/// <param name="s"> buffer, oversized to at least <code>len+1</code> </param>
/// <param name="len"> initial valid length of buffer </param>
/// <returns> new valid length, stemmed </returns>
public virtual int Stem(char[] s, int len)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(s.Length >= len + 1, "this stemmer requires an oversized array of at least 1");
}
len = plural.Apply(s, len);
len = unification.Apply(s, len);
len = adverb.Apply(s, len);
int oldlen;
do
{
oldlen = len;
len = augmentative.Apply(s, len);
} while (len != oldlen);
oldlen = len;
len = noun.Apply(s, len);
if (len == oldlen) // suffix not removed
{
len = verb.Apply(s, len);
}
len = vowel.Apply(s, len);
// RSLG accent removal
for (int i = 0; i < len; i++)
{
switch (s[i])
{
case 'á':
s[i] = 'a';
break;
case 'é':
case 'ê':
s[i] = 'e';
break;
case 'í':
s[i] = 'i';
break;
case 'ó':
s[i] = 'o';
break;
case 'ú':
s[i] = 'u';
break;
}
}
return(len);
}
public virtual int Stem(char[] s, int len)
{
return(pluralStep.Apply(s, len));
}
/// <param name="s"> buffer, oversized to at least <code>len+1</code> </param>
/// <param name="len"> initial valid length of buffer </param>
/// <returns> new valid length, stemmed </returns>
public virtual int Stem(char[] s, int len)
{
Debug.Assert(s.Length >= len + 1, "this stemmer requires an oversized array of at least 1");
len = plural.Apply(s, len);
len = adverb.Apply(s, len);
len = feminine.Apply(s, len);
len = augmentative.Apply(s, len);
int oldlen = len;
len = noun.Apply(s, len);
if (len == oldlen) // suffix not removed
{
oldlen = len;
len = verb.Apply(s, len);
if (len == oldlen) // suffix not removed
{
len = vowel.Apply(s, len);
}
}
// rslp accent removal
for (int i = 0; i < len; i++)
{
switch (s[i])
{
case 'à':
case 'á':
case 'â':
case 'ã':
case 'ä':
case 'å':
s[i] = 'a';
break;
case 'ç':
s[i] = 'c';
break;
case 'è':
case 'é':
case 'ê':
case 'ë':
s[i] = 'e';
break;
case 'ì':
case 'í':
case 'î':
case 'ï':
s[i] = 'i';
break;
case 'ñ':
s[i] = 'n';
break;
case 'ò':
case 'ó':
case 'ô':
case 'õ':
case 'ö':
s[i] = 'o';
break;
case 'ù':
case 'ú':
case 'û':
case 'ü':
s[i] = 'u';
break;
case 'ý':
case 'ÿ':
s[i] = 'y';
break;
}
}
return(len);
}