Changeset: d2b933f5fa50 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/d2b933f5fa50
Modified Files:
sql/backends/monet5/copy.c
sql/backends/monet5/copy.h
sql/backends/monet5/copy_convert.c
sql/backends/monet5/copy_convert_num.h
sql/backends/monet5/rel_copy.c
Branch: copyparpipe
Log Message:
Add specialized integer parsing
diffs (258 lines):
diff --git a/sql/backends/monet5/copy.c b/sql/backends/monet5/copy.c
--- a/sql/backends/monet5/copy.c
+++ b/sql/backends/monet5/copy.c
@@ -968,6 +968,30 @@ static mel_func copy_init_funcs[] = {
batarg("block", bte), batarg("offsets", int), arg("digits", int),
arg("scale", int), arg("type", hge)
)),
#endif
+
+ command("copy", "parse_integer", COPYparse_integer_bte, false, "Parse as an
integer", args(1, 4,
+ batarg("", bte),
+ batarg("block", bte), batarg("offsets", int), arg("type", bte)
+ )),
+ command("copy", "parse_integer", COPYparse_integer_sht, false, "Parse as an
integer", args(1, 4,
+ batarg("", sht),
+ batarg("block", bte), batarg("offsets", int), arg("type", sht)
+ )),
+ command("copy", "parse_integer", COPYparse_integer_int, false, "Parse as an
integer", args(1, 4,
+ batarg("", int),
+ batarg("block", bte), batarg("offsets", int), arg("type", int)
+ )),
+ command("copy", "parse_integer", COPYparse_integer_lng, false, "Parse as an
integer", args(1, 4,
+ batarg("", lng),
+ batarg("block", bte), batarg("offsets", int), arg("type", lng)
+ )),
+ #ifdef HAVE_HGE
+ command("copy", "parse_integer", COPYparse_integer_hge, false, "Parse as an
integer", args(1, 4,
+ batarg("", hge),
+ batarg("block", bte), batarg("offsets", int), arg("type", hge)
+ )),
+#endif
+
command("copy", "set_blocksize", COPYset_blocksize, true, "set the COPY block
size", args(1, 2,
arg("blocksize", int)
)),
diff --git a/sql/backends/monet5/copy.h b/sql/backends/monet5/copy.h
--- a/sql/backends/monet5/copy.h
+++ b/sql/backends/monet5/copy.h
@@ -33,6 +33,13 @@ str parse_fixed_width_column(bat *ret, c
extern str COPYparse_generic(Client cntxt, MalBlkPtr mb, MalStkPtr stk,
InstrPtr pci);
+extern str COPYparse_integer_bte(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *dummy);
+extern str COPYparse_integer_sht(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *dummy);
+extern str COPYparse_integer_int(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *dummy);
+extern str COPYparse_integer_lng(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *dummy);
+#ifdef HAVE_HGE
+extern str COPYparse_integer_hge(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *dummy);
+#endif
extern str COPYparse_decimal_bte(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *digits_p, int *scale_p, int *dummy);
extern str COPYparse_decimal_sht(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *digits_p, int *scale_p, int *dummy);
extern str COPYparse_decimal_int(bat *parsed_bat_id, bat *block_bat_id, bat
*offsets_bat_id, int *digits_p, int *scale_p, int *dummy);
@@ -48,4 +55,15 @@ struct decimal_parms {
};
+#ifdef __GNUC__
+/* __builtin_expect returns its first argument; it is expected to be
+ * equal to the second argument */
+#define unlikely(expr) __builtin_expect((expr) != 0, 0)
+#define likely(expr) __builtin_expect((expr) != 0, 1)
+#else
+#define unlikely(expr) (expr)
+#define likely(expr) (expr)
+#endif
+
+
#endif /*_COPY_H_*/
diff --git a/sql/backends/monet5/copy_convert.c
b/sql/backends/monet5/copy_convert.c
--- a/sql/backends/monet5/copy_convert.c
+++ b/sql/backends/monet5/copy_convert.c
@@ -132,23 +132,33 @@ end:
}
#define TMPL_TYPE bte
+#define TMPL_NIL bte_nil
+#define TMPL_MAX GDK_bte_max
#define TMPL_SUFFIXED(s) s##_bte
#include "copy_convert_num.h"
#define TMPL_TYPE sht
+#define TMPL_NIL sht_nil
+#define TMPL_MAX GDK_sht_max
#define TMPL_SUFFIXED(s) s##_sht
#include "copy_convert_num.h"
#define TMPL_TYPE int
+#define TMPL_NIL int_nil
+#define TMPL_MAX GDK_int_max
#define TMPL_SUFFIXED(s) s##_int
#include "copy_convert_num.h"
#define TMPL_TYPE lng
+#define TMPL_NIL lng_nil
+#define TMPL_MAX GDK_lng_max
#define TMPL_SUFFIXED(s) s##_lng
#include "copy_convert_num.h"
#ifdef HAVE_HGE
#define TMPL_TYPE hge
+#define TMPL_NIL hge_nil
+#define TMPL_MAX GDK_hge_max
#define TMPL_SUFFIXED(s) s##_hge
#include "copy_convert_num.h"
#endif
diff --git a/sql/backends/monet5/copy_convert_num.h
b/sql/backends/monet5/copy_convert_num.h
--- a/sql/backends/monet5/copy_convert_num.h
+++ b/sql/backends/monet5/copy_convert_num.h
@@ -1,14 +1,64 @@
-#if !defined(TMPL_TYPE) || !defined(TMPL_SUFFIXED)
+#if !defined(TMPL_TYPE) || !defined(TMPL_SUFFIXED) || !defined TMPL_NIL ||
!defined(TMPL_MAX)
#error "This file is a template, it cannot be included standalone"
#endif
+static TMPL_TYPE
+TMPL_SUFFIXED(parse_one_integer) (struct error_handling *errors, int rel_row,
const char *value)
+{
+ bool pos = true;
+ TMPL_TYPE acc = 0;
+ const char *s = value;
+
+ while(isspace((unsigned char) *s))
+ s++;
+
+ if (*s == '-') {
+ pos = false;
+ s++;
+ } else if (*s == '+') {
+ s++;
+ }
+
+ while (isdigit((unsigned char) *s)) {
+ // int is safe because of promotion rules
+ int digit = *s - '0';
+ if (unlikely(acc >= (TMPL_MAX / 10))) {
+ copy_report_error(errors, rel_row, "overflow: %s",
value);
+ return TMPL_NIL;
+ }
+ TMPL_TYPE new_acc = 10 * acc + digit;
+ acc = new_acc;
+ s++;
+ }
+
+ if (*s == '.') {
+ s++;
+ while (*s == '0')
+ s++;
+ }
+
+ while (isspace((unsigned char) *s))
+ s++;
+
+ if (*s != '\0') {
+ copy_report_error(errors, rel_row, "trailing garbage: %s", s);
+ acc = TMPL_NIL;
+ }
+
+ if (!pos)
+ acc = -acc;
+
+ return acc;
+}
+
static TMPL_TYPE
-TMPL_SUFFIXED(parse_one_decimal) (struct error_handling *errors, struct
decimal_parms *parms, int rel_row, const char *s)
+TMPL_SUFFIXED(parse_one_decimal) (struct error_handling *errors, struct
decimal_parms *parms, int rel_row, const char *value)
{
+ const char *s = value;
int digits = parms->digits;
int scale = parms->scale;
int integer_digits = digits - scale;
@@ -49,7 +99,7 @@ TMPL_SUFFIXED(parse_one_decimal) (struct
}
if (*s) {
copy_report_error(errors, rel_row, "trailing garbage: %s", s);
- res = int_nil;
+ res = TMPL_NIL;
}
if (neg)
res = -res;
@@ -59,7 +109,7 @@ TMPL_SUFFIXED(parse_one_decimal) (struct
static str
-TMPL_SUFFIXED(parse_many_decimal) (struct error_handling *errors, void
*parms_, int count, void *dest_, char *data, int *offsets)
+TMPL_SUFFIXED(parse_many_decimals) (struct error_handling *errors, void
*parms_, int count, void *dest_, char *data, int *offsets)
{
struct decimal_parms *parms = parms_;
TMPL_TYPE *dest = dest_;
@@ -76,6 +126,24 @@ TMPL_SUFFIXED(parse_many_decimal) (struc
return MAL_SUCCEED;
}
+static str
+TMPL_SUFFIXED(parse_many_integers) (struct error_handling *errors, void
*parms, int count, void *dest_, char *data, int *offsets)
+{
+ (void)parms;
+ TMPL_TYPE *dest = dest_;
+
+ for (int i = 0; i < count; i++) {
+ int offset = offsets[i];
+ if (is_int_nil(offset)) {
+ dest[i] = int_nil;
+ continue;
+ }
+ dest[i] = TMPL_SUFFIXED(parse_one_integer)(errors, i, data +
offset);
+ }
+
+ return MAL_SUCCEED;
+}
+
str
TMPL_SUFFIXED(COPYparse_decimal) (
@@ -91,9 +159,23 @@ TMPL_SUFFIXED(COPYparse_decimal) (
return parse_fixed_width_column(
parsed_bat_id, "copy.parse_decimal",
*block_bat_id, *offsets_bat_id,
- TMPL_SUFFIXED(TYPE), TMPL_SUFFIXED(parse_many_decimal),
&myparms);
+ TMPL_SUFFIXED(TYPE), TMPL_SUFFIXED(parse_many_decimals),
&myparms);
+}
+
+str
+TMPL_SUFFIXED(COPYparse_integer) (
+ bat *parsed_bat_id,
+ bat *block_bat_id, bat *offsets_bat_id,
+ int *dummy)
+{
+ return parse_fixed_width_column(
+ parsed_bat_id, "copy.parse_integer",
+ *block_bat_id, *offsets_bat_id,
+ TMPL_SUFFIXED(TYPE), TMPL_SUFFIXED(parse_many_integers), NULL);
}
#undef TMPL_TYPE
+#undef TMPL_NIL
+#undef TMPL_MAX
#undef TMPL_SUFFIXED
diff --git a/sql/backends/monet5/rel_copy.c b/sql/backends/monet5/rel_copy.c
--- a/sql/backends/monet5/rel_copy.c
+++ b/sql/backends/monet5/rel_copy.c
@@ -298,6 +298,12 @@ rel2bin_copyparpipe(backend *be, sql_rel
const char *column_name = col->base.name;
switch (type->eclass) {
+ case EC_NUM:
+ q = newStmt(mb, "copy", "parse_integer");
+ q = pushArgument(mb, q, loop_vars.our_block);
+ q = pushArgument(mb, q, var_indices);
+ q = pushNil(mb, q, col->type.type->localtype);
+ break;
case EC_DEC:
q = newStmt(mb, "copy", "parse_decimal");
q = pushArgument(mb, q, loop_vars.our_block);
_______________________________________________
checkin-list mailing list -- [email protected]
To unsubscribe send an email to [email protected]
