public static List <ShortValueLabels> GenerateLabelIndexes(IEnumerable <VariableWrapper> variables)
{
var result = new List <ShortValueLabels>();
var dictionaryIndex = 1;
foreach (var variable in variables)
{
var isString = variable.FormatType == FormatType.A;
if (variable.ValueLabels != null && variable.ValueLabels.Any() && (!isString || variable.ValueLength <= 8))
{
var valueLabel = new ShortValueLabels
{
Labels = variable.ValueLabels !.ToDictionary(p => p.Key, p => p.Value),
VariableIndex = new List <int> {
dictionaryIndex
}
};
result.Add(valueLabel);
}
dictionaryIndex += isString ? SpssMath.GetNumberOf32ByteBlocks(variable.ValueLength) : 1;
}
return(result);
}
}
public void ReadVariableRecord()
{
var valueLength = _reader.ReadInt32();
var hasVariableLabel = _reader.ReadInt32() == 1;
var missingValueType = _reader.ReadInt32();
_reader.ReadInt32(); //print format
var decimalPlaces = _reader.ReadByte();
var spssWidth = _reader.ReadByte();
var formatType = _reader.ReadByte();
_reader.ReadByte();
var shortName = _reader.ReadBytes(8);
var properties = new VariableProperties {
FormatType = (FormatType)formatType, Index = _currentIndex, MissingValueType = missingValueType, ShortName = shortName, DecimalPlaces = decimalPlaces
};
if (hasVariableLabel)
{
properties.Label = ReadLabel();
}
if (Math.Abs(missingValueType) != 0)
{
ReadMissing(properties);
}
var blockWidth = ReadBlankRecords(valueLength);
properties.ValueLength = blockWidth;
properties.SpssWidth = formatType == (int)FormatType.A ? blockWidth : spssWidth;
_metadataInfo.Variables.Add(properties);
_currentIndex += SpssMath.GetNumberOf32ByteBlocks(blockWidth);
}
public static List <DisplayParameter> GenerateDisplayValues(List <VariableWrapper> variables)
{
List <DisplayParameter> displayValues = new List <DisplayParameter>();
foreach (var variable in variables)
{
var displayValue = new DisplayParameter {
Measure = variable.Measure, Columns = variable.Columns, Alignment = variable.Alignment
};
var namedVariables = SpssMath.GetNumberOfGhostVariables(variable.ValueLength) + 1;
displayValues.AddRange(Enumerable.Range(1, namedVariables).Select(_ => displayValue));
}
return(displayValues);
}
private int ReadBlankRecords(int valueLength)
{
var lengthTotal = 0;
if (valueLength <= 8)
{
return(8);
}
var skip = SpssMath.GetAllocatedSize(valueLength);
lengthTotal += skip == 256 ? 252 : valueLength;
skip -= 8;
_reader.BaseStream.Seek(skip * 4, SeekOrigin.Current);
return(lengthTotal);
}
private void WriteString(string value, int valueLength)
{
var rawVariableLength = SpssMath.GetAllocatedSize(valueLength);
var bytes = _encoding.GetBytes(value.TrimEnd(), valueLength);
var byteLength = bytes.Length;
WriteStringBytes(bytes);
_writer.AddPadding();
var writtenBytes = byteLength / 255 * 256 + (byteLength % 255 + 7) / 8 * 8;
while (writtenBytes < rawVariableLength)
{
_writer.WriteCompressedCode(CompressedCode.SpaceCharsBlock);
writtenBytes += 8;
}
}
public void WriteHeaderRecord(Metadata metadata, List <VariableWrapper> variablesCount)
{
var count = variablesCount.Select(x => SpssMath.GetNumberOf32ByteBlocks(x.ValueLength)).Aggregate((result, blocks) => result + blocks);
_writer.Write((int)RecordType.HeaderRecord);
_writer.Write(_encoding.GetPaddedValueAsByteArray("@(#) IBM SPSS STATISTICS MS Windows 25.0.0.0", 60));
_writer.Write(2); //LayoutCode
_writer.Write(count);
_writer.Write(1); //compressed
_writer.Write(0); //WeightIndex
_writer.Write(metadata.Cases);
_writer.Write((double)metadata.Bias);
_writer.Write(DateTime.Now.ToString("dd MMM yyHH:mm:ss", _invariantCultureDateTimeFormat).ToCharArray());
_writer.Write(_encoding.GetPaddedValueAsByteArray(string.Empty, 64));
_writer.Write(new byte[3]);
}
private void WriteBlankAndGhostRecords(VariableWrapper variable)
{
var extraRecordCount = variable.FormatType != FormatType.A ? 0 : SpssMath.GetNumberOf32ByteBlocks(variable.ValueLength);
for (var i = 1; i < extraRecordCount; i++)
{
if (i % 32 != 0)
{
WriteBlankRecord();
continue;
}
var ghostIndex = i / 32;
var len = ghostIndex == variable.GhostNames.Count ? variable.LastGhostVariableLength : 255;
WriteGhostRecord(variable.GhostNames[ghostIndex - 1], len, variable.Label, variable.MissingValuesObject);
}
}
private void GenerateVariableGhostNames(List <VariableWrapper> records)
{
foreach (var record in records)
{
if (record.ValueLength < 256)
{
continue;
}
var ghostVariables = SpssMath.GetNumberOfGhostVariables(record.ValueLength);
record.LastGhostVariableLength = record.ValueLength % 252;
for (var j = 0; j < ghostVariables; j++)
{
var sn = _encoding.GetString(GetShortNameByteArray(record.Name)).TrimEnd(); const int maxLengthPrefix = 5;
var prefix = _encoding.GetString(sn, maxLengthPrefix);
var shortName = _encoding.GetPaddedValueAsByteArray(prefix + '0', 8);
record.GhostNames.Add(GetUniqueShortName(shortName, index => $"{prefix}{GetGhostSuffix(index)}"));
}
}
}
