Changeset: 9d1d04ffa824 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=9d1d04ffa824
Modified Files:
monetdb5/modules/atoms/str.c
monetdb5/modules/atoms/str.h
monetdb5/modules/kernel/batstr.c
Branch: alloc-less-str
Log Message:
Cleaned unicode conversion string functions
diffs (truncated from 423 to 300 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
@@ -3499,51 +3499,65 @@ STRSubString(str *res, const str *arg1,
}
str
+str_from_wchr(str *buf, int *buflen, int c)
+{
+ if (is_int_nil(c)) {
+ strcpy(*buf, str_nil);
+ return MAL_SUCCEED;
+ }
+ CHECK_BUFFER_LENGTH(buf, buflen, 5, "str.unicode");
+ str s = *buf;
+ UTF8_PUTCHAR(c, s);
+ *s = 0;
+ return MAL_SUCCEED;
+illegal:
+ throw(MAL, "str.unicode", SQLSTATE(42000) "Illegal Unicode code point");
+}
+
+str
STRFromWChr(str *res, const int *c)
{
- str s;
+ int buflen = INITIAL_STR_BUFFER_LENGTH;
+ str buf = GDKmalloc(buflen), msg;
- if (is_int_nil(*c)) {
- *res = GDKstrdup(str_nil);
- if (*res == NULL)
- throw(MAL, "str.unicode", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
- return MAL_SUCCEED;
+ *res = NULL;
+ if (!buf)
+ throw(SQL, "str.unicode", SQLSTATE(HY013) MAL_MALLOC_FAIL);
+ msg = str_from_wchr(&buf, &buflen, *c);
+ if (!msg && !(*res = GDKstrdup(buf))) {
+ msg = createException(MAL, "str.unicode", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
}
- s = *res = GDKmalloc(5);
- if (*res == NULL)
- throw(MAL, "str.unicode", SQLSTATE(HY013) MAL_MALLOC_FAIL);
- UTF8_PUTCHAR(*c, s);
- *s = 0;
- return MAL_SUCCEED;
- illegal:
- GDKfree(*res);
- *res = NULL;
- throw(MAL, "str.unicode", SQLSTATE(42000) "Illegal Unicode code point");
+ GDKfree(buf);
+ return msg;
}
/* return the Unicode code point of arg1 at position at */
str
-STRWChrAt(int *res, const str *arg1, const int *at)
+str_wchr_at(int *res, const char *s, int at)
{
-/* 64bit: should have lng arg */
- const char *s = *arg1;
-
- if (strNil(s) || is_int_nil(*at) || *at < 0) {
+ /* 64bit: should have lng arg */
+ if (strNil(s) || is_int_nil(at) || at < 0) {
*res = int_nil;
return MAL_SUCCEED;
}
- s = UTF8_strtail(s, *at);
+ s = UTF8_strtail(s, at);
if (s == NULL || *s == 0) {
*res = int_nil;
return MAL_SUCCEED;
}
UTF8_GETCHAR(*res, s);
return MAL_SUCCEED;
- illegal:
+illegal:
throw(MAL, "str.unicodeAt", SQLSTATE(42000) "Illegal Unicode code
point");
}
+str
+STRWChrAt(int *res, const str *arg1, const int *at)
+{
+ return str_wchr_at(res, *arg1, *at);
+}
+
/* returns whether arg1 starts with arg2 */
str
STRPrefix(bit *res, const str *arg1, const str *arg2)
diff --git a/monetdb5/modules/atoms/str.h b/monetdb5/modules/atoms/str.h
--- a/monetdb5/modules/atoms/str.h
+++ b/monetdb5/modules/atoms/str.h
@@ -18,6 +18,9 @@
int str_length(const char *s);
int str_bytes(const char *s);
+str str_from_wchr(str *buf, int *buflen, int c);
+str str_wchr_at(int *res, const char *s, int at);
+
str str_tail(str *buf, int *buflen, const char *s, int off);
str str_Sub_String(str *buf, int *buflen, const char *s, int off, int l);
str str_substring_tail(str *buf, int *buflen, const char *s, int start);
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
@@ -169,85 +169,98 @@ bunins_failed:
}
static str
-STRbatFromWChr(bat *ret, const bat *l)
+STRbatFromWChr(bat *res, const bat *l)
{
- BAT *bn, *b;
- BUN q;
- str y;
- int *restrict input;
- str msg = MAL_SUCCEED;
-
- prepareOperand(b, l, "batstr.unicode");
- prepareResult(bn, b, TYPE_str, "batstr.unicode");
- input = Tloc(b, 0);
- q = BATcount(b);
+ BAT *bn = NULL, *b = NULL;
+ BUN p, q;
+ int buflen = INITIAL_STR_BUFFER_LENGTH, *restrict vals;
+ str buf = GDKmalloc(buflen), msg = MAL_SUCCEED;
+ bool nils = false;
- for (BUN p = 0 ; p < q; p++) {
- int x = input[p];
- gdk_return res;
+ if ((b = BATdescriptor(*l)) == NULL) {
+ msg = createException(MAL, "batstr.unicode", SQLSTATE(HY005)
RUNTIME_OBJECT_MISSING);
+ goto bailout;
+ }
+ q = BATcount(b);
+ if (!(bn = COLnew(b->hseqbase, TYPE_str, q, TRANSIENT))) {
+ msg = createException(MAL, "batstr.unicode", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
+ goto bailout;
+ }
- if ((msg = STRFromWChr(&y, &x)) != MAL_SUCCEED)
+ vals = Tloc(b, 0);
+ for (p = 0; p < q ; p++) {
+ if ((msg = str_from_wchr(&buf, &buflen, vals[p])) !=
MAL_SUCCEED)
goto bailout;
- res = bunfastappVAR(bn, y);
- GDKfree(y);
- if (res != GDK_SUCCEED)
+ if (tfastins_nocheckVAR(bn, p, buf, Tsize(bn)) != GDK_SUCCEED) {
+ msg = createException(MAL, "batstr.unicode",
SQLSTATE(HY013) MAL_MALLOC_FAIL);
goto bailout;
+ }
+ nils |= strNil(buf);
}
- bn->tnonil = b->tnonil;
- bn->tnil = b->tnonil;
- finalizeResult(ret, bn, b);
- return MAL_SUCCEED;
bailout:
- BBPunfix(b->batCacheid);
- BBPunfix(bn->batCacheid);
- if (msg != MAL_SUCCEED)
- return msg;
- throw(MAL, "batstr.unicode", OPERATION_FAILED " During bulk operation");
+ GDKfree(buf);
+ if (b)
+ BBPunfix(b->batCacheid);
+ if (bn && !msg) {
+ BATsetcount(bn, q);
+ bn->tnil = nils;
+ bn->tnonil = !nils;
+ bn->tkey = BATcount(bn) <= 1;
+ bn->tsorted = BATcount(bn) <= 1;
+ bn->trevsorted = BATcount(bn) <= 1;
+ BBPkeepref(*res = bn->batCacheid);
+ } else if (bn)
+ BBPreclaim(bn);
+ return msg;
}
static str
-STRbatSpace(bat *ret, const bat *l)
+STRbatSpace(bat *res, const bat *l)
{
- BAT *bn, *b;
- BUN q;
- str y, msg = MAL_SUCCEED;
- int *restrict input;
+ BAT *bn = NULL, *b = NULL;
+ BUN p, q;
+ int buflen = INITIAL_STR_BUFFER_LENGTH, *restrict vals;
+ str buf = GDKmalloc(buflen), msg = MAL_SUCCEED;
+ bool nils = false;
- prepareOperand(b, l, "batstr.Space");
- prepareResult(bn, b, TYPE_str, "batstr.Space");
- bn->tnonil=true;
- bn->tnil=false;
-
- input = Tloc(b, 0);
+ if ((b = BATdescriptor(*l)) == NULL) {
+ msg = createException(MAL, "batstr.space", SQLSTATE(HY005)
RUNTIME_OBJECT_MISSING);
+ goto bailout;
+ }
q = BATcount(b);
+ if (!(bn = COLnew(b->hseqbase, TYPE_str, q, TRANSIENT))) {
+ msg = createException(MAL, "batstr.space", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
+ goto bailout;
+ }
- for (BUN p = 0; p < q; p++) {
- if ((msg = STRspace(&y, &input[p])) != MAL_SUCCEED)
+ vals = Tloc(b, 0);
+ for (p = 0; p < q ; p++) {
+ char space[]= " ", *s = space;
+
+ if ((msg = str_repeat(&buf, &buflen, s, vals[p])) !=
MAL_SUCCEED)
goto bailout;
- if (bunfastappVAR(bn, y) != GDK_SUCCEED) {
- GDKfree(y);
+ if (tfastins_nocheckVAR(bn, p, buf, Tsize(bn)) != GDK_SUCCEED) {
+ msg = createException(MAL, "batstr.space",
SQLSTATE(HY013) MAL_MALLOC_FAIL);
goto bailout;
}
- if (strNil(y)) {
- bn->tnonil = false;
- bn->tnil = true;
- }
- GDKfree(y);
+ nils |= strNil(buf);
}
bailout:
- if (msg) {
- BBPreclaim(bn);
- } else {
+ GDKfree(buf);
+ if (b)
+ BBPunfix(b->batCacheid);
+ if (bn && !msg) {
BATsetcount(bn, q);
+ bn->tnil = nils;
+ bn->tnonil = !nils;
bn->tkey = BATcount(bn) <= 1;
bn->tsorted = BATcount(bn) <= 1;
bn->trevsorted = BATcount(bn) <= 1;
- BBPkeepref(*ret = bn->batCacheid);
- }
- if (b)
- BBPunfix(b->batCacheid);
+ BBPkeepref(*res = bn->batCacheid);
+ } else if (bn)
+ BBPreclaim(bn);
return msg;
}
@@ -1130,66 +1143,114 @@ STRbatRstrSearchcst(bat *ret, const bat
}
static str
-STRbatWChrAt(bat *ret, const bat *l, const bat *r)
+STRbatWChrAt(bat *res, const bat *l, const bat *r)
{
- BATiter lefti, righti;
- BAT *bn, *left, *right;
- BUN p,q;
- int v;
+ BATiter lefti;
+ BAT *bn = NULL, *left = NULL, *right = NULL;
+ BUN p, q;
+ int buflen = INITIAL_STR_BUFFER_LENGTH, *restrict righti, *restrict
vals, next;
+ str x, buf = GDKmalloc(buflen), msg = MAL_SUCCEED;
+ bool nils = false;
- prepareOperand2(left,l,right,r,"batstr.unicodeAt");
- if(BATcount(left) != BATcount(right)) {
- BBPunfix(left->batCacheid);
- BBPunfix(right->batCacheid);
- throw(MAL, "batstr.unicodeAt", ILLEGAL_ARGUMENT " Requires bats
of identical size");
+ if (!buf) {
+ msg = createException(MAL, "batstr.unicodeAt", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
+ goto bailout;
+ }
+ if (!(left = BATdescriptor(*l)) || !(right = BATdescriptor(*r))) {
+ msg = createException(MAL, "batstr.unicodeAt", SQLSTATE(HY005)
RUNTIME_OBJECT_MISSING);
+ goto bailout;
}
- prepareResult2(bn,left,right,TYPE_int,"batstr.unicodeAt");
+ if (BATcount(left) != BATcount(right)) {
+ msg = createException(MAL, "batstr.unicodeAt", ILLEGAL_ARGUMENT
" Requires bats of identical size");
+ goto bailout;
+ }
+ q = BATcount(left);
+ if (!(bn = COLnew(left->hseqbase, TYPE_int, q, TRANSIENT))) {
+ msg = createException(MAL, "batstr.unicodeAt", SQLSTATE(HY013)
MAL_MALLOC_FAIL);
_______________________________________________
checkin-list mailing list
[email protected]
https://www.monetdb.org/mailman/listinfo/checkin-list
