int CPorterStemming::Stem(char *word) /* in/out: the word stemmed */
{
int rule; /* which rule is fired in replacing an end */
/* Part 1: Check to ensure the word is all alphabetic */
for ( end = word; *end != '\0'; end++ )
{
if ( !isalpha(*end) )
return( 0 );
else
*end = tolower( *end );
}
end--;
/* Part 2: Run through the Porter algorithm */
(void)ReplaceEnd( word, step1a_rules );
rule = ReplaceEnd( word, step1b_rules );
if ( (106 == rule) || (107 == rule) )
(void)ReplaceEnd( word, step1b1_rules );
(void)ReplaceEnd( word, step1c_rules );
(void)ReplaceEnd( word, step2_rules );
(void)ReplaceEnd( word, step3_rules );
(void)ReplaceEnd( word, step4_rules );
(void)ReplaceEnd( word, step5a_rules );
(void)ReplaceEnd( word, step5b_rules );
/* Part 3: Return an indication of successful stemming */
return( 1 );
} /* Stem */
static FUZZY_WORD *Stem( FUZZY_OBJECT *fi, const char *inword)
{
char *end; /* pointer to the end of the word */
char word[MAXWORDLEN+1];
int length;
int rule_result; /* which rule is fired in replacing an end */
int i;
FUZZY_WORD *fw = create_fuzzy_word( inword, 1 );
/* Make sure the word is not too large from the start. */
if ( strlen( inword ) >= MAXWORDLEN )
{
fw->error = STEM_WORD_TOO_BIG;
return fw;
}
/* make working copy */
strcpy( word, inword );
/* Part 1: Check to ensure the word is all alphabetic */
/* no longer converts to lower case -- word should be lower before calling */
for ( end = word; *end; end++ )
if ( !isalpha( (unsigned int) *end ) )
{
fw->error = STEM_NOT_ALPHA;
return fw;
}
/* Part 2: Run through the Porter algorithm */
for (i = 0; i < (int)(sizeof(all_steps)/sizeof(all_steps[0])); i++)
{
rule_result = ReplaceEnd(word, all_steps[i]);
if ((rule_result == 106) || (rule_result == 107))
rule_result = ReplaceEnd(word, step1b1_rules);
if ( rule_result == STEM_WORD_TOO_BIG )
{
fw->error = rule_result;
return fw;
}
}
length = strlen( word );
/* Stem must be two chars or more in length */
if ( length <= 1 )
{
fw->error = STEM_TO_NOTHING;
return fw;
}
if ( length >= MAXWORDLEN )
{
fw->error = STEM_WORD_TOO_BIG;
return fw;
}
fw->free_strings = 1;
fw->string_list[0] = estrdup( word );
return fw;
}
