Revision 51020
Added by Michele Artini about 6 years ago
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/xml/ISTIUtilityFunction.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.xml; |
|
| 2 |
|
|
| 3 |
import java.util.Arrays; |
|
| 4 |
import java.util.regex.Matcher; |
|
| 5 |
import java.util.regex.Pattern; |
|
| 6 |
import java.util.stream.Collectors; |
|
| 7 |
|
|
| 8 |
import org.apache.commons.lang3.StringUtils; |
|
| 9 |
import org.apache.commons.logging.Log; |
|
| 10 |
import org.apache.commons.logging.LogFactory; |
|
| 11 |
|
|
| 12 |
import eu.dnetlib.data.mdstore.plugins.objects.Project; |
|
| 13 |
|
|
| 14 |
/** |
|
| 15 |
* Created by sandro on 12/6/16. |
|
| 16 |
*/ |
|
| 17 |
public class ISTIUtilityFunction {
|
|
| 18 |
|
|
| 19 |
private static final Log log = LogFactory.getLog(ISTIUtilityFunction.class); |
|
| 20 |
|
|
| 21 |
public static String cleanNames(final String s) {
|
|
| 22 |
if (StringUtils.isBlank(s)) { return ""; }
|
|
| 23 |
return cleanNames(s, StringUtils.countMatches(s, ",") > 1 ? "," : ";"); |
|
| 24 |
} |
|
| 25 |
|
|
| 26 |
private static String cleanNames(final String s, final String sep) {
|
|
| 27 |
return Arrays.stream(s.split(sep)) |
|
| 28 |
.map(String::trim) |
|
| 29 |
.map(ISTIUtilityFunction::clean) |
|
| 30 |
.map(ISTIUtilityFunction::capitalize) |
|
| 31 |
.filter(StringUtils::isNotBlank) |
|
| 32 |
.collect(Collectors.joining("#"));
|
|
| 33 |
} |
|
| 34 |
|
|
| 35 |
private static String clean(final String s) {
|
|
| 36 |
return s.replaceAll("\\(.*\\)", "")
|
|
| 37 |
.replaceAll("\\[.*\\]", "")
|
|
| 38 |
.replaceAll("(?i)^et\\.? al(\\.|\\s)*$", "")
|
|
| 39 |
.replaceAll("(?i)\\s*et\\.? al(\\.|\\s)*$", "")
|
|
| 40 |
.replaceAll("\\d|\\*", "")
|
|
| 41 |
.replaceAll("^(\\s|\\-|\\.)+", "")
|
|
| 42 |
.replaceAll("\\,", "")
|
|
| 43 |
.replaceAll("\\.", ". ")
|
|
| 44 |
.trim(); |
|
| 45 |
} |
|
| 46 |
|
|
| 47 |
public static String cleanDoi(final String doi) {
|
|
| 48 |
final String x = doi.replaceAll("\\?", "")
|
|
| 49 |
.replaceAll("\\s*\\/\\s*", "/")
|
|
| 50 |
.replaceAll("\\s*\\-\\s*", "-")
|
|
| 51 |
.trim() |
|
| 52 |
.replaceAll("\\s", "_");
|
|
| 53 |
|
|
| 54 |
if (!x.equals(doi)) {
|
|
| 55 |
log.info("Cleaning doi: " + doi + " -> " + x);
|
|
| 56 |
} |
|
| 57 |
|
|
| 58 |
return x; |
|
| 59 |
} |
|
| 60 |
|
|
| 61 |
public static String capitalize(final String s) {
|
|
| 62 |
return Arrays.stream(s.split(" "))
|
|
| 63 |
.map(String::toLowerCase) |
|
| 64 |
.map(StringUtils::capitalize) |
|
| 65 |
.map(ISTIUtilityFunction::fixApostrophes) |
|
| 66 |
.collect(Collectors.joining(" "));
|
|
| 67 |
} |
|
| 68 |
|
|
| 69 |
public static String fixApostrophes(final String s) {
|
|
| 70 |
return s.replaceAll("'a", "'A").replaceAll("'e", "'E").replaceAll("'i", "'I").replaceAll("'o", "'O").replaceAll("'u", "'U")
|
|
| 71 |
.replaceAll("a'", "à").replaceAll("e'", "è").replaceAll("i'", "ì").replaceAll("o'", "ò").replaceAll("u'", "ù");
|
|
| 72 |
} |
|
| 73 |
|
|
| 74 |
public static boolean isValidProject(final String id) {
|
|
| 75 |
return Project.isValid(id); |
|
| 76 |
} |
|
| 77 |
|
|
| 78 |
public static String calculatePersonName(final String s) {
|
|
| 79 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+):(.+)\\/(.+)\\/(.+)");
|
|
| 80 |
final Matcher matcher = pattern.matcher(s); |
|
| 81 |
return matcher.find() ? capitalize(String.format("%s %s", matcher.group(4), matcher.group(3))) : "";
|
|
| 82 |
} |
|
| 83 |
|
|
| 84 |
// <xsl:for-each select="tokenize(istiFunction:cleanName(.), ';')"> |
|
| 85 |
// <xsl:choose> |
|
| 86 |
// <xsl:when test="matches(normalize-space(.), ',(\s*[a-zA-Z]\.)+$')"> |
|
| 87 |
// <creator> |
|
| 88 |
// <creatorName> |
|
| 89 |
// <xsl:value-of select="normalize-space(translate(.,',',' '))"/> |
|
| 90 |
// </creatorName> |
|
| 91 |
// </creator> |
|
| 92 |
// </xsl:when> |
|
| 93 |
// |
|
| 94 |
// <xsl:when test="matches(normalize-space(.), '(^[a-zA-Z\.]+,\s?[a-zA-Z\.\s]+$)|(^[a-zA-Z\.\s]+,\s?[a-zA-Z\.]+$)')"> |
|
| 95 |
// <creator> |
|
| 96 |
// <creatorName> |
|
| 97 |
// <xsl:value-of select="normalize-space(translate(.,',',' '))"/> |
|
| 98 |
// </creatorName> |
|
| 99 |
// </creator> |
|
| 100 |
// </xsl:when> |
|
| 101 |
// |
|
| 102 |
// <xsl:otherwise> |
|
| 103 |
// <xsl:for-each select="tokenize(., ',')"> |
|
| 104 |
// <creator> |
|
| 105 |
// <creatorName> |
|
| 106 |
// <xsl:value-of select="normalize-space(.)"/> |
|
| 107 |
// </creatorName> |
|
| 108 |
// </creator> |
|
| 109 |
// </xsl:for-each> |
|
| 110 |
// </xsl:otherwise> |
|
| 111 |
// </xsl:choose> |
|
| 112 |
// </xsl:for-each> |
|
| 113 |
// |
|
| 114 |
|
|
| 115 |
} |
|
| modules/dnet-isti/trunk/src/main/resources/eu/dnetlib/bootstrap/profiles/MDFormatDSResources/datacite_mdformat.xml | ||
|---|---|---|
| 1 |
<RESOURCE_PROFILE> |
|
| 2 |
<HEADER> |
|
| 3 |
<RESOURCE_IDENTIFIER |
|
| 4 |
value="98d4c3c4-4274-4250-89b5-434195b20ba8_TURGb3JtYXREU1Jlc291cmNlcy9NREZvcm1hdERTUmVzb3VyY2VUeXBl"/> |
|
| 5 |
<RESOURCE_TYPE value="MDFormatDSResourceType"/> |
|
| 6 |
<RESOURCE_KIND value="MDFormatDSResources"/> |
|
| 7 |
<RESOURCE_URI value=""/> |
|
| 8 |
<DATE_OF_CREATION value="2016-02-01T15:51:45+01:00"/> |
|
| 9 |
</HEADER> |
|
| 10 |
<BODY> |
|
| 11 |
<CONFIGURATION> |
|
| 12 |
<NAME>PMF</NAME> |
|
| 13 |
<DESCRIPTION>ISTI Portal Metadata Format</DESCRIPTION> |
|
| 14 |
<INTERPRETATION>final</INTERPRETATION> |
|
| 15 |
<SCHEMA uri=""/> |
|
| 16 |
</CONFIGURATION> |
|
| 17 |
<STATUS> |
|
| 18 |
<LAYOUTS> |
|
| 19 |
<LAYOUT name="index"> |
|
| 20 |
<FIELDS> |
|
| 21 |
<FIELD indexable="true" name="objidentifier" result="false" stored="true" stat="false" tokenizable="false" xpath="//dri:objIdentifier"/> |
|
| 22 |
<FIELD indexable="true" name="creator" result="false" stored="true" stat="false" tokenizable="true" xpath="//*[local-name() = 'creator']/*[local-name()='creatorName']"/> |
|
| 23 |
<FIELD indexable="true" name="cnrcreator" result="false" stored="true" stat="false" tokenizable="true" xpath="//*[local-name() = 'person']/*[local-name()='personName']"/> |
|
| 24 |
<FIELD indexable="true" name="title" result="false" stored="true" stat="false" tokenizable="true" xpath="//*[local-name() = 'title']"/> |
|
| 25 |
<FIELD indexable="true" name="description" result="false" stored="true" stat="false" tokenizable="true" xpath="//*[local-name() = 'description']"/> |
|
| 26 |
<FIELD indexable="true" name="subject" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'subject']"/> |
|
| 27 |
<FIELD indexable="true" name="language" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'language']"/> |
|
| 28 |
<FIELD indexable="true" name="type" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'resourceType']"/> |
|
| 29 |
<FIELD indexable="true" name="date" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'date']"/> |
|
| 30 |
<FIELD indexable="true" name="affiliation" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'affiliation']"/> |
|
| 31 |
<FIELD indexable="true" name="group" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'affiliation' and @type = 'Servizio']/@group"/> |
|
| 32 |
<FIELD indexable="true" name="laboratory" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'affiliation' and (@type = 'Laboratorio' or @type = 'Centro')]/@group"/> |
|
| 33 |
<FIELD indexable="true" name="creatorforbrowsing" result="false" stored="false" stat="false" tokenizable="false" xpath="//*[local-name() = 'creator']/*[local-name()='creatorName']"/> |
|
| 34 |
<FIELD indexable="true" name="cnrcreatorforbrowsing" result="false" stored="false" stat="false" tokenizable="false" xpath="//*[local-name() = 'person']/*[local-name()='personName']"/> |
|
| 35 |
<FIELD indexable="true" name="doi" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'alternateIdentifier' and @alternateIdentifierType='doi']"/> |
|
| 36 |
<FIELD indexable="true" name="url" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'alternateIdentifier' and @alternateIdentifierType='url']/concat(@licence, '||', @hostedBy, '||', text())"/> |
|
| 37 |
<FIELD indexable="true" name="rights" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'bestLicence']"/> |
|
| 38 |
<FIELD indexable="true" name="projectacronym" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'project']/*[local-name() = 'acronym']"/> |
|
| 39 |
<FIELD indexable="true" name="projectname" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'project']/concat(./*[local-name() = 'acronym'], ' - ',./*[local-name() = 'name'])"/> |
|
| 40 |
<FIELD indexable="true" name="projectopenaireid" result="false" stat="false" stored="true" tokenizable="false" xpath="//*[local-name() = 'project']/concat(./*[local-name() = 'acronym'], '||', ./*[local-name() = 'openaireId'])"/> |
|
| 41 |
<FIELD indexable="true" name="source" result="false" stored="true" stat="false" tokenizable="true" xpath="//*[local-name() = 'source']"/> |
|
| 42 |
<FIELD indexable="true" name="labgroupacronym" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'affiliation' and (@type = 'Laboratorio' or @type = 'Servizio' or @type = 'Centro')]/@groupAcronym"/> |
|
| 43 |
<FIELD indexable="true" name="publisher" result="false" stored="true" stat="false" tokenizable="false" xpath="//*[local-name() = 'publisher']"/> |
|
| 44 |
</FIELDS> |
|
| 45 |
</LAYOUT> |
|
| 46 |
</LAYOUTS> |
|
| 47 |
</STATUS> |
|
| 48 |
</BODY> |
|
| 49 |
</RESOURCE_PROFILE> |
|
| modules/dnet-isti/trunk/src/test/java/eu/dnetlib/isti/transform/TransformTest.java | ||
|---|---|---|
| 18 | 18 |
import org.junit.Ignore; |
| 19 | 19 |
import org.junit.Test; |
| 20 | 20 |
|
| 21 |
import eu.dnetlib.xml.ISTIUtilityFunction;
|
|
| 21 |
import eu.dnetlib.data.utils.XsltFunctions;
|
|
| 22 | 22 |
|
| 23 | 23 |
/** |
| 24 | 24 |
* Created by sandro on 9/30/16. |
| ... | ... | |
| 45 | 45 |
|
| 46 | 46 |
// System.out.println("record = " + record);
|
| 47 | 47 |
|
| 48 |
System.out.println("name :" + ISTIUtilityFunction.cleanNames("(eds.)"));
|
|
| 48 |
System.out.println("name :" + XsltFunctions.cleanNames("(eds.)"));
|
|
| 49 | 49 |
|
| 50 | 50 |
} |
| 51 | 51 |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/msro/workflows/nodes/PimpaUpdateJobNode.java | ||
|---|---|---|
| 11 | 11 |
import org.springframework.beans.factory.annotation.Autowired; |
| 12 | 12 |
|
| 13 | 13 |
import eu.dnetlib.clients.pimpa.PimpaStubFactory; |
| 14 |
import eu.dnetlib.data.db.Affiliation; |
|
| 15 | 14 |
import eu.dnetlib.data.db.AffiliationsDao; |
| 15 |
import eu.dnetlib.data.utils.IstiConstants; |
|
| 16 |
import eu.dnetlib.data.utils.XsltFunctions; |
|
| 16 | 17 |
import eu.dnetlib.msro.workflows.graph.Arc; |
| 17 | 18 |
import eu.dnetlib.msro.workflows.procs.Env; |
| 18 |
import eu.dnetlib.xml.ISTIUtilityFunction; |
|
| 19 | 19 |
|
| 20 | 20 |
public class PimpaUpdateJobNode extends SimpleJobNode {
|
| 21 | 21 |
|
| ... | ... | |
| 43 | 43 |
.collect(Collectors.toSet()); |
| 44 | 44 |
|
| 45 | 45 |
pimpa.getStub() |
| 46 |
.getDataRange(Affiliation.START_YEAR, Calendar.getInstance().get(Calendar.YEAR))
|
|
| 46 |
.getDataRange(IstiConstants.PIMPA_START_YEAR, Calendar.getInstance().get(Calendar.YEAR))
|
|
| 47 | 47 |
.getYears() |
| 48 | 48 |
.forEach(y -> {
|
| 49 | 49 |
y.getPersons().getList().forEach(p -> {
|
| ... | ... | |
| 81 | 81 |
if (capitalizeFullnames) {
|
| 82 | 82 |
log.info("Capitalizing person fullnames");
|
| 83 | 83 |
dao.listPersons().forEach(p -> {
|
| 84 |
dao.updatePerson(p.getId(), ISTIUtilityFunction.capitalize(p.getName()));
|
|
| 84 |
dao.updatePerson(p.getId(), XsltFunctions.capitalize(p.getName()));
|
|
| 85 | 85 |
});; |
| 86 | 86 |
} |
| 87 | 87 |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/utils/XsltFunctions.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.utils; |
|
| 2 |
|
|
| 3 |
import java.util.Arrays; |
|
| 4 |
import java.util.regex.Matcher; |
|
| 5 |
import java.util.regex.Pattern; |
|
| 6 |
import java.util.stream.Collectors; |
|
| 7 |
|
|
| 8 |
import org.apache.commons.lang3.StringUtils; |
|
| 9 |
import org.apache.commons.logging.Log; |
|
| 10 |
import org.apache.commons.logging.LogFactory; |
|
| 11 |
|
|
| 12 |
import eu.dnetlib.data.mdstore.plugins.objects.Project; |
|
| 13 |
|
|
| 14 |
/** |
|
| 15 |
* Created by sandro on 12/6/16. |
|
| 16 |
*/ |
|
| 17 |
public class XsltFunctions {
|
|
| 18 |
|
|
| 19 |
private static final Log log = LogFactory.getLog(XsltFunctions.class); |
|
| 20 |
|
|
| 21 |
public static String cleanNames(final String s) {
|
|
| 22 |
if (StringUtils.isBlank(s)) { return ""; }
|
|
| 23 |
return cleanNames(s, StringUtils.countMatches(s, ",") > 1 ? "," : ";"); |
|
| 24 |
} |
|
| 25 |
|
|
| 26 |
private static String cleanNames(final String s, final String sep) {
|
|
| 27 |
return Arrays.stream(s.split(sep)) |
|
| 28 |
.map(String::trim) |
|
| 29 |
.map(XsltFunctions::clean) |
|
| 30 |
.map(XsltFunctions::capitalize) |
|
| 31 |
.filter(StringUtils::isNotBlank) |
|
| 32 |
.collect(Collectors.joining("#"));
|
|
| 33 |
} |
|
| 34 |
|
|
| 35 |
private static String clean(final String s) {
|
|
| 36 |
return s.replaceAll("\\(.*\\)", "")
|
|
| 37 |
.replaceAll("\\[.*\\]", "")
|
|
| 38 |
.replaceAll("(?i)^et\\.? al(\\.|\\s)*$", "")
|
|
| 39 |
.replaceAll("(?i)\\s*et\\.? al(\\.|\\s)*$", "")
|
|
| 40 |
.replaceAll("\\d|\\*", "")
|
|
| 41 |
.replaceAll("^(\\s|\\-|\\.)+", "")
|
|
| 42 |
.replaceAll("\\,", "")
|
|
| 43 |
.replaceAll("\\.", ". ")
|
|
| 44 |
.trim(); |
|
| 45 |
} |
|
| 46 |
|
|
| 47 |
public static String cleanDoi(final String doi) {
|
|
| 48 |
final String x = doi.replaceAll("\\?", "")
|
|
| 49 |
.replaceAll("\\s*\\/\\s*", "/")
|
|
| 50 |
.replaceAll("\\s*\\-\\s*", "-")
|
|
| 51 |
.trim() |
|
| 52 |
.replaceAll("\\s", "_");
|
|
| 53 |
|
|
| 54 |
if (!x.equals(doi)) {
|
|
| 55 |
log.info("Cleaning doi: " + doi + " -> " + x);
|
|
| 56 |
} |
|
| 57 |
|
|
| 58 |
return x; |
|
| 59 |
} |
|
| 60 |
|
|
| 61 |
public static String capitalize(final String s) {
|
|
| 62 |
return Arrays.stream(s.split(" "))
|
|
| 63 |
.map(String::toLowerCase) |
|
| 64 |
.map(StringUtils::capitalize) |
|
| 65 |
.map(XsltFunctions::fixApostrophes) |
|
| 66 |
.collect(Collectors.joining(" "));
|
|
| 67 |
} |
|
| 68 |
|
|
| 69 |
public static String fixApostrophes(final String s) {
|
|
| 70 |
return s.replaceAll("'a", "'A").replaceAll("'e", "'E").replaceAll("'i", "'I").replaceAll("'o", "'O").replaceAll("'u", "'U")
|
|
| 71 |
.replaceAll("a'", "à").replaceAll("e'", "è").replaceAll("i'", "ì").replaceAll("o'", "ò").replaceAll("u'", "ù");
|
|
| 72 |
} |
|
| 73 |
|
|
| 74 |
public static boolean isValidProject(final String id) {
|
|
| 75 |
return Project.isValid(id); |
|
| 76 |
} |
|
| 77 |
|
|
| 78 |
public static String calculatePersonName(final String s) {
|
|
| 79 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+):(.+)\\/(.+)\\/(.+)");
|
|
| 80 |
final Matcher matcher = pattern.matcher(s); |
|
| 81 |
return matcher.find() ? capitalize(String.format("%s %s", matcher.group(4), matcher.group(3))) : "";
|
|
| 82 |
} |
|
| 83 |
|
|
| 84 |
// <xsl:for-each select="tokenize(istiFunction:cleanName(.), ';')"> |
|
| 85 |
// <xsl:choose> |
|
| 86 |
// <xsl:when test="matches(normalize-space(.), ',(\s*[a-zA-Z]\.)+$')"> |
|
| 87 |
// <creator> |
|
| 88 |
// <creatorName> |
|
| 89 |
// <xsl:value-of select="normalize-space(translate(.,',',' '))"/> |
|
| 90 |
// </creatorName> |
|
| 91 |
// </creator> |
|
| 92 |
// </xsl:when> |
|
| 93 |
// |
|
| 94 |
// <xsl:when test="matches(normalize-space(.), '(^[a-zA-Z\.]+,\s?[a-zA-Z\.\s]+$)|(^[a-zA-Z\.\s]+,\s?[a-zA-Z\.]+$)')"> |
|
| 95 |
// <creator> |
|
| 96 |
// <creatorName> |
|
| 97 |
// <xsl:value-of select="normalize-space(translate(.,',',' '))"/> |
|
| 98 |
// </creatorName> |
|
| 99 |
// </creator> |
|
| 100 |
// </xsl:when> |
|
| 101 |
// |
|
| 102 |
// <xsl:otherwise> |
|
| 103 |
// <xsl:for-each select="tokenize(., ',')"> |
|
| 104 |
// <creator> |
|
| 105 |
// <creatorName> |
|
| 106 |
// <xsl:value-of select="normalize-space(.)"/> |
|
| 107 |
// </creatorName> |
|
| 108 |
// </creator> |
|
| 109 |
// </xsl:for-each> |
|
| 110 |
// </xsl:otherwise> |
|
| 111 |
// </xsl:choose> |
|
| 112 |
// </xsl:for-each> |
|
| 113 |
// |
|
| 114 |
|
|
| 115 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/utils/CsvConverter.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.utils; |
|
| 2 |
|
|
| 3 |
public class CsvConverter {
|
|
| 4 |
|
|
| 5 |
public static String asCsv(final String string) {
|
|
| 6 |
// TODO Auto-generated method stub |
|
| 7 |
return null; |
|
| 8 |
} |
|
| 9 |
|
|
| 10 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/utils/BibTexConverter.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.utils; |
|
| 2 |
|
|
| 3 |
import java.io.IOException; |
|
| 4 |
import java.io.StringReader; |
|
| 5 |
import java.io.StringWriter; |
|
| 6 |
import java.io.Writer; |
|
| 7 |
import java.util.List; |
|
| 8 |
import java.util.stream.Collectors; |
|
| 9 |
|
|
| 10 |
import org.apache.commons.lang3.StringUtils; |
|
| 11 |
import org.apache.commons.logging.Log; |
|
| 12 |
import org.apache.commons.logging.LogFactory; |
|
| 13 |
import org.dom4j.Document; |
|
| 14 |
import org.dom4j.DocumentException; |
|
| 15 |
import org.dom4j.Node; |
|
| 16 |
import org.dom4j.io.SAXReader; |
|
| 17 |
import org.jbibtex.BibTeXDatabase; |
|
| 18 |
import org.jbibtex.BibTeXEntry; |
|
| 19 |
import org.jbibtex.BibTeXFormatter; |
|
| 20 |
import org.jbibtex.Key; |
|
| 21 |
import org.jbibtex.StringValue; |
|
| 22 |
import org.jbibtex.StringValue.Style; |
|
| 23 |
|
|
| 24 |
public class BibTexConverter {
|
|
| 25 |
|
|
| 26 |
private static final Log log = LogFactory.getLog(BibTexConverter.class); |
|
| 27 |
|
|
| 28 |
public static String asBibTex(final String xml) {
|
|
| 29 |
try {
|
|
| 30 |
final Document doc = new SAXReader().read(new StringReader(xml)); |
|
| 31 |
|
|
| 32 |
final BibTeXEntry dbEntry = new BibTeXEntry( |
|
| 33 |
findBibTexType(doc.valueOf("//*[local-name() = 'resourceType']")),
|
|
| 34 |
new Key(doc.valueOf("//*[local-name() = 'identifier']")));
|
|
| 35 |
|
|
| 36 |
addSimpleStringValue(dbEntry, BibTeXEntry.KEY_TITLE, doc, "//*[local-name() = 'title']"); |
|
| 37 |
addMultipleStringValue(dbEntry, BibTeXEntry.KEY_AUTHOR, doc, "//*[local-name() = 'creator']"); |
|
| 38 |
addSimpleStringValue(dbEntry, BibTeXEntry.KEY_YEAR, doc, "//*[local-name() = 'date']"); |
|
| 39 |
|
|
| 40 |
final BibTeXFormatter formatter = new BibTeXFormatter(); |
|
| 41 |
final Writer res = new StringWriter(); |
|
| 42 |
final BibTeXDatabase db = new BibTeXDatabase(); |
|
| 43 |
db.addObject(dbEntry); |
|
| 44 |
formatter.format(db, res); |
|
| 45 |
return res.toString(); |
|
| 46 |
} catch (final DocumentException | IOException e) {
|
|
| 47 |
log.warn("Error generating bibtex", e);
|
|
| 48 |
return ""; |
|
| 49 |
} |
|
| 50 |
} |
|
| 51 |
|
|
| 52 |
private static void addSimpleStringValue(final BibTeXEntry dbEntry, final Key key, final Document doc, final String xpath) {
|
|
| 53 |
final String val = doc.valueOf(xpath).trim(); |
|
| 54 |
if (StringUtils.isNotBlank(val)) {
|
|
| 55 |
dbEntry.addField(key, new StringValue(val, Style.BRACED)); |
|
| 56 |
} |
|
| 57 |
} |
|
| 58 |
|
|
| 59 |
@SuppressWarnings("unchecked")
|
|
| 60 |
private static void addMultipleStringValue(final BibTeXEntry dbEntry, final Key key, final Document doc, final String xpath) {
|
|
| 61 |
final String val = ((List<Node>) doc.selectNodes(xpath)) |
|
| 62 |
.stream() |
|
| 63 |
.map(Node::getText) |
|
| 64 |
.map(String::trim) |
|
| 65 |
.filter(StringUtils::isNotEmpty) |
|
| 66 |
.collect(Collectors.joining(" and "));
|
|
| 67 |
|
|
| 68 |
if (StringUtils.isNotBlank(val)) {
|
|
| 69 |
dbEntry.addField(key, new StringValue(val, Style.BRACED)); |
|
| 70 |
} |
|
| 71 |
} |
|
| 72 |
|
|
| 73 |
private static Key findBibTexType(final String resourceType) {
|
|
| 74 |
switch (resourceType) {
|
|
| 75 |
case "Conference object": |
|
| 76 |
return BibTeXEntry.TYPE_INPROCEEDINGS; |
|
| 77 |
case "Report": |
|
| 78 |
return BibTeXEntry.TYPE_TECHREPORT; |
|
| 79 |
case "Article": |
|
| 80 |
return BibTeXEntry.TYPE_ARTICLE; |
|
| 81 |
case "Other": |
|
| 82 |
return BibTeXEntry.TYPE_MISC; |
|
| 83 |
case "Part of book or chapter of book": |
|
| 84 |
return BibTeXEntry.TYPE_INBOOK; |
|
| 85 |
case "Book": |
|
| 86 |
return BibTeXEntry.TYPE_BOOK; |
|
| 87 |
case "Bachelor thesis": |
|
| 88 |
return BibTeXEntry.TYPE_MASTERSTHESIS; |
|
| 89 |
case "Doctoral thesis": |
|
| 90 |
return BibTeXEntry.TYPE_PHDTHESIS; |
|
| 91 |
case "Patent": |
|
| 92 |
return BibTeXEntry.TYPE_MISC; |
|
| 93 |
case "Master thesis": |
|
| 94 |
return BibTeXEntry.TYPE_MASTERSTHESIS; |
|
| 95 |
default: |
|
| 96 |
return BibTeXEntry.TYPE_MISC; |
|
| 97 |
} |
|
| 98 |
} |
|
| 99 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/utils/IstiConstants.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.utils; |
|
| 2 |
|
|
| 3 |
public class IstiConstants {
|
|
| 4 |
|
|
| 5 |
public static final int PIMPA_START_YEAR = 1985; |
|
| 6 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/db/Affiliation.java | ||
|---|---|---|
| 2 | 2 |
|
| 3 | 3 |
public class Affiliation {
|
| 4 | 4 |
|
| 5 |
public static final int START_YEAR = 1985; |
|
| 6 |
|
|
| 7 | 5 |
private String pid; |
| 8 | 6 |
private String gid; |
| 9 | 7 |
private int year; |
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/EnrichDatasetsPlugin.java | ||
|---|---|---|
| 2 | 2 |
|
| 3 | 3 |
import java.net.URI; |
| 4 | 4 |
import java.net.URISyntaxException; |
| 5 |
import java.util.HashMap;
|
|
| 5 |
import java.util.ArrayList;
|
|
| 6 | 6 |
import java.util.List; |
| 7 |
import java.util.Map; |
|
| 8 | 7 |
|
| 9 | 8 |
import org.apache.commons.lang3.StringUtils; |
| 10 |
import org.apache.commons.logging.Log; |
|
| 11 |
import org.apache.commons.logging.LogFactory; |
|
| 12 |
import org.dom4j.Document; |
|
| 13 |
import org.dom4j.Element; |
|
| 14 |
import org.dom4j.Namespace; |
|
| 15 |
import org.dom4j.QName; |
|
| 16 | 9 |
import org.springframework.beans.factory.annotation.Value; |
| 17 | 10 |
|
| 18 | 11 |
import com.google.gson.Gson; |
| 19 | 12 |
import com.google.gson.reflect.TypeToken; |
| 20 | 13 |
|
| 14 |
import eu.dnetlib.data.mdstore.plugins.objects.Dataset; |
|
| 15 |
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord; |
|
| 21 | 16 |
import eu.dnetlib.data.mdstore.plugins.objects.dli.DliIdentifier; |
| 22 | 17 |
import eu.dnetlib.data.mdstore.plugins.objects.dli.DliRelation; |
| 23 | 18 |
|
| 24 | 19 |
public class EnrichDatasetsPlugin extends GenericDoiMdstorePlugin {
|
| 25 | 20 |
|
| 26 |
private static final Log log = LogFactory.getLog(EnrichDatasetsPlugin.class);
|
|
| 21 |
private static final Gson gson = new Gson();
|
|
| 27 | 22 |
|
| 28 | 23 |
@Value("${plugin.enrich.dataset.dli.url}")
|
| 29 | 24 |
private String baseUrl; |
| ... | ... | |
| 34 | 29 |
} |
| 35 | 30 |
|
| 36 | 31 |
@Override |
| 37 |
protected boolean updateDocument(final Document doc, final String response) {
|
|
| 32 |
protected void reconfigure() {}
|
|
| 38 | 33 |
|
| 39 |
final Gson gson = new Gson(); |
|
| 34 |
@Override |
|
| 35 |
protected boolean updateDocument(final MdRecord doc, final String response) {
|
|
| 36 |
|
|
| 37 |
doc.getDatasets().clear(); |
|
| 38 |
|
|
| 39 |
final List<Dataset> datasets = new ArrayList<>(); |
|
| 40 |
|
|
| 40 | 41 |
final List<DliRelation> rels = gson.fromJson(response, new TypeToken<List<DliRelation>>() {}.getType());
|
| 41 | 42 |
|
| 42 |
final Map<String, String> datasets = new HashMap<>(); |
|
| 43 |
|
|
| 44 | 43 |
for (final DliRelation rel : rels) {
|
| 45 | 44 |
final String title = rel.getTarget().getTitle(); |
| 46 | 45 |
for (final DliIdentifier id : rel.getTarget().getIdentifiers()) {
|
| 47 | 46 |
if (id.getSchema().equalsIgnoreCase("doi") && StringUtils.isNoneBlank(id.getIdentifier()) && StringUtils.isNotBlank(title)) {
|
| 48 |
datasets.put(id.getIdentifier(), title);
|
|
| 47 |
datasets.add(new Dataset(id.getIdentifier(), title));
|
|
| 49 | 48 |
} |
| 50 | 49 |
} |
| 51 | 50 |
} |
| 52 | 51 |
|
| 53 |
if (datasets.isEmpty()) { return false; }
|
|
| 52 |
if (datasets.isEmpty()) {
|
|
| 53 |
return false; |
|
| 54 |
} else {
|
|
| 55 |
doc.getDatasets().addAll(datasets); |
|
| 56 |
return true; |
|
| 57 |
} |
|
| 58 |
} |
|
| 54 | 59 |
|
| 55 |
final Element node = (Element) doc.selectSingleNode("//*[local-name() = 'datasets']");
|
|
| 56 |
|
|
| 57 |
datasets.entrySet().forEach(e -> {
|
|
| 58 |
final Element ds = node.addElement(new QName("dataset", new Namespace("isti", "http://www.isti.cnr.it/")));
|
|
| 59 |
ds.addAttribute("doi", e.getKey());
|
|
| 60 |
ds.addAttribute("url", "https://dx.doi.org/" + e.getKey());
|
|
| 61 |
ds.setText(e.getValue()); |
|
| 62 |
log.info("Found dataset: " + e.getKey() + " -> " + e.getValue());
|
|
| 63 |
}); |
|
| 64 |
|
|
| 65 |
return true; |
|
| 66 |
|
|
| 67 |
} |
|
| 68 | 60 |
} |
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/EnrichProjectsPlugin.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.mdstore.plugins; |
| 2 | 2 |
|
| 3 |
import java.io.StringReader; |
|
| 4 | 3 |
import java.net.URI; |
| 5 | 4 |
import java.net.URISyntaxException; |
| 6 |
import java.util.ArrayList; |
|
| 7 | 5 |
import java.util.HashMap; |
| 8 | 6 |
import java.util.List; |
| 9 | 7 |
import java.util.Map; |
| 8 |
import java.util.stream.Collectors; |
|
| 10 | 9 |
|
| 11 | 10 |
import org.apache.commons.lang3.StringUtils; |
| 12 | 11 |
import org.apache.commons.logging.Log; |
| 13 | 12 |
import org.apache.commons.logging.LogFactory; |
| 14 |
import org.dom4j.Document; |
|
| 15 |
import org.dom4j.DocumentException; |
|
| 16 |
import org.dom4j.Element; |
|
| 17 |
import org.dom4j.io.SAXReader; |
|
| 18 | 13 |
import org.springframework.beans.factory.annotation.Value; |
| 19 | 14 |
|
| 20 |
import com.mongodb.BasicDBObject; |
|
| 21 |
import com.mongodb.DBObject; |
|
| 22 |
import com.mongodb.client.MongoCollection; |
|
| 23 |
|
|
| 24 |
import eu.dnetlib.data.mdstore.modular.mongodb.MongoMDStore; |
|
| 15 |
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord; |
|
| 25 | 16 |
import eu.dnetlib.data.mdstore.plugins.objects.Project; |
| 26 |
import eu.dnetlib.rmi.data.MDStoreServiceException; |
|
| 27 | 17 |
|
| 28 |
public class EnrichProjectsPlugin extends AbstractIstiMDStorePlugin {
|
|
| 18 |
public class EnrichProjectsPlugin extends MdRecordPlugin {
|
|
| 29 | 19 |
|
| 30 | 20 |
private static final Log log = LogFactory.getLog(EnrichProjectsPlugin.class); |
| 31 | 21 |
|
| 32 | 22 |
@Value("${plugin.enrich.projects.openaire.url}")
|
| 33 | 23 |
private String baseUrl; |
| 34 | 24 |
|
| 25 |
private Map<String, Project> mapProjects = new HashMap<>(); |
|
| 26 |
|
|
| 35 | 27 |
@Override |
| 36 |
public void process(final MongoMDStore store, final Map<String, String> params) throws MDStoreServiceException {
|
|
| 37 |
log.info("*****************************************");
|
|
| 38 |
log.info("* ENRICH PROJECTS ...");
|
|
| 39 |
log.info("*****************************************");
|
|
| 28 |
protected void reconfigure() {
|
|
| 29 |
log.info("Cleaning projects cache");
|
|
| 30 |
mapProjects.clear();
|
|
| 31 |
}
|
|
| 40 | 32 |
|
| 41 |
final MongoCollection<DBObject> collPubs = store.getCollection(); |
|
| 33 |
@Override |
|
| 34 |
protected boolean updateRecord(final MdRecord record) {
|
|
| 35 |
log.debug("Found " + record.getProjects().size() + " projects");
|
|
| 42 | 36 |
|
| 43 |
final Map<String, List<Project>> pubs = resolveProjects(collPubs);
|
|
| 37 |
if (record.getProjects().isEmpty()) { return false; }
|
|
| 44 | 38 |
|
| 45 |
// update the publications |
|
| 46 |
pubs.entrySet().forEach(e -> patchRecord(collPubs, e.getKey(), e.getValue())); |
|
| 39 |
final List<Project> list = record.getProjects() |
|
| 40 |
.stream() |
|
| 41 |
.map(Project::getInfoId) |
|
| 42 |
.filter(StringUtils::isNotBlank) |
|
| 43 |
.map(Project::newInstance) |
|
| 44 |
.map(this::findEnrichedProject) |
|
| 45 |
.collect(Collectors.toList()); |
|
| 47 | 46 |
|
| 48 |
touch(store); |
|
| 49 |
} |
|
| 47 |
log.debug("Processed " + list.size() + " projects");
|
|
| 50 | 48 |
|
| 51 |
private Map<String, List<Project>> resolveProjects(final MongoCollection<DBObject> collPubs) {
|
|
| 52 |
final Map<String, List<Project>> pubs = new HashMap<>(); |
|
| 53 |
final Map<String, Project> mapProjects = new HashMap<>(); |
|
| 54 |
for (final DBObject obj : collPubs.find()) {
|
|
| 55 |
final List<Project> projects = projectsFromPublication(obj, mapProjects); |
|
| 56 |
pubs.put(obj.get("id").toString(), projects);
|
|
| 57 |
} |
|
| 58 |
return pubs; |
|
| 59 |
} |
|
| 49 |
if (list.isEmpty()) { return false; }
|
|
| 60 | 50 |
|
| 61 |
private List<Project> projectsFromPublication(final DBObject obj, final Map<String, Project> mapProjects) {
|
|
| 62 |
final List<Project> projects = new ArrayList<>(); |
|
| 63 |
try {
|
|
| 64 |
final Document doc = (new SAXReader()).read(new StringReader(obj.get("body").toString()));
|
|
| 65 |
for (final Object o : doc.selectNodes("//*[local-name()='project']")) {
|
|
| 66 |
final Element p = (Element) o; |
|
| 67 |
final String code = p.valueOf("./*[local-name() = 'code']").trim();
|
|
| 68 |
final String infoId = p.valueOf("./*[local-name() = 'infoId']").trim();
|
|
| 51 |
record.getProjects().clear(); |
|
| 52 |
record.getProjects().addAll(list); |
|
| 69 | 53 |
|
| 70 |
if (StringUtils.isEmpty(code) && StringUtils.isNotEmpty(infoId)) {
|
|
| 71 |
final Project simpleProject = Project.newInstance(infoId); |
|
| 72 |
if (simpleProject != null) {
|
|
| 73 |
final String s = simpleProject.getFunder() + "/" + simpleProject.getProgram() + "/" + simpleProject.getCode(); |
|
| 74 |
if (!mapProjects.containsKey(s)) {
|
|
| 75 |
final Project project = resolveProject(s); |
|
| 76 |
mapProjects.put(s, project != null ? project : simpleProject); |
|
| 77 |
} |
|
| 78 |
projects.add(mapProjects.get(s)); |
|
| 79 |
} |
|
| 80 |
} |
|
| 81 |
|
|
| 82 |
} |
|
| 83 |
} catch (final DocumentException e) {
|
|
| 84 |
log.warn("Problem parsing a mdstore record");
|
|
| 85 |
} |
|
| 86 |
|
|
| 87 |
return projects; |
|
| 54 |
return true; |
|
| 88 | 55 |
} |
| 89 | 56 |
|
| 90 |
private void patchRecord(final MongoCollection<DBObject> collPubs, final String id, final List<Project> projects) {
|
|
| 91 |
log.debug("Adding to " + id + " " + projects.size() + " project(s)");
|
|
| 92 |
|
|
| 93 |
try {
|
|
| 94 |
final SAXReader reader = new SAXReader(); |
|
| 95 |
final DBObject obj = collPubs.find(new BasicDBObject("id", id)).first();
|
|
| 96 |
|
|
| 97 |
final Document doc = reader.read(new StringReader(obj.get("body").toString()));
|
|
| 98 |
|
|
| 99 |
for (final Object o : doc.selectNodes("//*[local-name()='project']")) {
|
|
| 100 |
final Element n = (Element) o; |
|
| 101 |
final String infoId = n.valueOf("./*[local-name()='infoId']");
|
|
| 102 |
|
|
| 103 |
projects.stream() |
|
| 104 |
.filter(p -> (p != null) && p.match(infoId)) |
|
| 105 |
.forEach(p -> {
|
|
| 106 |
n.selectSingleNode("./*[local-name()='code']").setText(p.getCode());
|
|
| 107 |
n.selectSingleNode("./*[local-name()='name']").setText(p.getName());
|
|
| 108 |
n.selectSingleNode("./*[local-name()='acronym']").setText(p.getAcronym());
|
|
| 109 |
n.selectSingleNode("./*[local-name()='funder']").setText(p.getFunder());
|
|
| 110 |
n.selectSingleNode("./*[local-name()='program']").setText(p.getProgram());
|
|
| 111 |
n.selectSingleNode("./*[local-name()='jurisdiction']").setText(p.getJurisdiction());
|
|
| 112 |
n.selectSingleNode("./*[local-name()='openaireId']").setText(p.getOpenaireId());
|
|
| 113 |
}); |
|
| 114 |
} |
|
| 115 |
|
|
| 116 |
collPubs.updateOne(new BasicDBObject("id", id), new BasicDBObject("$set", new BasicDBObject("body", doc.asXML())));
|
|
| 117 |
} catch (final DocumentException e) {
|
|
| 118 |
log.warn("Error patching record: " + id);
|
|
| 57 |
private Project findEnrichedProject(final Project simpleProject) {
|
|
| 58 |
final String s = simpleProject.getFunder() + "/" + simpleProject.getProgram() + "/" + simpleProject.getCode(); |
|
| 59 |
if (!mapProjects.containsKey(s)) {
|
|
| 60 |
final Project project = resolveProject(s); |
|
| 61 |
mapProjects.put(s, project != null ? project : simpleProject); |
|
| 119 | 62 |
} |
| 63 |
return mapProjects.get(s); |
|
| 120 | 64 |
} |
| 121 | 65 |
|
| 122 | 66 |
private Project resolveProject(final String s) {
|
| ... | ... | |
| 129 | 73 |
throw new RuntimeException("Error resolving project: " + s, e);
|
| 130 | 74 |
} |
| 131 | 75 |
} |
| 76 |
|
|
| 132 | 77 |
} |
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/GenericDoiMdstorePlugin.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.mdstore.plugins; |
| 2 | 2 |
|
| 3 |
import java.io.StringReader; |
|
| 4 | 3 |
import java.net.URI; |
| 5 | 4 |
import java.net.URISyntaxException; |
| 6 | 5 |
import java.security.KeyManagementException; |
| 7 | 6 |
import java.security.KeyStoreException; |
| 8 | 7 |
import java.security.NoSuchAlgorithmException; |
| 9 |
import java.util.Map; |
|
| 10 | 8 |
import java.util.concurrent.TimeUnit; |
| 11 | 9 |
|
| 12 | 10 |
import org.apache.commons.logging.Log; |
| ... | ... | |
| 14 | 12 |
import org.apache.http.conn.ssl.SSLConnectionSocketFactory; |
| 15 | 13 |
import org.apache.http.impl.client.HttpClientBuilder; |
| 16 | 14 |
import org.apache.http.ssl.SSLContextBuilder; |
| 17 |
import org.dom4j.Document; |
|
| 18 |
import org.dom4j.DocumentException; |
|
| 19 |
import org.dom4j.Node; |
|
| 20 |
import org.dom4j.io.SAXReader; |
|
| 21 | 15 |
import org.springframework.http.client.HttpComponentsClientHttpRequestFactory; |
| 22 | 16 |
import org.springframework.web.client.RestTemplate; |
| 23 | 17 |
|
| 24 |
import com.mongodb.BasicDBObject; |
|
| 25 |
import com.mongodb.DBObject; |
|
| 26 |
import com.mongodb.client.MongoCollection; |
|
| 18 |
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord; |
|
| 27 | 19 |
|
| 28 |
import eu.dnetlib.data.mdstore.modular.mongodb.MongoMDStore; |
|
| 29 |
import eu.dnetlib.rmi.data.MDStoreServiceException; |
|
| 20 |
public abstract class GenericDoiMdstorePlugin extends MdRecordPlugin {
|
|
| 30 | 21 |
|
| 31 |
public abstract class GenericDoiMdstorePlugin extends AbstractIstiMDStorePlugin {
|
|
| 32 |
|
|
| 33 | 22 |
private static final Log log = LogFactory.getLog(GenericDoiMdstorePlugin.class); |
| 34 | 23 |
|
| 35 | 24 |
private static final int MAX_NUMBER_OF_ATTEMPTS = 10; |
| ... | ... | |
| 57 | 46 |
.build()); |
| 58 | 47 |
|
| 59 | 48 |
@Override |
| 60 |
public void process(final MongoMDStore store, final Map<String, String> params) throws MDStoreServiceException {
|
|
| 61 |
|
|
| 62 |
final MongoCollection<DBObject> collPubs = store.getCollection(); |
|
| 63 |
|
|
| 64 |
int count = 0; |
|
| 65 |
|
|
| 66 |
for (final DBObject obj : collPubs.find()) {
|
|
| 67 |
|
|
| 68 |
try {
|
|
| 69 |
final String recordId = obj.get("id").toString();
|
|
| 70 |
final Document doc = (new SAXReader()).read(new StringReader(obj.get("body").toString()));
|
|
| 71 |
for (final Object o : doc.selectNodes("//*[local-name()='alternateIdentifier' and @alternateIdentifierType='doi']")) {
|
|
| 72 |
final String doi = ((Node) o).getText().trim(); |
|
| 73 |
|
|
| 74 |
log.debug(" Record " + recordId + " has doi " + doi);
|
|
| 75 |
final String response = download(doi); |
|
| 76 |
if ((response != null) && updateDocument(doc, response)) {
|
|
| 77 |
collPubs.updateOne(new BasicDBObject("id", recordId), new BasicDBObject("$set", new BasicDBObject("body", doc.asXML())));
|
|
| 78 |
count++; |
|
| 79 |
} |
|
| 80 |
} |
|
| 81 |
} catch (final DocumentException e) {
|
|
| 82 |
log.warn("Problem parsing a mdstore record");
|
|
| 83 |
} |
|
| 49 |
protected final boolean updateRecord(final MdRecord record) {
|
|
| 50 |
for (final String doi : record.getDois()) {
|
|
| 51 |
log.debug(" Record " + record.getId() + " has doi " + doi);
|
|
| 52 |
final String response = download(doi); |
|
| 53 |
if ((response != null) && updateDocument(record, response)) { return true; }
|
|
| 84 | 54 |
} |
| 85 |
|
|
| 86 |
log.info("Number of patched records: " + count);
|
|
| 87 |
|
|
| 88 |
touch(store); |
|
| 55 |
return false; |
|
| 89 | 56 |
} |
| 90 | 57 |
|
| 91 |
abstract protected boolean updateDocument(Document doc, String response);
|
|
| 58 |
abstract protected boolean updateDocument(MdRecord doc, String response);
|
|
| 92 | 59 |
|
| 93 | 60 |
abstract protected URI prepareURI(String doi) throws URISyntaxException; |
| 94 | 61 |
|
| ... | ... | |
| 116 | 83 |
} |
| 117 | 84 |
} |
| 118 | 85 |
} |
| 86 |
|
|
| 119 | 87 |
} |
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/EnrichLabsPlugin.java | ||
|---|---|---|
| 1 | 1 |
package eu.dnetlib.data.mdstore.plugins; |
| 2 | 2 |
|
| 3 |
import java.io.StringReader;
|
|
| 3 |
import java.util.HashMap;
|
|
| 4 | 4 |
import java.util.Map; |
| 5 | 5 |
import java.util.Set; |
| 6 | 6 |
import java.util.regex.Matcher; |
| 7 | 7 |
import java.util.regex.Pattern; |
| 8 | 8 |
import java.util.stream.Collectors; |
| 9 | 9 |
|
| 10 |
import org.apache.commons.lang3.math.NumberUtils; |
|
| 11 | 10 |
import org.apache.commons.logging.Log; |
| 12 | 11 |
import org.apache.commons.logging.LogFactory; |
| 13 |
import org.dom4j.Document; |
|
| 14 |
import org.dom4j.DocumentException; |
|
| 15 |
import org.dom4j.Element; |
|
| 16 |
import org.dom4j.Node; |
|
| 17 |
import org.dom4j.QName; |
|
| 18 |
import org.dom4j.io.SAXReader; |
|
| 19 | 12 |
import org.springframework.beans.factory.annotation.Autowired; |
| 20 | 13 |
|
| 21 | 14 |
import com.google.common.collect.Sets; |
| 22 |
import com.mongodb.BasicDBObject; |
|
| 23 |
import com.mongodb.DBObject; |
|
| 24 |
import com.mongodb.client.MongoCollection; |
|
| 25 | 15 |
|
| 26 |
import eu.dnetlib.data.db.Affiliation; |
|
| 27 | 16 |
import eu.dnetlib.data.db.AffiliationsDao; |
| 28 | 17 |
import eu.dnetlib.data.db.Group; |
| 29 | 18 |
import eu.dnetlib.data.db.Person; |
| 30 |
import eu.dnetlib.data.mdstore.modular.mongodb.MongoMDStore; |
|
| 31 |
import eu.dnetlib.rmi.data.MDStoreServiceException; |
|
| 32 |
import eu.dnetlib.xml.ISTIUtilityFunction; |
|
| 19 |
import eu.dnetlib.data.mdstore.plugins.objects.Affiliation; |
|
| 20 |
import eu.dnetlib.data.mdstore.plugins.objects.CnrPerson; |
|
| 21 |
import eu.dnetlib.data.mdstore.plugins.objects.MdRecord; |
|
| 22 |
import eu.dnetlib.data.utils.IstiConstants; |
|
| 23 |
import eu.dnetlib.data.utils.XsltFunctions; |
|
| 33 | 24 |
|
| 34 |
public class EnrichLabsPlugin extends AbstractIstiMDStorePlugin {
|
|
| 25 |
public class EnrichLabsPlugin extends MdRecordPlugin {
|
|
| 35 | 26 |
|
| 36 | 27 |
private static final Log log = LogFactory.getLog(EnrichLabsPlugin.class); |
| 37 | 28 |
|
| ... | ... | |
| 40 | 31 |
@Autowired |
| 41 | 32 |
private AffiliationsDao dao; |
| 42 | 33 |
|
| 34 |
private Map<String, Person> affiliations = new HashMap<>(); |
|
| 35 |
|
|
| 43 | 36 |
@Override |
| 44 |
public void process(final MongoMDStore store, final Map<String, String> params) throws MDStoreServiceException {
|
|
| 45 |
log.warn("********************************************");
|
|
| 46 |
log.warn("* ENRICH LABS");
|
|
| 47 |
log.warn("********************************************");
|
|
| 48 |
|
|
| 49 |
// CODE -> year -> Lab |
|
| 50 |
final Map<String, Person> affiliations = dao.listPersonsWithAffiliations() |
|
| 37 |
protected void reconfigure() {
|
|
| 38 |
affiliations = dao.listPersonsWithAffiliations() |
|
| 51 | 39 |
.stream() |
| 52 | 40 |
.collect(Collectors.toMap(Person::getId, a -> a)); |
| 41 |
} |
|
| 53 | 42 |
|
| 54 |
final MongoCollection<DBObject> pubsCollection = store.getCollection(); |
|
| 43 |
@Override |
|
| 44 |
protected boolean updateRecord(final MdRecord doc) {
|
|
| 45 |
final int year = doc.getDate(); |
|
| 55 | 46 |
|
| 56 |
for (final DBObject obj : pubsCollection.find()) {
|
|
| 57 |
final String id = obj.get("id").toString();
|
|
| 58 |
log.debug("**********************************************");
|
|
| 59 |
log.debug("Record " + id);
|
|
| 47 |
for (final CnrPerson cp : doc.getCnrPersons()) {
|
|
| 48 |
cp.getAffiliations().clear(); |
|
| 60 | 49 |
|
| 61 |
try {
|
|
| 62 |
final Document doc = (new SAXReader()).read(new StringReader(obj.get("body").toString()));
|
|
| 50 |
if (year >= IstiConstants.PIMPA_START_YEAR) {
|
|
| 51 |
final String infoId = cp.getId(); |
|
| 52 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+)\\/(.+)\\/(.+)");
|
|
| 53 |
final Matcher matcher = pattern.matcher(infoId); |
|
| 54 |
if (matcher.find()) {
|
|
| 55 |
final String code = matcher.group(1); |
|
| 56 |
final String fullname = XsltFunctions.capitalize(String.format("%s %s", matcher.group(3), matcher.group(2)));
|
|
| 63 | 57 |
|
| 64 |
final int year = NumberUtils.toInt(doc.valueOf("//*[local-name()='date' and @dateType='Accepted']"), 0);
|
|
| 58 |
final Person p = affiliations.get(code);
|
|
| 65 | 59 |
|
| 66 |
if (year >= Affiliation.START_YEAR) {
|
|
| 67 |
resetAffiliations(doc); |
|
| 68 |
|
|
| 69 |
for (final Object o : doc.selectNodes("//*[local-name()='person']")) {
|
|
| 70 |
final Element node = (Element) o; |
|
| 71 |
final String infoId = node.valueOf("./*[local-name()='infoId']");
|
|
| 72 |
final Pattern pattern = Pattern.compile("info:cnr-pdr\\/author\\/(.+)\\/(.+)\\/(.+)");
|
|
| 73 |
final Matcher matcher = pattern.matcher(infoId); |
|
| 74 |
if (matcher.find()) {
|
|
| 75 |
final String code = matcher.group(1); |
|
| 76 |
final String fullname = ISTIUtilityFunction.capitalize(String.format("%s %s", matcher.group(3), matcher.group(2)));
|
|
| 77 |
|
|
| 78 |
final Person p = affiliations.get(code); |
|
| 79 |
|
|
| 80 |
if (p != null) {
|
|
| 81 |
final Set<Group> groups = p.getAffiliations().get(year); |
|
| 82 |
if (groups != null) {
|
|
| 83 |
node.selectSingleNode("./*[local-name() = 'personName']").setText(p.getName());
|
|
| 84 |
for (final Group g : groups) {
|
|
| 85 |
if (!g.getId().equals(UNKNOWN_GROUP.getId())) {
|
|
| 86 |
final Element affNode = node.addElement(new QName("affiliation", node.getNamespace()));
|
|
| 87 |
affNode.addAttribute("code", code);
|
|
| 88 |
affNode.addAttribute("groupAcronym", g.getId());
|
|
| 89 |
affNode.addAttribute("group", g.getName());
|
|
| 90 |
affNode.addAttribute("type", g.getType());
|
|
| 91 |
} |
|
| 92 |
} |
|
| 93 |
} else {
|
|
| 94 |
log.info("Affiliation not found, infoId: " + infoId + ", year: " + year);
|
|
| 95 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 96 |
p.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 60 |
if (p != null) {
|
|
| 61 |
final Set<Group> groups = p.getAffiliations().get(year); |
|
| 62 |
if (groups != null) {
|
|
| 63 |
cp.setName(p.getName()); |
|
| 64 |
for (final Group g : groups) {
|
|
| 65 |
if (!g.getId().equals(UNKNOWN_GROUP.getId())) {
|
|
| 66 |
cp.getAffiliations().add(new Affiliation(code, g.getType(), g.getId(), g.getName())); |
|
| 97 | 67 |
} |
| 98 |
} else {
|
|
| 99 |
log.info("Person not found, infoId: " + infoId + ", year: " + year);
|
|
| 100 |
dao.registerPerson(code, fullname); |
|
| 101 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 102 |
|
|
| 103 |
final Person np = new Person(code, fullname); |
|
| 104 |
np.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 105 |
|
|
| 106 |
affiliations.put(code, np); |
|
| 107 | 68 |
} |
| 108 | 69 |
} else {
|
| 109 |
log.warn("Invalid infoId: " + infoId);
|
|
| 70 |
log.info("Affiliation not found, infoId: " + infoId + ", year: " + year);
|
|
| 71 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 72 |
p.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 110 | 73 |
} |
| 74 |
} else {
|
|
| 75 |
log.info("Person not found, infoId: " + infoId + ", year: " + year);
|
|
| 76 |
dao.registerPerson(code, fullname); |
|
| 77 |
dao.registerAffiliation(code, UNKNOWN_GROUP.getId(), year); |
|
| 78 |
|
|
| 79 |
final Person np = new Person(code, fullname); |
|
| 80 |
np.getAffiliations().put(year, Sets.newHashSet(UNKNOWN_GROUP)); |
|
| 81 |
|
|
| 82 |
affiliations.put(code, np); |
|
| 111 | 83 |
} |
| 112 |
|
|
| 113 |
pubsCollection.updateOne(new BasicDBObject("id", id), new BasicDBObject("$set", new BasicDBObject("body", doc.asXML())));
|
|
| 84 |
} else {
|
|
| 85 |
log.warn("Invalid infoId: " + infoId);
|
|
| 114 | 86 |
} |
| 115 |
} catch (final DocumentException e) {
|
|
| 116 |
log.warn("Problem parsing a mdstore record");
|
|
| 117 | 87 |
} |
| 118 |
} |
|
| 119 |
} |
|
| 120 | 88 |
|
| 121 |
private void resetAffiliations(final Document doc) {
|
|
| 122 |
for (final Object n : doc.selectNodes("//*[local-name() = 'person']/*[local-name() = 'affiliation']")) {
|
|
| 123 |
((Node) n).detach(); |
|
| 124 | 89 |
} |
| 90 |
return true; |
|
| 125 | 91 |
} |
| 126 | 92 |
|
| 127 | 93 |
} |
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/objects/MdRecord.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.mdstore.plugins.objects; |
|
| 2 |
|
|
| 3 |
import java.util.LinkedHashSet; |
|
| 4 |
import java.util.Set; |
|
| 5 |
import java.util.TreeSet; |
|
| 6 |
|
|
| 7 |
import javax.xml.bind.annotation.XmlAccessType; |
|
| 8 |
import javax.xml.bind.annotation.XmlAccessorType; |
|
| 9 |
import javax.xml.bind.annotation.XmlElement; |
|
| 10 |
import javax.xml.bind.annotation.XmlElementWrapper; |
|
| 11 |
import javax.xml.bind.annotation.XmlRootElement; |
|
| 12 |
|
|
| 13 |
@XmlRootElement(name = "record") |
|
| 14 |
@XmlAccessorType(XmlAccessType.FIELD) |
|
| 15 |
public class MdRecord {
|
|
| 16 |
|
|
| 17 |
@XmlElement(name = "identifier") |
|
| 18 |
private String id; |
|
| 19 |
|
|
| 20 |
@XmlElement(name = "title") |
|
| 21 |
private String title; |
|
| 22 |
|
|
| 23 |
@XmlElementWrapper(name = "creators") |
|
| 24 |
@XmlElement(name = "creator") |
|
| 25 |
private Set<String> creators = new LinkedHashSet<>(); |
|
| 26 |
|
|
| 27 |
@XmlElement(name = "publisher") |
|
| 28 |
private String publisher; |
|
| 29 |
|
|
| 30 |
@XmlElementWrapper(name = "subjects") |
|
| 31 |
@XmlElement(name = "subject") |
|
| 32 |
private Set<String> subjects = new LinkedHashSet<>(); |
|
| 33 |
|
|
| 34 |
@XmlElementWrapper(name = "contributors") |
|
| 35 |
@XmlElement(name = "contributor") |
|
| 36 |
private Set<String> contributors = new LinkedHashSet<>(); |
|
| 37 |
|
|
| 38 |
@XmlElement(name = "date") |
|
| 39 |
private int date; |
|
| 40 |
|
|
| 41 |
@XmlElement(name = "language") |
|
| 42 |
private String language; |
|
| 43 |
|
|
| 44 |
@XmlElement(name = "type") |
|
| 45 |
private String type; |
|
| 46 |
|
|
| 47 |
@XmlElementWrapper(name = "urls") |
|
| 48 |
@XmlElement(name = "url") |
|
| 49 |
private Set<MyURL> urls = new TreeSet<>((u1, u2) -> {
|
|
| 50 |
final String r1 = u1.getRights(); |
|
| 51 |
final String r2 = u2.getRights(); |
|
| 52 |
final int n1 = r1.equalsIgnoreCase("Open Access") ? 0 : r1.equalsIgnoreCase("Unknown") ? 2 : 1;
|
|
| 53 |
final int n2 = r2.equalsIgnoreCase("Open Access") ? 0 : r2.equalsIgnoreCase("Unknown") ? 2 : 1;
|
|
| 54 |
return Integer.compare(n1, n2); |
|
| 55 |
}); |
|
| 56 |
|
|
| 57 |
@XmlElementWrapper(name = "dois") |
|
| 58 |
@XmlElement(name = "doi") |
|
| 59 |
private Set<String> dois = new LinkedHashSet<>(); |
|
| 60 |
|
|
| 61 |
@XmlElement(name = "bestRights") |
|
| 62 |
private String bestRights; |
|
| 63 |
|
|
| 64 |
@XmlElement(name = "abstract") |
|
| 65 |
private String descriptionAbstract; |
|
| 66 |
|
|
| 67 |
@XmlElement(name = "source") |
|
| 68 |
private String source; |
|
| 69 |
|
|
| 70 |
@XmlElementWrapper(name = "projects") |
|
| 71 |
@XmlElement(name = "project") |
|
| 72 |
private Set<Project> projects = new TreeSet<>((p1, p2) -> p1.getName().compareTo(p2.getName())); |
|
| 73 |
|
|
| 74 |
@XmlElementWrapper(name = "cnrPersons") |
|
| 75 |
@XmlElement(name = "cnrPerson") |
|
| 76 |
private Set<CnrPerson> cnrPersons = new TreeSet<>((p1, p2) -> p1.getName().compareTo(p2.getName())); |
|
| 77 |
|
|
| 78 |
@XmlElementWrapper(name = "datasets") |
|
| 79 |
@XmlElement(name = "dataset") |
|
| 80 |
private Set<Dataset> datasets = new TreeSet<>((d1, d2) -> d1.getTitle().compareTo(d2.getTitle())); |
|
| 81 |
|
|
| 82 |
@XmlElementWrapper(name = "citations") |
|
| 83 |
@XmlElement(name = "citation") |
|
| 84 |
private Set<String> citations = new LinkedHashSet<>(); |
|
| 85 |
|
|
| 86 |
public String getId() {
|
|
| 87 |
return id; |
|
| 88 |
} |
|
| 89 |
|
|
| 90 |
public void setId(final String id) {
|
|
| 91 |
this.id = id; |
|
| 92 |
} |
|
| 93 |
|
|
| 94 |
public String getTitle() {
|
|
| 95 |
return title; |
|
| 96 |
} |
|
| 97 |
|
|
| 98 |
public void setTitle(final String title) {
|
|
| 99 |
this.title = title; |
|
| 100 |
} |
|
| 101 |
|
|
| 102 |
public Set<String> getCreators() {
|
|
| 103 |
return creators; |
|
| 104 |
} |
|
| 105 |
|
|
| 106 |
public void setCreators(final Set<String> creators) {
|
|
| 107 |
this.creators = creators; |
|
| 108 |
} |
|
| 109 |
|
|
| 110 |
public String getPublisher() {
|
|
| 111 |
return publisher; |
|
| 112 |
} |
|
| 113 |
|
|
| 114 |
public void setPublisher(final String publisher) {
|
|
| 115 |
this.publisher = publisher; |
|
| 116 |
} |
|
| 117 |
|
|
| 118 |
public Set<String> getSubjects() {
|
|
| 119 |
return subjects; |
|
| 120 |
} |
|
| 121 |
|
|
| 122 |
public void setSubjects(final Set<String> subjects) {
|
|
| 123 |
this.subjects = subjects; |
|
| 124 |
} |
|
| 125 |
|
|
| 126 |
public Set<String> getContributors() {
|
|
| 127 |
return contributors; |
|
| 128 |
} |
|
| 129 |
|
|
| 130 |
public void setContributors(final Set<String> contributors) {
|
|
| 131 |
this.contributors = contributors; |
|
| 132 |
} |
|
| 133 |
|
|
| 134 |
public int getDate() {
|
|
| 135 |
return date; |
|
| 136 |
} |
|
| 137 |
|
|
| 138 |
public void setDate(final int date) {
|
|
| 139 |
this.date = date; |
|
| 140 |
} |
|
| 141 |
|
|
| 142 |
public String getLanguage() {
|
|
| 143 |
return language; |
|
| 144 |
} |
|
| 145 |
|
|
| 146 |
public void setLanguage(final String language) {
|
|
| 147 |
this.language = language; |
|
| 148 |
} |
|
| 149 |
|
|
| 150 |
public String getType() {
|
|
| 151 |
return type; |
|
| 152 |
} |
|
| 153 |
|
|
| 154 |
public void setType(final String type) {
|
|
| 155 |
this.type = type; |
|
| 156 |
} |
|
| 157 |
|
|
| 158 |
public Set<MyURL> getUrls() {
|
|
| 159 |
return urls; |
|
| 160 |
} |
|
| 161 |
|
|
| 162 |
public void setUrls(final Set<MyURL> urls) {
|
|
| 163 |
this.urls = urls; |
|
| 164 |
} |
|
| 165 |
|
|
| 166 |
public Set<String> getDois() {
|
|
| 167 |
return dois; |
|
| 168 |
} |
|
| 169 |
|
|
| 170 |
public void setDois(final Set<String> dois) {
|
|
| 171 |
this.dois = dois; |
|
| 172 |
} |
|
| 173 |
|
|
| 174 |
public String getBestRights() {
|
|
| 175 |
return bestRights; |
|
| 176 |
} |
|
| 177 |
|
|
| 178 |
public void setBestRights(final String bestRights) {
|
|
| 179 |
this.bestRights = bestRights; |
|
| 180 |
} |
|
| 181 |
|
|
| 182 |
public String getDescriptionAbstract() {
|
|
| 183 |
return descriptionAbstract; |
|
| 184 |
} |
|
| 185 |
|
|
| 186 |
public void setDescriptionAbstract(final String descriptionAbstract) {
|
|
| 187 |
this.descriptionAbstract = descriptionAbstract; |
|
| 188 |
} |
|
| 189 |
|
|
| 190 |
public String getSource() {
|
|
| 191 |
return source; |
|
| 192 |
} |
|
| 193 |
|
|
| 194 |
public void setSource(final String source) {
|
|
| 195 |
this.source = source; |
|
| 196 |
} |
|
| 197 |
|
|
| 198 |
public Set<Project> getProjects() {
|
|
| 199 |
return projects; |
|
| 200 |
} |
|
| 201 |
|
|
| 202 |
public void setProjects(final Set<Project> projects) {
|
|
| 203 |
this.projects = projects; |
|
| 204 |
} |
|
| 205 |
|
|
| 206 |
public Set<CnrPerson> getCnrPersons() {
|
|
| 207 |
return cnrPersons; |
|
| 208 |
} |
|
| 209 |
|
|
| 210 |
public void setCnrPerson(final Set<CnrPerson> cnrPersons) {
|
|
| 211 |
this.cnrPersons = cnrPersons; |
|
| 212 |
} |
|
| 213 |
|
|
| 214 |
public Set<Dataset> getDatasets() {
|
|
| 215 |
return datasets; |
|
| 216 |
} |
|
| 217 |
|
|
| 218 |
public void setDatasets(final Set<Dataset> datasets) {
|
|
| 219 |
this.datasets = datasets; |
|
| 220 |
} |
|
| 221 |
|
|
| 222 |
public Set<String> getCitations() {
|
|
| 223 |
return citations; |
|
| 224 |
} |
|
| 225 |
|
|
| 226 |
public void setCitations(final Set<String> citations) {
|
|
| 227 |
this.citations = citations; |
|
| 228 |
} |
|
| 229 |
|
|
| 230 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/objects/MdRecordConvertUtils.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.mdstore.plugins.objects; |
|
| 2 |
|
|
| 3 |
import java.io.StringReader; |
|
| 4 |
import java.io.StringWriter; |
|
| 5 |
|
|
| 6 |
import javax.xml.bind.JAXBContext; |
|
| 7 |
import javax.xml.bind.JAXBException; |
|
| 8 |
import javax.xml.bind.Marshaller; |
|
| 9 |
import javax.xml.bind.Unmarshaller; |
|
| 10 |
|
|
| 11 |
public class MdRecordConvertUtils {
|
|
| 12 |
|
|
| 13 |
private static JAXBContext jaxbContext; |
|
| 14 |
private static Marshaller marshaller; |
|
| 15 |
private static Unmarshaller unmarshaller; |
|
| 16 |
|
|
| 17 |
public static MdRecord fromString(final String s) throws JAXBException {
|
|
| 18 |
verifyInit(); |
|
| 19 |
return (MdRecord) unmarshaller.unmarshal(new StringReader(s)); |
|
| 20 |
|
|
| 21 |
} |
|
| 22 |
|
|
| 23 |
public static String toString(final MdRecord obj) throws JAXBException {
|
|
| 24 |
final StringWriter sw = new StringWriter(); |
|
| 25 |
marshaller.marshal(obj, sw); |
|
| 26 |
return sw.toString(); |
|
| 27 |
} |
|
| 28 |
|
|
| 29 |
private static void verifyInit() throws JAXBException {
|
|
| 30 |
if (jaxbContext == null) {
|
|
| 31 |
jaxbContext = JAXBContext.newInstance(MdRecord.class); |
|
| 32 |
|
|
| 33 |
marshaller = jaxbContext.createMarshaller(); |
|
| 34 |
unmarshaller = jaxbContext.createUnmarshaller(); |
|
| 35 |
|
|
| 36 |
marshaller.setProperty(Marshaller.JAXB_FORMATTED_OUTPUT, true); |
|
| 37 |
} |
|
| 38 |
} |
|
| 39 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/objects/Dataset.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.mdstore.plugins.objects; |
|
| 2 |
|
|
| 3 |
import javax.xml.bind.annotation.XmlAccessType; |
|
| 4 |
import javax.xml.bind.annotation.XmlAccessorType; |
|
| 5 |
import javax.xml.bind.annotation.XmlAttribute; |
|
| 6 |
import javax.xml.bind.annotation.XmlValue; |
|
| 7 |
|
|
| 8 |
@XmlAccessorType(XmlAccessType.FIELD) |
|
| 9 |
public class Dataset {
|
|
| 10 |
|
|
| 11 |
@XmlAttribute(name = "doi") |
|
| 12 |
private String doi; |
|
| 13 |
|
|
| 14 |
@XmlAttribute(name = "url") |
|
| 15 |
private String url; |
|
| 16 |
|
|
| 17 |
@XmlValue |
|
| 18 |
private String title; |
|
| 19 |
|
|
| 20 |
public Dataset() {}
|
|
| 21 |
|
|
| 22 |
public Dataset(final String doi, final String title) {
|
|
| 23 |
this.doi = doi; |
|
| 24 |
this.title = title; |
|
| 25 |
setUrl("https://dx.doi.org/" + doi);
|
|
| 26 |
} |
|
| 27 |
|
|
| 28 |
public Dataset(final String doi, final String url, final String title) {
|
|
| 29 |
this.doi = doi; |
|
| 30 |
this.url = url; |
|
| 31 |
this.title = title; |
|
| 32 |
} |
|
| 33 |
|
|
| 34 |
public String getDoi() {
|
|
| 35 |
return doi; |
|
| 36 |
} |
|
| 37 |
|
|
| 38 |
public void setDoi(final String doi) {
|
|
| 39 |
this.doi = doi; |
|
| 40 |
} |
|
| 41 |
|
|
| 42 |
public String getUrl() {
|
|
| 43 |
return url; |
|
| 44 |
} |
|
| 45 |
|
|
| 46 |
public void setUrl(final String url) {
|
|
| 47 |
this.url = url; |
|
| 48 |
} |
|
| 49 |
|
|
| 50 |
public String getTitle() {
|
|
| 51 |
return title; |
|
| 52 |
} |
|
| 53 |
|
|
| 54 |
public void setTitle(final String title) {
|
|
| 55 |
this.title = title; |
|
| 56 |
} |
|
| 57 |
|
|
| 58 |
@Override |
|
| 59 |
public int hashCode() {
|
|
| 60 |
return doi.toLowerCase().hashCode(); |
|
| 61 |
} |
|
| 62 |
|
|
| 63 |
@Override |
|
| 64 |
public boolean equals(final Object obj) {
|
|
| 65 |
if (this == obj) { return true; }
|
|
| 66 |
if (obj == null) { return false; }
|
|
| 67 |
return (obj instanceof Dataset) && doi.equalsIgnoreCase(((Dataset) obj).getDoi()); |
|
| 68 |
} |
|
| 69 |
|
|
| 70 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/objects/Affiliation.java | ||
|---|---|---|
| 1 |
package eu.dnetlib.data.mdstore.plugins.objects; |
|
| 2 |
|
|
| 3 |
import javax.xml.bind.annotation.XmlAccessType; |
|
| 4 |
import javax.xml.bind.annotation.XmlAccessorType; |
|
| 5 |
import javax.xml.bind.annotation.XmlAttribute; |
|
| 6 |
|
|
| 7 |
@XmlAccessorType(XmlAccessType.FIELD) |
|
| 8 |
public class Affiliation {
|
|
| 9 |
|
|
| 10 |
@XmlAttribute(name = "code") |
|
| 11 |
private String code; |
|
| 12 |
|
|
| 13 |
@XmlAttribute(name = "type") |
|
| 14 |
private String type; |
|
| 15 |
|
|
| 16 |
@XmlAttribute(name = "groupAcronym") |
|
| 17 |
private String groupAcronym; |
|
| 18 |
|
|
| 19 |
@XmlAttribute(name = "group") |
|
| 20 |
private String group; |
|
| 21 |
|
|
| 22 |
public Affiliation() {}
|
|
| 23 |
|
|
| 24 |
public Affiliation(final String code, final String type, final String groupAcronym, final String group) {
|
|
| 25 |
super(); |
|
| 26 |
this.code = code; |
|
| 27 |
this.type = type; |
|
| 28 |
this.groupAcronym = groupAcronym; |
|
| 29 |
this.group = group; |
|
| 30 |
} |
|
| 31 |
|
|
| 32 |
public String getGroupAcronym() {
|
|
| 33 |
return groupAcronym; |
|
| 34 |
} |
|
| 35 |
|
|
| 36 |
public void setGroupAcronym(final String groupAcronym) {
|
|
| 37 |
this.groupAcronym = groupAcronym; |
|
| 38 |
} |
|
| 39 |
|
|
| 40 |
public String getGroup() {
|
|
| 41 |
return group; |
|
| 42 |
} |
|
| 43 |
|
|
| 44 |
public void setGroup(final String group) {
|
|
| 45 |
this.group = group; |
|
| 46 |
} |
|
| 47 |
|
|
| 48 |
public String getType() {
|
|
| 49 |
return type; |
|
| 50 |
} |
|
| 51 |
|
|
| 52 |
public void setType(final String type) {
|
|
| 53 |
this.type = type; |
|
| 54 |
} |
|
| 55 |
|
|
| 56 |
public String getCode() {
|
|
| 57 |
return code; |
|
| 58 |
} |
|
| 59 |
|
|
| 60 |
public void setCode(final String code) {
|
|
| 61 |
this.code = code; |
|
| 62 |
} |
|
| 63 |
|
|
| 64 |
@Override |
|
| 65 |
public int hashCode() {
|
|
| 66 |
return groupAcronym == null ? 0 : groupAcronym.hashCode(); |
|
| 67 |
|
|
| 68 |
} |
|
| 69 |
|
|
| 70 |
@Override |
|
| 71 |
public boolean equals(final Object obj) {
|
|
| 72 |
if (this == obj) { return true; }
|
|
| 73 |
if (obj == null) { return false; }
|
|
| 74 |
if (getClass() != obj.getClass()) { return false; }
|
|
| 75 |
final Affiliation other = (Affiliation) obj; |
|
| 76 |
if (groupAcronym == null) {
|
|
| 77 |
if (other.groupAcronym != null) { return false; }
|
|
| 78 |
} else if (!groupAcronym.equals(other.groupAcronym)) { return false; }
|
|
| 79 |
return true; |
|
| 80 |
} |
|
| 81 |
|
|
| 82 |
} |
|
| modules/dnet-isti/trunk/src/main/java/eu/dnetlib/data/mdstore/plugins/objects/Project.java | ||
|---|---|---|
| 6 | 6 |
import java.util.regex.Matcher; |
| 7 | 7 |
import java.util.regex.Pattern; |
| 8 | 8 |
|
| 9 |
import javax.xml.bind.annotation.XmlAccessType; |
|
| 10 |
import javax.xml.bind.annotation.XmlAccessorType; |
|
| 11 |
import javax.xml.bind.annotation.XmlElement; |
|
| 12 |
|
|
| 9 | 13 |
import org.apache.commons.lang3.StringUtils; |
| 10 | 14 |
import org.apache.commons.logging.Log; |
| 11 | 15 |
import org.apache.commons.logging.LogFactory; |
| ... | ... | |
| 16 | 20 |
import org.springframework.http.client.HttpComponentsClientHttpRequestFactory; |
| 17 | 21 |
import org.springframework.web.client.RestTemplate; |
| 18 | 22 |
|
| 23 |
@XmlAccessorType(XmlAccessType.FIELD) |
|
| 19 | 24 |
public class Project {
|
| 20 | 25 |
|
| 26 |
@XmlElement(name = "infoId") |
|
| 27 |
private String infoId; |
|
| 28 |
@XmlElement(name = "openaireId") |
|
| 21 | 29 |
private String openaireId; |
| 30 |
@XmlElement(name = "code") |
|
| 22 | 31 |
private String code; |
| 32 |
@XmlElement(name = "name") |
|
| 23 | 33 |
private String name; |
| 34 |
@XmlElement(name = "acronym") |
|
| 24 | 35 |
private String acronym; |
| 36 |
@XmlElement(name = "funder") |
|
| 25 | 37 |
private String funder; |
| 38 |
@XmlElement(name = "program") |
|
| 26 | 39 |
private String program; |
| 40 |
@XmlElement(name = "jurisdiction") |
|
Also available in: Unified diff
refactoring using jaxb