public void ForeignWordConstructorTest()
{
var text = "Bonjour";
var target = new ForeignWord(text);
Check.That(target.Text).IsEqualTo(text);
Check.That(target.UsedAsType).IsNull();
}
public void UsedAsTypeTest()
{
var text = "Bonjour";
var target = new ForeignWord(text);
var expected = typeof(Interjection);
Type actual;
target.UsedAsType = expected;
actual = target.UsedAsType;
Check.That(actual).IsEqualTo(expected);
}
public string TryGloss(string language, string pos, Dialect dialect)
{
//HACK: This is wrong. Should be using Token or some other base class.
if (word.ContainsCheck("-"))
{
CompoundWord cw = new CompoundWord(Text);
return(cw.TryGloss(language, pos));
}
int digits;
bool isNumeric = int.TryParse(word, out digits);
if (word.StartCheck("#") || isNumeric)
{
Number n = new Number(Text);
return(n.TryGloss(language, pos, dialect));
}
if (Token.IsNeologism(LookupForm(word)))
{
//A neologism is unglossable.
return(word);
}
//HACK: This is a proper modifer
if (LookupForm(word).IsFirstUpperCased())
{
if (!ProperModifier.IsProperModifer(LookupForm(word)))
{
ForeignWord fw = new ForeignWord(word);
return(fw.TryGloss(language, pos));
}
ProperModifier cw = new ProperModifier(Text);
return(cw.TryGloss(language, pos));
}
Dictionary <string, Dictionary <string, string[]> > glossMap;
Words.Glosses.TryGetValue(LookupForm(word), out glossMap);
if (glossMap == null)
{
return("[missing map for " + Text + "]");
}
if (glossMap.ContainsKey(language))
{
if (glossMap[language].ContainsKey(pos))
{
Random r = new Random(DateTime.Now.Millisecond);//TODO: Make this a part of config
string[] possibilities = glossMap[language][pos];
return(possibilities[r.Next(possibilities.Length)]);
}
}
//TraceMissingGloss();
if (Text.ToUpper()[0] != Text[0])
{
return("[Error " + pos + " " + Text + " " + language + "]");
}
else
{
return(Text);
}
}
public void NewThings()
{
string[] samples =
new string[]
{
CorpusTexts.ProfesorAndMadMan,
CorpusTexts.UnpaText,
CorpusTexts.Gilgamesh,
CorpusTexts.SampleText1,
CorpusTexts.SampleText3,
CorpusTexts.Lao,
CorpusTexts.GeorgeSong,
CorpusTexts.CrazyAnimal,
CorpusTexts.CrazyAnimal2
//,CorpusTexts.JanSin //Too many neologisms to cope.
, CorpusTexts.RuneDanceSong
, CorpusTexts.janPusaRice
, CorpusTexts.janPend
};
foreach (string sample in samples)
{
//Split, normalize, tokenize, find words.
TokenParserUtils tp = new TokenParserUtils();
Dictionary <string, string> stuff = new Dictionary <string, string>();
foreach (Token toke in tp.ValidTokens(sample).Distinct())
{
if (toke.CheckIsCompoundWord(toke.Text))
{
if (!stuff.ContainsKey(toke.Text))
{
stuff.Add(toke.Text, "Compound");
}
}
if (toke.Text.StartCheck("#") && toke.CheckIsNumber(toke.Text))
{
//Should just have to verify we can parse. No need for dictionary.
if (!stuff.ContainsKey(toke.Text))
{
stuff.Add(toke.Text, "Number");
}
}
if (toke.CheckIsProperModifier(toke.Text))
{
if (!stuff.ContainsKey(toke.Text))
{
stuff.Add(toke.Text, "Proper");
}
}
if (ForeignWord.IsForeign(toke.Text))
{
if (!stuff.ContainsKey(toke.Text))
{
stuff.Add(toke.Text, "Proper");
}
}
if (Neologism.IsNeologism(toke.Text))
{
if (!stuff.ContainsKey(toke.Text))
{
stuff.Add(toke.Text, "Neologism");
}
}
}
foreach (var t in stuff)
{
Console.WriteLine(t.Value + " " + t.Key);
}
}
}
