litvinovg/vivo
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

[VIVO-1306] fixing LCSH lookup (#45)

Browse source
This commit is contained in:
hudajkhan 2016-12-06 09:16:50 -05:00 • committed by grahamtriggs
parent 8ee3e1e74f
commit da8943e64a
1 changed files with 23 additions and 29 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

52
api/src/main/java/edu/cornell/mannlib/semservices/service/impl/LCSHService.java
View file

@ -49,7 +49,7 @@ public class LCSHService implements ExternalConceptService {
String results = null; String results = null;
String dataUrl = baseUri + "?q=" + URLEncoder.encode(term, "UTF-8") String dataUrl = baseUri + "?q=" + URLEncoder.encode(term, "UTF-8")
+ "&q=cs%3Ahttp%3A%2F%2Fid.loc.gov%2Fauthorities%2Fsubjects" + "&q=cs%3Ahttp%3A%2F%2Fid.loc.gov%2Fauthorities%2Fsubjects"
+ "&format=XML"; + "&format=atom";
log.debug("dataURL " + dataUrl); log.debug("dataURL " + dataUrl);
try { try {
@ -81,7 +81,7 @@ public class LCSHService implements ExternalConceptService {
return conceptList; return conceptList;
} }
// Results are in json format (atom) - atom entries need to be extracted // Results are in XML format (atom) - atom entries need to be extracted
// retrieve the URIs and get the SKOS version of the entry, getting broader // retrieve the URIs and get the SKOS version of the entry, getting broader
// and narrower terms as applicable as well as any description (skos:note) // and narrower terms as applicable as well as any description (skos:note)
// that might exist // that might exist
@ -99,7 +99,7 @@ public class LCSHService implements ExternalConceptService {
//This is the URL for retrieving the concept - the pattern is http://id.loc.gov/authorities/subjects/sh85014203.skos.rdf //This is the URL for retrieving the concept - the pattern is http://id.loc.gov/authorities/subjects/sh85014203.skos.rdf
//This is not the URI itself which would be http://id.loc.gov/authorities/subjects/sh85014203 //This is not the URI itself which would be http://id.loc.gov/authorities/subjects/sh85014203
String conceptURLString = getSKOSURL(uri); String conceptURLString = getSKOSURL(uri);
String baseConceptURI = getConceptURI(uri); String baseConceptURI = uri;
URL conceptURL = null; URL conceptURL = null;
try { try {
conceptURL = new URL(conceptURLString); conceptURL = new URL(conceptURLString);
@ -154,23 +154,12 @@ public class LCSHService implements ExternalConceptService {
private String getSKOSURL(String uri) { private String getSKOSURL(String uri) {
// Strip .xml at the end and replace with .skos.rdf String skosURI = uri + skosSuffix;
String skosURI = uri;
if (uri.endsWith(".xml")) { return skosURI;
skosURI = uri.substring(0, uri.length() - 4);
skosURI += skosSuffix;
}
return hostUri + skosURI;
} }
//Given the URI from the xml, get just the base URI
private String getConceptURI(String uri) {
String skosURI = uri;
if (uri.endsWith(".xml")) {
skosURI = uri.substring(0, uri.length() - 4);
}
return hostUri + skosURI;
}
public List<String> getConceptURISFromJSON(String results) { public List<String> getConceptURISFromJSON(String results) {
List<String> uris = new ArrayList<String>(); List<String> uris = new ArrayList<String>();
@ -192,14 +181,25 @@ public class LCSHService implements ExternalConceptService {
String conceptUri = new String(); String conceptUri = new String();
try { try {
Document doc = XMLUtils.parse(rdf); Document doc = XMLUtils.parse(rdf);
NodeList nodes = doc.getElementsByTagName("search:result"); NodeList nodes = doc.getElementsByTagName("entry");
int len = nodes.getLength(); int len = nodes.getLength();
int i; int i, j;
for (i = 0; i < len; i++) { for (i = 0; i < len; i++) {
Node node = nodes.item(i); Node node = nodes.item(i);
NamedNodeMap attrs = node.getAttributes(); NodeList childNodes = node.getChildNodes();
Attr idAttr = (Attr) attrs.getNamedItem("uri"); for(j = 0; j < childNodes.getLength(); j++) {
conceptUri = idAttr.getTextContent(); Node childNode = childNodes.item(j);
if(childNode.getNodeName().equals("link")) {
NamedNodeMap attrs = childNode.getAttributes();
Attr hrefAttr = (Attr) attrs.getNamedItem("href");
//if type doesn't exist, this is the direct URL without extension
if((hrefAttr != null) && ((Attr)attrs.getNamedItem("type") == null)) {
conceptUri = hrefAttr.getTextContent();
}
}
}
log.debug("concept uri is " + conceptUri); log.debug("concept uri is " + conceptUri);
uris.add(conceptUri); uris.add(conceptUri);
} }
@ -250,10 +250,4 @@ public class LCSHService implements ExternalConceptService {
} }
} }