|
1
|
package eu.dnetlib.data.collector.plugins.HTTPWithFileName;
|
|
2
|
import java.util.Iterator;
|
|
3
|
|
|
4
|
import org.json.JSONObject;
|
|
5
|
import org.json.XML;
|
|
6
|
import org.junit.Ignore;
|
|
7
|
import org.junit.Test;
|
|
8
|
|
|
9
|
|
|
10
|
/**
|
|
11
|
* Created by miriam on 07/05/2018.
|
|
12
|
*/
|
|
13
|
public class HTTPWithFileNameTest {
|
|
14
|
|
|
15
|
private void iterate(Iterator<String> iterator, boolean exit){
|
|
16
|
try{
|
|
17
|
while (iterator.hasNext()){
|
|
18
|
|
|
19
|
System.out.println(iterator.next());
|
|
20
|
if(exit)
|
|
21
|
System.exit(0);
|
|
22
|
|
|
23
|
|
|
24
|
}
|
|
25
|
|
|
26
|
}catch(Exception ex){
|
|
27
|
ex.printStackTrace();
|
|
28
|
}
|
|
29
|
}
|
|
30
|
|
|
31
|
@Test
|
|
32
|
@Ignore
|
|
33
|
public void testRSCollectorFrontiers()
|
|
34
|
{
|
|
35
|
HTTPWithFileNameCollectorIterable rsc = new HTTPWithFileNameCollectorIterable("https://dev-openaire.d4science.org/RS/Frontiers/data/Frontiers/metadata/000/",null);
|
|
36
|
iterate(rsc.iterator(),false);
|
|
37
|
|
|
38
|
}
|
|
39
|
|
|
40
|
@Test
|
|
41
|
@Ignore
|
|
42
|
public void testRSCollectorPLOSCount()
|
|
43
|
{
|
|
44
|
HTTPWithFileNameCollectorIterable rsc = new HTTPWithFileNameCollectorIterable("https://dev-openaire.d4science.org/RS/PLOS/data/public_library_of_science/metadata/354/","article-type=\"correction\"");
|
|
45
|
Iterator<String> iterator = rsc.iterator();
|
|
46
|
int count = 0;
|
|
47
|
int body = 0;
|
|
48
|
int corrections = 0;
|
|
49
|
try{
|
|
50
|
while (iterator.hasNext()){
|
|
51
|
|
|
52
|
String meta = iterator.next();
|
|
53
|
if (!meta.contains("article-type=\"correction\"")){
|
|
54
|
count++;
|
|
55
|
int index = meta.indexOf("<body>");
|
|
56
|
if(meta.substring(index).contains("<sec"))
|
|
57
|
body++;
|
|
58
|
else {
|
|
59
|
System.out.println(meta);
|
|
60
|
System.out.println(count);
|
|
61
|
}
|
|
62
|
|
|
63
|
}else
|
|
64
|
corrections++;
|
|
65
|
|
|
66
|
}
|
|
67
|
System.out.println(count + " " + body + " " + corrections);
|
|
68
|
}catch(Exception ex){
|
|
69
|
ex.printStackTrace();
|
|
70
|
}
|
|
71
|
}
|
|
72
|
|
|
73
|
@Test
|
|
74
|
@Ignore
|
|
75
|
public void testRSCollectorPLOS()
|
|
76
|
{
|
|
77
|
HTTPWithFileNameCollectorIterable rsc = new HTTPWithFileNameCollectorIterable("https://dev-openaire.d4science.org/RS/PLOS/data/public_library_of_science/metadata/400/","article-type=\"correction\"");
|
|
78
|
|
|
79
|
|
|
80
|
iterate(rsc.iterator(),false);
|
|
81
|
}
|
|
82
|
|
|
83
|
@Test
|
|
84
|
@Ignore
|
|
85
|
public void testRSCollectorSpringer()
|
|
86
|
{
|
|
87
|
HTTPWithFileNameCollectorIterable rsc = new HTTPWithFileNameCollectorIterable("https://dev-openaire.d4science.org/RS/Springer-OA/data/Springer-OA/metadata/8a0/",null);
|
|
88
|
|
|
89
|
iterate(rsc.iterator(),false);
|
|
90
|
|
|
91
|
}
|
|
92
|
|
|
93
|
@Test
|
|
94
|
public void testEmptyCollection()
|
|
95
|
{
|
|
96
|
HTTPWithFileNameCollectorIterable rsc = new HTTPWithFileNameCollectorIterable("",null);
|
|
97
|
|
|
98
|
iterate(rsc.iterator(),true);
|
|
99
|
}
|
|
100
|
|
|
101
|
|
|
102
|
|
|
103
|
@Test
|
|
104
|
public void testJsonToXml() {
|
|
105
|
System.out.println(XML.toString(new JSONObject("{'resource':{'DOI':'JUNK','url':'asd'}}")));
|
|
106
|
|
|
107
|
}
|
|
108
|
|
|
109
|
}
|