private static int[] ConvertToVector_Text_Number(string text, char[] uniqueNonWhitespace)
{
if (string.IsNullOrEmpty(text))
{
return(new int[0]);
}
int[] retVal = new int[text.Length];
for (int cntr = 0; cntr < retVal.Length; cntr++)
{
char current = text[cntr];
if (!UtilityCore.IsWhitespace(current)) // just leave whitespace as zero
{
retVal[cntr] = Array.IndexOf <char>(uniqueNonWhitespace, current) + 1; // UniqueChars should hold all possible chars. If it doesn't, IndexOf will return -1, and adding one will make it look like whitespace (but that should never happen)
}
}
return(retVal);
}
public static SOMFieldStats GetFieldStats(IEnumerable <string> field, SOMFieldType?typeOverride = null)
{
// Dedupe
string[] deduped = field.
Select(o => o.Trim()).
Distinct().
ToArray();
// Unique Chars
char[] uniqueChars = deduped.
SelectMany(o => o).
Distinct().
OrderBy(o => o).
ToArray();
// FieldType
SOMFieldType type = typeOverride ?? GetFieldType(field);
// Numeric Stats
double?numericMin = null;
double?numericMax = null;
double?numericAvg = null;
double?numericStandDev = null;
if (type == SOMFieldType.Integer || type == SOMFieldType.FloatingPoint)
{
double[] numerics = field.
Select(o => string.IsNullOrWhiteSpace(o) ? 0d : double.Parse(o.Trim())).
ToArray();
numericMin = numerics.Min();
numericMax = numerics.Max();
var avg_stdev = Math1D.Get_Average_StandardDeviation(numerics);
numericAvg = avg_stdev.Item1;
numericStandDev = avg_stdev.Item2;
}
// Date Stats
DateTime?dateMin = null;
DateTime?dateMax = null;
DateTime?dateAvg = null;
TimeSpan?dateStandDev = null;
if (type == SOMFieldType.DateTime)
{
DateTime[] dates = field.
Where(o => !string.IsNullOrWhiteSpace(o)).
Select(o => DateTime.Parse(o.Trim())).
ToArray();
dateMin = dates.Min();
dateMax = dates.Max();
var avg_stdev = Math1D.Get_Average_StandardDeviation(dates);
dateAvg = avg_stdev.Item1;
dateStandDev = avg_stdev.Item2;
}
// Return
return(new SOMFieldStats()
{
Count = field.Count(),
UniqueCount = deduped.Length,
MinLength = deduped.Min(o => o.Length), // deduped are already trimmed
MaxLength = deduped.Max(o => o.Length),
UniqueChars = uniqueChars,
UniqueChars_NonWhitespace = uniqueChars.Where(o => !UtilityCore.IsWhitespace(o)).ToArray(),
FieldType = type,
Numeric_Min = numericMin,
Numeric_Max = numericMax,
Numeric_Avg = numericAvg,
Numeric_StandDev = numericStandDev,
Date_Min = dateMin,
Date_Max = dateMax,
Date_Avg = dateAvg,
Date_StandDev = dateStandDev,
});
}