1
|
package eu.dnetlib.usagestats.export;
|
2
|
|
3
|
import org.apache.hadoop.conf.Configuration;
|
4
|
import org.apache.hadoop.fs.FSDataOutputStream;
|
5
|
import org.apache.hadoop.fs.Path;
|
6
|
import org.apache.hadoop.fs.FileSystem;
|
7
|
import org.apache.log4j.Logger;
|
8
|
|
9
|
import java.io.*;
|
10
|
import java.net.URL;
|
11
|
import java.net.URLConnection;
|
12
|
import java.sql.PreparedStatement;
|
13
|
import java.sql.ResultSet;
|
14
|
import java.sql.Statement;
|
15
|
import java.text.SimpleDateFormat;
|
16
|
import java.util.Date;
|
17
|
import java.util.Calendar;
|
18
|
|
19
|
public class PiwikDownloadLogs {
|
20
|
|
21
|
private final String piwikUrl;
|
22
|
private Date startDate;
|
23
|
private final String tokenAuth;
|
24
|
|
25
|
/*
|
26
|
The Piwik's API method
|
27
|
*/
|
28
|
private final String APImethod = "?module=API&method=Live.getLastVisitsDetails";
|
29
|
private final String format = "&format=json";
|
30
|
|
31
|
private final Logger log = Logger.getLogger(this.getClass());
|
32
|
|
33
|
public PiwikDownloadLogs(String piwikUrl, String tokenAuth) {
|
34
|
this.piwikUrl = piwikUrl;
|
35
|
this.tokenAuth = tokenAuth;
|
36
|
|
37
|
}
|
38
|
|
39
|
private String getPiwikLogUrl() {
|
40
|
return "https://" + piwikUrl + "/";
|
41
|
}
|
42
|
|
43
|
private String getJson(String url) throws Exception {
|
44
|
try {
|
45
|
URL website = new URL(url);
|
46
|
URLConnection connection = website.openConnection();
|
47
|
|
48
|
//connection.setRequestProperty ("Authorization", "Basic "+encoded);
|
49
|
StringBuilder response;
|
50
|
try (BufferedReader in = new BufferedReader(new InputStreamReader(connection.getInputStream()))) {
|
51
|
response = new StringBuilder();
|
52
|
String inputLine;
|
53
|
while ((inputLine = in.readLine()) != null) {
|
54
|
response.append(inputLine);
|
55
|
response.append("\n");
|
56
|
}
|
57
|
}
|
58
|
return response.toString();
|
59
|
} catch (Exception e) {
|
60
|
log.error("Failed to get URL: " + e);
|
61
|
throw new Exception("Failed to get URL: " + e.toString(), e);
|
62
|
}
|
63
|
}
|
64
|
|
65
|
public void GetOpenAIRELogs(String repoLogsPath, String portalLogPath, String portalMatomoID) throws Exception {
|
66
|
|
67
|
Statement statement = ConnectDB.getConnection().createStatement();
|
68
|
|
69
|
ResultSet rs = statement.executeQuery("SELECT distinct piwik_id from public.datasource where piwik_id is not null order by piwik_id;");
|
70
|
while (rs.next()) {
|
71
|
int siteId = rs.getInt(1);
|
72
|
SimpleDateFormat simpleDateFormat = new SimpleDateFormat("YYYY-MM");
|
73
|
|
74
|
Calendar start = Calendar.getInstance();
|
75
|
start.set(Calendar.YEAR, 2016);
|
76
|
start.set(Calendar.MONTH, Calendar.MARCH);
|
77
|
//start.setTime(simpleDateFormat.parse("2016-01"));
|
78
|
|
79
|
Calendar end = Calendar.getInstance();
|
80
|
end.add(Calendar.DAY_OF_MONTH, -1);
|
81
|
|
82
|
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd");
|
83
|
PreparedStatement st = ConnectDB.DB_CONNECTION.prepareStatement("SELECT max(timestamp) FROM piwiklog WHERE source=? HAVING max(timestamp) is not null;");
|
84
|
st.setInt(1, siteId);
|
85
|
|
86
|
ResultSet rs_date = st.executeQuery();
|
87
|
while (rs_date.next()) {
|
88
|
if (rs_date.getString(1) != null && !rs_date.getString(1).equals("null") && !rs_date.getString(1).equals("")) {
|
89
|
start.setTime(sdf.parse(rs_date.getString(1)));
|
90
|
}
|
91
|
}
|
92
|
rs_date.close();
|
93
|
|
94
|
for (Date date = start.getTime(); start.before(end); start.add(Calendar.DATE, 1), date = start.getTime()) {
|
95
|
log.info("Downloading logs for repoid " + siteId + " and for " + sdf.format(date));
|
96
|
|
97
|
String period = "&period=day&date=" + sdf.format(date);
|
98
|
String outFolder = "";
|
99
|
//portal siteId = 109;
|
100
|
if (siteId == Integer.parseInt(portalMatomoID)) {
|
101
|
outFolder = portalLogPath;
|
102
|
} else {
|
103
|
outFolder = repoLogsPath;
|
104
|
}
|
105
|
FileSystem fs = FileSystem.get(new Configuration());
|
106
|
FSDataOutputStream fin = fs.create(new Path(outFolder + "/" + siteId + "_Piwiklog" + sdf.format((date)) + ".json"), true);
|
107
|
|
108
|
String baseApiUrl = getPiwikLogUrl() + APImethod + "&idSite=" + siteId + period + format + "&expanded=5&filter_limit=1000&token_auth=" + tokenAuth;
|
109
|
String content = "";
|
110
|
|
111
|
int i = 0;
|
112
|
|
113
|
while (!content.equals("[]\n")) {
|
114
|
String apiUrl = baseApiUrl;
|
115
|
|
116
|
if (i > 0) {
|
117
|
apiUrl += "&filter_offset=" + (i * 1000);
|
118
|
}
|
119
|
|
120
|
content = getJson(apiUrl);
|
121
|
|
122
|
fin.write(content.getBytes());
|
123
|
|
124
|
i++;
|
125
|
}
|
126
|
fin.close();
|
127
|
|
128
|
}
|
129
|
|
130
|
}
|
131
|
}
|