protected void Evaluate(int inputSetSize)
{
var inputRecords = GetRecords(inputSetSize);
InputRecords = inputRecords.Count();
var rawTokens = GetTokens(inputRecords);
if (rawTokens == null || !rawTokens.Any())
{
throw new FormatikException("Could not find any similarities between the input document and the example");
}
this.Separators = GetOutputSeparators(rawTokens);
Tokens = PurifyTokens(rawTokens, Separators);
// construct output template
SetTokenWrappers(Tokens, Separators);
var firstToken = Tokens.First();
var lastToken = Tokens.Last();
var headerSize = firstToken.Values
.Select(value => Example.IndexOf($"{firstToken.Prefix}{value.Value}{firstToken.Suffix}{(Separators.Count() > 1 ? Separators.Skip(1).First() : string.Empty)}"))
.Min(index => index);
this.Header = headerSize > 0 ? Example.Substring(0, headerSize) : string.Empty;
var footerStartAt = lastToken.Values
.Select(value =>
{
var lastTokenString = $"{(Separators.Count() > 1 ? Separators.Skip(1).First() : Separators.First())}{lastToken.Prefix}{value.Value}{lastToken.Suffix}";
return(Example.LastIndexOf(lastTokenString) + lastTokenString.Length);
})
.Max(index => index);
this.Footer = footerStartAt > 0 ? Example.Substring(footerStartAt) : string.Empty;
}
private int Process(IEnumerable <IInputRecord> records, Stream output, Encoding encoding)
{
var processed = 0;
using (var writer = new StreamWriter(output, encoding, 8192, true))
{
if (!string.IsNullOrEmpty(Header))
{
writer.Write(Header);
}
var isFirst = true;
var _1stOrderSeparator = Separators.First();
var _2ndOrderSeparator = Separators.Skip(1).FirstOrDefault() ?? string.Empty;
foreach (var record in records)
{
if (isFirst)
{
isFirst = false;
}
else
{
writer.Write(_1stOrderSeparator);
}
writer.Write(String.Join(_2ndOrderSeparator,
Tokens.Select(token => token.GetOutput(record))));
processed++;
}
if (!string.IsNullOrEmpty(Footer))
{
writer.Write(Footer);
}
}
return(processed);
}