protected override async Task RunJob(JobInfo job, UserAccount userAccount)
{
ProgressLog.Info("Starting import job processing for " + userAccount.UserId);
var targetDirectory = Path.Combine(_configuration.RepoBaseDir, job.JobId);
Log.Information("Using local directory {localDirPath}", targetDirectory);
// Clone the repository
await _git.CloneRepository(job.GitRepositoryUrl, targetDirectory, AuthenticationToken, userAccount);
// Retrieve CSV and CSVM files to src directory in the repository
await AddCsvFilesToRepository(targetDirectory,
job.DatasetId,
job.CsvFileName,
job.CsvFileId,
job.CsvmFileId);
var csvPath = Path.Combine(targetDirectory, "csv", job.DatasetId, job.CsvFileName);
var metaPath = Path.Combine(targetDirectory, "csv", job.DatasetId, job.CsvFileName + "-metadata.json");
// Parse the JSON metadata
JObject metadataJson;
using (var metadataReader = File.OpenText(metaPath))
{
var metadataString = metadataReader.ReadToEnd();
metadataJson = JObject.Parse(metadataString);
}
// Run the CSV to RDF conversion
var repositoryUri = new Uri(_dataDockUriService.GetRepositoryUri(job.OwnerId, job.RepositoryId));
var publisherIri = new Uri(_dataDockUriService.GetRepositoryPublisherIdentifier(job.OwnerId, job.RepositoryId));
var datasetUri = new Uri(job.DatasetIri);
var datasetMetadataGraphIri = new Uri(datasetUri + "/metadata");
var rootMetadataGraphIri = new Uri(_dataDockUriService.GetMetadataGraphIdentifier(job.OwnerId, job.RepositoryId));
var definitionsGraphIri = new Uri(_dataDockUriService.GetDefinitionsGraphIdentifier(job.OwnerId, job.RepositoryId));
var dateTag = DateTime.UtcNow.ToString("yyyyMMdd_HHmmss");
var releaseTag = MakeSafeTag(job.DatasetId + "_" + dateTag);
var publisher = await GetPublisherContactInfo(job.OwnerId, job.RepositoryId);
var ntriplesDownloadLink =
new Uri($"https://github.com/{job.OwnerId}/{job.RepositoryId}/releases/download/{releaseTag}/{releaseTag}.nt.gz");
var csvDownloadLink =
new Uri(repositoryUri + $"csv/{job.DatasetId}/{job.CsvFileName}");
using (var tmpReader = File.OpenText(csvPath))
{
var header = tmpReader.ReadLine();
Log.Information("CSV header: {CsvHeader}", header);
}
var metadataBaseUri = new Uri(datasetUri + "/csv/" + job.CsvFileName + "-metadata.json");
IGraph datasetGraph = await GenerateDatasetGraphAsync(csvPath, metadataJson, metadataBaseUri);
IGraph metadataGraph = GenerateMetadataGraph(datasetUri, publisherIri, metadataJson,
new[] { ntriplesDownloadLink, csvDownloadLink }, datasetGraph);
IGraph definitionsGraph = GenerateDefinitionsGraph(metadataJson);
var dataDataDockRepository = _dataDataDockRepositoryFactory.GetRepositoryForJob(job, ProgressLog);
dataDataDockRepository.UpdateDataset(
datasetGraph, datasetUri, job.OverwriteExistingData,
metadataGraph, datasetMetadataGraphIri,
definitionsGraph, definitionsGraphIri,
publisherIri, publisher,
"", "",
rootMetadataGraphIri);
await UpdateHtmlPagesAsync(dataDataDockRepository,
new[] { datasetUri, datasetMetadataGraphIri, rootMetadataGraphIri });
// Add and Commit all changes
if (await _git.CommitChanges(targetDirectory,
$"Added {job.CsvFileName} to dataset {job.DatasetIri}", userAccount))
{
await _git.PushChanges(job.GitRepositoryUrl, targetDirectory, AuthenticationToken);
await _git.MakeRelease(datasetGraph, releaseTag, job.OwnerId, job.RepositoryId, job.DatasetId, targetDirectory, AuthenticationToken);
}
// Update the dataset repository
try
{
var voidMetadataJson = ExtractVoidMetadata(metadataGraph);
var datasetInfo = new DatasetInfo
{
OwnerId = job.OwnerId,
RepositoryId = job.RepositoryId,
DatasetId = job.DatasetId,
LastModified = DateTime.UtcNow,
CsvwMetadata = metadataJson,
VoidMetadata = voidMetadataJson,
ShowOnHomePage = job.IsPublic,
Tags = metadataJson["dcat:keyword"]?.ToObject <List <string> >()
};
await _datasetStore.CreateOrUpdateDatasetRecordAsync(datasetInfo);
ProgressLog.DatasetUpdated(datasetInfo);
}
catch (Exception ex)
{
Log.Error(ex, "Failed to update dataset record");
throw new WorkerException(ex,
"Failed to update dataset record. Your repository is updated, but may not show in the portal.");
}
}
