1 |
36212
|
sandro.lab
|
/**
|
2 |
|
|
*
|
3 |
|
|
*/
|
4 |
|
|
package eu.dnetlib.download.plugin;
|
5 |
|
|
|
6 |
40523
|
claudio.at
|
import java.util.ArrayList;
|
7 |
|
|
import java.util.List;
|
8 |
|
|
|
9 |
36212
|
sandro.lab
|
import com.google.common.base.Function;
|
10 |
|
|
import com.google.common.collect.Iterables;
|
11 |
|
|
import com.google.gson.Gson;
|
12 |
40045
|
sandro.lab
|
import eu.dnetlib.data.download.rmi.AbstractDownloadPlugin;
|
13 |
36212
|
sandro.lab
|
import eu.dnetlib.data.download.rmi.DownloadItem;
|
14 |
|
|
import eu.dnetlib.data.download.rmi.DownloadPlugin;
|
15 |
40523
|
claudio.at
|
import eu.dnetlib.data.download.rmi.DownloadPluginException;
|
16 |
40045
|
sandro.lab
|
import org.apache.commons.logging.Log;
|
17 |
|
|
import org.apache.commons.logging.LogFactory;
|
18 |
36212
|
sandro.lab
|
|
19 |
|
|
|
20 |
|
|
/**
|
21 |
|
|
* @author sandro
|
22 |
|
|
*/
|
23 |
|
|
public class HALPdfDocumentPlugin extends AbstractDownloadPlugin implements DownloadPlugin {
|
24 |
|
|
|
25 |
40045
|
sandro.lab
|
private static final Log log = LogFactory.getLog(HALPdfDocumentPlugin.class); // NOPMD by marko on 11/24/08 5:02 PM
|
26 |
36410
|
sandro.lab
|
|
27 |
|
|
|
28 |
40045
|
sandro.lab
|
/**
|
29 |
|
|
* {@inheritDoc}
|
30 |
|
|
*
|
31 |
|
|
* @see eu.dnetlib.data.download.rmi.DownloadPlugin#getPluginName()
|
32 |
|
|
*/
|
33 |
|
|
@Override
|
34 |
|
|
public String getPluginName() {
|
35 |
|
|
// TODO Auto-generated method stub
|
36 |
|
|
return "HALPdfDocumentPlugin";
|
37 |
|
|
}
|
38 |
36212
|
sandro.lab
|
|
39 |
40045
|
sandro.lab
|
/**
|
40 |
|
|
* {@inheritDoc}
|
41 |
|
|
*
|
42 |
|
|
* @see eu.dnetlib.data.download.rmi.DownloadPlugin#retrieveUrls(java.lang.Iterable)
|
43 |
|
|
*/
|
44 |
|
|
@Override
|
45 |
|
|
public Iterable<DownloadItem> retrieveUrls(final Iterable<DownloadItem> urls) {
|
46 |
|
|
return Iterables.transform(urls, new Function<DownloadItem, DownloadItem>() {
|
47 |
36212
|
sandro.lab
|
|
48 |
40045
|
sandro.lab
|
@Override
|
49 |
|
|
public DownloadItem apply(final DownloadItem input) {
|
50 |
|
|
return retrieveUrl(input);
|
51 |
|
|
}
|
52 |
|
|
});
|
53 |
|
|
}
|
54 |
36212
|
sandro.lab
|
|
55 |
40045
|
sandro.lab
|
/**
|
56 |
|
|
* {@inheritDoc}
|
57 |
|
|
*
|
58 |
|
|
* @see eu.dnetlib.data.download.rmi.DownloadPlugin#retrieveUrl(eu.dnetlib.data.download.rmi.DownloadItem)
|
59 |
|
|
*/
|
60 |
|
|
@Override
|
61 |
|
|
public DownloadItem retrieveUrl(final DownloadItem input) {
|
62 |
|
|
try {
|
63 |
|
|
if (checkOpenAccess(input) == null) return null;
|
64 |
|
|
if (input == null) return null;
|
65 |
|
|
String url = input.getOriginalUrl();
|
66 |
36212
|
sandro.lab
|
|
67 |
40045
|
sandro.lab
|
if ((url == null) || (url.trim().length() == 0)) return input;
|
68 |
|
|
@SuppressWarnings("unchecked")
|
69 |
|
|
List<String> urls = new Gson().fromJson(url, ArrayList.class);
|
70 |
|
|
if ((urls == null) || (urls.size() == 0)) return input;
|
71 |
|
|
if (checkUrlsNotNull(input, urls)) return input;
|
72 |
|
|
input.setOriginalUrl(null);
|
73 |
|
|
input.setUrl(null);
|
74 |
|
|
return input;
|
75 |
|
|
} catch (Throwable e) {
|
76 |
|
|
log.error("Error on retreiving URL", e);
|
77 |
|
|
input.setOriginalUrl(null);
|
78 |
|
|
input.setUrl(null);
|
79 |
|
|
return input;
|
80 |
|
|
}
|
81 |
|
|
}
|
82 |
36212
|
sandro.lab
|
|
83 |
40045
|
sandro.lab
|
/**
|
84 |
|
|
* {@inheritDoc}
|
85 |
|
|
*
|
86 |
|
|
* @see eu.dnetlib.data.download.rmi.DownloadPlugin#setBasePath(java.lang.String)
|
87 |
|
|
*/
|
88 |
|
|
@Override
|
89 |
|
|
public void setBasePath(final String arg0) {
|
90 |
|
|
// TODO Auto-generated method stub
|
91 |
36212
|
sandro.lab
|
|
92 |
40045
|
sandro.lab
|
}
|
93 |
36212
|
sandro.lab
|
|
94 |
40045
|
sandro.lab
|
@Override
|
95 |
40523
|
claudio.at
|
public String extractURL(String baseURL) throws DownloadPluginException {
|
96 |
40045
|
sandro.lab
|
return baseURL.toLowerCase().endsWith("document") ? baseURL : null;
|
97 |
|
|
}
|
98 |
36212
|
sandro.lab
|
}
|