Update of /cvsroot/monetdb/MonetDB5/src/modules/mal
In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv15444
Modified Files:
txtsim.mx
Log Message:
fixes to the m5 interface (strings are always passed as str *)
fixed stringdiff
Index: txtsim.mx
===================================================================
RCS file: /cvsroot/monetdb/MonetDB5/src/modules/mal/txtsim.mx,v
retrieving revision 1.1
retrieving revision 1.2
diff -u -d -r1.1 -r1.2
--- txtsim.mx 10 Dec 2007 21:23:59 -0000 1.1
+++ txtsim.mx 18 Dec 2007 22:34:17 -0000 1.2
@@ -55,6 +55,10 @@
address soundex_impl
comment "Soundex function for phonetic matching";
+command stringdiff(s1:str, s2:str) :int
+address stringdiff_impl
+comment "calculate the soundexed editdistance";
+
command qgramnormalize(input:str): str
address CMDqgramnormalize
comment "'Normalizes' strings (eg. toUpper and replaces non-alphanumerics with
one space";
@@ -90,21 +94,24 @@
#define txtsim_export extern
#endif
-txtsim_export int levenshtein_impl(int *result, str s, str t, int
*insdel_cost, int *replace_cost, int *transpose_cost);
-txtsim_export int levenshteinbasic_impl(int *result, str s, str t);
-txtsim_export int levenshteinbasic2_impl(int *result, str s, str t);
-txtsim_export int fstrcmp_impl(dbl *ret, str string1, str string2, dbl
*minimum);
-txtsim_export int fstrcmp0_impl(dbl *ret, str string1, str string2);
-txtsim_export int soundex_impl(str *res, str Name);
-txtsim_export int CMDqgramnormalize(str *res, str input);
-txtsim_export int CMDqgramselfjoin(BAT **res, BAT *qgram, BAT *id, BAT *pos,
BAT *len, flt *c, int *k);
+txtsim_export str levenshtein_impl(int *result, str *s, str *t, int
*insdel_cost, int *replace_cost, int *transpose_cost);
+txtsim_export str levenshteinbasic_impl(int *result, str *s, str *t);
+txtsim_export str levenshteinbasic2_impl(int *result, str *s, str *t);
+txtsim_export str fstrcmp_impl(dbl *ret, str *string1, str *string2, dbl
*minimum);
+txtsim_export str fstrcmp0_impl(dbl *ret, str *string1, str *string2);
+txtsim_export str soundex_impl(str *res, str *Name);
+txtsim_export str stringdiff_impl(int *res, str *s1, str*s2);
+txtsim_export str CMDqgramnormalize(str *res, str *input);
+txtsim_export str CMDqgramselfjoin(BAT **res, BAT *qgram, BAT *id, BAT *pos,
BAT *len, flt *c, int *k);
#endif /*_TXTSIM_H*/
@c
+#include "mal_config.h"
#include "txtsim.h"
+#include "mal_exception.h"
#define RETURN_NIL_IF(b,t) \
@@ -114,7 +121,7 @@
} else {\
memcpy(res, ATOMnilptr(t), ATOMsize(t));\
}\
- return GDK_SUCCEED; \
+ return MAL_SUCCEED; \
}
/* =========================================================================
@@ -169,9 +176,11 @@
/******************************
* Compute Levenshtein distance
*****************************/
-int
-levenshtein_impl(int *result, str s, str t, int *insdel_cost, int
*replace_cost, int *transpose_cost)
+str
+levenshtein_impl(int *result, str *S, str *T, int *insdel_cost, int
*replace_cost, int *transpose_cost)
{
+ char *s = *S;
+ char *t = *T;
int *d; /* pointer to matrix */
int n; /* length of s */
int m; /* length of t */
@@ -191,10 +200,12 @@
n = (int) strlen(s); /* 64bit: assume strings are less than 2 GB */
m = (int) strlen(t);
if (n == 0) {
- return m;
+ *result = m;
+ return MAL_SUCCEED;
}
if (m == 0) {
- return n;
+ *result = n;
+ return MAL_SUCCEED;
}
sz = (n + 1) * (m + 1) * sizeof(int);
d = (int *) GDKmalloc(sz);
@@ -250,19 +261,19 @@
/* Step 7 */
*result = levenshtein_GetAt(d, n, m, n);
GDKfree(d);
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
}
-int
-levenshteinbasic_impl(int *result, str s, str t)
+str
+levenshteinbasic_impl(int *result, str *s, str *t)
{
int insdel = 1, replace = 1, transpose = 2;
return levenshtein_impl(result, s, t, &insdel, &replace, &transpose);
}
-int
-levenshteinbasic2_impl(int *result, str s, str t)
+str
+levenshteinbasic2_impl(int *result, str *s, str *t)
{
int insdel = 1, replace = 1, transpose = 1;
@@ -332,19 +343,32 @@
}
-int
-soundex_impl(str *res, str Name)
+str
+soundex_impl(str *res, str *Name)
{
- RETURN_NIL_IF(strNil(Name), TYPE_str);
+ RETURN_NIL_IF(strNil(*Name), TYPE_str);
*res = (str) GDKmalloc(sizeof(char) * (SoundexLen + 1));
/* calculate Key for Name */
- soundex_code(Name, *res);
+ soundex_code(*Name, *res);
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
}
+str
+stringdiff_impl(int *res, str *s1, str *s2)
+{
+ str r = MAL_SUCCEED;
+ char *S1 = NULL, *S2 = NULL;
+
+ soundex_impl(&S1, s1);
+ soundex_impl(&S2, s2);
+ r = levenshteinbasic_impl(res, &S1, &S2);
+ GDKfree(S1);
+ GDKfree(S2);
+ return r;
+}
/******************************
* QGRAMNORMALIZE
@@ -357,9 +381,10 @@
* qgramnormalize(" '' t ' est").print(); --> [ "T EST" ]
*
*****************************/
-int
-CMDqgramnormalize(str *res, str input)
+str
+CMDqgramnormalize(str *res, str *Input)
{
+ char *input = *Input;
int i, j = 0;
char c, last = ' ';
@@ -381,8 +406,7 @@
while (j > 0 && (*res)[--j] == ' ')
(*res)[j] = 0;
- return GDK_SUCCEED;
-
+ return MAL_SUCCEED;
}
/* =========================================================================
@@ -855,9 +879,11 @@
strings are identical, and a number in between if they are
similar. */
-int
-fstrcmp_impl(dbl *ret, str string1, str string2, dbl *minimum)
+str
+fstrcmp_impl(dbl *ret, str *S1, str *S2, dbl *minimum)
{
+ char *string1 = *S1;
+ char *string2 = *S2;
int i;
size_t fdiag_len;
@@ -873,11 +899,11 @@
/* short-circuit obvious comparisons */
if (string[0].data_length == 0 && string[1].data_length == 0) {
*ret = 1.0;
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
}
if (string[0].data_length == 0 || string[1].data_length == 0) {
*ret = 0.0;
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
}
/* Set TOO_EXPENSIVE to be approximate square root of input size,
@@ -914,11 +940,11 @@
*ret = ((double)
(string[0].data_length + string[1].data_length -
string[1].edit_count - string[0].edit_count)
/ (string[0].data_length + string[1].data_length));
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
}
-int
-fstrcmp0_impl(dbl *ret, str string1, str string2)
+str
+fstrcmp0_impl(dbl *ret, str *string1, str *string2)
{
double min = 0.0;
@@ -928,19 +954,17 @@
/* ============ Q-GRAM SELF JOIN ============== */
-int
+str
CMDqgramselfjoin(BAT **res, BAT *qgram, BAT *id, BAT *pos, BAT *len, flt *c,
int *k)
{
size_t n = BATcount(qgram);
unsigned int i, j;
BAT *bn;
- oid *qbuf;
- int *ibuf, *pbuf, *lbuf;
- qbuf = (oid *) Tloc(qgram, BUNfirst(qgram));
- ibuf = (int *) Tloc(id, BUNfirst(id));
- pbuf = (int *) Tloc(pos, BUNfirst(pos));
- lbuf = (int *) Tloc(len, BUNfirst(len));
+ oid *qbuf = (oid *) Tloc(qgram, BUNfirst(qgram));
+ int *ibuf = (int *) Tloc(id, BUNfirst(id));
+ int *pbuf = (int *) Tloc(pos, BUNfirst(pos));
+ int *lbuf = (int *) Tloc(len, BUNfirst(len));
ERRORcheck((qgram->ttype != TYPE_oid), "CMDqgramselfjoin: tail of BAT
qgram must be oid.\n");
ERRORcheck((id->ttype != TYPE_int), "CMDqgramselfjoin: tail of BAT id
must be int.\n");
@@ -955,10 +979,10 @@
ERRORcheck((ALIGNsynced(qgram, len) == 0), "CMDqgramselfjoin: qgram and
len are not synced");
ERRORcheck((Tsize(qgram) != ATOMsize(qgram->ttype)), "CMDqgramselfjoin:
qgram is not a true void bat");
- ERRORcheck((Tsize(qgram) != ATOMsize(id->ttype)), "CMDqgramselfjoin: id
is not a true void bat");
+ ERRORcheck((Tsize(id) != ATOMsize(id->ttype)), "CMDqgramselfjoin: id is
not a true void bat");
ERRORcheck((Tsize(pos) != ATOMsize(pos->ttype)), "CMDqgramselfjoin: pos
is not a true void bat");
- ERRORcheck((Tsize(qgram) != ATOMsize(len->ttype)), "CMDqgramselfjoin:
len is not a true void bat");
+ ERRORcheck((Tsize(len) != ATOMsize(len->ttype)), "CMDqgramselfjoin: len
is not a true void bat");
*res = bn = BATnew(TYPE_int, TYPE_int, n);
@@ -972,30 +996,16 @@
bn->hsorted = bn->tsorted = 0;
- return GDK_SUCCEED;
+ return MAL_SUCCEED;
bunins_failed:
- GDKerror("CMDqgramselfjoin: could not realloc\n");
BBPreclaim(bn);
- return GDK_FAIL;
+ throw(MAL, "txtsim.qgramselfjoin", "could not realloc\n");
}
@mal
#mil implementation
-#proc stringdiff( str s1, str s1 ) : int {
-# var sd1 := soundex(s1);
-# var sd2 := soundex(s2);
-# return editdistance(sd1,sd2);
-#}
-
-#mal implementation
-function stringdiff( s1:str, s2:str):int;
- sd1 := soundex(s1);
- sd2 := soundex(s2);
- return editdistance(sd1,sd2);
-end stringdiff;
-
#mil implementation
#proc str2qgrams(str s) : bat[oid,str]
#{
-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
_______________________________________________
Monetdb-checkins mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/monetdb-checkins
