/// <summary>
/// Crea un LiteralNode según el tipo de dato del contenido.
/// </summary>
/// <param name="pDataGraph">Grafo.</param>
/// <param name="pContenido">Contenido del nodo.</param>
/// <param name="pDatatype">Tipo del dato del contenido.</param>
/// <param name="pTransform">Transformación del contenido</param>
/// <returns>Nodo construido según el tipo de dato.</returns>
private ILiteralNode CreateILiteralNodeType(RohGraph pDataGraph, string pContenido, string pDatatype, string pTransform = null)
{
pContenido = pContenido.Trim();
if (!string.IsNullOrEmpty(pTransform))
{
if (pTransform.Contains("{value}"))
{
pContenido = pTransform.Replace("{value}", pContenido);
}
if (pTransform.Contains("{regex|"))
{
string regString = pTransform.Substring(pTransform.IndexOf("{regex|") + 7);
regString = regString.Substring(0, regString.IndexOf("|endregex}"));
Regex regex = new Regex(regString);
Match match = regex.Match(pContenido);
pContenido = pTransform.Replace("{regex|" + regString + "|endregex}", match.Value);
}
}
if (string.IsNullOrEmpty(pDatatype))
{
return(pDataGraph.CreateLiteralNode(pContenido, new Uri("http://www.w3.org/2001/XMLSchema#string")));
}
else
{
return(pDataGraph.CreateLiteralNode(pContenido, new Uri(pDatatype)));
}
}
/// <summary>
/// Publica un RDF en Asio aplicado todos losprocedimientos pertinentes
/// </summary>
/// <param name="pDataGraph">Grafo con los datos a cargar</param>
/// <param name="pOntologyGraph">Grafo con la ontología</param>
/// <param name="pAttributedTo">Sujeto y nombre para atribuir los triples de los apis externos</param>
/// <param name="pActivityStartedAtTime">Inicio del proceso</param>
/// <param name="pActivityEndedAtTime">Fin del proceso</param>
/// <param name="pDiscoverLinkData">Datos para trabajar con el descubrimiento de enlaces</param>
/// <param name="pCallUrisFactoryApiService">Servicio para hacer llamadas a los métodos del Uris Factory</param>
public void PublishRDF(RohGraph pDataGraph, RohGraph pOntologyGraph, KeyValuePair <string, string>?pAttributedTo, DateTime pActivityStartedAtTime, DateTime pActivityEndedAtTime, DiscoverLinkData pDiscoverLinkData, CallUrisFactoryApiService pCallUrisFactoryApiService)
{
RohGraph inferenceDataGraph = null;
if (pOntologyGraph != null)
{
inferenceDataGraph = pDataGraph.Clone();
RohRdfsReasoner reasoner = new RohRdfsReasoner();
reasoner.Initialise(pOntologyGraph);
reasoner.Apply(inferenceDataGraph);
}
// 1º Eliminamos de la BBDD las entidades principales que aparecen en el RDF
HashSet <string> graphs = RemovePrimaryTopics(ref pDataGraph);
graphs.Add(_Graph);
// 2º Eliminamos todos los triples de la BBDD cuyo sujeto y predicado estén en el RDF a cargar y estén marcados como monovaluados.
if (pOntologyGraph != null && inferenceDataGraph != null)
{
RemoveMonovaluatedProperties(pOntologyGraph, inferenceDataGraph);
}
//3º Insertamos los triples en la BBDD
if (pAttributedTo.HasValue)
{
//Añadimos triples del softwareagent
IUriNode t_subject = pDataGraph.CreateUriNode(UriFactory.Create(pAttributedTo.Value.Key));
IUriNode t_predicate_rdftype = pDataGraph.CreateUriNode(UriFactory.Create("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"));
IUriNode t_object_rdftype = pDataGraph.CreateUriNode(UriFactory.Create("http://www.w3.org/ns/prov#SoftwareAgent"));
pDataGraph.Assert(new Triple(t_subject, t_predicate_rdftype, t_object_rdftype));
IUriNode t_predicate_name = pDataGraph.CreateUriNode(UriFactory.Create("http://purl.org/roh/mirror/foaf#name"));
ILiteralNode t_object_name = pDataGraph.CreateLiteralNode(pAttributedTo.Value.Value, new Uri("http://www.w3.org/2001/XMLSchema#string"));
pDataGraph.Assert(new Triple(t_subject, t_predicate_name, t_object_name));
}
SparqlUtility.LoadTriples(SparqlUtility.GetTriplesFromGraph(pDataGraph), _SPARQLEndpoint, _QueryParam, _Graph, _Username, _Password);
//4º Insertamos los triples con provenance en la BBDD
if (pDiscoverLinkData != null && pDiscoverLinkData.entitiesProperties != null)
{
Dictionary <string, List <string> > graphDeletes = new Dictionary <string, List <string> >();
Dictionary <string, List <string> > graphTriples = new Dictionary <string, List <string> >();
foreach (string t_subject in pDiscoverLinkData.entitiesProperties.Keys)
{
foreach (DiscoverLinkData.PropertyData property in pDiscoverLinkData.entitiesProperties[t_subject])
{
string t_property = property.property;
foreach (var prop in property.valueProvenance)
{
string t_object = prop.Key;
HashSet <string> t_sourceids = prop.Value;
foreach (string sourceId in t_sourceids)
{
string graph = pCallUrisFactoryApiService.GetUri("Graph", sourceId);
if (!graphTriples.ContainsKey(graph))
{
graphTriples.Add(graph, new List <string>());
}
string bNodeid = "_:" + Guid.NewGuid().ToString();
graphTriples[graph].Add($@"<{t_subject}> <http://www.w3.org/ns/prov#wasUsedBy> {bNodeid} .");
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://www.w3.org/ns/prov#Activity> .");
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate> <{t_property}>.");
if (Uri.IsWellFormedUriString(t_object, UriKind.Absolute))
{
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/1999/02/22-rdf-syntax-ns#object> <{ t_object}>.");
}
else
{
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/1999/02/22-rdf-syntax-ns#object> ""{ t_object.Replace("\"", "\\\"").Replace("\n", "\\n") }""^^<http://www.w3.org/2001/XMLSchema#string>.");
}
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/ns/prov#startedAtTime> ""{ pActivityStartedAtTime.ToString("yyyy-MM-ddTHH:mm:ss.fffzzz") }""^^<http://www.w3.org/2001/XMLSchema#datetime>.");
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/ns/prov#endedAtTime> ""{ pActivityEndedAtTime.ToString("yyyy-MM-ddTHH:mm:ss.fffzzz") }""^^<http://www.w3.org/2001/XMLSchema#datetime>.");
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/ns/prov#wasAssociatedWith> <{pAttributedTo.Value.Key}>.");
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/ns/prov#wasAssociatedWith> <{pCallUrisFactoryApiService.GetUri("http://purl.org/roh/mirror/foaf#Organization", sourceId)}>.");
if (pAttributedTo.HasValue)
{
graphTriples[graph].Add($@"{bNodeid} <http://www.w3.org/ns/prov#wasAssociatedWith> <{pAttributedTo.Value.Key}>.");
}
if (!graphDeletes.ContainsKey(graph))
{
graphDeletes.Add(graph, new List <string>());
}
if (!Uri.IsWellFormedUriString(t_object, UriKind.Absolute))
{
string stringDelete = $@" {{
?s ?p ?o.
?o <http://www.w3.org/1999/02/22-rdf-syntax-ns#predicate> <{t_property}>.
?o <http://www.w3.org/1999/02/22-rdf-syntax-ns#object> ""{ t_object.Replace("\"", "\\\"").Replace("\n", "\\n") }""^^<http://www.w3.org/2001/XMLSchema#string>.
FILTER(?s = <{t_subject}>)
}}";
graphDeletes[graph].Add(stringDelete);
}
}
}
}
}
//Eliminamos aquellos triples de provenance que ya estén cargados
foreach (string graph in graphDeletes.Keys)
{
graphs.Add(graph);
string queryDeleteProvenance = $@" DELETE {{ ?s ?p ?o. }}
WHERE
{{
{{{string.Join("}UNION{", graphDeletes[graph])}}}
}}";
_SparqlUtility.SelectData(_SPARQLEndpoint, graph, queryDeleteProvenance, _QueryParam, _Username, _Password);
}
//Cargamos los nuevos triples
foreach (string graph in graphTriples.Keys)
{
SparqlUtility.LoadTriples(graphTriples[graph], _SPARQLEndpoint, _QueryParam, graph, _Username, _Password);
}
}
//5º Limpiamos los blanknodes huerfanos, o que no tengan triples
//TODO mover a una tarea que se ejecute continuamente
//DeleteOrphanNodes(graphs);
}
/// <summary>
/// Aplica el descubrimiento sobre las entidades cargadas en el SGI
/// </summary>
/// <param name="pSecondsSleep">Segundos para dormir después de procesar una entidad</param>
/// <param name="pCallUrisFactoryApiService">Servicio para hacer llamadas a los métodos del Uris Factory</param>
public void ApplyDiscoverLoadedEntities(int pSecondsSleep, CallUrisFactoryApiService pCallUrisFactoryApiService)
{
CallEtlApiService callEtlApiService = _serviceScopeFactory.CreateScope().ServiceProvider.GetRequiredService <CallEtlApiService>();
#region Cargamos configuraciones
ConfigSparql ConfigSparql = new ConfigSparql();
string SGI_SPARQLEndpoint = ConfigSparql.GetEndpoint();
string SGI_SPARQLGraph = ConfigSparql.GetGraph();
string SGI_SPARQLQueryParam = ConfigSparql.GetQueryParam();
string SGI_SPARQLUsername = ConfigSparql.GetUsername();
string SGI_SPARQLPassword = ConfigSparql.GetPassword();
string Unidata_SPARQLEndpoint = ConfigSparql.GetUnidataEndpoint();
string Unidata_SPARQLGraph = ConfigSparql.GetUnidataGraph();
string Unidata_SPARQLQueryParam = ConfigSparql.GetUnidataQueryParam();
string Unidata_SPARQLUsername = ConfigSparql.GetUnidataUsername();
string Unidata_SPARQLPassword = ConfigSparql.GetUnidataPassword();
ConfigService ConfigService = new ConfigService();
string UnidataDomain = ConfigService.GetUnidataDomain();
string UnidataUriTransform = ConfigService.GetUnidataUriTransform();
float MaxScore = ConfigService.GetMaxScore();
float MinScore = ConfigService.GetMinScore();
ConfigScopus ConfigScopus = new ConfigScopus();
string ScopusApiKey = ConfigScopus.GetScopusApiKey();
string ScopusUrl = ConfigScopus.GetScopusUrl();
ConfigCrossref ConfigCrossref = new ConfigCrossref();
string CrossrefUserAgent = ConfigCrossref.GetCrossrefUserAgent();
ConfigWOS ConfigWOS = new ConfigWOS();
string WOSAuthorization = ConfigWOS.GetWOSAuthorization();
#endregion
DiscoverUtility discoverUtility = new DiscoverUtility();
//Cargar todas las personas en la lista de manera aleatoria.
List <string> personList = discoverUtility.GetPersonList(SGI_SPARQLEndpoint, SGI_SPARQLGraph, SGI_SPARQLQueryParam, SGI_SPARQLUsername, SGI_SPARQLPassword);
List <string> randomPersonList = GetRandomOrderList(personList);
RohGraph ontologyGraph = callEtlApiService.CallGetOntology();
foreach (string person in randomPersonList)
{
try
{
//Hora de inicio de la ejecución
DateTime startTime = DateTime.Now;
//Obtener el RohGraph de una única persona.
RohGraph dataGraph = discoverUtility.GetDataGraphPersonLoadedForDiscover(person, SGI_SPARQLEndpoint, SGI_SPARQLGraph, SGI_SPARQLQueryParam, SGI_SPARQLUsername, SGI_SPARQLPassword);
//Clonamos el grafo original para hacer luego comprobaciones
RohGraph originalDataGraph = dataGraph.Clone();
RohRdfsReasoner reasoner = new RohRdfsReasoner();
reasoner.Initialise(ontologyGraph);
RohGraph dataInferenceGraph = dataGraph.Clone();
reasoner.Apply(dataInferenceGraph);
bool hasChanges = false;
//Dictionary<string, string> discoveredEntityList = new Dictionary<string, string>();
Dictionary <string, Dictionary <string, float> > discoveredEntitiesProbability = new Dictionary <string, Dictionary <string, float> >();
Dictionary <string, ReconciliationData.ReconciliationScore> entidadesReconciliadasConIntegracionExternaAux;
Dictionary <string, HashSet <string> > discardDissambiguations = new Dictionary <string, HashSet <string> >();
DiscoverCache discoverCache = new DiscoverCache();
DiscoverCacheGlobal discoverCacheGlobal = new DiscoverCacheGlobal();
//Obtención de la integración externa
ReconciliationData reconciliationData = new ReconciliationData();
DiscoverLinkData discoverLinkData = new DiscoverLinkData();
Dictionary <string, List <DiscoverLinkData.PropertyData> > integration = discoverUtility.ExternalIntegration(ref hasChanges, ref reconciliationData, ref discoverLinkData, ref discoveredEntitiesProbability, ref dataGraph, reasoner, null, ontologyGraph, out entidadesReconciliadasConIntegracionExternaAux, discardDissambiguations, discoverCache, discoverCacheGlobal, ScopusApiKey, ScopusUrl, CrossrefUserAgent, WOSAuthorization, MinScore, MaxScore, SGI_SPARQLEndpoint, SGI_SPARQLGraph, SGI_SPARQLQueryParam, SGI_SPARQLUsername, SGI_SPARQLPassword, pCallUrisFactoryApiService, false);
//Limpiamos 'integration' para no insertar triples en caso de que ya estén cargados
foreach (string entity in integration.Keys.ToList())
{
foreach (DiscoverLinkData.PropertyData propertyData in integration[entity].ToList())
{
string p = propertyData.property;
HashSet <string> objetos = new HashSet <string>(propertyData.valueProvenance.Keys.ToList());
foreach (string o in objetos)
{
if (((SparqlResultSet)originalDataGraph.ExecuteQuery($@"ASK
WHERE
{{
?s ?p ?o.
FILTER(?s=<{entity}>)
FILTER(?p=<{p}>)
FILTER(str(?o)='{o}')
}}")).Result)
{
//Elimiamos el valor porque ya estaba cargado
propertyData.valueProvenance.Remove(o);
}
}
if (propertyData.valueProvenance.Count == 0)
{
integration[entity].Remove(propertyData);
}
}
if (integration[entity].Count == 0)
{
integration.Remove(entity);
}
}
//Creación de dataGraph con el contenido de 'integration' + RdfTypes + SameAS
RohGraph dataGraphIntegration = new RohGraph();
foreach (string sujeto in integration.Keys)
{
IUriNode s = dataGraphIntegration.CreateUriNode(UriFactory.Create(sujeto));
//Agregamos SameAs y RDFType de las entidades
SparqlResultSet sparqlResultSet = (SparqlResultSet)dataGraph.ExecuteQuery("select ?rdftype ?sameas where {?s a ?rdftype. OPTIONAL{?s <http://www.w3.org/2002/07/owl#sameAs> ?sameAS} FILTER(?s=<" + sujeto + ">)}");
foreach (SparqlResult sparqlResult in sparqlResultSet.Results)
{
string rdfType = sparqlResult["rdftype"].ToString();
IUriNode pRdfType = dataGraphIntegration.CreateUriNode(UriFactory.Create("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"));
IUriNode oRdfType = dataGraphIntegration.CreateUriNode(UriFactory.Create(rdfType));
dataGraphIntegration.Assert(new Triple(s, pRdfType, oRdfType));
if (sparqlResult.Variables.Contains("sameas"))
{
string sameas = sparqlResult["sameas"].ToString();
IUriNode pSameAs = dataGraphIntegration.CreateUriNode(UriFactory.Create("http://www.w3.org/2002/07/owl#sameAs"));
IUriNode oSameAs = dataGraphIntegration.CreateUriNode(UriFactory.Create(sameas));
dataGraphIntegration.Assert(new Triple(s, pSameAs, oSameAs));
}
}
foreach (DiscoverLinkData.PropertyData propertyData in integration[sujeto])
{
foreach (string valor in propertyData.valueProvenance.Keys)
{
IUriNode p = dataGraphIntegration.CreateUriNode(UriFactory.Create(propertyData.property));
if (Uri.IsWellFormedUriString(valor, UriKind.Absolute))
{
IUriNode uriNode = dataGraphIntegration.CreateUriNode(UriFactory.Create(propertyData.property));
dataGraphIntegration.Assert(new Triple(s, p, uriNode));
}
else
{
ILiteralNode literalNode = dataGraphIntegration.CreateLiteralNode(valor, new Uri("http://www.w3.org/2001/XMLSchema#string"));
dataGraphIntegration.Assert(new Triple(s, p, literalNode));
}
foreach (string org in propertyData.valueProvenance[valor])
{
//Agregamos los datos de las organizaciones y los grafos
SparqlResultSet sparqlResultSetOrgs = (SparqlResultSet)dataGraph.ExecuteQuery("select ?s ?p ?o where {?s ?p ?o. FILTER(?s in(<" + pCallUrisFactoryApiService.GetUri("http://purl.org/roh/mirror/foaf#Organization", org) + ">,<" + pCallUrisFactoryApiService.GetUri("Graph", org) + "> ))}");
foreach (SparqlResult sparqlResult in sparqlResultSetOrgs.Results)
{
INode sOrg = dataGraphIntegration.CreateUriNode(UriFactory.Create(sparqlResult["s"].ToString()));
INode pOrg = dataGraphIntegration.CreateUriNode(UriFactory.Create(sparqlResult["p"].ToString()));
if (sparqlResult["o"] is UriNode)
{
INode oOrg = dataGraphIntegration.CreateUriNode(UriFactory.Create(sparqlResult["o"].ToString()));
dataGraphIntegration.Assert(new Triple(sOrg, pOrg, oOrg));
}
else if (sparqlResult["o"] is LiteralNode)
{
INode oOrg = dataGraphIntegration.CreateLiteralNode(((LiteralNode)sparqlResult["o"]).Value, ((LiteralNode)sparqlResult["o"]).DataType);
dataGraphIntegration.Assert(new Triple(sOrg, pOrg, oOrg));
}
}
}
}
}
}
//Hora fin de la ejecución
DateTime endTime = DateTime.Now;
if (integration.Count > 0)
{
//Si hay datos nuevos los cargamos
string urlDiscoverAgent = pCallUrisFactoryApiService.GetUri("Agent", "discover");
//Publicamos en el SGI
AsioPublication asioPublication = new AsioPublication(SGI_SPARQLEndpoint, SGI_SPARQLQueryParam, SGI_SPARQLGraph, SGI_SPARQLUsername, SGI_SPARQLPassword);
asioPublication.PublishRDF(dataGraphIntegration, null, new KeyValuePair <string, string>(urlDiscoverAgent, "Algoritmos de descubrimiento"), startTime, endTime, discoverLinkData, pCallUrisFactoryApiService);
//Preparamos los datos para cargarlos en Unidata
RohGraph unidataGraph = dataGraphIntegration.Clone();
#region Si no tiene un sameAs apuntando a Unidata lo eliminamos, no hay que cargar la entidad
SparqlResultSet sparqlResultSet = (SparqlResultSet)unidataGraph.ExecuteQuery("select ?s ?rdftype ?sameas where {?s a ?rdftype. OPTIONAL{?s <http://www.w3.org/2002/07/owl#sameAs> ?sameAS} }");
Dictionary <string, bool> entidadesConSameAsUnidata = new Dictionary <string, bool>();
foreach (SparqlResult sparqlResult in sparqlResultSet.Results)
{
string s = sparqlResult["s"].ToString();
if (!entidadesConSameAsUnidata.ContainsKey(s))
{
entidadesConSameAsUnidata.Add(s, false);
}
if (sparqlResult.Variables.Contains("sameas"))
{
if (sparqlResult["sameas"].ToString().StartsWith(UnidataDomain))
{
entidadesConSameAsUnidata[s] = true;
}
}
}
TripleStore store = new TripleStore();
store.Add(unidataGraph);
foreach (string entity in entidadesConSameAsUnidata.Keys)
{
if (!entidadesConSameAsUnidata[entity])
{
//Cambiamos candidato.Key por entityID
SparqlUpdateParser parser = new SparqlUpdateParser();
SparqlUpdateCommandSet delete = parser.ParseFromString(@"DELETE { ?s ?p ?o. }
WHERE
{
?s ?p ?o.
FILTER(?s = <" + entity + @">)
}");
LeviathanUpdateProcessor processor = new LeviathanUpdateProcessor(store);
processor.ProcessCommandSet(delete);
}
}
#endregion
//TODO descomentar cuando esté habilitaado Unidata
////Si hay triples para cargar en Unidata procedemos
//if (unidataGraph.Triples.ToList().Count > 0)
//{
// //Publicamos en UNIDATA
// AsioPublication asioPublicationUnidata = new AsioPublication(Unidata_SPARQLEndpoint, Unidata_SPARQLQueryParam, Unidata_SPARQLGraph, Unidata_SPARQLUsername, Unidata_SPARQLPassword);
// // Prepara el grafo para su carga en Unidata, para ello coge las URIs de Unidata del SameAs y la aplica a los sujetos y los antiguos sujetos se agregan al SameAs
// unidataGraph = AsioPublication.TransformUrisToUnidata(unidataGraph, UnidataDomain, UnidataUriTransform);
// asioPublicationUnidata.PublishRDF(unidataGraph, null, new KeyValuePair<string, string>(urlDiscoverAgent, "Algoritmos de descubrimiento"), startTime, endTime, discoverLinkData,pCallUrisFactoryApiService);
//}
}
}
catch (Exception exception)
{
Logging.Error(exception);
}
Thread.Sleep(pSecondsSleep * 1000);
}
}