Changeset: eb01f82e8f44 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/eb01f82e8f44
Modified Files:
monetdb5/modules/atoms/str.c
monetdb5/modules/kernel/batstr.c
Branch: default
Log Message:
Use more functions from utf8.h library.
This also involves fixing str_strlen to obey its comment: no nil argument.
diffs (255 lines):
diff --git a/monetdb5/modules/atoms/str.c b/monetdb5/modules/atoms/str.c
--- a/monetdb5/modules/atoms/str.c
+++ b/monetdb5/modules/atoms/str.c
@@ -3028,29 +3028,7 @@ STRepilogue(void *ret)
static inline void
UTF8_assert(const char *s)
{
- int c;
-
- if (s == NULL)
- return;
- if (*s == '\200' && s[1] == '\0')
- return; /* str_nil */
- while ((c = *s++) != '\0') {
- if ((c & 0x80) == 0)
- continue;
- if ((*s++ & 0xC0) != 0x80)
- assert(0);
- if ((c & 0xE0) == 0xC0)
- continue;
- if ((*s++ & 0xC0) != 0x80)
- assert(0);
- if ((c & 0xF0) == 0xE0)
- continue;
- if ((*s++ & 0xC0) != 0x80)
- assert(0);
- if ((c & 0xF8) == 0xF0)
- continue;
- assert(0);
- }
+ assert(strNil(s) || utf8valid(s) == 0);
}
#else
#define UTF8_assert(s) ((void) 0)
@@ -3060,19 +3038,12 @@ UTF8_assert(const char *s)
static inline int
UTF8_strpos(const char *s, const char *end)
{
- int pos = 0;
-
UTF8_assert(s);
if (s > end) {
return -1;
}
- while (s < end) {
- /* just count leading bytes of encoded code points; only works
- * for correctly encoded UTF-8 */
- pos += (*s++ & 0xC0) != 0x80;
- }
- return pos;
+ return (int) utf8nlen(s, (size_t) (end - s));
}
/* return a pointer to the byte that starts the pos'th (0-based)
@@ -3127,27 +3098,20 @@ UTF8_strncpy(char *restrict dst, const c
int
UTF8_strlen(const char *s)
{ /* This
function assumes s is never nil */
- size_t pos = 0;
-
UTF8_assert(s);
assert(!strNil(s));
- while (*s) {
- /* just count leading bytes of encoded code points; only works
- * for correctly encoded UTF-8 */
- pos += (*s++ & 0xC0) != 0x80;
- }
- assert(pos < INT_MAX);
- return (int) pos;
+ return (int) utf8len(s);
}
/* return (int) strlen(s); s is not nil */
int
str_strlen(const char *s)
{ /* This
function assumes s is never nil */
- size_t pos = strlen(s);
- assert(pos < INT_MAX);
- return (int) pos;
+ UTF8_assert(s);
+ assert(!strNil(s));
+
+ return (int) strlen(s);
}
/* return the display width of s */
@@ -3791,11 +3755,15 @@ STRstartswith(Client cntxt, MalBlkPtr mb
bit icase = pci->argc == 4
&& *getArgReference_bit(stk, pci, 3) ? true : false;
str s = *arg1, prefix = *arg2, msg = MAL_SUCCEED;
- int plen = str_strlen(prefix);
-
- *res = (strNil(s) || strNil(prefix)) ? bit_nil :
- icase ? str_is_iprefix(s, prefix, plen) :
str_is_prefix(s, prefix,
-
plen);
+ if (strNil(s) || strNil(prefix)) {
+ *res = bit_nil;
+ } else {
+ int plen = str_strlen(prefix);
+
+ *res = icase ?
+ str_is_iprefix(s, prefix, plen) :
+ str_is_prefix(s, prefix, plen);
+ }
return msg;
}
@@ -3834,11 +3802,15 @@ STRendswith(Client cntxt, MalBlkPtr mb,
bit icase = pci->argc == 4
&& *getArgReference_bit(stk, pci, 3) ? true : false;
str s = *arg1, suffix = *arg2, msg = MAL_SUCCEED;
- int sul = str_strlen(suffix);
-
- *res = (strNil(s) || strNil(suffix)) ? bit_nil :
- icase ? str_is_isuffix(s, suffix, sul) :
str_is_suffix(s, suffix,
-
sul);
+ if (strNil(s) || strNil(suffix)) {
+ *res = bit_nil;
+ } else {
+ int sul = str_strlen(suffix);
+
+ *res = icase ?
+ str_is_isuffix(s, suffix, sul) :
+ str_is_suffix(s, suffix, sul);
+ }
return msg;
}
@@ -3876,12 +3848,15 @@ STRcontains(Client cntxt, MalBlkPtr mb,
bit icase = pci->argc == 4
&& *getArgReference_bit(stk, pci, 3) ? true : false;
str haystack = *arg1, needle = *arg2, msg = MAL_SUCCEED;
- int needle_len = str_strlen(needle);
-
- *res = (strNil(haystack) || strNil(needle)) ? bit_nil :
- icase ? str_icontains(haystack, needle,
- needle_len) :
str_contains(haystack, needle,
-
needle_len);
+ if (strNil(haystack) || strNil(needle)) {
+ *res = bit_nil;
+ } else {
+ int needle_len = str_strlen(needle);
+
+ *res = icase ?
+ str_icontains(haystack, needle, needle_len) :
+ str_contains(haystack, needle, needle_len);
+ }
return msg;
}
@@ -3919,11 +3894,15 @@ STRstr_search(Client cntxt, MalBlkPtr mb
bit icase = pci->argc == 4
&& *getArgReference_bit(stk, pci, 3) ? true : false;
str s = *haystack, h = *needle, msg = MAL_SUCCEED;
- int needle_len = str_strlen(h);
-
- *res = (strNil(s) || strNil(h)) ? bit_nil :
- icase ? str_isearch(s, h, needle_len) : str_search(s, h,
-
needle_len);
+ if (strNil(s) || strNil(h)) {
+ *res = bit_nil;
+ } else {
+ int needle_len = str_strlen(h);
+
+ *res = icase ?
+ str_isearch(s, h, needle_len) :
+ str_search(s, h, needle_len);
+ }
return msg;
}
@@ -3977,12 +3956,15 @@ STRrevstr_search(Client cntxt, MalBlkPtr
bit icase = pci->argc == 4
&& *getArgReference_bit(stk, pci, 3) ? true : false;
str s = *haystack, h = *needle, msg = MAL_SUCCEED;
- int needle_len = str_strlen(h);
-
- *res = (strNil(s) || strNil(h)) ? bit_nil :
- icase ? str_reverse_str_isearch(s, h,
-
needle_len) :
+ if (strNil(s) || strNil(h)) {
+ *res = bit_nil;
+ } else {
+ int needle_len = str_strlen(h);
+
+ *res = icase ?
+ str_reverse_str_isearch(s, h, needle_len) :
str_reverse_str_search(s, h, needle_len);
+ }
return msg;
}
@@ -4902,8 +4884,9 @@ STRlocate3(int *ret, const str *needle,
const char *s = *needle, *s2 = *haystack;
int st = *start;
- *ret = (strNil(s) || strNil(s2)
- || is_int_nil(st)) ? int_nil : str_locate2(s, s2, st);
+ *ret = (strNil(s) || strNil(s2) || is_int_nil(st)) ?
+ int_nil :
+ str_locate2(s, s2, st);
return MAL_SUCCEED;
}
@@ -5163,6 +5146,9 @@ do_string_select(BAT *bn, BAT *b, BAT *s
bit (*str_cmp)(const char *, const char *,
int),
bool keep_nulls)
{
+ if (strNil(key))
+ return MAL_SUCCEED;
+
BATiter bi = bat_iterator(b);
BUN cnt = 0, ncands = ci->ncand;
oid off = b->hseqbase, *restrict vals = Tloc(bn, 0);
@@ -5370,6 +5356,8 @@ STRcontainsselect(Client cntxt, MalBlkPt
GDK_CHECK_TIMEOUT(qry_ctx, counter,
GOTO_LABEL_TIMEOUT_HANDLER(exit, qry_ctx)); \
ro = canditer_next(&rci);
\
vr = VALUE(r, ro - rbase);
\
+ if (strNil(vr))
\
+ continue;
\
rlen = STR_LEN;
\
nl = 0;
\
if (with_strimps)
\
@@ -5443,6 +5431,8 @@ STRcontainsselect(Client cntxt, MalBlkPt
GDK_CHECK_TIMEOUT(qry_ctx, counter,
GOTO_LABEL_TIMEOUT_HANDLER(exit, qry_ctx)); \
ro = canditer_next(&rci);
\
vr = VALUE(r, ro - rbase);
\
+ if (strNil(vr))
\
+ continue;
\
rlen = STR_LEN;
\
nl = 0;
\
canditer_init(&lci, l, sl);
\
diff --git a/monetdb5/modules/kernel/batstr.c b/monetdb5/modules/kernel/batstr.c
--- a/monetdb5/modules/kernel/batstr.c
+++ b/monetdb5/modules/kernel/batstr.c
@@ -1980,7 +1980,7 @@ prefix_or_suffix_cst(Client cntxt, MalBl
bi = bat_iterator(b);
vals = Tloc(bn, 0);
ynil = strNil(y);
- ylen = str_strlen(y);
+ ylen = ynil ? 0 : str_strlen(y); /* not used if nil */
if (ci1.tpe == cand_dense) {
for (BUN i = 0; i < ci1.ncand; i++) {
oid p1 = (canditer_next_dense(&ci1) - off1);
@@ -2356,7 +2356,7 @@ search_string_bat_cst(Client cntxt, MalB
bi = bat_iterator(b);
vals = Tloc(bn, 0);
ynil = strNil(y);
- ylen = str_strlen(y);
+ ylen = ynil ? 0 : str_strlen(y); /* not used if nil */
if (ci1.tpe == cand_dense) {
for (BUN i = 0; i < ci1.ncand; i++) {
oid p1 = (canditer_next_dense(&ci1) - off1);
_______________________________________________
checkin-list mailing list -- [email protected]
To unsubscribe send an email to [email protected]
