| 1 |
44356
|
claudio.at
|
package eu.dnetlib.dli.resolver;
|
| 2 |
|
|
|
| 3 |
|
|
import java.util.List;
|
| 4 |
|
|
|
| 5 |
|
|
import com.google.common.collect.Lists;
|
| 6 |
|
|
import com.google.gson.JsonArray;
|
| 7 |
|
|
import com.google.gson.JsonElement;
|
| 8 |
|
|
import com.google.gson.JsonObject;
|
| 9 |
|
|
import com.google.gson.JsonParser;
|
| 10 |
|
|
import eu.dnetlib.dli.resolver.model.*;
|
| 11 |
49184
|
sandro.lab
|
import eu.dnetlib.pid.resolver.AbstractPIDResolver;
|
| 12 |
|
|
import eu.dnetlib.pid.resolver.model.ObjectType;
|
| 13 |
49288
|
sandro.lab
|
import eu.dnetlib.pid.resolver.model.SubjectType;
|
| 14 |
44356
|
claudio.at
|
import org.apache.commons.logging.Log;
|
| 15 |
|
|
import org.apache.commons.logging.LogFactory;
|
| 16 |
|
|
|
| 17 |
|
|
/**
|
| 18 |
|
|
* The Class DataciteDOIResolver.
|
| 19 |
|
|
*/
|
| 20 |
|
|
|
| 21 |
|
|
public class DataciteResolver extends AbstractPIDResolver {
|
| 22 |
|
|
|
| 23 |
47294
|
sandro.lab
|
public static final String DATACITE_NS_PREFIX = "datacite____";
|
| 24 |
44356
|
claudio.at
|
|
| 25 |
47294
|
sandro.lab
|
/**
|
| 26 |
|
|
* The Constant log.
|
| 27 |
|
|
*/
|
| 28 |
|
|
private static final Log log = LogFactory.getLog(DataciteResolver.class); // NOPMD by marko on 11/24/08 5:02 PM
|
| 29 |
44356
|
claudio.at
|
|
| 30 |
47294
|
sandro.lab
|
/**
|
| 31 |
|
|
* The Constant baseURL.
|
| 32 |
|
|
*/
|
| 33 |
|
|
private final static String baseURL = "http://search.datacite.org/api?wt=json&q=doi:";
|
| 34 |
44356
|
claudio.at
|
|
| 35 |
49184
|
sandro.lab
|
private DLIResolvedObject parseResponse(final String response) {
|
| 36 |
47294
|
sandro.lab
|
if (response == null) return null;
|
| 37 |
|
|
JsonElement jElement = new JsonParser().parse(response);
|
| 38 |
|
|
JsonObject jobject = jElement.getAsJsonObject();
|
| 39 |
|
|
if (jobject.has("response")) {
|
| 40 |
44356
|
claudio.at
|
|
| 41 |
47294
|
sandro.lab
|
long total = jobject.get("response").getAsJsonObject().get("numFound").getAsLong();
|
| 42 |
44356
|
claudio.at
|
|
| 43 |
47294
|
sandro.lab
|
if (total == 0) return null;
|
| 44 |
49184
|
sandro.lab
|
DLIResolvedObject responseObj = new DLIResolvedObject();
|
| 45 |
44356
|
claudio.at
|
|
| 46 |
47294
|
sandro.lab
|
JsonElement hits = ((JsonObject) jobject.get("response")).get("docs");
|
| 47 |
44356
|
claudio.at
|
|
| 48 |
47294
|
sandro.lab
|
JsonArray hitsObject = hits.getAsJsonArray();
|
| 49 |
44356
|
claudio.at
|
|
| 50 |
47294
|
sandro.lab
|
for (final JsonElement elem : hitsObject) {
|
| 51 |
|
|
String doi = ((JsonObject) elem).get("doi").getAsString();
|
| 52 |
|
|
responseObj.setPid(doi);
|
| 53 |
|
|
responseObj.setPidType("DOI");
|
| 54 |
|
|
responseObj.setType(ObjectType.dataset);
|
| 55 |
|
|
List<String> authors = Lists.newArrayList();
|
| 56 |
|
|
List<String> titles = Lists.newArrayList();
|
| 57 |
|
|
List<SubjectType> subjects = Lists.newArrayList();
|
| 58 |
44356
|
claudio.at
|
|
| 59 |
|
|
|
| 60 |
47294
|
sandro.lab
|
// ADDING SUBJECTS
|
| 61 |
|
|
final JsonElement subjectElement = ((JsonObject) elem).get("subject");
|
| 62 |
|
|
if (subjectElement != null) {
|
| 63 |
|
|
JsonArray subjectArray = subjectElement.getAsJsonArray();
|
| 64 |
|
|
for (int i = 0; i < subjectArray.size(); i++) {
|
| 65 |
|
|
subjects.add(new SubjectType("unknown", subjectArray.get(i).getAsString()));
|
| 66 |
|
|
}
|
| 67 |
|
|
}
|
| 68 |
|
|
responseObj.setSubjects(subjects);
|
| 69 |
44356
|
claudio.at
|
|
| 70 |
47294
|
sandro.lab
|
// ADDING TITLES
|
| 71 |
|
|
JsonElement titleElement = ((JsonObject) elem).get("title");
|
| 72 |
|
|
if (titleElement != null) {
|
| 73 |
|
|
JsonArray titleArray = titleElement.getAsJsonArray();
|
| 74 |
|
|
for (int i = 0; i < titleArray.size(); i++) {
|
| 75 |
|
|
titles.add(titleArray.get(i).getAsString());
|
| 76 |
|
|
}
|
| 77 |
|
|
}
|
| 78 |
|
|
responseObj.setTitles(titles);
|
| 79 |
44356
|
claudio.at
|
|
| 80 |
47294
|
sandro.lab
|
//ADDING DESCRIPTION
|
| 81 |
44356
|
claudio.at
|
|
| 82 |
47294
|
sandro.lab
|
JsonElement descriptionElement = ((JsonObject) elem).get("description");
|
| 83 |
|
|
if (descriptionElement != null) {
|
| 84 |
|
|
JsonArray descriptionArray = descriptionElement.getAsJsonArray();
|
| 85 |
|
|
if (descriptionArray.size() > 0)
|
| 86 |
|
|
responseObj.setDescription(descriptionArray.get(0).getAsString());
|
| 87 |
|
|
}
|
| 88 |
44356
|
claudio.at
|
|
| 89 |
47294
|
sandro.lab
|
// ADDING AUTHORS
|
| 90 |
|
|
JsonElement contributorElement = ((JsonObject) elem).get("contributor");
|
| 91 |
|
|
if (contributorElement != null) {
|
| 92 |
|
|
JsonArray contributorArray = contributorElement.getAsJsonArray();
|
| 93 |
|
|
for (int i = 0; i < contributorArray.size(); i++) {
|
| 94 |
|
|
authors.add(contributorArray.get(i).getAsString());
|
| 95 |
|
|
}
|
| 96 |
|
|
}
|
| 97 |
|
|
// ADDING AUTHORS
|
| 98 |
|
|
JsonElement creatorElement = ((JsonObject) elem).get("creator");
|
| 99 |
|
|
if (creatorElement != null) {
|
| 100 |
|
|
JsonArray creatorArray = creatorElement.getAsJsonArray();
|
| 101 |
|
|
for (int i = 0; i < creatorArray.size(); i++) {
|
| 102 |
|
|
authors.add(creatorArray.get(i).getAsString());
|
| 103 |
|
|
}
|
| 104 |
|
|
}
|
| 105 |
|
|
responseObj.setAuthors(authors);
|
| 106 |
|
|
|
| 107 |
49184
|
sandro.lab
|
final DLIObjectProvenance provenance = new DLIObjectProvenance();
|
| 108 |
|
|
DLIPIDResolver.setDatasourceProvenance(provenance, DATACITE_NS_PREFIX);
|
| 109 |
44356
|
claudio.at
|
|
| 110 |
47294
|
sandro.lab
|
final String dataCenter_symbol = ((JsonObject) elem).get("datacentre_symbol").getAsString();
|
| 111 |
|
|
final String datacenter = ((JsonObject) elem).get("datacentre").getAsString();
|
| 112 |
|
|
if (datacenter != null) {
|
| 113 |
|
|
final String publisher = datacenter.replace(dataCenter_symbol, "").replace("-", "").trim();
|
| 114 |
|
|
if (publisher != null && publisher.length() > 0) {
|
| 115 |
|
|
provenance.setPublisher(publisher);
|
| 116 |
|
|
}
|
| 117 |
|
|
}
|
| 118 |
|
|
responseObj.setDatasourceProvenance(Lists.newArrayList(provenance));
|
| 119 |
|
|
responseObj.setCompletionStatus(CompletionStatus.complete.toString());
|
| 120 |
|
|
return responseObj;
|
| 121 |
|
|
}
|
| 122 |
|
|
return null;
|
| 123 |
|
|
}
|
| 124 |
|
|
return null;
|
| 125 |
44356
|
claudio.at
|
|
| 126 |
47294
|
sandro.lab
|
}
|
| 127 |
44356
|
claudio.at
|
|
| 128 |
47294
|
sandro.lab
|
@Override
|
| 129 |
|
|
protected boolean canResolvePid(final String pidType) {
|
| 130 |
|
|
return (pidType != null) && ("doi".equals(pidType.toLowerCase().trim()) || "handle".equals(pidType.toLowerCase().trim()));
|
| 131 |
44356
|
claudio.at
|
|
| 132 |
47294
|
sandro.lab
|
}
|
| 133 |
44356
|
claudio.at
|
|
| 134 |
47294
|
sandro.lab
|
@Override
|
| 135 |
49184
|
sandro.lab
|
protected DLIResolvedObject resolve(final String pid, final String pidType) {
|
| 136 |
47294
|
sandro.lab
|
try {
|
| 137 |
|
|
final String response = requestURL(baseURL + pid.replace(" ", "%20"));
|
| 138 |
|
|
return parseResponse(response);
|
| 139 |
|
|
} catch (Throwable e) {
|
| 140 |
|
|
log.error("Error on getting item from url " + baseURL + pid.replace(" ", "%20"));
|
| 141 |
|
|
}
|
| 142 |
|
|
return null;
|
| 143 |
|
|
}
|
| 144 |
44356
|
claudio.at
|
}
|