sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

extended OCRID cleaning procedure

This commit is contained in:
Claudio Atzori 2020-12-21 11:40:17 +01:00
parent 573a8a3272
commit 6cb0dc3f43
1 changed files with 23 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
dhp-workflows/dhp-graph-mapper/src/main/java/eu/dnetlib/dhp/oa/graph/clean/CleaningFunctions.java
View File

@ -16,7 +16,7 @@ import eu.dnetlib.dhp.schema.oaf.*;
public class CleaningFunctions {
public static final String DOI_URL_PREFIX_REGEX = "(^http(s?):\\/\\/)(((dx\\.)?doi\\.org)|(handle\\.test\\.datacite\\.org))\\/";
public static final String ORCID_PREFIX_REGEX = "^http(s?):\\/\\/orcid\\.org\\/";
public static final String ORCID_CLEANING_REGEX = ".*([0-9]{4}).*[-–—−=].*([0-9]{4}).*[-–—−=].*([0-9]{4}).*[-–—−=].*([0-9x]{4})";
public static final String CLEANING_REGEX = "(?:\\n|\\r|\\t)";
public static final Set<String> PID_BLACKLIST = new HashSet<>();
@ -211,14 +211,31 @@ public class CleaningFunctions {
.map(Qualifier::getClassid)
.orElse(""))
.orElse("");
if (pidProvenance.equals(ModelConstants.SYSIMPORT_CROSSWALK_ENTITYREGISTRY)) {
if (p
.getQualifier()
.getClassid()
.toLowerCase()
.contains(ModelConstants.ORCID)) {
if (pidProvenance
.equals(ModelConstants.SYSIMPORT_CROSSWALK_ENTITYREGISTRY)) {
p.getQualifier().setClassid(ModelConstants.ORCID);
} else {
p.getQualifier().setClassid(ModelConstants.ORCID_PENDING);
}
p.setValue(p.getValue().trim().replaceAll(ORCID_PREFIX_REGEX, ""));
final String orcid = p
.getValue()
.trim()
.toLowerCase()
.replaceAll(ORCID_CLEANING_REGEX, "$1-$2-$3-$4");
if (orcid.length() == 19) {
p.setValue(orcid);
} else {
p.setValue("");
}
}
return p;
})
.filter(p -> StringUtils.isNotBlank(p.getValue()))
.collect(
Collectors
.toMap(