Skip to content

Commit

Permalink
some fixes
Browse files Browse the repository at this point in the history
  • Loading branch information
@kermitt2
kermitt2 committed Aug 12, 2020
1 parent 1323638 commit f1c3cf9
Show file tree
Hide file tree
Showing 2 changed files with 56 additions and 89 deletions.
135 changes: 51 additions & 84 deletions grobid-core/src/main/java/org/grobid/core/data/BiblioItem.java
View file
Expand Up @@ -4253,12 +4253,12 @@ else if (bibo.getFullAuthors().size() == 1) {
List<Person> auts = bib.getFullAuthors();
if (auts != null) {
for (Person aut : auts) {
if (aut.getLastName() != null) {
if (aut.getLastName().equals(auto.getLastName())) {
if (StringUtils.isNotBlank(aut.getLastName()) && StringUtils.isNotBlank(auto.getLastName())) {
if (aut.getLastName().toLowerCase().equals(auto.getLastName().toLowerCase())) {
if (StringUtils.isBlank(aut.getFirstName()) ||
(auto.getFirstName() != null &&
aut.getFirstName().length() <= auto.getFirstName().length() &&
auto.getFirstName().startsWith(aut.getFirstName()))) {
auto.getFirstName().toLowerCase().startsWith(aut.getFirstName().toLowerCase()))) {
aut.setFirstName(auto.getFirstName());
aut.setCorresp(true);
if (StringUtils.isNotBlank(auto.getEmail()))
Expand All @@ -4271,101 +4271,68 @@ else if (bibo.getFullAuthors().size() == 1) {
}
}
}

/*List<Person> thePersons = new ArrayList<>();
for(Person thePerson : bib.getFullAuthors()) {
thePersons.add(thePerson.clonePerson());
}
thePersons.add(bibo.getFullAuthors().get(0));
Person.deduplicate(thePersons);
if (thePersons.size() == bib.getFullAuthors().size()) {
bib.setFullAuthors(thePersons);
}*/
} else if (bibo.getFullAuthors().size() > 1) {
// we have the complete list of authors so we can take them from the second
// biblio item and merge some possible extra from the first when a match is
// reliable
for (Person aut : bibo.getFullAuthors()) {
// try to find the author in the first item (we know it's not empty)
for (Person aut2 : bib.getFullAuthors()) {


if (StringUtils.isNotBlank(aut2.getLastName())) {
if (StringUtils.isNotBlank(aut.getLastName()) && aut.getLastName().equals(aut2.getLastName())) {
// check also first name if present - at least for the initial
if ( StringUtils.isBlank(aut2.getFirstName()) ||
(StringUtils.isNotBlank(aut2.getFirstName()) && StringUtils.isNotBlank(aut.getFirstName())) ) {
// we have no first name or a match (full first name)

if ( StringUtils.isBlank(aut2.getFirstName())
||
aut.getFirstName().equals(aut2.getFirstName())
||
( aut.getFirstName().length() == 1 &&
aut.getFirstName().equals(aut2.getFirstName().substring(0,1)))
) {
// we have a match (full or initial)
if (StringUtils.isNotBlank(aut2.getFirstName()) &&
aut2.getFirstName().length() > aut.getFirstName().length())
aut.setFirstName(aut2.getFirstName());
if (StringUtils.isBlank(aut.getMiddleName()))
aut.setMiddleName(aut2.getMiddleName());
if (StringUtils.isBlank(aut.getTitle()))
aut.setTitle(aut2.getTitle());
if (StringUtils.isBlank(aut.getSuffix()))
aut.setSuffix(aut2.getSuffix());
if (StringUtils.isBlank(aut.getORCID()))
aut.setORCID(aut2.getORCID());
if (StringUtils.isBlank(aut.getEmail()))
aut.setEmail(aut2.getEmail());
if(!CollectionUtils.isEmpty(aut2.getAffiliations()))
aut.setAffiliations(aut2.getAffiliations());
if (!CollectionUtils.isEmpty(aut2.getAffiliationBlocks()))
aut.setAffiliationBlocks(aut2.getAffiliationBlocks());
if (!CollectionUtils.isEmpty(aut2.getAffiliationMarkers()))
aut.setAffiliationMarkers(aut2.getAffiliationMarkers());
if (!CollectionUtils.isEmpty(aut2.getMarkers()))
aut.setMarkers(aut2.getMarkers());
if (!CollectionUtils.isEmpty(aut2.getLayoutTokens()))
aut.setLayoutTokens(aut2.getLayoutTokens());
break;
}
}
String aut2_lastname = aut2.getLastName().toLowerCase();

if (StringUtils.isNotBlank(aut.getLastName())) {
String aut_lastname = aut.getLastName().toLowerCase();

if (aut_lastname.equals(aut2_lastname)) {
// check also first name if present - at least for the initial
if ( StringUtils.isBlank(aut2.getFirstName()) ||
(StringUtils.isNotBlank(aut2.getFirstName()) && StringUtils.isNotBlank(aut.getFirstName())) ) {
// we have no first name or a match (full first name)

if ( StringUtils.isBlank(aut2.getFirstName())
||
aut.getFirstName().equals(aut2.getFirstName())
||
( aut.getFirstName().length() == 1 &&
aut.getFirstName().equals(aut2.getFirstName().substring(0,1)) )
) {
// we have a match (full or initial)
if (StringUtils.isNotBlank(aut2.getFirstName()) &&
aut2.getFirstName().length() > aut.getFirstName().length())
aut.setFirstName(aut2.getFirstName());
if (StringUtils.isBlank(aut.getMiddleName()))
aut.setMiddleName(aut2.getMiddleName());
if (StringUtils.isBlank(aut.getTitle()))
aut.setTitle(aut2.getTitle());
if (StringUtils.isBlank(aut.getSuffix()))
aut.setSuffix(aut2.getSuffix());
if (StringUtils.isBlank(aut.getORCID()))
aut.setORCID(aut2.getORCID());
if (StringUtils.isBlank(aut.getEmail()))
aut.setEmail(aut2.getEmail());
if(!CollectionUtils.isEmpty(aut2.getAffiliations()))
aut.setAffiliations(aut2.getAffiliations());
if (!CollectionUtils.isEmpty(aut2.getAffiliationBlocks()))
aut.setAffiliationBlocks(aut2.getAffiliationBlocks());
if (!CollectionUtils.isEmpty(aut2.getAffiliationMarkers()))
aut.setAffiliationMarkers(aut2.getAffiliationMarkers());
if (!CollectionUtils.isEmpty(aut2.getMarkers()))
aut.setMarkers(aut2.getMarkers());
if (!CollectionUtils.isEmpty(aut2.getLayoutTokens()))
aut.setLayoutTokens(aut2.getLayoutTokens());
break;
}
}
}
}
}
}
}
bib.setFullAuthors(bibo.getFullAuthors());

/*List<Person> correctedAuthors = new ArrayList<>();
for (Person aut : bib.getFullAuthors()) {
boolean found = false;
for (Person aut2 : bibo.getFullAuthors()) {
List<Person> thePersons = new ArrayList<>();
thePersons.add(aut2.clonePerson());
thePersons.add(aut.clonePerson());
Person.deduplicate(thePersons);
if (thePersons.size() == 1) {
correctedAuthors.add(thePersons.get(0));
found = true;
break;
}
}
if (!found) {
correctedAuthors.add(aut);
}
}
bib.setFullAuthors(correctedAuthors);*/
}

/*if (CollectionUtils.isEmpty(bib.getFullAuthors()))
bib.setFullAuthors(bibo.getFullAuthors());
else {
// we have the complete list of authors so we can take them from the second
// biblio item and merge some possible extra from the first when a match is
// reliable
List<Person> thePersons = bib.getFullAuthors();
thePersons.addAll(bibo.getFullAuthors());
Person.deduplicate(thePersons);
}*/
}
}

Expand Down
10 changes: 5 additions & 5 deletions grobid-core/src/main/java/org/grobid/core/data/Person.java
View file
Expand Up @@ -273,23 +273,23 @@ public String toTEI(boolean withCoordinates) {
XmlBuilderUtils.addCoords(persElement, LayoutTokensUtil.getCoordsString(getLayoutTokens()));
}
if (title != null) {
persElement.appendChild(XmlBuilderUtils.teiElement("roleName", TextUtilities.HTMLEncode(title)));
persElement.appendChild(XmlBuilderUtils.teiElement("roleName", title));
}
if (firstName != null) {
Element forename = XmlBuilderUtils.teiElement("forename", TextUtilities.HTMLEncode(firstName));
Element forename = XmlBuilderUtils.teiElement("forename", firstName);
forename.addAttribute(new Attribute("type", "first"));
persElement.appendChild(forename);
}
if (middleName != null) {
Element mn = XmlBuilderUtils.teiElement("forename", TextUtilities.HTMLEncode(middleName));
Element mn = XmlBuilderUtils.teiElement("forename", middleName);
mn.addAttribute(new Attribute("type", "middle"));
persElement.appendChild(mn);
}
if (lastName != null) {
persElement.appendChild(XmlBuilderUtils.teiElement("surname", TextUtilities.HTMLEncode(lastName)));
persElement.appendChild(XmlBuilderUtils.teiElement("surname", lastName));
}
if (suffix != null) {
persElement.appendChild(XmlBuilderUtils.teiElement("genName", TextUtilities.HTMLEncode(suffix)));
persElement.appendChild(XmlBuilderUtils.teiElement("genName", suffix));
}

return XmlBuilderUtils.toXml(persElement);
Expand Down

0 comments on commit f1c3cf9

Please sign in to comment.