1 |
33251
|
sandro.lab
|
package eu.dnetlib.download.plugin;
|
2 |
|
|
|
3 |
40523
|
claudio.at
|
import java.io.File;
|
4 |
|
|
import java.nio.file.Path;
|
5 |
|
|
import java.nio.file.Paths;
|
6 |
|
|
import java.util.List;
|
7 |
|
|
|
8 |
33251
|
sandro.lab
|
import com.google.common.collect.Iterables;
|
9 |
|
|
import com.google.gson.Gson;
|
10 |
40045
|
sandro.lab
|
import eu.dnetlib.data.download.rmi.AbstractDownloadPlugin;
|
11 |
33251
|
sandro.lab
|
import eu.dnetlib.data.download.rmi.DownloadItem;
|
12 |
|
|
import eu.dnetlib.data.download.rmi.DownloadPlugin;
|
13 |
40523
|
claudio.at
|
import eu.dnetlib.data.download.rmi.DownloadPluginException;
|
14 |
50842
|
sandro.lab
|
import org.apache.commons.lang3.StringUtils;
|
15 |
40045
|
sandro.lab
|
import org.apache.commons.logging.Log;
|
16 |
|
|
import org.apache.commons.logging.LogFactory;
|
17 |
33251
|
sandro.lab
|
|
18 |
|
|
public class ArxivImportFromFile extends AbstractDownloadPlugin implements DownloadPlugin {
|
19 |
|
|
|
20 |
40045
|
sandro.lab
|
private static final Log log = LogFactory.getLog(ArxivImportFromFile.class); // NOPMD by marko on 11/24/08 5:02 PM
|
21 |
33900
|
sandro.lab
|
|
22 |
40045
|
sandro.lab
|
/**
|
23 |
|
|
* The base path.
|
24 |
|
|
*/
|
25 |
|
|
private String basePath;
|
26 |
33251
|
sandro.lab
|
|
27 |
40045
|
sandro.lab
|
@Override
|
28 |
|
|
public DownloadItem retrieveUrl(final DownloadItem item) {
|
29 |
|
|
if (checkOpenAccess(item) == null) return null;
|
30 |
|
|
final String baseURLs = item.getUrl();
|
31 |
|
|
final List<String> urlsList = new Gson().fromJson(baseURLs, List.class);
|
32 |
|
|
for (final String baseURL : urlsList) {
|
33 |
33251
|
sandro.lab
|
|
34 |
40045
|
sandro.lab
|
if (baseURL.isEmpty() == false && baseURL.trim().startsWith("http://") == true) {
|
35 |
|
|
final String name = StringUtils.substringAfter(baseURL, "abs/").trim();
|
36 |
|
|
if (name == null) {
|
37 |
|
|
item.setUrl(null);
|
38 |
|
|
return item;
|
39 |
|
|
}
|
40 |
33575
|
sandro.lab
|
|
41 |
40045
|
sandro.lab
|
final String fileURL = createPath(name);
|
42 |
|
|
if (StringUtils.isBlank(fileURL)) {
|
43 |
|
|
item.setUrl(null);
|
44 |
|
|
return item;
|
45 |
|
|
}
|
46 |
33916
|
claudio.at
|
|
47 |
40045
|
sandro.lab
|
final File f = new File(fileURL);
|
48 |
|
|
if (f.exists()) {
|
49 |
|
|
if (log.isDebugEnabled()) {
|
50 |
|
|
log.debug("found path associated to " + item.getIdItemMetadata() + " with path : " + fileURL);
|
51 |
|
|
}
|
52 |
|
|
item.setUrl("file://" + fileURL);
|
53 |
|
|
} else {
|
54 |
|
|
if (log.isDebugEnabled()) {
|
55 |
51735
|
alessia.ba
|
log.debug("NOT found path associated to " + item.getIdItemMetadata()+": "+fileURL);
|
56 |
40045
|
sandro.lab
|
}
|
57 |
|
|
item.setUrl(null);
|
58 |
|
|
}
|
59 |
|
|
item.setOriginalUrl(baseURL);
|
60 |
|
|
return item;
|
61 |
33575
|
sandro.lab
|
|
62 |
40045
|
sandro.lab
|
} else {
|
63 |
|
|
item.setUrl(null);
|
64 |
|
|
}
|
65 |
33251
|
sandro.lab
|
|
66 |
40045
|
sandro.lab
|
}
|
67 |
|
|
return item;
|
68 |
|
|
}
|
69 |
33251
|
sandro.lab
|
|
70 |
40045
|
sandro.lab
|
public String createPath(final String name) {
|
71 |
51744
|
sandro.lab
|
final String regExp = "\\d+\\.\\d+";
|
72 |
40045
|
sandro.lab
|
if (name.matches(regExp)) {
|
73 |
|
|
final String[] values = name.split("\\.");
|
74 |
|
|
final Path bsPath = Paths.get(basePath);
|
75 |
|
|
final Path filePath = Paths.get(String.format("%s/%s.pdf", values[0], name));
|
76 |
33575
|
sandro.lab
|
|
77 |
40045
|
sandro.lab
|
final String fileURL = bsPath.resolve(filePath).toString();
|
78 |
|
|
return fileURL;
|
79 |
|
|
} else {
|
80 |
|
|
if (name.contains("/")) {
|
81 |
|
|
final String[] values = name.split("/");
|
82 |
|
|
if (values.length != 2) return null;
|
83 |
|
|
if (values[1].length() > 4) {
|
84 |
|
|
final String middle = values[1].substring(0, 4);
|
85 |
|
|
final Path bsPath = Paths.get(basePath);
|
86 |
|
|
final Path filePath = Paths.get(String.format("%s/%s.pdf", middle, name.replace("/", "")));
|
87 |
|
|
final String fileURL = bsPath.resolve(filePath).toString();
|
88 |
|
|
return fileURL;
|
89 |
|
|
}
|
90 |
|
|
}
|
91 |
|
|
}
|
92 |
|
|
return null;
|
93 |
|
|
}
|
94 |
33575
|
sandro.lab
|
|
95 |
40045
|
sandro.lab
|
@Override
|
96 |
|
|
public Iterable<DownloadItem> retrieveUrls(final Iterable<DownloadItem> items) {
|
97 |
51736
|
alessia.ba
|
return Iterables.transform(items, input -> retrieveUrl(input));
|
98 |
40045
|
sandro.lab
|
}
|
99 |
33251
|
sandro.lab
|
|
100 |
40045
|
sandro.lab
|
@Override
|
101 |
|
|
public String getPluginName() {
|
102 |
|
|
return "ArxivImportFromFile";
|
103 |
|
|
}
|
104 |
33251
|
sandro.lab
|
|
105 |
40045
|
sandro.lab
|
/**
|
106 |
|
|
* @return the basePath
|
107 |
|
|
*/
|
108 |
|
|
public String getBasePath() {
|
109 |
|
|
return basePath;
|
110 |
|
|
}
|
111 |
33251
|
sandro.lab
|
|
112 |
40045
|
sandro.lab
|
/**
|
113 |
|
|
* @param basePath the basePath to set
|
114 |
|
|
*/
|
115 |
|
|
@Override
|
116 |
|
|
public void setBasePath(final String basePath) {
|
117 |
|
|
this.basePath = basePath;
|
118 |
|
|
}
|
119 |
33251
|
sandro.lab
|
|
120 |
40045
|
sandro.lab
|
@Override
|
121 |
40523
|
claudio.at
|
public String extractURL(String baseURL) throws DownloadPluginException {
|
122 |
40045
|
sandro.lab
|
return null;
|
123 |
|
|
}
|
124 |
33251
|
sandro.lab
|
}
|