From 8424511e9b509ea1a7a9e8b4695bc20ed852bbe2 Mon Sep 17 00:00:00 2001
From: Georgy Litvinov <git@litvinovg.pro>
Date: Thu, 25 Mar 2021 14:20:08 +0100
Subject: [PATCH] Extended CSV metadata to support multiple authors with
affiliations
---
.../java/w2phtml/rdf/DocumentStructure.java | 117 ++++++++++--------
1 file changed, 68 insertions(+), 49 deletions(-)
diff --git a/src/main/java/w2phtml/rdf/DocumentStructure.java b/src/main/java/w2phtml/rdf/DocumentStructure.java
index 25f085a..c6decc5 100644
--- a/src/main/java/w2phtml/rdf/DocumentStructure.java
+++ b/src/main/java/w2phtml/rdf/DocumentStructure.java
@@ -18,6 +18,8 @@ import org.apache.jena.ontology.OntModel;
import org.apache.jena.rdf.model.ModelFactory;
import org.apache.jena.rdf.model.Resource;
import org.apache.jena.vocabulary.*;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
import w2phtml.xhtml.XhtmlConfig;
import w2phtml.xhtml.XhtmlDocument;
@@ -26,6 +28,24 @@ import org.apache.jena.rdf.model.Property;
public class DocumentStructure {
+
+ private static final String AFFILIATED_ORGANIZATION_POSTAL_CODE = "affiliated organization postal code";
+
+ private static final String AFFILIATED_ORGANIZATION_ADDRESS = "affiliated organization address";
+
+ private static final String AFFILIATED_ORGANIZATION_OFFICIAL_NAME = "affiliated organization official name";
+
+ private static final String AFFILIATED_ORGANIZATION_NAME = "affiliated organization name";
+
+ private static final String AUTHOR_INITIALS = "author initials";
+
+ private static final String AUTHOR_EMAIL = "author email";
+
+ private static final String AUTHOR_FAMILY = "author family";
+
+ private static final Logger logger = LoggerFactory.getLogger(DocumentStructure.class);
+
+ private static final String AUTHOR_GIVEN_NAME = "author given name";
private static final String MODIFICATION_TIME = "modificationTime";
private static final String HTML_EXCERPT_PROPERTY = "htmlExcerpt";
private static final String TOCITEM = "TOCItem";
@@ -211,80 +231,79 @@ public class DocumentStructure {
mainResource.addProperty( property, todayAsString);
}
- private void addAuthor(Resource resource, DocumentPart docPart) {
+ private void addAuthor(Resource excerpt, DocumentPart docPart) {
String order = docPart.getOrder();
ArrayList<Map<String, String>> sectionMeta = metadata.getSection(order);
for (Map<String, String> map : sectionMeta) {
Set<String> names = map.keySet();
- if (names.contains("author given name")) {
- attachAuthor(resource, map, order);
+ for (String givenNameKey : names) {
+ if (givenNameKey.startsWith(AUTHOR_GIVEN_NAME)) {
+ String postfix = givenNameKey.replaceFirst(AUTHOR_GIVEN_NAME, "");
+ if (!postfix.matches("[a-zA-Z0-9]*")) {
+ logger.error("Metadata author postfix is bad. Should contain only symbols a-zA-Z0-9, but "
+ + postfix + " found. Author givenName field is " + givenNameKey );
+ System.exit(1);
+ }
+ String participantURI = TS + PARTICIPANT + "/" + PARSERNAME + "_" + docID + order + "author" + postfix;
+ Resource participant = m.createResource(participantURI, participantClass);
+ Property hasAuthor = m.createProperty(TS + "hasAuthor");
+ excerpt.addProperty(hasAuthor, participant);
+ String givenName = map.get(givenNameKey);
+ Property givenNameProperty = m.createProperty(TS + "participantGivenName");
+ participant.addProperty( givenNameProperty, givenName.trim());
+
+ String family = map.get(AUTHOR_FAMILY + postfix);
+ if (family != null) {
+ Property familyProperty = m.createProperty(TS + "participantFamily");
+ participant.addProperty( familyProperty, family.trim());
+ }
+ String labelName = givenNameKey.trim();
+ if (family != null ) {
+ labelName += " " + family;
+ }
+ participant.addProperty(RDFS.label, labelName);
+
+ String email = map.get(AUTHOR_EMAIL + postfix);
+ if (email != null) {
+ Property emailProperty = m.createProperty(TS + "participantEmail");
+ participant.addProperty( emailProperty, email.trim());
+ }
+ String authorInitials = map.get(AUTHOR_INITIALS + postfix);
+ if (email != null) {
+ Property initialsProperty = m.createProperty(TS + "participantInitials");
+ participant.addProperty( initialsProperty, authorInitials.trim());
+ }
+ attachOrganization(participant, map, order, postfix);
+ }
}
}
}
- private void attachAuthor(Resource resource, Map<String, String> map, String order) {
- String participantURI = TS + PARTICIPANT + "/" + PARSERNAME + "_" + docID + order ;
- Resource participant = m.createResource(participantURI, participantClass);
- Property hasAuthor = m.createProperty(TS + "hasAuthor");
- resource.addProperty(hasAuthor, participant);
- createParticipant(map, order, participant);
- }
-
- private void createParticipant(Map<String, String> map, String order, Resource participant) {
- String givenName = map.get("author given name");
- if (givenName != null) {
- Property givenNameProperty = m.createProperty(TS + "participantGivenName");
- participant.addProperty( givenNameProperty, givenName.trim());
- }
- String family = map.get("author family");
- if (family != null) {
- Property familyProperty = m.createProperty(TS + "participantFamily");
- participant.addProperty( familyProperty, family.trim());
- }
- if (givenName != null ) {
- String labelName = givenName.trim();
- if (family != null ) {
- labelName += " " + family;
- }
- participant.addProperty(RDFS.label, labelName);
- }
- String email = map.get("author email");
- if (email != null) {
- Property emailProperty = m.createProperty(TS + "participantEmail");
- participant.addProperty( emailProperty, email.trim());
- }
- String authorInitials = map.get("author initials");
- if (email != null) {
- Property initialsProperty = m.createProperty(TS + "participantInitials");
- participant.addProperty( initialsProperty, authorInitials.trim());
- }
- attachOrganization(participant, map, order);
- }
-
- private void attachOrganization(Resource participant, Map<String, String> map, String order) {
- String orgName = map.get("affiliated organization name");
- System.out.println(orgName);
+ private void attachOrganization(Resource participant, Map<String, String> map, String order, String postfix) {
+ String orgName = map.get(AFFILIATED_ORGANIZATION_NAME + postfix);
if (orgName == null) {
+ System.out.println(orgName + " not found");
+ logger.warn("No " + AFFILIATED_ORGANIZATION_NAME + postfix + " found.");
return;
}
- String organizationUri = TS + ORGANIZATION + "/" + PARSERNAME + "_" + docID + order;
+ String organizationUri = TS + ORGANIZATION + "/" + PARSERNAME + "_" + docID + order + "author" + postfix;
Resource organization = m.createResource(organizationUri, organizationClass);
Property affiliatedWith = m.createProperty(TS + "affiliatedWith");
participant.addProperty(affiliatedWith, organization);
organization.addProperty(RDFS.label, orgName);
- String officialName = map.get("affiliated organization official name");
+ String officialName = map.get(AFFILIATED_ORGANIZATION_OFFICIAL_NAME + postfix);
if (officialName != null) {
Property officialNameProperty = m.createProperty(TS + "officialOrganizationName");
organization.addProperty(officialNameProperty, officialName);
}
- String orgAddress = map.get("affiliated organization address");
+ String orgAddress = map.get(AFFILIATED_ORGANIZATION_ADDRESS + postfix);
if (orgAddress != null) {
Property orgAddressProperty = m.createProperty(TS + "organizationAddress");
organization.addProperty(orgAddressProperty, orgAddress);
}
- String postalCode = map.get("affiliated organization postal code");
+ String postalCode = map.get(AFFILIATED_ORGANIZATION_POSTAL_CODE + postfix);
if (postalCode != null) {
Property postalCodeProperty = m.createProperty(TS + "organizationPostalCode");
organization.addProperty(postalCodeProperty, postalCode);