/**
* Get JSON map for a given resource by applying the river settings
*
* @param rs resource being processed
* @param properties properties to be indexed
* @param model model returned by the indexing query
* @param getPropLabel if set to true all URI property values will be indexed as their label. The
* label is taken as the value of one of the properties set in {@link #uriDescriptionList}.
* @return map of properties to be indexed for res
*/
private Map<String, ArrayList<String>> getJsonMap(
Resource rs, Set<Property> properties, Model model, boolean getPropLabel) {
Map<String, ArrayList<String>> jsonMap = new HashMap<String, ArrayList<String>>();
ArrayList<String> results = new ArrayList<String>();
if (addUriForResource) {
results.add("\"" + rs.toString() + "\"");
jsonMap.put("http://www.w3.org/1999/02/22-rdf-syntax-ns#about", results);
}
Set<String> rdfLanguages = new HashSet<String>();
for (Property prop : properties) {
NodeIterator niter = model.listObjectsOfProperty(rs, prop);
String property = prop.toString();
results = new ArrayList<String>();
String lang;
String currValue;
while (niter.hasNext()) {
RDFNode node = niter.next();
currValue = getStringForResult(node, getPropLabel);
if (addLanguage) {
if (node.isLiteral()) {
lang = node.asLiteral().getLanguage();
if (!lang.isEmpty()) {
rdfLanguages.add("\"" + lang + "\"");
}
}
}
String shortValue = currValue;
int currLen = currValue.length();
// Unquote string
if (currLen > 1) shortValue = currValue.substring(1, currLen - 1);
// If either whiteMap does contains shortValue
// or blackMap contains the value
// skip adding it to the index
boolean whiteMapCond =
whiteMap.containsKey(property) && !whiteMap.get(property).contains(shortValue);
boolean blackMapCond =
blackMap.containsKey(property) && blackMap.get(property).contains(shortValue);
if (whiteMapCond || blackMapCond) {
continue;
}
if (normalizeObj.containsKey(shortValue)) {
results.add("\"" + normalizeObj.get(shortValue) + "\"");
} else {
results.add(currValue);
}
}
// Do not index empty properties
if (results.isEmpty()) continue;
if (normalizeProp.containsKey(property)) {
property = normalizeProp.get(property);
if (jsonMap.containsKey(property)) {
jsonMap.get(property).addAll(results);
} else {
jsonMap.put(property, results);
}
} else {
jsonMap.put(property, results);
}
}
if (addLanguage) {
if (rdfLanguages.isEmpty() && !language.isEmpty()) rdfLanguages.add(language);
if (!rdfLanguages.isEmpty()) jsonMap.put("language", new ArrayList<String>(rdfLanguages));
}
for (Map.Entry<String, String> it : normalizeMissing.entrySet()) {
if (!jsonMap.containsKey(it.getKey())) {
ArrayList<String> res = new ArrayList<String>();
res.add("\"" + it.getValue() + "\"");
jsonMap.put(it.getKey(), res);
}
}
return jsonMap;
}
/**
* Starts a harvester with predefined queries to synchronize with the changes from the SPARQL
* endpoint
*/
public boolean sync() {
logger.info("Sync resources newer than {}", startTime);
String rdfQueryTemplate =
"PREFIX xsd:<http://www.w3.org/2001/XMLSchema#> "
+ "SELECT DISTINCT ?resource WHERE { "
+ " GRAPH ?graph { %s }"
+ " ?graph <%s> ?time . %s "
+ " FILTER (?time > xsd:dateTime(\"%s\")) }";
String queryStr =
String.format(
rdfQueryTemplate, syncConditions, syncTimeProp, graphSyncConditions, startTime);
Set<String> syncUris = executeSyncQuery(queryStr, "resource");
if (syncUris == null) {
logger.error("Errors occurred during sync procedure. Aborting!");
return false;
}
/**
* If desired, query for old data that has the sync conditions modified
*
* <p>This option is useful in the case in which the application indexes resources that match
* some conditions. In this case, if they are modified and no longer match the initial
* conditions, they will not be synchronized. When syncOldData is True, the modified resources
* that no longer match the conditions are deleted.
*/
int deleted = 0;
int count = 0;
if (this.syncOldData) {
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss");
queryStr =
String.format(
rdfQueryTemplate,
syncConditions,
syncTimeProp,
graphSyncConditions,
sdf.format(new Date(0)));
HashSet<String> allIndexURIs = executeSyncQuery(queryStr, "resource");
if (allIndexURIs == null) {
logger.error("Errors occurred during modified content sync query. Aborting!");
return false;
}
deleted = removeMissingUris(allIndexURIs);
}
/* Prepare a series of bulk uris to be described so we can make
* a smaller number of calls to the SPARQL endpoint. */
ArrayList<ArrayList<String>> bulks = new ArrayList<ArrayList<String>>();
ArrayList<String> currentBulk = new ArrayList<String>();
for (String uri : syncUris) {
currentBulk.add(uri);
if (currentBulk.size() == EEASettings.DEFAULT_BULK_SIZE) {
bulks.add(currentBulk);
currentBulk = new ArrayList<String>();
}
}
if (currentBulk.size() > 0) {
bulks.add(currentBulk);
}
/* Execute RDF queries for the resources in each bulk */
for (ArrayList<String> bulk : bulks) {
String syncQuery = getSyncQueryStr(bulk);
try {
Query query = QueryFactory.create(syncQuery);
QueryExecution qExec = QueryExecutionFactory.sparqlService(rdfEndpoint, query);
try {
Model constructModel = ModelFactory.createDefaultModel();
qExec.execConstruct(constructModel);
BulkRequestBuilder bulkRequest = client.prepareBulk();
/**
* When adding the model to ES do not use toDescribeURIs as the query already returned the
* correct labels.
*/
addModelToES(constructModel, bulkRequest, false);
count += bulk.size();
} catch (Exception e) {
logger.error("Error while querying for modified content. {}", e.getLocalizedMessage());
return false;
} finally {
qExec.close();
}
} catch (QueryParseException qpe) {
logger.warn(
"Could not parse Sync query. Please provide a relevant query. {}",
qpe.getLocalizedMessage());
return false;
}
}
logger.info(
"Finished synchronisation: Deleted {}, Updated {}/{}", deleted, count, syncUris.size());
return true;
}