 it.remove();
45 else {
46 s = s.replaceAll("<[^<]+?>", "");
47 //s = hexToChar(s);
48 it.set(s);
49 }
50 }
51 twTest.setText(""+readHTML.size() + " " + readHTML.toString());
52 twCount.setText(""+readHTML.size() + " " + readHTML.get(5));
53 }
54
55 /*public String hexToChar(String input) {
56 input = input.replaceAll("&#228;", "ä");
57 input = input.replaceAll("–", "-");
58 input = input.replaceAll("&auml;", "ä");
59 input = input.replaceAll("&euro;", "€");
60 return input;
61 }*/
62
63 public void readMensaplan() {
64 try {
65 URL mensaPlanBs1 = new URL("http://www.sw-bs.de/braunschweig/essen/menus/mensa-1");
66 BufferedReader in = new BufferedReader(
67 new InputStreamReader(
68 mensaPlanBs1.openStream()));
69 String inputLine;
70 while ((inputLine = in.readLine()) != null) {
71 inputLine.trim();
72 readHTML.add(inputLine);
73 }
74 twTest.setText(""+readHTML.size());
75 in.close();
76 }
77 catch (IOException e) {
78 e.printStackTrace();
79 }
80 }
81}

1package de.code.worker.lotto.helferlein;
2
3import java.io.BufferedReader;
4import java.io.DataInputStream;
5import java.io.InputStreamReader;
6import java.net.URL;
7import java.net.URLConnection;
8import java.util.regex.Matcher;
9import java.util.regex.Pattern;
10
11import android.app.Activity;
12import android.os.Bundle;
13import android.widget.TextView;
14
15public class WelcomeScreenActivity extends Activity {
16
17 private TextView contentView;
18
19 /** Called when the activity is first created. */
20 @Override
21 public void onCreate(Bundle savedInstanceState) {
22 super.onCreate(savedInstanceState);
23 setContentView(R.layout.welcomescreenactivity);
24
25 contentView = (TextView)findViewById(R.id.twTest);
26
27 try {
28 URL url_ = new URL("http://m.lotto.de");
29 URLConnection conn = url_.openConnection();
30 DataInputStream in = new DataInputStream ( conn.getInputStream ( ) ) ;
31 BufferedReader d = new BufferedReader(new InputStreamReader(in, "utf-8"));
32 String content = "";
33 while(d.ready()) {
34 content += d.readLine();
35 }
36 String source = content;
37 Pattern p = Pattern.compile("(<span\b[^>]*>(.*?)</span>) (<span\b[^>]*>(.*?)</span>)");
38 Matcher m = p.matcher(source);
39 while (m.find()) {
40 String lottozahlen1 = m.group(1);
41 String lottozahlen2 = m.group(2);
42 contentView.setText(lottozahlen1 + lottozahlen2);
43 }
44
45 }
46 catch(Exception ex) {
47 contentView.setText("Exception:" + ex.toString() + " = " + ex.getMessage());
48 }
49
50 }
51}