public ResourcePackage Import(string path, string agencyId)
{
this.harmonizingCache = new HarmonizingCache(MultilingualString.CurrentCulture);
var resourcePackage = new ResourcePackage();
resourcePackage.AgencyId = agencyId;
logger.Debug("Importing RData");
if (string.IsNullOrEmpty(path))
{
throw new ArgumentNullException("fileName");
}
if (!File.Exists(path))
{
throw new ArgumentException("The specified file must exist");
}
string fileNameWithExtension = Path.GetFileName(path);
string fileNameOnly = Path.GetFileNameWithoutExtension(path);
logger.Debug("RData import file: " + fileNameOnly);
resourcePackage.DublinCoreMetadata.Title.Current = fileNameOnly;
// Create the PhysicalInstance.
var physicalInstance = new PhysicalInstance()
{
AgencyId = agencyId
};
resourcePackage.PhysicalInstances.Add(physicalInstance);
physicalInstance.DublinCoreMetadata.Title.Current = fileNameOnly;
// File location
if (path != null)
{
DataFileIdentification fileID = new DataFileIdentification();
Uri uri;
if (Uri.TryCreate(path, UriKind.RelativeOrAbsolute, out uri))
{
fileID.Uri = uri;
}
fileID.Path = path;
physicalInstance.FileIdentifications.Add(fileID);
}
// Create the DataRelationship.
var dataRelationship = new DataRelationship();
physicalInstance.DataRelationships.Add(dataRelationship);
dataRelationship.AgencyId = agencyId;
dataRelationship.Label.Current = fileNameOnly;
// Load the file into R.
string pathForR = path.Replace("\\", "/");
engine.Evaluate(string.Format("load('{0}')", pathForR));
// Find all the data frames.
var dataFrames = GetDataFrames();
// For each data frame in the RData file, create a LogicalRecord.
foreach (var pair in dataFrames)
{
string name = pair.Key;
var dataFrame = pair.Value;
// TODO This should be tracked per record, not PhysicalInstance.
physicalInstance.FileStructure.CaseQuantity = dataFrame.RowCount;
var logicalRecord = new LogicalRecord()
{
AgencyId = agencyId
};
dataRelationship.LogicalRecords.Add(logicalRecord);
logicalRecord.Label.Current = name;
List <string> variableLabels = null;
var variableLabelsExpr = dataFrame.GetAttribute("var.labels");
if (variableLabelsExpr != null)
{
var labelVector = variableLabelsExpr.AsVector();
variableLabels = new List <string>(labelVector.Select(x => (string)x));
}
for (int i = 0; i < dataFrame.ColumnCount; i++)
{
string columnName = dataFrame.ColumnNames[i];
var column = dataFrame[i];
var variable = new Variable()
{
AgencyId = agencyId
};
logicalRecord.VariablesInRecord.Add(variable);
// Name
variable.ItemName.Current = columnName;
// Label
if (variableLabels != null)
{
variable.Label.Current = variableLabels[i];
}
// Type
if (column.Type == RDotNet.Internals.SymbolicExpressionType.NumericVector)
{
variable.RepresentationType = RepresentationType.Numeric;
variable.Additivity = AdditivityType.Stock;
}
else if (column.Type == RDotNet.Internals.SymbolicExpressionType.IntegerVector)
{
if (column.IsFactor())
{
variable.RepresentationType = RepresentationType.Code;
string[] factors = column.AsFactor().GetLevels();
variable.CodeRepresentation.Codes = GetCodeList(factors, agencyId, resourcePackage);
}
else
{
variable.RepresentationType = RepresentationType.Numeric;
variable.NumericRepresentation.NumericType = NumericType.Integer;
variable.Additivity = AdditivityType.Stock;
}
}
else if (column.Type == RDotNet.Internals.SymbolicExpressionType.CharacterVector)
{
variable.RepresentationType = RepresentationType.Text;
}
}
}
return(resourcePackage);
}
public void Map(ManagedFile file, PhysicalInstance pi)
{
if (file == null)
{
throw new ArgumentNullException("file");
}
if (pi == null)
{
throw new ArgumentNullException("physicalInstance");
}
// Map properties from the ManagedFile to DDI.
pi.SetUserId("FileNumber", file.Number?.ToString());
pi.DublinCoreMetadata.Title.Current = file.Title;
pi.DublinCoreMetadata.AlternateTitle.Current = file.PublicName;
var fileId = pi.FileIdentifications.FirstOrDefault();
if (fileId == null)
{
fileId = new DataFileIdentification();
pi.FileIdentifications.Add(fileId);
}
Uri uri;
bool gotUri = Uri.TryCreate(file.PersistentLink, UriKind.RelativeOrAbsolute, out uri);
if (gotUri)
{
fileId.Uri = uri;
}
pi.SetUserAttribute("PersistentLinkDate", file.PersistentLinkDate?.ToString());
pi.SetUserAttribute("FileType", file.Type);
pi.SetUserAttribute("FormatName", file.FormatName);
pi.SetUserAttribute("FormatId", file.FormatId);
pi.SetUserAttribute("Size", file.Size.ToString());
pi.SetUserAttribute("CreationDate", file.CreationDate);
pi.SetUserAttribute("KindOfData", file.KindOfData);
pi.DublinCoreMetadata.Source.Current = file.Source;
pi.SetUserAttribute("SourceInformation", file.SourceInformation);
pi.DublinCoreMetadata.Rights.Current = file.Rights;
fileId.IsPublic = file.IsPublicAccess;
pi.SetUserAttribute("UploadedDate", file.UploadedDate);
pi.SetUserAttribute("ExternalDatabase", file.ExternalDatabase);
pi.SetUserAttribute("Software", file.Software);
pi.SetUserAttribute("SoftwareVersion", file.SoftwareVersion);
pi.SetUserAttribute("Hardware", file.Hardware);
var fingerprint = pi.Fingerprints.FirstOrDefault();
if (fingerprint == null)
{
fingerprint = new Fingerprint();
pi.Fingerprints.Add(fingerprint);
}
fingerprint.FingerprintValue = file.Checksum;
fingerprint.AlgorithmSpecification = file.ChecksumMethod;
pi.SetUserAttribute("ChecksumDate", file.ChecksumDate);
pi.SetUserAttribute("VirusCheckOutcome", file.VirusCheckOutcome);
pi.SetUserAttribute("VirusCheckMethod", file.VirusCheckMethod);
pi.SetUserAttribute("VirusCheckDate", file.VirusCheckDate);
pi.SetUserAttribute("AcceptedDate", file.AcceptedDate);
pi.SetUserAttribute("CertifiedDate", file.CertifiedDate);
}
