Revision 41818
Added by Eri Katsari over 8 years ago
| modules/dnet-openaire-stats/branches/extended/src/test/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/GeneralTests.java | ||
|---|---|---|
| 106 | 106 |
} |
| 107 | 107 |
|
| 108 | 108 |
|
| 109 |
|
|
| 110 | 109 |
@Test |
| 111 | 110 |
public void testPath() {
|
| 112 | 111 |
|
| ... | ... | |
| 120 | 119 |
@Test |
| 121 | 120 |
public void test() {
|
| 122 | 121 |
|
| 123 |
Integer s = new Integer(5); |
|
| 124 |
// s="lala"; |
|
| 125 |
System.out.println(s); |
|
| 126 |
test2(s); |
|
| 127 |
System.out.println("s outside " + s);
|
|
| 122 |
String value = "d'Analisi,"; |
|
| 123 |
value = value.replaceAll("'", " ");
|
|
| 124 |
value = value.replaceAll("\\r\\n|\\r|\\n", " ");
|
|
| 125 |
value = value.replaceAll("\\s+", " ");
|
|
| 128 | 126 |
|
| 127 |
value = value.replaceAll("(\\r|\\n)", " ");
|
|
| 128 |
value = value.replaceAll("\\t", " ");
|
|
| 129 |
value = value.replace("\n", " ");
|
|
| 130 |
value = value.replaceAll("[\"\\r\\\\;]", "");
|
|
| 131 |
|
|
| 132 |
value.replace(">", " ");
|
|
| 133 |
value.replace("<", " ");
|
|
| 134 |
value.replace(",", " ");
|
|
| 135 |
value.replace("\"", " ");
|
|
| 136 |
value.replace("'", " ");
|
|
| 137 |
value = value.replaceAll("[^a-zA-Z0-9 .-_:/@+=]+", " ");
|
|
| 138 |
System.out.println(value); |
|
| 139 |
|
|
| 129 | 140 |
} |
| 130 | 141 |
|
| 131 | 142 |
public void test2(Integer s) {
|
| 132 | 143 |
Integer a = new Integer(10); |
| 133 | 144 |
s = 10; |
| 134 | 145 |
|
| 146 |
|
|
| 135 | 147 |
System.out.println("s inside " + s);
|
| 136 | 148 |
|
| 137 | 149 |
|
| modules/dnet-openaire-stats/branches/extended/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/daos/UsageStatsDAO.java | ||
|---|---|---|
| 6 | 6 |
import org.apache.hadoop.fs.FileSystem; |
| 7 | 7 |
import org.apache.hadoop.fs.Path; |
| 8 | 8 |
import org.apache.log4j.Logger; |
| 9 |
|
|
| 9 | 10 |
import org.springframework.jdbc.datasource.DriverManagerDataSource; |
| 10 | 11 |
|
| 12 |
import javax.sql.DataSource; |
|
| 11 | 13 |
import java.io.File; |
| 12 |
import java.sql.Connection; |
|
| 13 |
import java.sql.ResultSet; |
|
| 14 |
import java.sql.ResultSetMetaData; |
|
| 15 |
import java.sql.Statement; |
|
| 14 |
import java.sql.*; |
|
| 16 | 15 |
|
| 17 | 16 |
/** |
| 18 | 17 |
* @author eri |
| ... | ... | |
| 26 | 25 |
private Logger log = Logger.getLogger(this.getClass()); |
| 27 | 26 |
private String delim; |
| 28 | 27 |
private String outputPath; |
| 28 |
private String dbUrl; |
|
| 29 | 29 |
|
| 30 | 30 |
public UsageStatsDAO(String dbUrl, String dbDriver, String delim, String outputPath) throws Exception {
|
| 31 |
usageStatsDatasource = new DriverManagerDataSource(); |
|
| 31 |
|
|
| 32 |
usageStatsDatasource = new DriverManagerDataSource(dbUrl); |
|
| 32 | 33 |
usageStatsDatasource.setDriverClassName(dbDriver); |
| 33 |
usageStatsDatasource.setUrl(dbUrl); |
|
| 34 |
|
|
| 34 | 35 |
this.delim = delim; |
| 35 | 36 |
this.outputPath = outputPath; |
| 37 |
this.dbUrl = dbUrl; |
|
| 36 | 38 |
|
| 37 | 39 |
log.info("Usage Stats DB url" + dbUrl);
|
| 38 | 40 |
|
| 39 |
String dbName = dbUrl.substring(dbUrl.indexOf("/"), dbUrl.length());
|
|
| 40 |
File file = new File(dbName); |
|
| 41 |
|
|
| 42 |
if (file.exists()) //here's how to check |
|
| 43 |
{
|
|
| 44 |
log.info("Database " + dbName + "exists");
|
|
| 45 |
|
|
| 46 |
} else {
|
|
| 47 |
log.error("Database " + dbName + "not found");
|
|
| 48 |
throw new Exception("Database not found");
|
|
| 49 |
} |
|
| 50 | 41 |
} |
| 51 | 42 |
|
| 52 | 43 |
public void getUsageStatistics(String table) throws Exception {
|
| 53 | 44 |
|
| 54 | 45 |
log.info("Getting Usage Statistics for " + table);
|
| 55 | 46 |
// String q = "select * from " + table + "Stats"; |
| 56 |
String q = "SELECT * from datasourceStats"; |
|
| 47 |
String q = "SELECT * from datasourceStats; ";
|
|
| 57 | 48 |
Connection con = null; |
| 58 | 49 |
|
| 50 |
//con = DriverManager.getConnection(dbUrl); |
|
| 51 |
|
|
| 59 | 52 |
con = usageStatsDatasource.getConnection(); |
| 60 | 53 |
|
| 61 | 54 |
if (con == null) {
|
| 62 |
log.error("Database not found");
|
|
| 63 |
throw new Exception("Database not found");
|
|
| 55 |
log.error("cannot open connection to Database ");
|
|
| 56 |
throw new Exception("cannot open connection to Database");
|
|
| 64 | 57 |
} |
| 65 | 58 |
|
| 66 | 59 |
Statement st = con.createStatement(); |
| ... | ... | |
| 132 | 125 |
} |
| 133 | 126 |
|
| 134 | 127 |
|
| 135 |
public DriverManagerDataSource getUsageStatsDatasource() {
|
|
| 136 |
return usageStatsDatasource; |
|
| 137 |
} |
|
| 138 |
|
|
| 139 |
public void setUsageStatsDatasource(DriverManagerDataSource usageStatsDatasource) {
|
|
| 140 |
this.usageStatsDatasource = usageStatsDatasource; |
|
| 141 |
} |
|
| 142 |
|
|
| 143 | 128 |
public String getDelim() {
|
| 144 | 129 |
return delim; |
| 145 | 130 |
} |
| modules/dnet-openaire-stats/branches/extended/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/drivers/SqoopImportDriver.java | ||
|---|---|---|
| 63 | 63 |
"--target-dir", outputPath, |
| 64 | 64 |
"--verbose", |
| 65 | 65 |
"--input-fields-terminated-by", delim, |
| 66 |
"--driver", driver, |
|
| 67 |
"--mapreduce-job-name", "Sqoop Usage Stats Import Job for " + table.getKey() |
|
| 66 |
"--driver", driver |
|
| 67 |
// "--relaxed-isolation", |
|
| 68 |
//"--mapreduce-job-name", "Sqoop Usage Stats Import Job for " + table.getKey() |
|
| 68 | 69 |
}; |
| 69 | 70 |
|
| 70 | 71 |
int ret = Sqoop.runTool(str); |
| modules/dnet-openaire-stats/branches/full_export/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/StaticSerializer.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.mapreduce.hbase.statsExport.utils; |
|
| 2 |
|
|
| 3 |
import com.google.common.collect.ArrayListMultimap; |
|
| 4 |
import com.google.common.collect.Multimap; |
|
| 5 |
import eu.dnetlib.data.mapreduce.util.LicenseComparator; |
|
| 6 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource; |
|
| 7 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata; |
|
| 8 |
import eu.dnetlib.data.proto.FieldTypeProtos; |
|
| 9 |
import eu.dnetlib.data.proto.FieldTypeProtos.Qualifier; |
|
| 10 |
import eu.dnetlib.data.proto.FieldTypeProtos.StringField; |
|
| 11 |
import eu.dnetlib.data.proto.FieldTypeProtos.StructuredProperty; |
|
| 12 |
import eu.dnetlib.data.proto.OafProtos.Oaf; |
|
| 13 |
import eu.dnetlib.data.proto.OafProtos.OafEntity; |
|
| 14 |
import eu.dnetlib.data.proto.OafProtos.OafRel; |
|
| 15 |
import eu.dnetlib.data.proto.OrganizationProtos.Organization; |
|
| 16 |
import eu.dnetlib.data.proto.PersonProtos; |
|
| 17 |
import eu.dnetlib.data.proto.ProjectProtos.Project; |
|
| 18 |
import eu.dnetlib.data.proto.RelTypeProtos.RelType; |
|
| 19 |
import eu.dnetlib.data.proto.ResultProtos.Result; |
|
| 20 |
import eu.dnetlib.data.proto.ResultProtos.Result.Instance; |
|
| 21 |
import org.apache.log4j.Logger; |
|
| 22 |
import org.jsoup.Jsoup; |
|
| 23 |
|
|
| 24 |
import java.text.SimpleDateFormat; |
|
| 25 |
import java.util.Date; |
|
| 26 |
import java.util.List; |
|
| 27 |
|
|
| 28 |
/** |
|
| 29 |
* @author eri Simple serializer that parses input Oaf Protos and prepares them |
|
| 30 |
* for sqoop |
|
| 31 |
*/ |
|
| 32 |
public class StaticSerializer {
|
|
| 33 |
|
|
| 34 |
public static String serialize(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 35 |
|
|
| 36 |
switch (oaf.getKind()) {
|
|
| 37 |
case entity: |
|
| 38 |
OafEntity valueEntity = oaf.getEntity(); |
|
| 39 |
|
|
| 40 |
switch (valueEntity.getType()) {
|
|
| 41 |
case datasource: |
|
| 42 |
|
|
| 43 |
return buildDatasource(oaf, DELIM, ENCLOSING); |
|
| 44 |
|
|
| 45 |
case organization: |
|
| 46 |
|
|
| 47 |
return buildOrganization(oaf, DELIM, ENCLOSING); |
|
| 48 |
|
|
| 49 |
case project: |
|
| 50 |
|
|
| 51 |
return buildProject(oaf, DELIM, ENCLOSING); |
|
| 52 |
case result: |
|
| 53 |
|
|
| 54 |
return buildResult(oaf, DELIM, ENCLOSING); |
|
| 55 |
case person: |
|
| 56 |
return buildPerson(oaf, DELIM, ENCLOSING); |
|
| 57 |
default: |
|
| 58 |
break; |
|
| 59 |
} |
|
| 60 |
break; |
|
| 61 |
case relation: |
|
| 62 |
OafRel valueRel = oaf.getRel(); |
|
| 63 |
return buildRel(valueRel, DELIM, ENCLOSING); |
|
| 64 |
|
|
| 65 |
} |
|
| 66 |
|
|
| 67 |
return null; |
|
| 68 |
|
|
| 69 |
} |
|
| 70 |
|
|
| 71 |
public static String serialize(OafRel oaf, String DELIM, String ENCLOSING) {
|
|
| 72 |
|
|
| 73 |
switch (oaf.getRelType()) {
|
|
| 74 |
case resultProject: |
|
| 75 |
return getResultProject(oaf, DELIM, ENCLOSING); |
|
| 76 |
default: |
|
| 77 |
return buildRel(oaf, DELIM, ENCLOSING); |
|
| 78 |
} |
|
| 79 |
} |
|
| 80 |
|
|
| 81 |
private static String buildRel(OafRel Rel, String DELIM, String ENCLOSING) {
|
|
| 82 |
return getStringField(Rel.getTarget(), DELIM, ENCLOSING); |
|
| 83 |
} |
|
| 84 |
|
|
| 85 |
public static Multimap<String, String> extractRelations(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 86 |
OafEntity valueEntity = oaf.getEntity(); |
|
| 87 |
Multimap<String, String> relations = ArrayListMultimap.create(); |
|
| 88 |
|
|
| 89 |
getOriginalId(valueEntity, relations, DELIM, ENCLOSING); |
|
| 90 |
|
|
| 91 |
switch (valueEntity.getType()) {
|
|
| 92 |
case datasource: |
|
| 93 |
getDatasourceLanguages(valueEntity, relations, DELIM, ENCLOSING); |
|
| 94 |
case result: |
|
| 95 |
getResultTopics(valueEntity, relations, DELIM, ENCLOSING); |
|
| 96 |
getResultLanguages(valueEntity, relations, DELIM, ENCLOSING); |
|
| 97 |
getResultClassifications(valueEntity, relations, DELIM, ENCLOSING); |
|
| 98 |
getResultDatasources(valueEntity, relations, DELIM, ENCLOSING); |
|
| 99 |
getResultConcepts(valueEntity, relations, DELIM, ENCLOSING); |
|
| 100 |
getResultDois(valueEntity, relations, DELIM, ENCLOSING); |
|
| 101 |
getResultCitations(valueEntity, relations, DELIM, ENCLOSING); |
|
| 102 |
getResultDescriptions(valueEntity, relations, DELIM, ENCLOSING); |
|
| 103 |
return relations; |
|
| 104 |
|
|
| 105 |
case project: |
|
| 106 |
getProjectKeywords(valueEntity, relations, DELIM, ENCLOSING); |
|
| 107 |
getProjectSubjects(valueEntity, relations, DELIM, ENCLOSING); |
|
| 108 |
|
|
| 109 |
default: |
|
| 110 |
return null; |
|
| 111 |
} |
|
| 112 |
|
|
| 113 |
} |
|
| 114 |
|
|
| 115 |
|
|
| 116 |
private static void getOriginalId(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
|
| 117 |
|
|
| 118 |
String relName = oafEntity.getType().toString().toLowerCase() + "Oid"; |
|
| 119 |
for (String oid : oafEntity.getOriginalIdList()) {
|
|
| 120 |
relations.put(relName, ENCLOSING + oid.replace(DELIM, " ").replace("\n", "") + ENCLOSING + DELIM);
|
|
| 121 |
} |
|
| 122 |
|
|
| 123 |
} |
|
| 124 |
|
|
| 125 |
private static void getProjectKeywords(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
|
| 126 |
relations.put("projectKeyword", getStringField(oafEntity.getProject().getMetadata().getKeywords().getValue(), DELIM, ENCLOSING));
|
|
| 127 |
|
|
| 128 |
} |
|
| 129 |
|
|
| 130 |
private static void getProjectSubjects(OafEntity oafEntity, Multimap<String, String> relations, String DELIM, String ENCLOSING) {
|
|
| 131 |
for (StructuredProperty subj : oafEntity.getProject().getMetadata().getSubjectsList()) {
|
|
| 132 |
relations.put("projectSubject", getStringField(subj.getValue(), DELIM, ENCLOSING));
|
|
| 133 |
} |
|
| 134 |
} |
|
| 135 |
|
|
| 136 |
private static String getResultProject(OafRel oaf, String DELIM, String ENCLOSING) {
|
|
| 137 |
String buff = new String(); |
|
| 138 |
String result = oaf.getTarget(); |
|
| 139 |
|
|
| 140 |
buff += getStringField(result, DELIM, ENCLOSING); |
|
| 141 |
// TODO is declared as int!!! |
|
| 142 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
|
| 143 |
if (diff < 0) {
|
|
| 144 |
diff = 0; |
|
| 145 |
} |
|
| 146 |
buff += getNumericField(String.valueOf(diff), DELIM, ENCLOSING); |
|
| 147 |
return buff; |
|
| 148 |
} |
|
| 149 |
|
|
| 150 |
|
|
| 151 |
private static void getDatasourceLanguages(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 152 |
Datasource d = valueEntity.getDatasource(); |
|
| 153 |
Metadata metadata = d.getMetadata(); |
|
| 154 |
|
|
| 155 |
for (StringField lang : metadata.getOdlanguagesList()) {
|
|
| 156 |
rels.put("datasourceLanguage", getStringField(lang.getValue(), DELIM, ENCLOSING));
|
|
| 157 |
} |
|
| 158 |
} |
|
| 159 |
|
|
| 160 |
private static void getResultLanguages(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 161 |
|
|
| 162 |
Result d = valueEntity.getResult(); |
|
| 163 |
Result.Metadata metadata = d.getMetadata(); |
|
| 164 |
if (metadata.getLanguage().getClassname() != null && !metadata.getLanguage().getClassname().isEmpty()) {
|
|
| 165 |
rels.put("resultLanguage", getStringField(metadata.getLanguage().getClassname(), DELIM, ENCLOSING));
|
|
| 166 |
} |
|
| 167 |
|
|
| 168 |
} |
|
| 169 |
|
|
| 170 |
private static void getResultDois(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 171 |
|
|
| 172 |
for (StructuredProperty pid : valueEntity.getPidList()) {
|
|
| 173 |
|
|
| 174 |
rels.put("resultPid",
|
|
| 175 |
getStringField(pid.getQualifier().getClassname(), DELIM, ENCLOSING) + getStringField(pid.getValue(), DELIM, ENCLOSING)); |
|
| 176 |
} |
|
| 177 |
} |
|
| 178 |
|
|
| 179 |
private static void getResultClassifications(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 180 |
|
|
| 181 |
Result result = valueEntity.getResult(); |
|
| 182 |
|
|
| 183 |
for (Instance instance : (result.getInstanceList())) {
|
|
| 184 |
String classification = instance.getInstancetype().getClassname(); |
|
| 185 |
if (classification != null && !classification.isEmpty()) {
|
|
| 186 |
rels.put("resultClassification", getStringField(instance.getInstancetype().getClassname(), DELIM, ENCLOSING));
|
|
| 187 |
// TODO HERE KEEP ONLY ONE CLASSIFICATIONS PER RESULT |
|
| 188 |
break; |
|
| 189 |
} |
|
| 190 |
} |
|
| 191 |
} |
|
| 192 |
|
|
| 193 |
private static void getResultDescriptions(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 194 |
Result result = valueEntity.getResult(); |
|
| 195 |
//description |
|
| 196 |
for (StringField s : result.getMetadata().getDescriptionList()) {
|
|
| 197 |
|
|
| 198 |
rels.put("resultDescription", getStringField(Jsoup.parse(s.getValue()).text(), DELIM, ENCLOSING));
|
|
| 199 |
} |
|
| 200 |
} |
|
| 201 |
|
|
| 202 |
private static void getResultConcepts(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 203 |
|
|
| 204 |
Result result = valueEntity.getResult(); |
|
| 205 |
|
|
| 206 |
for (Result.Context context : result.getMetadata().getContextList()) {
|
|
| 207 |
|
|
| 208 |
rels.put("resultConcept", getStringField(context.getId(), DELIM, ENCLOSING));
|
|
| 209 |
} |
|
| 210 |
} |
|
| 211 |
|
|
| 212 |
|
|
| 213 |
private static void getResultDatasources(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 214 |
Result result = valueEntity.getResult(); |
|
| 215 |
|
|
| 216 |
//TODO hosted by |
|
| 217 |
for (Instance instance : (result.getInstanceList())) {
|
|
| 218 |
String hostedBy = instance.getHostedby().getKey(); |
|
| 219 |
if (hostedBy != null && !hostedBy.isEmpty()) {
|
|
| 220 |
rels.put("resultDatasource", getStringField(hostedBy, DELIM, ENCLOSING));
|
|
| 221 |
} |
|
| 222 |
} |
|
| 223 |
|
|
| 224 |
//TODO collected froms |
|
| 225 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
|
| 226 |
|
|
| 227 |
String collectedFrom = collectedFromValue.getKey(); |
|
| 228 |
if (collectedFrom != null && !collectedFrom.isEmpty()) {
|
|
| 229 |
rels.put("resultDatasource", getStringField(collectedFrom, DELIM, ENCLOSING));
|
|
| 230 |
} |
|
| 231 |
} |
|
| 232 |
} |
|
| 233 |
|
|
| 234 |
private static void getResultTopics(OafEntity valueEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 235 |
|
|
| 236 |
Result d = valueEntity.getResult(); |
|
| 237 |
Result.Metadata metadata = d.getMetadata(); |
|
| 238 |
|
|
| 239 |
List<StructuredProperty> Topics = metadata.getSubjectList(); |
|
| 240 |
|
|
| 241 |
for (StructuredProperty topic : Topics) {
|
|
| 242 |
// TODO result topics |
|
| 243 |
rels.put("resultTopic", getStringField(topic.getValue(), DELIM, ENCLOSING));
|
|
| 244 |
} |
|
| 245 |
} |
|
| 246 |
|
|
| 247 |
|
|
| 248 |
private static void getResultCitations(OafEntity oafEntity, Multimap<String, String> rels, String DELIM, String ENCLOSING) {
|
|
| 249 |
for (FieldTypeProtos.ExtraInfo extraInfo : oafEntity.getExtraInfoList()) {
|
|
| 250 |
if (extraInfo.getName().equals("result citations")) {
|
|
| 251 |
rels.put("resultCitation", getStringField(extraInfo.getTrust(), DELIM, ENCLOSING) +
|
|
| 252 |
getStringField(extraInfo.getProvenance(), DELIM, ENCLOSING) + getStringField(extraInfo.getValue(), DELIM, ENCLOSING)); |
|
| 253 |
} |
|
| 254 |
|
|
| 255 |
} |
|
| 256 |
} |
|
| 257 |
|
|
| 258 |
private static String buildDatasource(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 259 |
OafEntity data = oaf.getEntity(); |
|
| 260 |
|
|
| 261 |
String buff = new String(); |
|
| 262 |
|
|
| 263 |
Datasource d = data.getDatasource(); |
|
| 264 |
Metadata metadata = d.getMetadata(); |
|
| 265 |
|
|
| 266 |
|
|
| 267 |
// name |
|
| 268 |
if (metadata.getOfficialname().getValue().equalsIgnoreCase("unknown")) {
|
|
| 269 |
buff += getStringField("Unknown Repository", DELIM, ENCLOSING);
|
|
| 270 |
} else {
|
|
| 271 |
buff += getStringField(metadata.getOfficialname().getValue(), DELIM, ENCLOSING); |
|
| 272 |
} |
|
| 273 |
// type |
|
| 274 |
|
|
| 275 |
if (metadata.hasDatasourcetype()) |
|
| 276 |
|
|
| 277 |
{
|
|
| 278 |
buff += getStringField(metadata.getDatasourcetype().getClassname().replaceFirst(".*::", ""), DELIM, ENCLOSING);
|
|
| 279 |
|
|
| 280 |
} else {
|
|
| 281 |
buff += getStringField(null, DELIM, ENCLOSING); |
|
| 282 |
; |
|
| 283 |
} |
|
| 284 |
|
|
| 285 |
// compatibility, |
|
| 286 |
buff += getStringField(metadata.getOpenairecompatibility().getClassname(), DELIM, ENCLOSING); |
|
| 287 |
|
|
| 288 |
// latitude |
|
| 289 |
buff += getStringField(metadata.getLatitude().getValue(), DELIM, ENCLOSING); |
|
| 290 |
|
|
| 291 |
// longtitude |
|
| 292 |
buff += getStringField(metadata.getLongitude().getValue(), DELIM, ENCLOSING); |
|
| 293 |
|
|
| 294 |
// dateofvalidation, |
|
| 295 |
buff += getStringField(metadata.getDateofvalidation().getValue(), DELIM, ENCLOSING); |
|
| 296 |
; |
|
| 297 |
|
|
| 298 |
// yearofvalidation, |
|
| 299 |
buff += getYearInt(metadata.getDateofvalidation().getValue(), DELIM, ENCLOSING); |
|
| 300 |
; |
|
| 301 |
|
|
| 302 |
//website |
|
| 303 |
buff += getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING); |
|
| 304 |
; |
|
| 305 |
|
|
| 306 |
//harvested |
|
| 307 |
buff += getStringField("false", DELIM, ENCLOSING);
|
|
| 308 |
; |
|
| 309 |
|
|
| 310 |
// deletedByInference |
|
| 311 |
buff += getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING); |
|
| 312 |
; |
|
| 313 |
|
|
| 314 |
return buff; |
|
| 315 |
} |
|
| 316 |
|
|
| 317 |
private static String buildOrganization(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 318 |
OafEntity data = oaf.getEntity(); |
|
| 319 |
String buff = new String(); |
|
| 320 |
|
|
| 321 |
Organization organization = data.getOrganization(); |
|
| 322 |
Organization.Metadata metadata = organization.getMetadata(); |
|
| 323 |
|
|
| 324 |
// `name`, |
|
| 325 |
buff += getStringField(metadata.getLegalname().getValue(), DELIM, ENCLOSING); |
|
| 326 |
; |
|
| 327 |
// `country`, |
|
| 328 |
buff += getStringField(metadata.getCountry().getClassname(), DELIM, ENCLOSING); |
|
| 329 |
; |
|
| 330 |
|
|
| 331 |
//website |
|
| 332 |
buff += getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING); |
|
| 333 |
; |
|
| 334 |
|
|
| 335 |
|
|
| 336 |
// deletedByInference |
|
| 337 |
buff += getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING); |
|
| 338 |
; |
|
| 339 |
|
|
| 340 |
|
|
| 341 |
return buff; |
|
| 342 |
|
|
| 343 |
} |
|
| 344 |
|
|
| 345 |
private static String buildResult(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 346 |
OafEntity data = oaf.getEntity(); |
|
| 347 |
String buff = new String(); |
|
| 348 |
|
|
| 349 |
Result result = data.getResult(); |
|
| 350 |
Result.Metadata metadata = result.getMetadata(); |
|
| 351 |
|
|
| 352 |
// pubtitle |
|
| 353 |
|
|
| 354 |
String titleString = new String(); |
|
| 355 |
|
|
| 356 |
|
|
| 357 |
for (int i = 0; i < metadata.getTitleList().size(); i++) {
|
|
| 358 |
StructuredProperty title = metadata.getTitleList().get(i); |
|
| 359 |
if (i == 0) {
|
|
| 360 |
titleString = title.getValue().replaceAll("\\s+", " ");
|
|
| 361 |
titleString = titleString.replaceAll("\n", " ");
|
|
| 362 |
} |
|
| 363 |
break; |
|
| 364 |
} |
|
| 365 |
// pubtitle |
|
| 366 |
buff += getStringField(titleString, DELIM, ENCLOSING); //!#Osvaldo Raineri: Gli Atti etiopici del martire egiziano Giorgio il nuovo (β 978)#!# |
|
| 367 |
|
|
| 368 |
// format |
|
| 369 |
String formatString = new String(); |
|
| 370 |
|
|
| 371 |
for (StringField format : metadata.getFormatList()) {
|
|
| 372 |
formatString += format.getValue() + ";"; |
|
| 373 |
|
|
| 374 |
} |
|
| 375 |
|
|
| 376 |
buff += getStringField(formatString, DELIM, ENCLOSING); |
|
| 377 |
// publisher |
|
| 378 |
|
|
| 379 |
buff += getStringField(metadata.getPublisher().getValue(), DELIM, ENCLOSING); |
|
| 380 |
// journal |
|
| 381 |
|
|
| 382 |
buff += getStringField(metadata.getJournal().getName().replaceAll("\n", " "), DELIM, ENCLOSING); //#null#!
|
|
| 383 |
|
|
| 384 |
// year |
|
| 385 |
buff += getYearInt(metadata.getDateofacceptance().getValue(), DELIM, ENCLOSING); |
|
| 386 |
|
|
| 387 |
// date CHANGED THIS TO DATE FORMAT |
|
| 388 |
buff += getStringDateField(metadata.getDateofacceptance().getValue(), DELIM, ENCLOSING); |
|
| 389 |
|
|
| 390 |
// access_mode, |
|
| 391 |
buff += getStringField(getAccessMode(result), DELIM, ENCLOSING); |
|
| 392 |
|
|
| 393 |
// bestlicense |
|
| 394 |
|
|
| 395 |
buff += getStringField(getBestLicense(result), DELIM, ENCLOSING); |
|
| 396 |
; |
|
| 397 |
// type |
|
| 398 |
buff += getStringField(metadata.getResulttype().getClassname(), DELIM, ENCLOSING); |
|
| 399 |
// embargo_end_date |
|
| 400 |
buff += getStringField(metadata.getEmbargoenddate().getValue(), DELIM, ENCLOSING); |
|
| 401 |
|
|
| 402 |
// `authors`, |
|
| 403 |
int authors = 0; |
|
| 404 |
String delayed = "no"; |
|
| 405 |
|
|
| 406 |
for (OafRel rel : data.getCachedRelList()) {
|
|
| 407 |
|
|
| 408 |
if (rel.getRelType().equals(RelType.personResult)) {
|
|
| 409 |
|
|
| 410 |
authors++; |
|
| 411 |
} else if (rel.getRelType().equals(RelType.resultProject)) |
|
| 412 |
// TODO remember : in result Project, first id is project, second is |
|
| 413 |
// result. |
|
| 414 |
{
|
|
| 415 |
|
|
| 416 |
String daysfromend = getYearDifferenceInteger(rel.getResultProject().getOutcome().getRelMetadata().getEnddate(), |
|
| 417 |
rel.getResultProject().getOutcome().getRelMetadata().getStartdate(), DELIM, ENCLOSING); |
|
| 418 |
if (Integer.parseInt(daysfromend) > 0) {
|
|
| 419 |
delayed = "yes"; |
|
| 420 |
} |
|
| 421 |
} |
|
| 422 |
} |
|
| 423 |
// `delayed`, |
|
| 424 |
buff += getStringField(delayed, DELIM, ENCLOSING); |
|
| 425 |
|
|
| 426 |
buff += getNumericField(String.valueOf(authors), DELIM, ENCLOSING); |
|
| 427 |
|
|
| 428 |
|
|
| 429 |
// deletedByInference |
|
| 430 |
buff += getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING); |
|
| 431 |
|
|
| 432 |
|
|
| 433 |
return buff; |
|
| 434 |
|
|
| 435 |
} |
|
| 436 |
|
|
| 437 |
|
|
| 438 |
private static String getBestLicense(Result result) {
|
|
| 439 |
Qualifier bestLicense = null; |
|
| 440 |
LicenseComparator lc = new LicenseComparator(); |
|
| 441 |
for (Instance instance : (result.getInstanceList())) {
|
|
| 442 |
if (lc.compare(bestLicense, instance.getLicence()) > 0) {
|
|
| 443 |
bestLicense = instance.getLicence(); |
|
| 444 |
} |
|
| 445 |
} |
|
| 446 |
if (bestLicense != null) {
|
|
| 447 |
return bestLicense.getClassname(); |
|
| 448 |
} else {
|
|
| 449 |
return null; |
|
| 450 |
} |
|
| 451 |
} |
|
| 452 |
|
|
| 453 |
// TODO here iterate over all values |
|
| 454 |
private static String getAccessMode(Result result) {
|
|
| 455 |
String accessMode = " "; |
|
| 456 |
for (Instance instance : (result.getInstanceList())) {
|
|
| 457 |
if (instance.getLicence().getClassname() != null && !instance.getLicence().getClassname().isEmpty()) {
|
|
| 458 |
accessMode = instance.getLicence().getClassname(); |
|
| 459 |
break; |
|
| 460 |
} |
|
| 461 |
|
|
| 462 |
} |
|
| 463 |
|
|
| 464 |
return accessMode; |
|
| 465 |
} |
|
| 466 |
|
|
| 467 |
private static String buildProject(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 468 |
OafEntity data = oaf.getEntity(); |
|
| 469 |
String buff = new String(); |
|
| 470 |
|
|
| 471 |
Project project = data.getProject(); |
|
| 472 |
Project.Metadata metadata = project.getMetadata(); |
|
| 473 |
|
|
| 474 |
|
|
| 475 |
// `acronym`, |
|
| 476 |
String acronym = metadata.getAcronym().getValue(); |
|
| 477 |
if (acronym.equalsIgnoreCase("UNKNOWN")) {
|
|
| 478 |
acronym = metadata.getTitle().getValue(); |
|
| 479 |
} |
|
| 480 |
|
|
| 481 |
buff += getStringField(acronym, DELIM, ENCLOSING); |
|
| 482 |
|
|
| 483 |
//title! |
|
| 484 |
String title = getStringField(metadata.getTitle().getValue(), DELIM, ENCLOSING); |
|
| 485 |
buff += getStringField(title, DELIM, ENCLOSING); |
|
| 486 |
|
|
| 487 |
List<StringField> fundList = metadata.getFundingtreeList(); |
|
| 488 |
|
|
| 489 |
|
|
| 490 |
if (!fundList.isEmpty()) // `funding_lvl0`, |
|
| 491 |
{
|
|
| 492 |
//TODO funder + 3 funding levels |
|
| 493 |
/* funder text, |
|
| 494 |
funding_lvl0 text, |
|
| 495 |
funding_lvl1 text, |
|
| 496 |
funding_lvl2 text, |
|
| 497 |
funding_lvl3 text,*/ |
|
| 498 |
buff += FundingParser.getFundingInfo(fundList.get(0).getValue(), DELIM, ENCLOSING); |
|
| 499 |
; |
|
| 500 |
|
|
| 501 |
} else {
|
|
| 502 |
|
|
| 503 |
buff += FundingParser.getFundingInfo("", DELIM, ENCLOSING);
|
|
| 504 |
; |
|
| 505 |
} |
|
| 506 |
|
|
| 507 |
|
|
| 508 |
String sc39 = metadata.getEcsc39().getValue().toString(); |
|
| 509 |
if (sc39.equalsIgnoreCase("true") || sc39.equalsIgnoreCase("t") || sc39.contains("yes")) {
|
|
| 510 |
sc39 = "yes"; |
|
| 511 |
} else if (sc39.equalsIgnoreCase("false") || sc39.equalsIgnoreCase("f") || sc39.contains("no")) {
|
|
| 512 |
sc39 = "no"; |
|
| 513 |
} |
|
| 514 |
|
|
| 515 |
buff += getStringField(sc39, DELIM, ENCLOSING); |
|
| 516 |
|
|
| 517 |
// `url`, |
|
| 518 |
buff += getStringField(metadata.getWebsiteurl().getValue(), DELIM, ENCLOSING); |
|
| 519 |
|
|
| 520 |
// start_year |
|
| 521 |
|
|
| 522 |
buff += getYearInt(metadata.getStartdate().getValue(), DELIM, ENCLOSING); |
|
| 523 |
|
|
| 524 |
// end_year |
|
| 525 |
buff += getYearInt(metadata.getEnddate().getValue(), DELIM, ENCLOSING); |
|
| 526 |
|
|
| 527 |
// duration enddate-startdate |
|
| 528 |
|
|
| 529 |
buff += getYearDifferenceInteger(metadata.getEnddate().getValue(), metadata.getStartdate().getValue(), DELIM, ENCLOSING); |
|
| 530 |
|
|
| 531 |
// haspubs |
|
| 532 |
buff += getStringField("no", DELIM, ENCLOSING);
|
|
| 533 |
|
|
| 534 |
// numpubs |
|
| 535 |
buff += getNumericField("0", DELIM, ENCLOSING);
|
|
| 536 |
// enddate |
|
| 537 |
buff += getNumericField(metadata.getEnddate().getValue(), DELIM, ENCLOSING); |
|
| 538 |
// startdate |
|
| 539 |
buff += getNumericField(metadata.getStartdate().getValue(), DELIM, ENCLOSING); |
|
| 540 |
|
|
| 541 |
// `daysforlastpub`, |
|
| 542 |
buff += getNumericField("", DELIM, ENCLOSING);
|
|
| 543 |
// `delayedpubs`, |
|
| 544 |
buff += getNumericField("", DELIM, ENCLOSING);
|
|
| 545 |
|
|
| 546 |
//call identifier |
|
| 547 |
buff += getStringField(metadata.getCallidentifier().getValue(), DELIM, ENCLOSING); |
|
| 548 |
//code |
|
| 549 |
buff += getStringField(metadata.getCode().getValue(), DELIM, ENCLOSING); |
|
| 550 |
//esc39 |
|
| 551 |
buff += getStringField(metadata.getEcsc39().getValue(), DELIM, ENCLOSING); |
|
| 552 |
|
|
| 553 |
|
|
| 554 |
// deletedByInference |
|
| 555 |
buff += getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING); |
|
| 556 |
return buff; |
|
| 557 |
|
|
| 558 |
} |
|
| 559 |
|
|
| 560 |
private static String buildPerson(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 561 |
OafEntity data = oaf.getEntity(); |
|
| 562 |
String buff = new String(); |
|
| 563 |
|
|
| 564 |
PersonProtos.Person person = data.getPerson(); |
|
| 565 |
PersonProtos.Person.Metadata metadata = person.getMetadata(); |
|
| 566 |
|
|
| 567 |
//firstName |
|
| 568 |
buff += getStringField(metadata.getFirstname().getValue(), DELIM, ENCLOSING); |
|
| 569 |
//LastName |
|
| 570 |
|
|
| 571 |
String secondName = new String(); |
|
| 572 |
for (StringField s : metadata.getSecondnamesList()) {
|
|
| 573 |
secondName += s.getValue() + " "; |
|
| 574 |
} |
|
| 575 |
|
|
| 576 |
buff += getStringField(secondName, DELIM, ENCLOSING); |
|
| 577 |
|
|
| 578 |
// `fullname`, |
|
| 579 |
buff += getStringField(metadata.getFullname().getValue(), DELIM, ENCLOSING); |
|
| 580 |
|
|
| 581 |
// `Nationality`, |
|
| 582 |
buff += getStringField(metadata.getNationality().getClassname(), DELIM, ENCLOSING); |
|
| 583 |
// `Email`, |
|
| 584 |
buff += getStringField(metadata.getEmail().getValue(), DELIM, ENCLOSING); |
|
| 585 |
// `Phone`, |
|
| 586 |
buff += getStringField(metadata.getPhone().getValue(), DELIM, ENCLOSING); |
|
| 587 |
// deletedByInference |
|
| 588 |
buff += getStringField(String.valueOf(oaf.getDataInfo().getDeletedbyinference()), DELIM, ENCLOSING); |
|
| 589 |
|
|
| 590 |
return buff; |
|
| 591 |
|
|
| 592 |
} |
|
| 593 |
|
|
| 594 |
|
|
| 595 |
private static String getYearDifferenceInteger(String enddate, String startdate, String DELIM, String ENCLOSING) {
|
|
| 596 |
|
|
| 597 |
if (enddate != null && !enddate.isEmpty() && startdate != null && !startdate.isEmpty()) {
|
|
| 598 |
|
|
| 599 |
String[] split = startdate.split("-");
|
|
| 600 |
|
|
| 601 |
if (split == null || split.length == 0) {
|
|
| 602 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 603 |
} |
|
| 604 |
|
|
| 605 |
int Startdate = Integer.parseInt(split[0]); |
|
| 606 |
|
|
| 607 |
split = enddate.split("-");
|
|
| 608 |
|
|
| 609 |
if (split == null || split.length == 0) {
|
|
| 610 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 611 |
} |
|
| 612 |
|
|
| 613 |
int Enddate = Integer.parseInt(split[0]); |
|
| 614 |
|
|
| 615 |
int diff = Enddate - Startdate; |
|
| 616 |
|
|
| 617 |
return ENCLOSING + diff + ENCLOSING + DELIM; |
|
| 618 |
|
|
| 619 |
} |
|
| 620 |
|
|
| 621 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 622 |
} |
|
| 623 |
|
|
| 624 |
private static String getYearInt(String data, String DELIM, String ENCLOSING) {
|
|
| 625 |
if (data == null || data.isEmpty() || data.equals("-1")) {
|
|
| 626 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 627 |
} |
|
| 628 |
|
|
| 629 |
String[] split = data.split("-");
|
|
| 630 |
|
|
| 631 |
if (split == null || split.length == 0) {
|
|
| 632 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 633 |
} |
|
| 634 |
|
|
| 635 |
String year = split[0]; |
|
| 636 |
|
|
| 637 |
year = cleanNumber(year); |
|
| 638 |
|
|
| 639 |
return ENCLOSING + year + ENCLOSING + DELIM; |
|
| 640 |
|
|
| 641 |
|
|
| 642 |
} |
|
| 643 |
|
|
| 644 |
private static String cleanNumber(String number) {
|
|
| 645 |
number = number.replaceAll("[^A-Za-z0-9:,____]", "");
|
|
| 646 |
|
|
| 647 |
return number; |
|
| 648 |
} |
|
| 649 |
|
|
| 650 |
private static String getStringField(String data, String DELIM, String ENCLOSING) {
|
|
| 651 |
|
|
| 652 |
if (data == null || data.isEmpty()) {
|
|
| 653 |
|
|
| 654 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 655 |
} else {
|
|
| 656 |
String field = clean(data, DELIM, ENCLOSING); |
|
| 657 |
|
|
| 658 |
if (field == null) {
|
|
| 659 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 660 |
} else {
|
|
| 661 |
|
|
| 662 |
return ENCLOSING + field + ENCLOSING + DELIM; |
|
| 663 |
} |
|
| 664 |
} |
|
| 665 |
} |
|
| 666 |
|
|
| 667 |
private static String getStringDateField(String data, String DELIM, String ENCLOSING) {
|
|
| 668 |
|
|
| 669 |
if (data == null || data.isEmpty() || data.equals("") || data.equals("-1")) {
|
|
| 670 |
|
|
| 671 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 672 |
} else {
|
|
| 673 |
|
|
| 674 |
String field = clean(data, DELIM, ENCLOSING); |
|
| 675 |
if (field == null) {
|
|
| 676 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 677 |
} else {
|
|
| 678 |
return ENCLOSING + field + ENCLOSING + DELIM; |
|
| 679 |
} |
|
| 680 |
} |
|
| 681 |
} |
|
| 682 |
|
|
| 683 |
private static String getNumericField(String data, String DELIM, String ENCLOSING) {
|
|
| 684 |
if (data == null || data.isEmpty() || data.equals("")) {
|
|
| 685 |
return ENCLOSING + "0" + ENCLOSING + DELIM; |
|
| 686 |
} else {
|
|
| 687 |
return ENCLOSING + data + ENCLOSING + DELIM; |
|
| 688 |
} |
|
| 689 |
} |
|
| 690 |
|
|
| 691 |
public static String getId(Oaf oaf, String DELIM, String ENCLOSING) {
|
|
| 692 |
switch (oaf.getKind()) {
|
|
| 693 |
case entity: |
|
| 694 |
|
|
| 695 |
return cleanId(oaf.getEntity().getId(), DELIM, ENCLOSING); |
|
| 696 |
case relation: |
|
| 697 |
|
|
| 698 |
return cleanId(oaf.getRel().getSource(), DELIM, ENCLOSING); |
|
| 699 |
|
|
| 700 |
} |
|
| 701 |
return null; |
|
| 702 |
|
|
| 703 |
} |
|
| 704 |
|
|
| 705 |
public static String getId(OafRel relOaf, String DELIM, String ENCLOSING) {
|
|
| 706 |
return cleanId(relOaf.getSource(), DELIM, ENCLOSING); |
|
| 707 |
} |
|
| 708 |
|
|
| 709 |
public static String clean(String value, String DELIM, String ENCLOSING) {
|
|
| 710 |
if (value != null) {
|
|
| 711 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
| 712 |
// to datacite____:: ) |
|
| 713 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
| 714 |
value = value.replaceFirst(".*\\|", "");
|
|
| 715 |
|
|
| 716 |
value = value.replace(DELIM, " "); |
|
| 717 |
value = value.replaceAll("'", " ");
|
|
| 718 |
value = value.replace("#", " ");
|
|
| 719 |
|
|
| 720 |
value = value.replaceAll("\\r\\n|\\r|\\n", " ");
|
|
| 721 |
value = value.replaceAll("\\s+", " ");
|
|
| 722 |
value = value.replaceAll("(\\r|\\n)", " ");
|
|
| 723 |
value = value.replaceAll("\\t", " ");
|
|
| 724 |
value = value.replace("\n", " ");
|
|
| 725 |
|
|
| 726 |
value = value.replaceAll("[\"\\r\\\\;]", "");
|
|
| 727 |
|
|
| 728 |
value = value.replace("\"", " ");
|
|
| 729 |
|
|
| 730 |
value = value.replaceAll("[^a-zA-Z0-9 .-_:/!@+=]+", " ");
|
|
| 731 |
|
|
| 732 |
} |
|
| 733 |
if (value == null) {
|
|
| 734 |
return null; |
|
| 735 |
} |
|
| 736 |
return value; |
|
| 737 |
|
|
| 738 |
} |
|
| 739 |
|
|
| 740 |
public static String cleanId(String value, String DELIM, String ENCLOSING) {
|
|
| 741 |
if (value != null) {
|
|
| 742 |
// TODO DO NOT CHANGE THIS: IT REMOVES ID PREFIX ( 5|datacite____:: |
|
| 743 |
|
|
| 744 |
// to datacite____:: ) |
|
| 745 |
// AND REPLACES OCCURENCES OF DELIM CHARS IN DATA |
|
| 746 |
value = value.replaceFirst(".*\\|", "");
|
|
| 747 |
value = value.replace("\n", "");
|
|
| 748 |
value = value.replace(DELIM, ""); |
|
| 749 |
value = value.replace("'", " ");
|
|
| 750 |
value = value.replace(ENCLOSING, " "); |
|
| 751 |
|
|
| 752 |
|
|
| 753 |
value = value.trim(); |
|
| 754 |
|
|
| 755 |
} |
|
| 756 |
if (value == null) {
|
|
| 757 |
return null; |
|
| 758 |
} |
|
| 759 |
return ENCLOSING + value + ENCLOSING; |
|
| 760 |
|
|
| 761 |
} |
|
| 762 |
|
|
| 763 |
public static long DATEDIFF(String startDate, String endDate) {
|
|
| 764 |
long MILLISECS_PER_DAY = 24 * 60 * 60 * 1000; |
|
| 765 |
long days = 0l; |
|
| 766 |
SimpleDateFormat format = new SimpleDateFormat("yyyy-MM-dd"); // "dd/MM/yyyy HH:mm:ss");
|
|
| 767 |
// <startdate>2011-09-01</startdate> |
|
| 768 |
// <enddate>2015-08-31</enddate> |
|
| 769 |
Date dateIni = null; |
|
| 770 |
Date dateFin = null; |
|
| 771 |
|
|
| 772 |
if (startDate == null || startDate.isEmpty() || endDate == null || endDate.isEmpty()) {
|
|
| 773 |
return 0; |
|
| 774 |
} |
|
| 775 |
try {
|
|
| 776 |
dateIni = (Date) format.parse(startDate); |
|
| 777 |
dateFin = (Date) format.parse(endDate); |
|
| 778 |
days = (dateFin.getTime() - dateIni.getTime()) / MILLISECS_PER_DAY; |
|
| 779 |
} catch (Exception e) {
|
|
| 780 |
|
|
| 781 |
return 0; |
|
| 782 |
} |
|
| 783 |
|
|
| 784 |
return days; |
|
| 785 |
} |
|
| 786 |
|
|
| 787 |
|
|
| 788 |
} |
|
| 0 | 789 | |
| modules/dnet-openaire-stats/branches/full_export/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/FundingParser.java | ||
|---|---|---|
| 7 | 7 |
*/ |
| 8 | 8 |
|
| 9 | 9 |
public class FundingParser {
|
| 10 |
private String NULL_STRING; |
|
| 11 |
private String ENCLOSED; |
|
| 12 |
private String DELIM ; |
|
| 13 | 10 |
|
| 14 | 11 |
|
| 15 |
public FundingParser(String delim, String enclosed, String nullString) |
|
| 16 |
{
|
|
| 17 |
this.setNULL_STRING(nullString); |
|
| 18 |
this.setENCLOSED(enclosed); |
|
| 19 |
this.setDELIM(delim); |
|
| 12 |
public static String getFundingLevel(String funding_level, int level, String DELIM, String ENCLOSING) {
|
|
| 20 | 13 |
|
| 21 |
} |
|
| 14 |
if (funding_level.isEmpty()) {
|
|
| 15 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 16 |
} |
|
| 22 | 17 |
|
| 23 |
private Logger log = Logger.getLogger(this.getClass()); |
|
| 18 |
if (!funding_level.contains("<funding_level_" + level + ">")) {
|
|
| 19 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 20 |
} |
|
| 24 | 21 |
|
| 25 |
String getFundingLevel(String funding_level, int level) {
|
|
| 22 |
String[] split = funding_level.split("<funding_level_" + level + ">");
|
|
| 26 | 23 |
|
| 27 |
if (funding_level.isEmpty()) {
|
|
| 28 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
| 29 |
} |
|
| 24 |
funding_level = split[1]; |
|
| 30 | 25 |
|
| 31 |
if (!funding_level.contains("<funding_level_" + level + ">")) {
|
|
| 32 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
| 33 |
} |
|
| 26 |
split = funding_level.split("<name>");
|
|
| 27 |
funding_level = split[1]; |
|
| 34 | 28 |
|
| 35 |
String[] split = funding_level.split("<funding_level_" + level + ">");
|
|
| 36 |
|
|
| 37 |
funding_level = split[1]; |
|
| 38 |
|
|
| 39 |
split = funding_level.split("<name>");
|
|
| 40 |
funding_level = split[1]; |
|
| 41 |
|
|
| 42 |
funding_level = funding_level.substring(0, funding_level.indexOf("</name>"));
|
|
| 43 |
funding_level = funding_level.replaceAll("\"", "");
|
|
| 29 |
funding_level = funding_level.substring(0, funding_level.indexOf("</name>"));
|
|
| 30 |
funding_level = funding_level.replaceAll("\"", "");
|
|
| 44 | 31 |
funding_level = funding_level.replaceAll("/>", "");
|
| 45 | 32 |
funding_level = funding_level.replaceAll("<", "");
|
| 46 |
funding_level = funding_level.replaceAll("&", "");
|
|
| 33 |
funding_level = funding_level.replaceAll("&", "");
|
|
| 47 | 34 |
|
| 48 |
if (level == 1) {
|
|
| 49 |
if (funding_level.equalsIgnoreCase("SP1")) {
|
|
| 50 |
funding_level = "SP1-Cooperation";
|
|
| 51 |
} else if (funding_level.equalsIgnoreCase("SP2")) {
|
|
| 52 |
funding_level = "SP2-Ideas";
|
|
| 53 |
}
|
|
| 54 |
if (funding_level.equalsIgnoreCase("SP3")) {
|
|
| 55 |
funding_level = "SP3-People";
|
|
| 56 |
} else if (funding_level.equalsIgnoreCase("SP4")) {
|
|
| 57 |
funding_level = "SP4-Capacities";
|
|
| 35 |
if (level == 1) {
|
|
| 36 |
if (funding_level.equalsIgnoreCase("SP1")) {
|
|
| 37 |
funding_level = "SP1-Cooperation";
|
|
| 38 |
} else if (funding_level.equalsIgnoreCase("SP2")) {
|
|
| 39 |
funding_level = "SP2-Ideas";
|
|
| 40 |
}
|
|
| 41 |
if (funding_level.equalsIgnoreCase("SP3")) {
|
|
| 42 |
funding_level = "SP3-People";
|
|
| 43 |
} else if (funding_level.equalsIgnoreCase("SP4")) {
|
|
| 44 |
funding_level = "SP4-Capacities";
|
|
| 58 | 45 |
|
| 59 |
} else if (funding_level.equalsIgnoreCase("SP5")) {
|
|
| 60 |
funding_level = "SP5-Euratom";
|
|
| 61 |
}
|
|
| 62 |
}
|
|
| 46 |
} else if (funding_level.equalsIgnoreCase("SP5")) {
|
|
| 47 |
funding_level = "SP5-Euratom";
|
|
| 48 |
}
|
|
| 49 |
}
|
|
| 63 | 50 |
|
| 64 | 51 |
|
| 65 |
funding_level = funding_level.replaceAll(">", "");
|
|
| 52 |
funding_level = funding_level.replaceAll(">", "");
|
|
| 66 | 53 |
|
| 67 |
funding_level = funding_level.replaceAll("</", "");
|
|
| 68 |
funding_level=funding_level.replace(DELIM," ");
|
|
| 69 |
funding_level=funding_level.replace(ENCLOSED,"");
|
|
| 54 |
funding_level = funding_level.replaceAll("</", "");
|
|
| 55 |
funding_level = funding_level.replace(DELIM, " ");
|
|
| 56 |
funding_level = funding_level.replace(ENCLOSING, " ");
|
|
| 70 | 57 |
|
| 58 |
return ENCLOSING + funding_level + ENCLOSING + DELIM; |
|
| 59 |
} |
|
| 71 | 60 |
|
| 72 |
return ENCLOSED + funding_level + ENCLOSED + DELIM; |
|
| 73 |
} |
|
| 61 |
public static String getFundingInfo(String buff, String DELIM, String ENCLOSING) {
|
|
| 62 |
return getFunder(buff, DELIM, ENCLOSING) + |
|
| 63 |
getFundingLevel(buff, 0, DELIM, ENCLOSING) + (getFundingLevel(buff, 1, DELIM, ENCLOSING) + getFundingLevel(buff, 2, DELIM, ENCLOSING) |
|
| 64 |
+ getFundingLevel(buff, 3, DELIM, ENCLOSING)); |
|
| 65 |
} |
|
| 74 | 66 |
|
| 75 |
public String getFundingInfo(String buff) {
|
|
| 76 |
return getFunder(buff) + getFundingLevel(buff, 0) + (getFundingLevel(buff, 1) + getFundingLevel(buff, 2) + getFundingLevel(buff, 3)); |
|
| 77 |
} |
|
| 67 |
public static String getFunder(String buff, String DELIM, String ENCLOSING) {
|
|
| 78 | 68 |
|
| 79 |
public String getFunder(String buff) {
|
|
| 69 |
if (buff.isEmpty()) {
|
|
| 70 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 80 | 71 |
|
| 72 |
} |
|
| 73 |
if (!buff.contains("<funder>")) {
|
|
| 74 |
return ENCLOSING + " " + ENCLOSING + DELIM; |
|
| 75 |
} |
|
| 81 | 76 |
|
| 82 |
if (buff.isEmpty()) {
|
|
| 83 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
| 77 |
String[] split = buff.split("<funder>");
|
|
| 78 |
String funder = split[1]; |
|
| 79 |
split = funder.split("<name>");
|
|
| 84 | 80 |
|
| 85 |
}
|
|
| 81 |
funder = split[1];
|
|
| 86 | 82 |
|
| 87 |
if (!buff.contains("<funder>")) {
|
|
| 88 |
return ENCLOSED + NULL_STRING + ENCLOSED + DELIM; |
|
| 89 |
} |
|
| 90 |
String[] split = buff.split("<funder>");
|
|
| 91 |
String funder = split[1]; |
|
| 83 |
funder = funder.substring(0, funder.indexOf("</name>"));
|
|
| 92 | 84 |
|
| 93 |
split = funder.split("<name>");
|
|
| 85 |
funder = funder.replaceAll(">", "");
|
|
| 94 | 86 |
|
| 95 |
funder = split[1];
|
|
| 87 |
funder = funder.replaceAll("</", "");
|
|
| 96 | 88 |
|
| 97 |
funder = funder.substring(0, funder.indexOf("</name>"));
|
|
| 89 |
funder = funder.replaceAll("\"", "");
|
|
| 90 |
funder = funder.replaceAll("&", "");
|
|
| 91 |
funder = funder.replace(ENCLOSING, " "); |
|
| 98 | 92 |
|
| 99 |
funder = funder.replaceAll(">", "");
|
|
| 93 |
return ENCLOSING + funder + ENCLOSING + DELIM; |
|
| 94 |
} |
|
| 100 | 95 |
|
| 101 |
funder = funder.replaceAll("</", "");
|
|
| 102 |
|
|
| 103 |
funder = funder.replaceAll("\"", "");
|
|
| 104 |
funder = funder.replaceAll("&", "");
|
|
| 105 |
|
|
| 106 |
|
|
| 107 |
return ENCLOSED + funder + ENCLOSED + DELIM; |
|
| 108 |
} |
|
| 109 |
|
|
| 110 |
public String getNULL_STRING() {
|
|
| 111 |
return NULL_STRING; |
|
| 112 |
} |
|
| 113 |
|
|
| 114 |
public void setNULL_STRING(String NULL_STRING) {
|
|
| 115 |
this.NULL_STRING = NULL_STRING; |
|
| 116 |
} |
|
| 117 |
|
|
| 118 |
public String getENCLOSED() {
|
|
| 119 |
return ENCLOSED; |
|
| 120 |
} |
|
| 121 |
|
|
| 122 |
public void setENCLOSED(String ENCLOSED) {
|
|
| 123 |
this.ENCLOSED = ENCLOSED; |
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
public String getDELIM() {
|
|
| 127 |
return DELIM; |
|
| 128 |
} |
|
| 129 |
|
|
| 130 |
public void setDELIM(String DELIM) {
|
|
| 131 |
this.DELIM = DELIM; |
|
| 132 |
} |
|
| 133 | 96 |
} |
| 134 | 97 |
|
| modules/dnet-openaire-stats/branches/full_export/src/main/java/eu/dnetlib/data/mapreduce/hbase/statsExport/utils/Serializer.java | ||
|---|---|---|
| 1 |
/* |
|
| 1 | 2 |
package eu.dnetlib.data.mapreduce.hbase.statsExport.utils; |
| 2 | 3 |
|
| 4 |
import com.google.common.collect.ArrayListMultimap; |
|
| 5 |
import com.google.common.collect.Multimap; |
|
| 3 | 6 |
import eu.dnetlib.data.mapreduce.util.LicenseComparator; |
| 4 | 7 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource; |
| 5 | 8 |
import eu.dnetlib.data.proto.DatasourceProtos.Datasource.Metadata; |
| ... | ... | |
| 26 | 29 |
import java.util.HashMap; |
| 27 | 30 |
import java.util.List; |
| 28 | 31 |
|
| 32 |
*/ |
|
| 29 | 33 |
/** |
| 30 | 34 |
* @author eri Simple serializer that parses input Oaf Protos and prepares them |
| 31 | 35 |
* for sqoop |
| 32 |
*/ |
|
| 36 |
*//* |
|
| 37 |
|
|
| 33 | 38 |
public class Serializer {
|
| 34 | 39 |
|
| 35 | 40 |
private static String DELIM; |
| 41 |
//TODO replae here later |
|
| 42 |
private String ENCLOSING = "'"; |
|
| 36 | 43 |
private Logger log = Logger.getLogger(this.getClass()); |
| 37 |
private String NULL_STRING; |
|
| 38 |
private String NULL_NUM; |
|
| 44 |
private String NULL_STRING = "";
|
|
| 45 |
private String NULL_NUM = "0";
|
|
| 39 | 46 |
//TODO no longer used |
| 47 |
*/ |
|
| 48 |
/* |
|
| 40 | 49 |
private static String ENCLOSED; |
| 50 |
*//* |
|
| 41 | 51 |
|
| 52 |
|
|
| 42 | 53 |
private FundingParser fundingParser; |
| 43 | 54 |
|
| 44 |
|
|
| 45 | 55 |
public Serializer(String delim, String nullNum, String nullString, String enclosed) {
|
| 46 |
|
|
| 47 | 56 |
this.setDELIM(delim); |
| 48 |
this.setNULL_NUM(nullNum);
|
|
| 49 |
this.setENCLOSED(enclosed);
|
|
| 50 |
this.setNULL_STRING(nullString); |
|
| 57 |
*/
|
|
| 58 |
/*this.setNULL_NUM(nullNum);
|
|
| 59 |
this.setNULL_STRING(nullString);*//*
|
|
| 51 | 60 |
|
| 52 |
fundingParser=new FundingParser(delim, enclosed,nullString);
|
|
| 61 |
fundingParser = new FundingParser(delim, enclosed, nullString);
|
|
| 53 | 62 |
|
| 54 | 63 |
} |
| 55 | 64 |
|
| ... | ... | |
| 63 | 72 |
switch (valueEntity.getType()) {
|
| 64 | 73 |
case datasource: |
| 65 | 74 |
|
| 66 |
return buildDatasource(valueEntity);
|
|
| 75 |
return buildDatasource(oaf);
|
|
| 67 | 76 |
|
| 68 | 77 |
case organization: |
| 69 | 78 |
|
| 70 |
return buildOrganization(valueEntity);
|
|
| 79 |
return buildOrganization(oaf);
|
|
| 71 | 80 |
|
| 72 | 81 |
case project: |
| 73 | 82 |
|
| 74 |
return buildProject(valueEntity);
|
|
| 75 |
case result: |
|
| 83 |
return buildProject(oaf);
|
|
| 84 |
case result:`
|
|
| 76 | 85 |
|
| 77 |
return buildResult(valueEntity);
|
|
| 86 |
return buildResult(oaf);
|
|
| 78 | 87 |
case person: |
| 79 |
return buildPerson(valueEntity);
|
|
| 88 |
return buildPerson(oaf);
|
|
| 80 | 89 |
default: |
| 81 | 90 |
log.error("wrong type");
|
| 82 | 91 |
break; |
| ... | ... | |
| 109 | 118 |
} |
| 110 | 119 |
|
| 111 | 120 |
|
| 112 |
public HashMap<String, List<String>> extractRelations(Oaf oaf) {
|
|
| 121 |
public Multimap<String, String> extractRelations(Oaf oaf) {
|
|
| 113 | 122 |
OafEntity valueEntity = oaf.getEntity(); |
| 114 |
HashMap<String, List<String>> relations = new HashMap<String, List<String>>(); |
|
| 115 |
relations.putAll(getOriginalId(valueEntity)); |
|
| 123 |
Multimap<String, String> relations = ArrayListMultimap.create(); |
|
| 116 | 124 |
|
| 125 |
getOriginalId(valueEntity, relations); |
|
| 126 |
|
|
| 117 | 127 |
switch (valueEntity.getType()) {
|
| 118 | 128 |
case datasource: |
| 119 |
relations.putAll(getDatasourceLanguages(valueEntity));
|
|
| 129 |
getDatasourceLanguages(valueEntity, relations);
|
|
| 120 | 130 |
case result: |
| 121 |
relations.putAll(getResultTopics(valueEntity));
|
|
| 122 |
relations.putAll(getResultLanguages(valueEntity));
|
|
| 123 |
relations.putAll(getResultClassifications(valueEntity));
|
|
| 124 |
relations.putAll(getResultDatasources(valueEntity));
|
|
| 125 |
relations.putAll(getResultConcepts(valueEntity));
|
|
| 126 |
relations.putAll(getResultDois(valueEntity));
|
|
| 127 |
relations.putAll(getResultCitations(valueEntity));
|
|
| 128 |
relations.putAll(getResultDescriptions(valueEntity));
|
|
| 131 |
getResultTopics(valueEntity, relations);
|
|
| 132 |
getResultLanguages(valueEntity, relations);
|
|
| 133 |
getResultClassifications(valueEntity, relations);
|
|
| 134 |
getResultDatasources(valueEntity, relations);
|
|
| 135 |
getResultConcepts(valueEntity, relations);
|
|
| 136 |
getResultDois(valueEntity, relations);
|
|
| 137 |
getResultCitations(valueEntity, relations);
|
|
| 138 |
getResultDescriptions(valueEntity, relations);
|
|
| 129 | 139 |
return relations; |
| 130 | 140 |
|
| 131 | 141 |
case project: |
| 132 |
relations.putAll(getProjectKeywords(valueEntity));
|
|
| 133 |
relations.putAll(getProjectSubjects(valueEntity));
|
|
| 142 |
getProjectKeywords(valueEntity, relations);
|
|
| 143 |
getProjectSubjects(valueEntity, relations);
|
|
| 134 | 144 |
|
| 135 | 145 |
default: |
| 136 | 146 |
return null; |
| ... | ... | |
| 139 | 149 |
} |
| 140 | 150 |
|
| 141 | 151 |
|
| 152 |
private void getOriginalId(OafEntity oafEntity, Multimap<String, String> relations) {
|
|
| 142 | 153 |
|
| 143 |
private HashMap<String, List<String>> getOriginalId(OafEntity oafEntity) {
|
|
| 144 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 145 |
List<String> buffs = new ArrayList<String>(); |
|
| 146 |
|
|
| 147 |
|
|
| 154 |
String relName = oafEntity.getType().toString().toLowerCase() + "Oid"; |
|
| 148 | 155 |
for (String oid : oafEntity.getOriginalIdList()) {
|
| 149 |
buffs.add(oid.replace(DELIM, " ").replace("\n", "") + DELIM);
|
|
| 156 |
relations.put(relName, ENCLOSING + oid.replace(DELIM, " ").replace("\n", "") + ENCLOSING + DELIM);
|
|
| 150 | 157 |
} |
| 151 | 158 |
|
| 152 |
rels.put(oafEntity.getType().toString().toLowerCase() + "Oid", buffs); |
|
| 153 |
|
|
| 154 |
return rels; |
|
| 155 | 159 |
} |
| 156 | 160 |
|
| 157 |
private HashMap<String, List<String>> getProjectKeywords(OafEntity oafEntity) {
|
|
| 158 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 159 |
List<String> buffs = new ArrayList<String>(); |
|
| 161 |
private void getProjectKeywords(OafEntity oafEntity, Multimap<String, String> relations) {
|
|
| 162 |
relations.put("projectKeyword", getStringField(oafEntity.getProject().getMetadata().getKeywords().getValue()));
|
|
| 160 | 163 |
|
| 161 |
// String[] keywords = oafEntity.getProject().getMetadata().getKeywords().getValue().replace(";", ",").split(",");
|
|
| 162 |
|
|
| 163 |
/* for (String keyword : keywords) {
|
|
| 164 |
buffs.add(getStringField(keyword)); |
|
| 165 |
}*/ |
|
| 166 |
|
|
| 167 |
|
|
| 168 |
buffs.add(getStringField(oafEntity.getProject().getMetadata().getKeywords().getValue())); |
|
| 169 |
rels.put("projectKeyword", buffs);
|
|
| 170 |
return rels; |
|
| 171 | 164 |
} |
| 172 | 165 |
|
| 173 |
private HashMap<String, List<String>> getProjectSubjects(OafEntity oafEntity) {
|
|
| 174 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 175 |
List<String> buffs = new ArrayList<String>(); |
|
| 176 |
|
|
| 177 |
|
|
| 166 |
private void getProjectSubjects(OafEntity oafEntity, Multimap<String, String> relations) {
|
|
| 178 | 167 |
for (StructuredProperty subj : oafEntity.getProject().getMetadata().getSubjectsList()) {
|
| 179 |
buffs.add(getStringField(subj.getValue()));
|
|
| 168 |
relations.put("projectSubject", getStringField(subj.getValue()));
|
|
| 180 | 169 |
} |
| 181 |
|
|
| 182 |
rels.put("projectSubject", buffs);
|
|
| 183 |
|
|
| 184 |
return rels; |
|
| 185 | 170 |
} |
| 186 | 171 |
|
| 187 |
|
|
| 188 | 172 |
private String getResultProject(OafRel oaf) {
|
| 189 | 173 |
String buff = new String(); |
| 190 | 174 |
String result = oaf.getTarget(); |
| 191 | 175 |
|
| 192 |
|
|
| 193 | 176 |
buff += getStringField(result); |
| 194 | 177 |
// TODO is declared as int!!! |
| 195 | 178 |
long diff = DATEDIFF(oaf.getResultProject().getOutcome().getRelMetadata().getEnddate(), oaf.getResultProject().getOutcome().getRelMetadata().getStartdate()); |
| ... | ... | |
| 201 | 184 |
} |
| 202 | 185 |
|
| 203 | 186 |
|
| 204 |
private HashMap<String, List<String>> getDatasourceLanguages(OafEntity valueEntity) {
|
|
| 205 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 206 |
List<String> buffs = new ArrayList<String>(); |
|
| 207 |
|
|
| 208 |
|
|
| 187 |
private void getDatasourceLanguages(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 209 | 188 |
Datasource d = valueEntity.getDatasource(); |
| 210 |
|
|
| 211 | 189 |
Metadata metadata = d.getMetadata(); |
| 212 | 190 |
|
| 213 | 191 |
for (StringField lang : metadata.getOdlanguagesList()) {
|
| 214 |
|
|
| 215 |
buffs.add(getStringField(lang.getValue())); |
|
| 192 |
rels.put("datasourceLanguage", getStringField(lang.getValue()));
|
|
| 216 | 193 |
} |
| 217 |
rels.put("datasourceLanguage", buffs);
|
|
| 218 |
return rels; |
|
| 219 | 194 |
} |
| 220 | 195 |
|
| 221 |
private HashMap<String, List<String>> getResultLanguages(OafEntity valueEntity) {
|
|
| 222 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 223 |
List<String> buffs = new ArrayList<String>(); |
|
| 196 |
private void getResultLanguages(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 197 |
|
|
| 224 | 198 |
Result d = valueEntity.getResult(); |
| 225 |
|
|
| 226 | 199 |
eu.dnetlib.data.proto.ResultProtos.Result.Metadata metadata = d.getMetadata(); |
| 227 | 200 |
|
| 228 | 201 |
if (metadata.getLanguage().getClassname() != null && !metadata.getLanguage().getClassname().isEmpty()) {
|
| 229 | 202 |
|
| 230 |
buffs.add(getStringField(metadata.getLanguage().getClassname()));
|
|
| 203 |
rels.put("resultLanguage", getStringField(metadata.getLanguage().getClassname()));
|
|
| 231 | 204 |
} |
| 232 |
rels.put("resultLanguage", buffs);
|
|
| 233 |
return rels; |
|
| 234 | 205 |
|
| 235 | 206 |
} |
| 236 | 207 |
|
| 208 |
private void getResultDois(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 237 | 209 |
|
| 238 |
private HashMap<String, List<String>> getResultDois(OafEntity valueEntity) {
|
|
| 239 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 240 |
List<String> buffs = new ArrayList<String>(); |
|
| 241 |
|
|
| 242 | 210 |
for (StructuredProperty pid : valueEntity.getPidList()) {
|
| 243 |
buffs.add(getStringField(pid.getQualifier().getClassname()) + getStringField(pid.getValue())); |
|
| 244 | 211 |
|
| 212 |
rels.put("resultPid",
|
|
| 213 |
getStringField(pid.getQualifier().getClassname()) + getStringField(pid.getValue())); |
|
| 245 | 214 |
} |
| 246 |
|
|
| 247 |
rels.put("resultPid", buffs);
|
|
| 248 |
return rels; |
|
| 249 |
|
|
| 250 | 215 |
} |
| 251 | 216 |
|
| 252 |
private HashMap<String, List<String>> getResultClassifications(OafEntity valueEntity) {
|
|
| 217 |
private void getResultClassifications(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 253 | 218 |
|
| 254 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 255 |
List<String> buffs = new ArrayList<String>(); |
|
| 256 | 219 |
Result result = valueEntity.getResult(); |
| 257 | 220 |
|
| 258 | 221 |
for (Instance instance : (result.getInstanceList())) {
|
| 259 | 222 |
String classification = instance.getInstancetype().getClassname(); |
| 260 | 223 |
if (classification != null && !classification.isEmpty()) {
|
| 261 |
buffs.add(getStringField(instance.getInstancetype().getClassname()));
|
|
| 224 |
rels.put("resultClassification", getStringField(instance.getInstancetype().getClassname()));
|
|
| 262 | 225 |
// TODO HERE KEEP ONLY ONE CLASSIFICATIONS PER RESULT |
| 263 | 226 |
break; |
| 264 | 227 |
} |
| 265 | 228 |
} |
| 266 |
rels.put("resultClassification", buffs);
|
|
| 267 |
return rels; |
|
| 268 |
|
|
| 269 | 229 |
} |
| 270 | 230 |
|
| 271 |
private HashMap<String, List<String>> getResultDescriptions(OafEntity valueEntity) {
|
|
| 231 |
private void getResultDescriptions(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 232 |
Result result = valueEntity.getResult(); |
|
| 233 |
//description |
|
| 272 | 234 |
|
| 273 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 274 |
List<String> buffs = new ArrayList<String>(); |
|
| 275 |
Result result = valueEntity.getResult(); |
|
| 235 |
for (StringField s : result.getMetadata().getDescriptionList()) {
|
|
| 276 | 236 |
|
| 237 |
rels.put("resultDescription", getStringField(Jsoup.parse(s.getValue()).text()));
|
|
| 238 |
} |
|
| 239 |
} |
|
| 277 | 240 |
|
| 278 |
//description |
|
| 279 |
for (StringField s : result.getMetadata().getDescriptionList()) {
|
|
| 280 |
buffs.add(getStringField( Jsoup.parse(s.getValue()).text())); |
|
| 281 |
} |
|
| 241 |
private void getResultConcepts(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 282 | 242 |
|
| 283 |
|
|
| 284 |
rels.put("resultDescription", buffs);
|
|
| 285 |
return rels; |
|
| 286 |
|
|
| 287 |
} |
|
| 288 |
|
|
| 289 |
|
|
| 290 |
|
|
| 291 |
|
|
| 292 |
private HashMap<String, List<String>> getResultConcepts(OafEntity valueEntity) {
|
|
| 293 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 294 |
List<String> buffs = new ArrayList<String>(); |
|
| 295 |
|
|
| 296 | 243 |
Result result = valueEntity.getResult(); |
| 297 | 244 |
|
| 298 | 245 |
for (eu.dnetlib.data.proto.ResultProtos.Result.Context context : result.getMetadata().getContextList()) {
|
| 299 | 246 |
|
| 300 |
buffs.add(getStringField(context.getId()));
|
|
| 247 |
rels.put("resultConcept", getStringField(context.getId()));
|
|
| 301 | 248 |
} |
| 302 |
rels.put("resultConcept", buffs);
|
|
| 303 |
return rels; |
|
| 304 |
|
|
| 305 | 249 |
} |
| 306 | 250 |
|
| 307 |
private HashMap<String, List<String>> getResultDatasources(OafEntity valueEntity) {
|
|
| 308 | 251 |
|
| 309 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
| 310 |
List<String> buffs = new ArrayList<String>(); |
|
| 252 |
private void getResultDatasources(OafEntity valueEntity, Multimap<String, String> rels) {
|
|
| 311 | 253 |
Result result = valueEntity.getResult(); |
| 312 | 254 |
|
| 313 | 255 |
//TODO hosted by |
| 314 | 256 |
for (Instance instance : (result.getInstanceList())) {
|
| 315 | 257 |
String hostedBy = instance.getHostedby().getKey(); |
| 258 |
|
|
| 316 | 259 |
if (hostedBy != null && !hostedBy.isEmpty()) {
|
| 317 |
buffs.add((getStringField(hostedBy)));
|
|
| 260 |
rels.put("resultDatasource", getStringField(hostedBy));
|
|
| 318 | 261 |
} |
| 319 | 262 |
} |
| 320 | 263 |
|
| ... | ... | |
| 322 | 265 |
for (FieldTypeProtos.KeyValue collectedFromValue : (valueEntity.getCollectedfromList())) {
|
| 323 | 266 |
|
| 324 | 267 |
String collectedFrom = collectedFromValue.getKey(); |
| 325 |
if (collectedFrom != null && !collectedFrom.isEmpty()) buffs.add((getStringField(collectedFrom))); |
|
| 326 |
|
|
| 268 |
if (collectedFrom != null && !collectedFrom.isEmpty()) {
|
|
| 269 |
rels.put("resultDatasource", getStringField(collectedFrom));
|
|
| 270 |
} |
|
| 327 | 271 |
} |
| 328 |
rels.put("resultDatasource", buffs);
|
|
| 329 |
return rels; |
|
| 330 |
|
|
| 331 | 272 |
} |
| 332 | 273 |
|
| 333 |
private HashMap<String, List<String>> getResultTopics(OafEntity valueEntity) {
|
|
| 334 |
HashMap<String, List<String>> rels = new HashMap<String, List<String>>(); |
|
Also available in: Unified diff
bundle commit