private static string GetEnumString(string value, EnumCase enumCase)
{
switch (enumCase)
{
case EnumCase.lower:
return(value.ToLowerInvariant());
case EnumCase.upper:
return(value.ToUpperInvariant());
default:
return(value);
}
}
public static TAttr GetEnumAttribute <TAttr>(this XmlReader reader, string name, TAttr defaultValue, EnumCase enumCase)
{
string value = reader.GetAttribute(name);
if (!string.IsNullOrEmpty(value))
{
try
{
return((TAttr)Enum.Parse(typeof(TAttr), GetEnumString(value, enumCase)));
}
catch (ArgumentException x)
{
throw new InvalidDataException(string.Format(Resources.XmlUtil_GetAttribute_The_value__0__is_not_valid_for_the_attribute__1__, value, name), x);
}
}
return(defaultValue);
}
private static async Task <List <List <SentenceDetector.SentenceDetectorToken> > > ReadCorpusAsync(List <string> trainDocuments, EnumCase ConvertCase, SentenceDetector sentenceDetector)
{
var allLines = new List <string>();
foreach (var file in trainDocuments)
{
allLines.AddRange(await File.ReadAllLinesAsync(file));
}
var sentences = allLines.Where(l => l.StartsWith("# text =")).Select(l => l.Split(new char[] { '=' }, 2).Last().Trim()).ToList();
if (ConvertCase == EnumCase.ForceUpper)
{
sentences = sentences.Select(s => s.ToUpperInvariant()).ToList();
}
if (ConvertCase == EnumCase.ForceLower)
{
sentences = sentences.Select(s => s.ToLowerInvariant()).ToList();
}
return(sentences.Select(s =>
{
var tk = sentenceDetector.SentenceDetectorTokenizer(s).Select(t => new SentenceDetector.SentenceDetectorToken(t.Value, t.Begin, t.End)).ToList();
tk.Last().IsSentenceEnd = true;
return tk;
}).ToList());
}
public static TAttr GetEnumAttribute <TAttr>(this XmlReader reader, string name, TAttr defaultValue, EnumCase enumCase)
{
return(reader.GetEnumAttribute(name, null, defaultValue, enumCase));
}
private static List <List <SentenceDetector.SentenceDetectorToken> > ReadCorpus(List <string> trainDocuments, EnumCase ConvertCase, SentenceDetector sentenceDetector)
{
var allLines = trainDocuments.SelectMany(f => File.ReadAllLines(f));
var sentences = allLines.Where(l => l.StartsWith("# text =")).Select(l => l.Split(new char[] { '=' }, 2).Last().Trim()).ToList();
if (ConvertCase == EnumCase.ForceUpper)
{
sentences = sentences.Select(s => s.ToUpperInvariant()).ToList();
}
if (ConvertCase == EnumCase.ForceLower)
{
sentences = sentences.Select(s => s.ToLowerInvariant()).ToList();
}
return(sentences.Select(s =>
{
var tk = sentenceDetector.SentenceDetectorTokenizer(s).Select(t => new SentenceDetector.SentenceDetectorToken(t.Value, t.Begin, t.End)).ToList();
tk.Last().IsSentenceEnd = true;
return tk;
}).ToList());
}
public static TAttr GetEnumAttribute <TAttr>(this XmlReader reader, string name, TAttr defaultValue, EnumCase enumCase)
{
string value = reader.GetAttribute(name);
if (!string.IsNullOrEmpty(value))
{
try
{
return((TAttr)Enum.Parse(typeof(TAttr), GetEnumString(value, enumCase)));
}
catch (ArgumentException x)
{
throw new InvalidDataException(string.Format("The value {0} is not valid for the attribute {1}", value, name), x);
}
}
return(defaultValue);
}
