Esempio n. 1
0
 private static void DeleteCsvAndMetadata(string baseDirectory, string datasetId, IProgressLog progressLog)
 {
     Log.Information("DeleteCsvAndMetadata: {baseDirectory}, {datasetId}", baseDirectory, datasetId);
     try
     {
         progressLog.Info("Deleting source CSV and CSV metadata files");
         var csvPath = Path.Combine(baseDirectory, "csv", datasetId);
         Directory.Delete(csvPath, true);
     }
     catch (Exception ex)
     {
         progressLog.Exception(ex, "Error deleting source CSV and CSV metadata files");
         throw;
     }
 }
Esempio n. 2
0
        public bool HandleTripleCollection(IList <Triple> tripleCollection)
        {
            var subject = (tripleCollection[0].Subject as IUriNode)?.Uri;

            try
            {
                if (_resourceMap.CanMap(subject))
                {
                    if (_noFilter || tripleCollection.Any(t => _graphFilter.Contains(t.GraphUri)))
                    {
                        // There is some data in the graph that was updated, so we should regenerate the files for this resource
                        var targetPath = _resourceMap.GetPathFor(subject);
                        var targetDir  = Path.GetDirectoryName(targetPath);
                        if (targetDir != null && !Directory.Exists(targetDir))
                        {
                            Directory.CreateDirectory(targetDir);
                        }
                        var graph = new Graph();
                        graph.Assert(tripleCollection);

#if GENERATE_RDFXML
                        GenerateRdfXmlFile(subject, targetPath, graph);
#endif

                        GenerateNQuadsFile(subject, targetPath, graph);
                    }
                }
                _subjectCount++;
                if (_subjectCount % _reportInterval == 0)
                {
                    _progressLog.Info("Generating static files - {0} subjects processed.", _subjectCount);
                }
            }
            catch (Exception ex)
            {
                _progressLog.Exception(ex, "Error generating static files for subject {0}: {1}", subject, ex.Message);
            }
            return(true);
        }
Esempio n. 3
0
        public bool HandleResource(INode resourceNode, IList <Triple> subjectStatements, IList <Triple> objectStatements)
        {
            if (!(resourceNode is IUriNode))
            {
                return(true);
            }
            if ((subjectStatements == null || subjectStatements.Count == 0) && (objectStatements == null || objectStatements.Count == 0))
            {
                return(true);
            }
            var subject = (resourceNode as IUriNode)?.Uri;
            var nquads  = subject == null ? null : _uriService.GetSubjectDataUrl(subject.ToString(), "nq");

            try
            {
                var targetPath = _resourceMap.GetPathFor(subject);
                if (targetPath != null)
                {
                    targetPath += ".html";
                    var targetDir = Path.GetDirectoryName(targetPath);
                    if (!Directory.Exists(targetDir))
                    {
                        Directory.CreateDirectory(targetDir);
                    }

                    var parentDataset = string.Format("{0}://{1}", subject.Scheme, subject.Authority);
                    if (subject.ToString().IndexOf("id/resource", StringComparison.InvariantCultureIgnoreCase) > 0)
                    {
                        var datasetSegments = subject.Segments.Take(subject.Segments.Length - 1).ToArray();
                        foreach (var segment in datasetSegments)
                        {
                            if (segment.Equals("resource/"))
                            {
                                parentDataset += "dataset/";
                            }
                            else
                            {
                                parentDataset += segment;
                            }
                        }
                        parentDataset = parentDataset.Trim("/".ToCharArray());
                    }
                    UpdateVariables(nquads, parentDataset);

                    var html = _viewEngine.Render(subject, subjectStatements, objectStatements, _addVariables);
                    using (var stream = File.Open(targetPath, FileMode.Create, FileAccess.ReadWrite))
                    {
                        using (var writer = new StreamWriter(stream, Encoding.UTF8))
                        {
                            writer.Write(html);
                        }
                        stream.Close();
                    }
                    _numFilesGenerated++;
                    if (_numFilesGenerated % _reportInterval == 0)
                    {
                        _progressLog.Info("Generating static HTML files - {0} files created/updated.", _numFilesGenerated);
                    }
                }
            }
            catch (Exception ex)
            {
                _progressLog.Exception(ex, "Error generating HTML file for subject {0}: {1}", subject, ex.Message);
            }
            return(true);
        }
Esempio n. 4
0
        /// <summary>
        /// Adds content to a dataset graph in the DataDock repository, optionally overwriting existing dataset content
        /// </summary>
        /// <param name="insertTriples"></param>
        /// <param name="datasetIri"></param>
        /// <param name="dropExistingGraph"></param>
        /// <param name="metadataGraph"></param>
        /// <param name="metadataGraphIri"></param>
        /// <param name="definitionsGraph"></param>
        /// <param name="definitionsGraphIri"></param>
        /// <param name="publisherIri"></param>
        /// <param name="publisherInfo"></param>
        /// <param name="repositoryTitle"></param>
        /// <param name="repositoryDescription"></param>
        /// <param name="rootMetadataGraphIri"></param>
        public void UpdateDataset(
            IGraph insertTriples, Uri datasetIri, bool dropExistingGraph,
            IGraph metadataGraph, Uri metadataGraphIri,
            IGraph definitionsGraph, Uri definitionsGraphIri,
            Uri publisherIri, ContactInfo publisherInfo,
            string repositoryTitle, string repositoryDescription,
            Uri rootMetadataGraphIri)
        {
            try
            {
                _progressLog.Info("Updating RDF repository");

                if (dropExistingGraph)
                {
                    // Drop dataset data graph
                    _progressLog.Info("Dropping all existing RDF data in graph {0}", datasetIri);
                    _quinceStore.DropGraph(datasetIri);
                }

                // Drop dataset metadata graph
                _progressLog.Info("Dropping all metadata in graph {0}", metadataGraphIri);
                _quinceStore.DropGraph(metadataGraphIri);

                // Add triples to dataset data graph
                _progressLog.Info("Adding new RDF data to graph {0} ({1} triples)", datasetIri, insertTriples.Triples.Count);
                var addCount = 0;
                foreach (var t in insertTriples.Triples)
                {
                    _quinceStore.Assert(t.Subject, t.Predicate, t.Object, datasetIri);
                    addCount++;
                    if (addCount % 1000 == 0)
                    {
                        _progressLog.Info("Added {0} / {1} triples ({2}%)", addCount, insertTriples.Triples.Count,
                                          addCount * 100 / insertTriples.Triples.Count);
                    }
                }

                // Add triples to dataset metadata graph
                _progressLog.Info("Adding dataset metadata to graph {0} ({1} triples)", metadataGraphIri, metadataGraph.Triples.Count);
                foreach (var t in metadataGraph.Triples)
                {
                    _quinceStore.Assert(t.Subject, t.Predicate, t.Object, metadataGraphIri);
                }

                // Add triples to the definitions graph
                _progressLog.Info("Adding column definition metadata to graph {0} ({1} triples)", definitionsGraphIri, definitionsGraph.Triples.Count);
                foreach (var t in definitionsGraph.Triples)
                {
                    _quinceStore.Assert(t.Subject, t.Predicate, t.Object, definitionsGraphIri);
                }

                // Update the root metadata graph to ensure it includes this dataset as a subset
                _progressLog.Info("Updating root metadata");
                var repositoryNode     = insertTriples.CreateUriNode(_repositoryUri);
                var subset             = insertTriples.CreateUriNode(new Uri("http://rdfs.org/ns/void#subset"));
                var rdfType            = insertTriples.CreateUriNode(new Uri("http://www.w3.org/1999/02/22-rdf-syntax-ns#type"));
                var rdfsLabel          = insertTriples.CreateUriNode(new Uri("http://www.w3.org/2000/01/rdf-schema#label"));
                var foafName           = insertTriples.CreateUriNode(new Uri("http://xmlns.com/foaf/0.1/name"));
                var foafMbox           = insertTriples.CreateUriNode(new Uri("http://xmlns.com/foaf/0.1/mbox"));
                var foafHomepage       = insertTriples.CreateUriNode(new Uri("http://xmlns.com/foaf/0.1/homepage"));
                var dctermsPublisher   = insertTriples.CreateUriNode(new Uri("http://purl.org/dc/terms/publisher"));
                var dctermsTitle       = insertTriples.CreateUriNode(new Uri("http://purl.org/dc/terms/title"));
                var dctermsDescription = insertTriples.CreateUriNode(new Uri("http://purl.org/dc/terms/description"));
                var publisherNode      = insertTriples.CreateUriNode(publisherIri);
                _quinceStore.Assert(repositoryNode, rdfType, insertTriples.CreateUriNode(new Uri("http://rdfs.org/ns/void#Dataset")), rootMetadataGraphIri);
                _quinceStore.Assert(repositoryNode, subset, insertTriples.CreateUriNode(datasetIri), rootMetadataGraphIri);
                _quinceStore.Assert(repositoryNode, dctermsPublisher, publisherNode, rootMetadataGraphIri);


                // Update repository title and description
                foreach (var t in _quinceStore.GetTriplesForSubject(_repositoryUri))
                {
                    if (t.GraphUri.Equals(rootMetadataGraphIri))
                    {
                        if (t.Predicate.Equals(dctermsTitle) || t.Predicate.Equals(dctermsDescription))
                        {
                            _quinceStore.Retract(t.Subject, t.Predicate, t.Object, rootMetadataGraphIri);
                        }
                    }
                }
                if (!string.IsNullOrEmpty(repositoryTitle))
                {
                    _quinceStore.Assert(repositoryNode, dctermsTitle, insertTriples.CreateLiteralNode(repositoryTitle), rootMetadataGraphIri);
                }
                if (!string.IsNullOrEmpty(repositoryDescription))
                {
                    _quinceStore.Assert(repositoryNode, dctermsDescription, insertTriples.CreateLiteralNode(repositoryDescription), rootMetadataGraphIri);
                }

                // Update publisher information
                foreach (var t in _quinceStore.GetTriplesForSubject(publisherIri))
                {
                    if (t.GraphUri.Equals(rootMetadataGraphIri))
                    {
                        _quinceStore.Retract(t.Subject, t.Predicate, t.Object, rootMetadataGraphIri);
                    }
                }
                if (!string.IsNullOrEmpty(publisherInfo?.Type))
                {
                    var publisherType = insertTriples.CreateUriNode(new Uri(publisherInfo.Type));
                    _quinceStore.Assert(publisherNode, rdfType, publisherType, rootMetadataGraphIri);
                }
                if (!string.IsNullOrEmpty(publisherInfo?.Label))
                {
                    var publisherLabel = insertTriples.CreateLiteralNode(publisherInfo.Label);
                    _quinceStore.Assert(publisherNode, rdfsLabel, publisherLabel, rootMetadataGraphIri);
                    _quinceStore.Assert(publisherNode, foafName, publisherLabel, rootMetadataGraphIri);
                }
                if (!string.IsNullOrEmpty(publisherInfo?.Email))
                {
                    _quinceStore.Assert(publisherNode, foafMbox, insertTriples.CreateLiteralNode(publisherInfo.Email), rootMetadataGraphIri);
                }
                if (!string.IsNullOrEmpty(publisherInfo?.Website))
                {
                    _quinceStore.Assert(publisherNode, foafHomepage, insertTriples.CreateUriNode(new Uri(publisherInfo.Website)), rootMetadataGraphIri);
                }

                // Flush all data to disk
                _quinceStore.Flush();
            }
            catch (Exception ex)
            {
                Log.Error(ex, "CSV conversion failed");
                throw new WorkerException(ex, "CSV conversion failed. Error updating RDF repository.");
            }
        }