/******************************
* QGRAMNORMALIZE
*
* This function 'normalizes' a string so valid q-grams can be made of it:
* All characters are transformed to uppercase, and all characters
* which are not letters or digits are stripped to a single space.
*
* qgramnormalize("Hallo, allemaal!").print(); --> "HALLO ALLEMAAL"
* qgramnormalize(" '' t ' est").print(); --> [ "T EST" ]
*
*****************************/
str
CMDqgramnormalize(str *res, str *Input)
{
char *input = *Input;
int i, j = 0;
char c, last = ' ';
RETURN_NIL_IF(strNil(input), TYPE_str);
*res = (str) GDKmalloc(sizeof(char) * (strlen(input) + 1)); /* normalized strings are never longer than original */
for (i = 0; input[i]; i++) {
c = toupper(input[i]);
if (!(('A' <= c && c <= 'Z') || ('0' <= c && c <= '9')))
c = ' ';
if (c != ' ' || last != ' ') {
(*res)[j++] = c;
}
last = c;
}
(*res)[j] = 0;
/* strip final whitespace */
while (j > 0 && (*res)[--j] == ' ')
(*res)[j] = 0;
return MAL_SUCCEED;
}
str
soundex_impl(str *res, str *Name)
{
RETURN_NIL_IF(strNil(*Name), TYPE_str);
*res = (str) GDKmalloc(sizeof(char) * (SoundexLen + 1));
/* calculate Key for Name */
soundex_code(*Name, *res);
return MAL_SUCCEED;
}
str
soundex_impl(str *res, str *Name)
{
RETURN_NIL_IF(strNil(*Name), TYPE_str);
*res = (str) GDKmalloc(sizeof(char) * (SoundexLen + 1));
if( *res == NULL)
throw(MAL,"soundex", MAL_MALLOC_FAIL);
/* calculate Key for Name */
soundex_code(*Name, *res);
return MAL_SUCCEED;
}
