public override bool IsValidToken(Token t)
{
if (t is HebrewToken)
{
HebrewToken ht = t as HebrewToken;
// Pose a minimum score limit for words
if (ht.Score < 0.7f)
{
return(false);
}
// Pose a higher threshold to verbs (easier to get irrelevant verbs from toleration)
if ((ht.Mask & DMask.D_TYPEMASK) == DMask.D_VERB && ht.Score < 0.85f)
{
return(false);
}
}
return(true);
}
internal HebrewValue(HebrewToken token, short value)
{
this.token = token;
this.value = value;
}
internal HebrewValue(HebrewToken token, int value) {
this.token = token;
this.value = value;
}
public NounWord(HebrewToken ht) : base(ht.Text)
{
this.ht = ht;
}
private void btnCheck_Click(object sender, EventArgs e)
{
if (m_lemmatizer == null || !m_lemmatizer.IsInitialized || !(m_lemmatizer is StreamLemmatizer))
{
string hspellPath = SelectHSpellFolderPath();
if (hspellPath == null)
{
return;
}
var radix = Loader.LoadDictionaryFromHSpellFolder(hspellPath, true);
m_lemmatizer = new HebMorph.StreamLemmatizer(radix, false);
}
(m_lemmatizer as StreamLemmatizer).SetStream(new System.IO.StringReader(txbCheck.Text));
string word = string.Empty;
List <Token> tokens = new List <Token>();
while ((m_lemmatizer as StreamLemmatizer).LemmatizeNextToken(out word, tokens) > 0)
{
if (tokens.Count == 0)
{
LoggerWriteLine("{0}: Unrecognized word{1}{2}", word, Environment.NewLine, "------");
continue;
}
if (tokens.Count == 1 && !(tokens[0] is HebrewToken))
{
LoggerWriteLine("{0}: Not a Hebrew word; detected as {1}{2}{3}", word,
tokens[0].IsNumeric ? "Numeric" : "NonHebrew", Environment.NewLine, "------");
continue;
}
int curPrefix = -1;
string curWord = string.Empty;
foreach (Token r in tokens)
{
HebrewToken ht = r as HebrewToken;
if (ht == null)
{
continue;
}
if (curPrefix != ht.PrefixLength || !curWord.Equals(ht.Text))
{
curPrefix = ht.PrefixLength;
curWord = ht.Text;
if (curPrefix == 0)
{
LoggerWriteLine("Legal word: {0} (score: {1})", ht.Text, ht.Score);
}
else
{
LoggerWriteLine("Legal combination: {0}+{1} (score: {2})", ht.Text.Substring(0, curPrefix),
ht.Text.Substring(curPrefix), ht.Score);
}
}
LoggerWriteLine(ht.ToString());
}
LoggerWriteLine("------");
}
}
public unknownWord(HebrewToken ht) : base(ht.Text)
{
this.ht = ht;
}
public GufWord(HebrewToken ht) : base(ht.Text)
{
if (ht.Mask.HasFlag(DMask.D_FIRST | DMask.D_NUMBASE))
{
word = "אנחנו";
}
else if (ht.Mask.HasFlag(DMask.D_FIRST))
{
word = "אני";
}
else if (ht.Mask.HasFlag(DMask.D_SECOND | DMask.D_NUMBASE | DMask.D_MASCULINE))
{
word = "אתם";
}
else if (ht.Mask.HasFlag(DMask.D_SECOND | DMask.D_MASCULINE))
{
word = "אתה";
}
else if (ht.Mask.HasFlag(DMask.D_SECOND | DMask.D_NUMBASE | DMask.D_FEMININE))
{
word = "אתן";
}
else if (ht.Mask.HasFlag(DMask.D_SECOND | DMask.D_MASCULINE))
{
word = "את";
}
else if (ht.Mask.HasFlag(DMask.D_THIRD | DMask.D_NUMBASE | DMask.D_MASCULINE))
{
word = "הם";
}
else if (ht.Mask.HasFlag(DMask.D_THIRD | DMask.D_MASCULINE))
{
word = "הוא";
}
else if (ht.Mask.HasFlag(DMask.D_THIRD | DMask.D_NUMBASE | DMask.D_FEMININE))
{
word = "הן";
}
else if (ht.Mask.HasFlag(DMask.D_THIRD | DMask.D_MASCULINE))
{
word = "היא";
}
else if (ht.Mask.HasFlag(DMask.D_OFIRST | DMask.D_NUMBASE))
{
word = "שלנו";
}
else if (ht.Mask.HasFlag(DMask.D_OFIRST))
{
word = "שלי";
}
else if (ht.Mask.HasFlag(DMask.D_OSECOND | DMask.D_NUMBASE | DMask.D_MASCULINE))
{
word = "שלכם";
}
else if (ht.Mask.HasFlag(DMask.D_OSECOND | DMask.D_MASCULINE))
{
word = "שלך";
}
else if (ht.Mask.HasFlag(DMask.D_OSECOND | DMask.D_NUMBASE | DMask.D_FEMININE))
{
word = "שלכן";
}
else if (ht.Mask.HasFlag(DMask.D_OSECOND | DMask.D_MASCULINE))
{
word = "שלך";
}
else if (ht.Mask.HasFlag(DMask.D_OTHIRD | DMask.D_NUMBASE | DMask.D_MASCULINE))
{
word = "שלהם";
}
else if (ht.Mask.HasFlag(DMask.D_OTHIRD | DMask.D_MASCULINE))
{
word = "שלו";
}
else if (ht.Mask.HasFlag(DMask.D_OTHIRD | DMask.D_NUMBASE | DMask.D_FEMININE))
{
word = "שלהן";
}
else if (ht.Mask.HasFlag(DMask.D_OTHIRD | DMask.D_MASCULINE))
{
word = "שלה";
}
else
{
word = ht.Mask.ToString();
}
}
