Skip to content

Feature: extend int128 to numeric interface #381

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
jiaqizho wants to merge 1 commit into from
Closed

Feature: extend int128 to numeric interface #381

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
143 changes: 132 additions & 11 deletions src/backend/utils/adt/numeric.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -557,7 +557,8 @@ static void int64_to_numericvar(int64 val, NumericVar *var);
static bool numericvar_to_uint64(const NumericVar *var, uint64 *result);
#ifdef HAVE_INT128
static bool numericvar_to_int128(const NumericVar *var, int128 *result);
static void int128_to_numericvar(int128 val, NumericVar *var);
static void int128_to_numericvar_zero_scale(int128 val, NumericVar *var);
static void int128_to_numericvar_with_scale(int128 val, NumericVar *var, bool neg, int scale);
#endif
static double numericvar_to_double_no_overflow(const NumericVar *var);

Expand Down Expand Up @@ -4423,6 +4424,41 @@ int64_to_numeric(int64 val)
return res;
}

#ifdef HAVE_INT128
Numeric
int128_to_numeric(int128 val, int scale, bool neg,
Copy link
Copy Markdown
Contributor

@my-ship-it my-ship-it Mar 5, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No need to change in numeric.c?

bool nan, bool pinf, bool ninf)
{
Numeric res;
NumericVar result;

/* Make sure the parameter passed in is already a positive integer */
Assert(val >= 0);
AssertImply(nan, !(pinf || ninf));
AssertImply(pinf, !(nan || ninf));
AssertImply(ninf, !(nan || pinf));

if (nan)
{
res = make_result(&const_nan);
} else if (pinf) {
res = make_result(&const_pinf);
} else if (ninf) {
res = make_result(&const_ninf);
} else {
init_var(&result);

int128_to_numericvar_with_scale(val, &result, neg, scale);

res = make_result(&result);

free_var(&result);
}

return res;
}
#endif

/*
* Convert val1/(10**val2) to numeric. This is much faster than normal
* numeric division.
Expand Down Expand Up @@ -5858,7 +5894,7 @@ numeric_poly_serialize(PG_FUNCTION_ARGS)
init_var(&num);

#ifdef HAVE_INT128
int128_to_numericvar(state->sumX, &num);
int128_to_numericvar_zero_scale(state->sumX, &num);
#else
accum_sum_final(&state->sumX, &num);
#endif
Expand All @@ -5867,7 +5903,7 @@ numeric_poly_serialize(PG_FUNCTION_ARGS)
sumX = DatumGetByteaPP(temp);

#ifdef HAVE_INT128
int128_to_numericvar(state->sumX2, &num);
int128_to_numericvar_zero_scale(state->sumX2, &num);
#else
accum_sum_final(&state->sumX2, &num);
#endif
Expand Down Expand Up @@ -6089,7 +6125,7 @@ int8_avg_serialize(PG_FUNCTION_ARGS)
init_var(&num);

#ifdef HAVE_INT128
int128_to_numericvar(state->sumX, &num);
int128_to_numericvar_zero_scale(state->sumX, &num);
#else
accum_sum_final(&state->sumX, &num);
#endif
Expand Down Expand Up @@ -6281,7 +6317,7 @@ numeric_poly_sum(PG_FUNCTION_ARGS)

init_var(&result);

int128_to_numericvar(state->sumX, &result);
int128_to_numericvar_zero_scale(state->sumX, &result);

res = make_result(&result);

Expand Down Expand Up @@ -6310,7 +6346,7 @@ numeric_poly_avg(PG_FUNCTION_ARGS)

init_var(&result);

int128_to_numericvar(state->sumX, &result);
int128_to_numericvar_zero_scale(state->sumX, &result);

countd = NumericGetDatum(int64_to_numeric(state->N));
sumd = NumericGetDatum(make_result(&result));
Expand Down Expand Up @@ -6573,10 +6609,10 @@ numeric_poly_stddev_internal(Int128AggState *state,

init_var(&tmp_var);

int128_to_numericvar(state->sumX, &tmp_var);
int128_to_numericvar_zero_scale(state->sumX, &tmp_var);
accum_sum_add(&numstate.sumX, &tmp_var);

int128_to_numericvar(state->sumX2, &tmp_var);
int128_to_numericvar_zero_scale(state->sumX2, &tmp_var);
accum_sum_add(&numstate.sumX2, &tmp_var);

free_var(&tmp_var);
Expand Down Expand Up @@ -8208,7 +8244,7 @@ numericvar_to_int128(const NumericVar *var, int128 *result)
* Convert 128 bit integer to numeric.
*/
static void
int128_to_numericvar(int128 val, NumericVar *var)
int128_to_numericvar_zero_scale(int128 val, NumericVar *var)
{
uint128 uval,
newuval;
Expand Down Expand Up @@ -8248,6 +8284,91 @@ int128_to_numericvar(int128 val, NumericVar *var)
var->ndigits = ndigits;
var->weight = ndigits - 1;
}

/*
* Convert 128 bit integer to numeric with scale.
*/
static void
int128_to_numericvar_with_scale(int128 val, NumericVar *var,
bool neg, int scale)
{
NumericDigit *ptr;
int ndigits = 0;
int dweight, weight, nweight = 0;
int offset, scale_padding;
bool padding_done;
uint128 temp;

var->sign = neg ? NUMERIC_NEG : NUMERIC_POS;
var->dscale = scale;
if (val == 0)
{
var->ndigits = 0;
var->weight = 0;
return;
}

/* int128 can require at most 39 decimal digits; add one for safety */
alloc_var(var, 40 / DEC_DIGITS);
Assert(val > 0);

temp = val;
while (temp != 0) {
temp /= 10;
nweight++;
}

dweight = nweight - scale - 1;

ptr = var->digits + var->ndigits;

if (dweight >= 0)
weight = (dweight + 1 + DEC_DIGITS - 1) / DEC_DIGITS - 1;
else
weight = -((-dweight - 1) / DEC_DIGITS + 1);

/*
* offset is the number of decimal zeroes to insert before
* the first given digit to have a correctly aligned first NBASE digit.
*
* scale_padding determines the number of decimal places to complement.
*/
offset = (weight + 1) * DEC_DIGITS - (dweight + 1);
scale_padding = (offset + nweight) % DEC_DIGITS;
padding_done = scale_padding == 0;

while (val > 0) {
ptr--;
ndigits++;

if (!padding_done) {
int16 pow_padding = 10;
int16 pow_padding_remain = 10;
for (int i = 1; i < scale_padding; i++) {
pow_padding *= 10;
}

for (int i = 1; i < (DEC_DIGITS - scale_padding); i++) {
pow_padding_remain *= 10;
}

temp = val / pow_padding;
*ptr = (val - (temp * pow_padding)) * pow_padding_remain;
val = temp;

padding_done = true;
} else {
temp = val / NBASE;
*ptr = (val - (temp * NBASE));
val = temp;
}
}

var->digits = ptr;
var->ndigits = ndigits;
var->weight = weight;
}

#endif

/*
Expand Down Expand Up @@ -10019,9 +10140,9 @@ sqrt_var(const NumericVar *arg, NumericVar *result, int rscale)
* iteration we don't need the remainder, so we can save a few cycles
* there by not fully computing it.
*/
int128_to_numericvar(s_int128, &s_var);
int128_to_numericvar_zero_scale(s_int128, &s_var);
if (step >= 0)
int128_to_numericvar(r_int128, &r_var);
int128_to_numericvar_zero_scale(r_int128, &r_var);
}
else
{
Expand Down
5 changes: 5 additions & 0 deletions src/include/utils/numeric.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -73,6 +73,11 @@ extern char *numeric_normalize(Numeric num);
extern Numeric int64_to_numeric(int64 val);
extern Numeric int64_div_fast_to_numeric(int64 val1, int log10val2);

#ifdef HAVE_INT128
extern Numeric int128_to_numeric(int128 val, int scale, bool neg,
bool nan, bool pinf, bool ninf);
#endif

extern Numeric numeric_add_opt_error(Numeric num1, Numeric num2,
bool *have_error);
extern Numeric numeric_sub_opt_error(Numeric num1, Numeric num2,
Expand Down