public void IdentifyDatesFormatsUS()
{
var dateSep = "/";
var culture = new CultureInfo("en-US");
// Try the date formats
foreach (var fmt in StringConversion.StandardDateTimeFormats.Keys)
{
// Fill Samples
var samples = new HashSet <string>();
for (var month = 9; month < 10; month++)
{
for (var day = 10; day < 15; day++)
{
for (var hrs = 11; hrs < 13; hrs++)
{
for (var min = 24; min < 26; min++)
{
samples.Add(new DateTime(2010, month, day, hrs, min, 10, 876, DateTimeKind.Local).ToString(fmt, culture));
}
}
}
}
Assert.IsNotNull(
StringConversion.CheckDate(samples, fmt, dateSep, ":", CultureInfo.CurrentCulture).FoundValueFormat,
$"Test format {fmt}\nFirst not matching: {samples.First()}");
}
}
public static CheckResult GuessDateTime(IList <string> samples, bool checkNamedDates,
CancellationToken cancellationToken)
{
if (samples == null || samples.IsEmpty())
{
throw new ArgumentNullException(nameof(samples));
}
var checkResult = new CheckResult();
var firstValue = samples[0];
foreach (var fmt in StringConversion.StandardDateTimeFormats.MatchingforLength(firstValue.Length, checkNamedDates))
{
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
if (fmt.IndexOf('/') > 0)
{
foreach (var sep in StringConversion.DateSeparators)
{
var res = StringConversion.CheckDate(samples, fmt, sep, ":", CultureInfo.CurrentCulture);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
}
else
{
var res = StringConversion.CheckDate(samples, fmt,
CultureInfo.CurrentCulture.DateTimeFormat.DateSeparator, ":", CultureInfo.CurrentCulture);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
}
return(checkResult);
}
/// <summary>
/// Guesses the value format.
/// </summary>
/// <param name="cancellationToken">A cancellation token</param>
/// <param name="samples">The samples.</param>
/// <param name="minRequiredSamples">The minimum required samples.</param>
/// <param name="trueValue">The text to be regarded as <c>true</c></param>
/// <param name="falseValue">The text to be regarded as <c>false</c></param>
/// <param name="guessBoolean">Try to identify a boolean</param>
/// <param name="guessGuid">Try to determine if its a GUID</param>
/// <param name="guessNumeric">Try to determine if its a Number</param>
/// <param name="guessDateTime">Try to determine if it is a date time</param>
/// <param name="guessPercentage">Accept percentage values</param>
/// <param name="serialDateTime">Allow serial Date time</param>
/// <param name="checkNamedDates">if set to <c>true</c> [check named dates].</param>
/// <returns><c>Null</c> if no format could be determined otherwise a <see cref="ValueFormat" /></returns>
public static CheckResult GuessValueFormat(IList <string> samples, int minRequiredSamples,
string trueValue, string falseValue, bool guessBoolean, bool guessGuid, bool guessNumeric, bool guessDateTime,
bool guessPercentage, bool serialDateTime, bool checkNamedDates, ValueFormat othersValueFormatDate, CancellationToken cancellationToken)
{
Contract.Requires(samples != null);
if (samples.IsEmpty())
{
return(null);
}
var count = samples.Count();
var checkResult = new CheckResult {
FoundValueFormat = new ValueFormat()
};
// if it only one sample value and its false, assume its a boolean
if (guessBoolean && count == 1 && !string.IsNullOrEmpty(falseValue))
{
foreach (var value in samples)
{
if (value.Equals(falseValue, StringComparison.OrdinalIgnoreCase))
{
checkResult.FoundValueFormat.DataType = DataType.Boolean;
return(checkResult);
}
break;
}
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// this could be a boolean
if (guessBoolean && count <= 2)
{
var allParsed = true;
string usedTrueValue = null;
string usedFalseValue = null;
foreach (var value in samples)
{
var result = StringConversion.StringToBooleanStrict(value, trueValue, falseValue);
if (result == null)
{
allParsed = false;
break;
}
if (result.Item1)
{
usedTrueValue = result.Item2;
}
else
{
usedFalseValue = result.Item2;
}
}
if (allParsed)
{
checkResult.FoundValueFormat.DataType = DataType.Boolean;
if (!string.IsNullOrEmpty(usedTrueValue))
{
checkResult.FoundValueFormat.True = usedTrueValue;
}
if (!string.IsNullOrEmpty(usedFalseValue))
{
checkResult.FoundValueFormat.False = usedFalseValue;
}
return(checkResult);
}
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
if (guessGuid && StringConversion.CheckGuid(samples))
{
checkResult.FoundValueFormat.DataType = DataType.Guid;
return(checkResult);
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// in case we have named dates, this is not feasible
if (!checkNamedDates)
{
// Trying some chars, if they are in, assume its a string
var valuesWithChars = 0;
foreach (var value in samples)
{
// Not having AM PM or T as it might be part of a date Not having E in there as might be
// part of a number u 1.487% o 6.264% n 2.365% i 6.286% h 7.232% s 6.327% This adds to a
// 30% chance for each position in the text to determine if a text a regular text,
if (value.IndexOfAny(new[] { 'u', 'U', 'o', 'O', 'i', 'I', 'n', 'N', 's', 'S', 'h', 'H' }) <= -1)
{
continue;
}
valuesWithChars++;
// Only do so if more then half of the samples are string
if (valuesWithChars < count / 2 && valuesWithChars < 10)
{
continue;
}
checkResult.FoundValueFormat.DataType = DataType.String;
return(checkResult);
}
}
if (count < minRequiredSamples && guessDateTime && othersValueFormatDate != null)
{
var res = StringConversion.CheckDate(samples, othersValueFormatDate.DateFormat, othersValueFormatDate.DateSeparator, othersValueFormatDate.TimeSeparator, CultureInfo.CurrentCulture);
if (res.FoundValueFormat != null)
{
return(res);
}
}
// if we have less than the required samples values do not try and try to get a type
if (count < minRequiredSamples || cancellationToken.IsCancellationRequested)
{
return(null);
}
var firstValue = samples.First();
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// Guess a date format that could be interpreted as number before testing numbers
if (guessDateTime && firstValue.Length == 8)
{
var res = StringConversion.CheckDate(samples, "yyyyMMdd", string.Empty, ":", CultureInfo.InvariantCulture);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// We need to have at least 10 sample values here its too dangerous to assume it is a date
if (guessDateTime && serialDateTime && count > 10 && count > minRequiredSamples)
{
var res = StringConversion.CheckSerialDate(samples, true);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// assume dates are of the same format across the files we check if the dates
// we have would possibly match no matter how many samples we have
if (guessDateTime && othersValueFormatDate != null)
{
var res = StringConversion.CheckDate(samples, othersValueFormatDate.DateFormat, othersValueFormatDate.DateSeparator, othersValueFormatDate.TimeSeparator, CultureInfo.CurrentCulture);
if (res.FoundValueFormat != null)
{
return(res);
}
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
if (guessNumeric)
{
var res = GuessNumeric(samples, guessPercentage, false, cancellationToken);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// Minimum length of a date is 4 characters
if (guessDateTime && firstValue.Length > 3)
{
var res = GuessDateTime(samples, checkNamedDates, cancellationToken);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
// if we have dates and allow serial dates, but do not guess numeric (this would be a fit) try
// if the dates are all serial
if (!guessDateTime || !serialDateTime || guessNumeric)
{
return(checkResult);
}
{
var res = StringConversion.CheckSerialDate(samples, false);
if (res.FoundValueFormat != null)
{
return(res);
}
checkResult.KeepBestPossibleMatch(res);
}
return(checkResult);
}