private boolean is200AnRDF(CachedHTTPResource resource) {
if (resource.isContainsRDF() != null) return resource.isContainsRDF();
if (resource != null && resource.getResponses() != null) {
for (SerialisableHttpResponse response : resource.getResponses()) {
if (response != null && response.getHeaders("Content-Type") != null) {
if (LinkedDataContent.contentTypes.contains(response.getHeaders("Content-Type"))) {
if (response.getHeaders("Content-Type").equals(WebContent.contentTypeTextPlain)) {
Model m = this.tryRead(resource.getUri());
if (m != null && m.size() == 0) {
this.createProblemQuad(resource.getUri(), DQM.SC200WithoutRDF);
resource.setContainsRDF(false);
return false;
}
}
this.createProblemQuad(resource.getUri(), DQM.SC200WithRDF);
resource.setContainsRDF(true);
return true;
}
}
}
}
this.createProblemQuad(resource.getUri(), DQM.SC200WithoutRDF);
resource.setContainsRDF(false);
return false;
}
// Private Method for checking forward linking
private void checkForForwardLinking() {
for (String uri : uriSet) {
CachedHTTPResource httpResource =
(CachedHTTPResource)
DiachronCacheManager.getInstance()
.getFromCache(DiachronCacheManager.HTTP_RESOURCE_CACHE, uri);
if (httpResource == null
|| (httpResource.getResponses() == null
&& httpResource.getDereferencabilityStatusCode() != StatusCode.BAD)) {
this.notFetchedQueue.add(uri);
} else {
logger.info("Checking resource: {}. URIs left: {}.", httpResource.getUri(), uriSet.size());
// We perform a semantic lookup using heuristics to check if we
// really need to try parsing or not
if (HTTPResourceUtils.semanticURILookup(httpResource)) {
logger.info(
"Trying to find any dereferencable forward links for {}.", httpResource.getUri());
if (Dereferencer.hasValidDereferencability(httpResource)) {
logger.info("Dereferencable resource {}.", httpResource.getUri());
// boolean isValid = ModelParser.snapshotParserForForwardDereference(httpResource,
// (Lang) null, httpResource.getUri());
// if (isValid){
// //ok
// logger.info("A description exists for resource {}.", httpResource.getUri());
//
// totalDerefDataWithSub++;
// } else {
// //not ok
// this.createNotValidForwardLink(httpResource.getUri());
// }
Model m = RDFDataMgr.loadModel(httpResource.getUri()); // load partial model
Resource r = m.createResource(httpResource.getUri());
List<Statement> stmtList =
m.listStatements(r, (Property) null, (RDFNode) null).toList();
if (stmtList.size() > 1) {
// ok
logger.info("A description exists for resource {}.", httpResource.getUri());
totalDerefDataWithSub++;
} else {
// not ok
this.createNotValidForwardLink(httpResource.getUri());
}
}
} else {
logger.info("Non-meaningful dereferencable resource {}.", httpResource.getUri());
this.createNotValidForwardLink(httpResource.getUri());
}
}
}
}
// Private Methods for Dereferenceability Process
private boolean isDereferenceable(CachedHTTPResource httpResource) {
if (httpResource.getDereferencabilityStatusCode() == null) {
List<Integer> statusCode = this.getStatusCodes(httpResource.getStatusLines());
if (httpResource.getUri().contains("#") && statusCode.contains(200))
httpResource.setDereferencabilityStatusCode(StatusCode.HASH);
else if (statusCode.contains(200)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC200);
if (statusCode.contains(303)) httpResource.setDereferencabilityStatusCode(StatusCode.SC303);
else {
if (statusCode.contains(301))
httpResource.setDereferencabilityStatusCode(StatusCode.SC301);
else if (statusCode.contains(302))
httpResource.setDereferencabilityStatusCode(StatusCode.SC302);
else if (statusCode.contains(307))
httpResource.setDereferencabilityStatusCode(StatusCode.SC307);
}
}
if (has4xxCode(statusCode)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC4XX);
}
if (has5xxCode(statusCode)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC5XX);
}
}
StatusCode scode = httpResource.getDereferencabilityStatusCode();
return this.mapDerefStatusCode(scode);
}
// Private Method to check content type
private Model getMeaningfulData(CachedHTTPResource resource) {
Model m = null;
if (resource != null && resource.getResponses() != null) {
for (SerialisableHttpResponse response : resource.getResponses()) {
if (response != null && response.getHeaders("Content-Type") != null) {
if (CommonDataStructures.ldContentTypes.contains(response.getHeaders("Content-Type"))) {
m = this.tryRead(resource.getUri());
}
}
}
}
return m;
}
/**
* Tries to dereference all the URIs contained in the parameter, by retrieving them from the
* cache. URIs not found in the cache are added to the queue containing the URIs to be fetched by
* the async HTTP retrieval process
*
* @param uriSet Set of URIs to be dereferenced
* @return list with the results of the dereferenceability operations, for those URIs that were
* found in the cache
*/
private List<DerefResult> deReferenceUris(List<String> uriSet) {
// Start the dereferencing process, which will be run in parallel
httpRetriever.addListOfResourceToQueue(uriSet);
httpRetriever.start();
List<DerefResult> lstDerefUris = new ArrayList<DerefResult>();
List<String> lstToDerefUris = new ArrayList<String>(uriSet);
// Dereference each and every one of the URIs contained in the specified set
while (lstToDerefUris.size() > 0) {
// Remove the URI at the head of the queue of URIs to be dereferenced
String headUri = lstToDerefUris.remove(0);
// First, search for the URI in the cache
CachedHTTPResource httpResource =
(CachedHTTPResource)
dcmgr.getFromCache(DiachronCacheManager.HTTP_RESOURCE_CACHE, headUri);
if (httpResource == null || httpResource.getStatusLines() == null) {
// URIs not found in the cache, is still to be fetched via HTTP, add it to the end of the
// list
lstToDerefUris.add(headUri);
} else {
// URI found in the cache (which means that was fetched at some point), check if
// successfully dereferenced
DerefResult curUrlResult = new DerefResult(headUri, false, false);
lstDerefUris.add(curUrlResult);
if (this.isDereferenceable(httpResource)) {
curUrlResult.isDeref = true;
if (this.is200AnRDF(httpResource)) {
curUrlResult.isRdfXml = true;
} else this.createProblemQuad(httpResource.getUri(), DQM.NotMeaningful);
} else if (httpResource.getDereferencabilityStatusCode() == StatusCode.SC200) {
curUrlResult.isDeref = true;
// Check if the resource contains RDF on XML
if (this.is200AnRDF(httpResource)) {
curUrlResult.isRdfXml = true;
}
}
logger.trace(
"Resource fetched: {}. Deref. status: {}. Is RDF: {}",
headUri,
httpResource.getDereferencabilityStatusCode(),
curUrlResult.isRdfXml);
}
}
return lstDerefUris;
}
private boolean isDereferenceable(CachedHTTPResource httpResource) {
if (httpResource.getDereferencabilityStatusCode() == null) {
List<Integer> statusCode = this.getStatusCodes(httpResource.getStatusLines());
if (httpResource.getUri().contains("#") && statusCode.contains(200))
httpResource.setDereferencabilityStatusCode(StatusCode.HASH);
else if (statusCode.contains(200)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC200);
if (statusCode.contains(303)) httpResource.setDereferencabilityStatusCode(StatusCode.SC303);
else {
if (statusCode.contains(301)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC301);
this.createProblemQuad(httpResource.getUri(), DQM.SC301MovedPermanently);
} else if (statusCode.contains(302)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC302);
this.createProblemQuad(httpResource.getUri(), DQM.SC302Found);
} else if (statusCode.contains(307)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC307);
this.createProblemQuad(httpResource.getUri(), DQM.SC307TemporaryRedirectory);
} else {
if (hasBad3xxCode(statusCode))
this.createProblemQuad(httpResource.getUri(), DQM.SC3XXRedirection);
}
}
}
if (has4xxCode(statusCode)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC4XX);
this.createProblemQuad(httpResource.getUri(), DQM.SC4XXClientError);
}
if (has5xxCode(statusCode)) {
httpResource.setDereferencabilityStatusCode(StatusCode.SC5XX);
this.createProblemQuad(httpResource.getUri(), DQM.SC5XXServerError);
}
}
StatusCode scode = httpResource.getDereferencabilityStatusCode();
return this.mapDerefStatusCode(scode);
}