/// <summary>
/// Creates a new SPSS data document, initializing its dictionary
/// by copying the dictionary from an existing SPSS data file.
/// </summary>
/// <param name="filename">
/// The filename of the new document to create.
/// </param>
/// <param name="copyDictionaryFromFileName">
/// The filename of the existing SPSS data file to copy the dictionary from.
/// </param>
/// <returns>
/// The newly created <see cref="SpssDataDocument">SPSS data document</see>.
/// </returns>
public static SpssDataDocument Create(string filename, string copyDictionaryFromFileName)
{
if (File.Exists(filename))
{
throw new InvalidOperationException("File to create already exists.");
}
if (!File.Exists(copyDictionaryFromFileName))
{
throw new FileNotFoundException("File to copy does not exist.", copyDictionaryFromFileName);
}
using (SpssDataDocument read = SpssDataDocument.Open(copyDictionaryFromFileName, SpssFileAccess.Read))
{
SpssDataDocument toReturn = new SpssDataDocument(filename, SpssFileAccess.Create);
foreach (SpssVariable var in read.Variables)
{
toReturn.Variables.Add(var.Clone());
}
toReturn.CommitDictionary();
return(toReturn);
}
}
/// <summary>
/// Converts the metadata in an SPSS .SAV data file into a DataTable.
/// </summary>
/// <param name="spssSavFilename">
/// The filename of the SPSS .SAV data file.
/// </param>
/// <returns>
/// The <see cref="DataTable"/> containing all the metadata.
/// </returns>
public static DataTable ToDataTable(string spssSavFilename)
{
if (spssSavFilename == null)
{
throw new ArgumentNullException("spssSavFilename");
}
DataTable dataTable = new DataTable();
using (SpssDataDocument doc = SpssDataDocument.Open(spssSavFilename, SpssFileAccess.Read))
{
ToDataTable(doc, dataTable);
}
// Return the completed DataTable.
return(dataTable);
}
/// <summary>
/// Converts the metadata in an SPSS .SAV data file into a DDI codebook.
/// </summary>
/// <param name="spssSavFilename">
/// The filename of the SPSS .SAV data file.
/// </param>
/// <returns>
/// The <see cref="XmlDocument"/> containing all the metadata.
/// </returns>
public static XmlDocument ToDdi(string spssSavFilename)
{
const string ddiNamespace = "http://www.icpsr.umich.edu/DDI";
if (spssSavFilename == null)
{
throw new ArgumentNullException("spssSavFilename");
}
XmlDocument ddi = new XmlDocument();
// Build initial ddi document up.
// Open up SPSS file and fill in the ddi var tags.
using (SpssDataDocument doc = SpssDataDocument.Open(spssSavFilename, SpssFileAccess.Read))
{
ddi.PreserveWhitespace = true;
// Read from the embedded xml file: blankDdi.xml into the ddi document
ddi.LoadXml(EmbeddedResources.LoadFileFromAssemblyWithNamespace("/blankDdi.xml", Project.DefaultNamespace));
// This is where the hard coding ends and methods are called to extract data from the sav file
XmlElement xmlRoot = ddi.DocumentElement;
XmlNamespaceManager xmlNS = new XmlNamespaceManager(ddi.NameTable);
xmlNS.AddNamespace("ddi", ddiNamespace);
XmlNode nData = xmlRoot.SelectSingleNode(@"ddi:dataDscr", xmlNS);
foreach (SpssVariable var in doc.Variables)
{
string nameOfVar = var.Name;
// variable name and its ID and then if its a numeric : its interval
XmlElement variable = ddi.CreateElement("ddi:var", ddiNamespace);
variable.SetAttribute("ID", string.Empty, nameOfVar);
variable.SetAttribute("name", string.Empty, nameOfVar);
// This is the variable that holds the characteristic whether the variable has discrete or continuous interval
int Dec;
if (var is SpssNumericVariable)
{
Dec = ((SpssNumericVariable)var).PrintDecimal;
string interval = string.Empty;
if (Dec == 0)
{
interval = "discrete";
}
else
{
interval = "contin";
}
variable.SetAttribute("intrvl", string.Empty, interval);
}
// for the location width part
XmlElement location = ddi.CreateElement("ddi:location", ddiNamespace);
int Wid = var.ColumnWidth;
location.SetAttribute("width", Wid.ToString());
variable.AppendChild(location);
// label of the variable is set in "varlabel" and extracted using var.Label
XmlElement varLabel = ddi.CreateElement("ddi:labl", ddiNamespace);
varLabel.InnerText = var.Label;
variable.AppendChild(varLabel);
foreach (var response in var.GetValueLabels())
{
XmlElement answer = ddi.CreateElement("ddi:catgry", ddiNamespace);
// catValue(category Value) is the element storing the text i.e. option number
XmlElement catValue = ddi.CreateElement("ddi:catValu", ddiNamespace);
catValue.InnerText = response.Key;
answer.AppendChild(catValue);
// catLabel(category Label) is the element storing the text i.e. name of answer
XmlElement catLabel = ddi.CreateElement("ddi:labl", ddiNamespace);
catLabel.InnerText = response.Value;
answer.AppendChild(catLabel);
// appending the answer option to the parent "variable" node i.e. the question node
variable.AppendChild(answer);
}
// end of extracting the response values for each variable
XmlElement varFormat = ddi.CreateElement("ddi:varFormat", ddiNamespace);
if (var is SpssNumericVariable)
{
varFormat.SetAttribute("type", "numeric");
}
else if (var is SpssStringVariable)
{
varFormat.SetAttribute("type", "character");
}
else
{
throw new NotSupportedException("Variable " + nameOfVar + " is not a string or a numeric variable type.");
}
variable.AppendChild(varFormat);
nData.AppendChild(variable);
}
// end of extraction of each variable and now we have put all the variable data into ndata
// Return the completed ddi file.
return(ddi);
}
}
