1 | |
package org.paneris.bibliomania.metasearch.bob; |
2 | |
|
3 | |
import java.io.FileOutputStream; |
4 | |
import java.io.IOException; |
5 | |
import java.io.OutputStream; |
6 | |
import java.net.URL; |
7 | |
|
8 | |
import org.apache.oro.text.regex.MalformedPatternException; |
9 | |
import org.apache.oro.text.regex.Perl5Compiler; |
10 | |
import org.apache.oro.text.regex.Perl5Matcher; |
11 | |
import org.apache.oro.text.regex.Perl5Pattern; |
12 | |
import org.apache.oro.text.regex.StringSubstitution; |
13 | |
import org.apache.oro.text.regex.Util; |
14 | |
import org.melati.util.IoUtils; |
15 | |
import org.melati.util.MelatiRuntimeException; |
16 | |
import org.melati.util.UnexpectedExceptionException; |
17 | |
import org.paneris.bibliomania.metasearch.util.HackParser; |
18 | |
import org.paneris.bibliomania.metasearch.util.TimeoutThread; |
19 | |
|
20 | 0 | public class Session { |
21 | |
public static final String frontPageURL = |
22 | |
"http://bookshop.blackwell.co.uk/cgi-bin/BOB2?BV_Operation=Dyn_LoginReceive&BV_ServiceName=BOB&form%25login_type=guest&form%25failure=/service_error.html&form%25success=/bob/bob_main.html.tmpl"; |
23 | |
|
24 | |
public String sessionID, engineID; |
25 | |
|
26 | 0 | public Session(String sessionID, String engineID) { |
27 | 0 | this.sessionID = sessionID; |
28 | 0 | this.engineID = engineID; |
29 | 0 | } |
30 | |
|
31 | |
public String queryParams() { |
32 | 0 | return "BV_SessionID=" + sessionID + "&BV_EngineID=" + engineID; |
33 | |
} |
34 | |
|
35 | |
public String toString() { |
36 | 0 | return queryParams(); |
37 | |
} |
38 | |
|
39 | 0 | private static final Object bobSessionMon = new Object(); |
40 | |
|
41 | 0 | private static Session bobSession = null; |
42 | |
|
43 | 0 | private static Exception bobSessionException = null; |
44 | |
|
45 | |
public static class NoSessionIDException extends MelatiRuntimeException { |
46 | |
|
47 | |
|
48 | |
|
49 | |
private static final long serialVersionUID = 1L; |
50 | |
|
51 | |
public String getMessage() { |
52 | |
return "Couldn't extract session ID from front page"; |
53 | |
} |
54 | |
} |
55 | |
|
56 | |
private static void newSession() throws OpenException { |
57 | 0 | System.err.println("BOB: establishing new session"); |
58 | |
try { |
59 | 0 | TimeoutThread timeout = TimeoutThread.forCurrentThread(20000); |
60 | |
try { |
61 | 0 | FrontPage fp = |
62 | |
new FrontPage(IoUtils.slurp(new URL(frontPageURL), 32768)); |
63 | 0 | if (fp.sessionID == null) |
64 | 0 | throw new NoSessionIDException(); |
65 | |
|
66 | 0 | bobSession = new Session(fp.sessionID, fp.engineID); |
67 | 0 | bobSessionException = null; |
68 | |
} finally { |
69 | 0 | timeout.stop(); |
70 | 0 | } |
71 | 0 | } catch (Exception e) { |
72 | 0 | bobSessionException = e; |
73 | 0 | throw new OpenException(e); |
74 | 0 | } |
75 | 0 | } |
76 | |
|
77 | |
private static class SessionRenewer extends Thread { |
78 | 0 | public SessionRenewer() { |
79 | 0 | setDaemon(true); |
80 | 0 | } |
81 | |
|
82 | |
public void run() { |
83 | |
try { |
84 | |
for (;;) { |
85 | |
try { |
86 | 0 | newSession(); |
87 | 0 | } catch (Exception e) { |
88 | 0 | System.err.println("BOB SessionRenewer:"); |
89 | 0 | e.printStackTrace(); |
90 | |
} finally { |
91 | 0 | synchronized (bobSessionMon) { |
92 | 0 | bobSessionMon.notifyAll(); |
93 | 0 | } |
94 | 0 | } |
95 | |
|
96 | 0 | Thread.sleep(5 * 60 * 1000); |
97 | |
} |
98 | 0 | } catch (InterruptedException e) { |
99 | |
} finally { |
100 | 0 | sessionRenewer = null; |
101 | 0 | } |
102 | 0 | } |
103 | |
} |
104 | |
|
105 | 0 | private static SessionRenewer sessionRenewer = null; |
106 | |
|
107 | |
private static synchronized void checkSessionRenewer() { |
108 | 0 | if (sessionRenewer == null) |
109 | 0 | (sessionRenewer = new SessionRenewer()).start(); |
110 | 0 | } |
111 | |
|
112 | |
static Session bobSession() { |
113 | 0 | checkSessionRenewer(); |
114 | |
|
115 | 0 | if (bobSession == null) { |
116 | |
try { |
117 | 0 | synchronized (bobSessionMon) { |
118 | 0 | if (bobSession == null) |
119 | 0 | bobSessionMon.wait(); |
120 | 0 | } |
121 | 0 | } catch (InterruptedException e) { |
122 | 0 | throw new OpenException(e); |
123 | 0 | } |
124 | |
} |
125 | |
|
126 | 0 | if (bobSession == null) |
127 | 0 | throw new OpenException(bobSessionException); |
128 | |
|
129 | 0 | return bobSession; |
130 | |
} |
131 | |
|
132 | |
private static Perl5Pattern bv__IDParam; |
133 | |
|
134 | |
static { |
135 | |
try { |
136 | 0 | bv__IDParam = |
137 | |
(Perl5Pattern) (new Perl5Compiler()).compile( |
138 | |
"BV_(Session|Engine)ID=[^&]*&?", |
139 | |
0); |
140 | 0 | } catch (MalformedPatternException e) { |
141 | 0 | throw new UnexpectedExceptionException(e); |
142 | 0 | } |
143 | |
} |
144 | |
|
145 | 0 | private static final StringSubstitution empty = new StringSubstitution(); |
146 | |
|
147 | |
public static String urlInCurrentSession(String url) { |
148 | 0 | String urlWithoutSession = |
149 | |
Util.substitute( |
150 | |
new Perl5Matcher(), |
151 | |
bv__IDParam, |
152 | |
empty, |
153 | |
url, |
154 | |
Util.SUBSTITUTE_ALL); |
155 | |
|
156 | 0 | return urlWithoutSession + '&' + bobSession().queryParams(); |
157 | |
} |
158 | |
|
159 | 0 | private static String contentDumpPrefix = null; |
160 | 0 | private static int contentDumpIndex = 0; |
161 | |
|
162 | |
private static byte[] contentUsingCurrentSession( |
163 | |
String url, |
164 | |
int estimate, |
165 | |
int limit) |
166 | |
throws IOException { |
167 | 0 | byte[] content = IoUtils.slurp(new URL(urlInCurrentSession(url)), 32768); |
168 | |
|
169 | 0 | String dp = contentDumpPrefix; |
170 | 0 | if (dp != null) { |
171 | |
try { |
172 | 0 | OutputStream o = |
173 | |
new FileOutputStream(contentDumpPrefix + contentDumpIndex++ +".html"); |
174 | |
try { |
175 | 0 | o.write(content); |
176 | |
} finally { |
177 | 0 | try { |
178 | 0 | o.close(); |
179 | 0 | } catch (IOException e) { |
180 | 0 | } |
181 | 0 | } |
182 | 0 | } catch (Exception e) { |
183 | 0 | System.err.println("Can't write dump!"); |
184 | 0 | } |
185 | |
} |
186 | |
|
187 | 0 | return content; |
188 | |
} |
189 | |
|
190 | 0 | private static final byte[] timeoutExplan = |
191 | |
"current session timing out".getBytes(); |
192 | |
|
193 | |
public static byte[] content(String url, int estimate, int limit) |
194 | |
throws IOException { |
195 | 0 | byte[] it = contentUsingCurrentSession(url, estimate, limit); |
196 | 0 | if (HackParser.indexOf(it, timeoutExplan) != -1) { |
197 | 0 | newSession(); |
198 | 0 | return contentUsingCurrentSession(url, estimate, limit); |
199 | |
} else |
200 | 0 | return it; |
201 | |
} |
202 | |
} |