postgres/contrib/hstore/hstore_op.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1274 lines
28 KiB
C
Raw Permalink Normal View History

/*
2010-09-21 00:08:53 +04:00
* contrib/hstore/hstore_op.c
*/
#include "postgres.h"
#include "access/htup_details.h"
#include "catalog/pg_type.h"
#include "common/hashfn.h"
#include "funcapi.h"
#include "hstore.h"
#include "utils/builtins.h"
#include "utils/memutils.h"
/* old names for C functions */
HSTORE_POLLUTE(hstore_fetchval, fetchval);
HSTORE_POLLUTE(hstore_exists, exists);
HSTORE_POLLUTE(hstore_defined, defined);
HSTORE_POLLUTE(hstore_delete, delete);
HSTORE_POLLUTE(hstore_concat, hs_concat);
HSTORE_POLLUTE(hstore_contains, hs_contains);
HSTORE_POLLUTE(hstore_contained, hs_contained);
HSTORE_POLLUTE(hstore_akeys, akeys);
HSTORE_POLLUTE(hstore_avals, avals);
HSTORE_POLLUTE(hstore_skeys, skeys);
HSTORE_POLLUTE(hstore_svals, svals);
HSTORE_POLLUTE(hstore_each, each);
/*
* We're often finding a sequence of keys in ascending order. The
* "lowbound" parameter is used to cache lower bounds of searches
* between calls, based on this assumption. Pass NULL for it for
* one-off or unordered searches.
*/
int
hstoreFindKey(HStore *hs, int *lowbound, char *key, int keylen)
{
HEntry *entries = ARRPTR(hs);
int stopLow = lowbound ? *lowbound : 0;
int stopHigh = HS_COUNT(hs);
int stopMiddle;
char *base = STRPTR(hs);
2006-10-04 04:30:14 +04:00
while (stopLow < stopHigh)
{
int difference;
stopMiddle = stopLow + (stopHigh - stopLow) / 2;
if (HSTORE_KEYLEN(entries, stopMiddle) == keylen)
difference = memcmp(HSTORE_KEY(entries, base, stopMiddle), key, keylen);
else
difference = (HSTORE_KEYLEN(entries, stopMiddle) > keylen) ? 1 : -1;
if (difference == 0)
{
if (lowbound)
*lowbound = stopMiddle + 1;
return stopMiddle;
}
else if (difference < 0)
stopLow = stopMiddle + 1;
else
stopHigh = stopMiddle;
}
2006-10-04 04:30:14 +04:00
if (lowbound)
*lowbound = stopLow;
return -1;
}
Pairs *
hstoreArrayToPairs(ArrayType *a, int *npairs)
{
Datum *key_datums;
bool *key_nulls;
int key_count;
Pairs *key_pairs;
int bufsiz;
int i,
j;
deconstruct_array_builtin(a, TEXTOID, &key_datums, &key_nulls, &key_count);
if (key_count == 0)
{
*npairs = 0;
return NULL;
}
/*
* A text array uses at least eight bytes per element, so any overflow in
* "key_count * sizeof(Pairs)" is small enough for palloc() to catch.
* However, credible improvements to the array format could invalidate
* that assumption. Therefore, use an explicit check rather than relying
* on palloc() to complain.
*/
if (key_count > MaxAllocSize / sizeof(Pairs))
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("number of pairs (%d) exceeds the maximum allowed (%d)",
key_count, (int) (MaxAllocSize / sizeof(Pairs)))));
key_pairs = palloc(sizeof(Pairs) * key_count);
for (i = 0, j = 0; i < key_count; i++)
{
if (!key_nulls[i])
{
key_pairs[j].key = VARDATA(key_datums[i]);
key_pairs[j].keylen = VARSIZE(key_datums[i]) - VARHDRSZ;
key_pairs[j].val = NULL;
key_pairs[j].vallen = 0;
key_pairs[j].needfree = 0;
key_pairs[j].isnull = 1;
j++;
}
}
*npairs = hstoreUniquePairs(key_pairs, j, &bufsiz);
return key_pairs;
}
PG_FUNCTION_INFO_V1(hstore_fetchval);
Datum
hstore_fetchval(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
text *key = PG_GETARG_TEXT_PP(1);
HEntry *entries = ARRPTR(hs);
text *out;
int idx = hstoreFindKey(hs, NULL,
VARDATA_ANY(key), VARSIZE_ANY_EXHDR(key));
2006-10-04 04:30:14 +04:00
if (idx < 0 || HSTORE_VALISNULL(entries, idx))
PG_RETURN_NULL();
out = cstring_to_text_with_len(HSTORE_VAL(entries, STRPTR(hs), idx),
HSTORE_VALLEN(entries, idx));
PG_RETURN_TEXT_P(out);
}
PG_FUNCTION_INFO_V1(hstore_exists);
Datum
hstore_exists(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
text *key = PG_GETARG_TEXT_PP(1);
int idx = hstoreFindKey(hs, NULL,
VARDATA_ANY(key), VARSIZE_ANY_EXHDR(key));
PG_RETURN_BOOL(idx >= 0);
}
PG_FUNCTION_INFO_V1(hstore_exists_any);
Datum
hstore_exists_any(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
ArrayType *keys = PG_GETARG_ARRAYTYPE_P(1);
int nkeys;
Pairs *key_pairs = hstoreArrayToPairs(keys, &nkeys);
int i;
int lowbound = 0;
bool res = false;
/*
* we exploit the fact that the pairs list is already sorted into strictly
* increasing order to narrow the hstoreFindKey search; each search can
* start one entry past the previous "found" entry, or at the lower bound
* of the last search.
*/
for (i = 0; i < nkeys; i++)
{
int idx = hstoreFindKey(hs, &lowbound,
key_pairs[i].key, key_pairs[i].keylen);
if (idx >= 0)
{
res = true;
break;
}
}
PG_RETURN_BOOL(res);
}
PG_FUNCTION_INFO_V1(hstore_exists_all);
Datum
hstore_exists_all(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
ArrayType *keys = PG_GETARG_ARRAYTYPE_P(1);
int nkeys;
Pairs *key_pairs = hstoreArrayToPairs(keys, &nkeys);
int i;
int lowbound = 0;
bool res = true;
/*
* we exploit the fact that the pairs list is already sorted into strictly
* increasing order to narrow the hstoreFindKey search; each search can
* start one entry past the previous "found" entry, or at the lower bound
* of the last search.
*/
for (i = 0; i < nkeys; i++)
{
int idx = hstoreFindKey(hs, &lowbound,
key_pairs[i].key, key_pairs[i].keylen);
if (idx < 0)
{
res = false;
break;
}
}
PG_RETURN_BOOL(res);
}
PG_FUNCTION_INFO_V1(hstore_defined);
Datum
hstore_defined(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
text *key = PG_GETARG_TEXT_PP(1);
HEntry *entries = ARRPTR(hs);
int idx = hstoreFindKey(hs, NULL,
VARDATA_ANY(key), VARSIZE_ANY_EXHDR(key));
bool res = (idx >= 0 && !HSTORE_VALISNULL(entries, idx));
PG_RETURN_BOOL(res);
}
PG_FUNCTION_INFO_V1(hstore_delete);
Datum
hstore_delete(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
text *key = PG_GETARG_TEXT_PP(1);
char *keyptr = VARDATA_ANY(key);
int keylen = VARSIZE_ANY_EXHDR(key);
HStore *out = palloc(VARSIZE(hs));
char *bufs,
*bufd,
*ptrd;
HEntry *es,
*ed;
int i;
int count = HS_COUNT(hs);
int outcount = 0;
2006-10-04 04:30:14 +04:00
SET_VARSIZE(out, VARSIZE(hs));
HS_SETCOUNT(out, count); /* temporary! */
2006-10-04 04:30:14 +04:00
bufs = STRPTR(hs);
es = ARRPTR(hs);
bufd = ptrd = STRPTR(out);
ed = ARRPTR(out);
2006-10-04 04:30:14 +04:00
for (i = 0; i < count; ++i)
{
int len = HSTORE_KEYLEN(es, i);
char *ptrs = HSTORE_KEY(es, bufs, i);
if (!(len == keylen && memcmp(ptrs, keyptr, keylen) == 0))
{
int vallen = HSTORE_VALLEN(es, i);
2010-02-26 05:01:40 +03:00
HS_COPYITEM(ed, bufd, ptrd, ptrs, len, vallen,
HSTORE_VALISNULL(es, i));
++outcount;
}
}
HS_FINALIZE(out, outcount, bufd, ptrd);
PG_RETURN_POINTER(out);
}
PG_FUNCTION_INFO_V1(hstore_delete_array);
Datum
hstore_delete_array(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
HStore *out = palloc(VARSIZE(hs));
int hs_count = HS_COUNT(hs);
char *ps,
*bufd,
*pd;
HEntry *es,
*ed;
int i,
j;
int outcount = 0;
ArrayType *key_array = PG_GETARG_ARRAYTYPE_P(1);
int nkeys;
Pairs *key_pairs = hstoreArrayToPairs(key_array, &nkeys);
SET_VARSIZE(out, VARSIZE(hs));
HS_SETCOUNT(out, hs_count); /* temporary! */
ps = STRPTR(hs);
es = ARRPTR(hs);
bufd = pd = STRPTR(out);
ed = ARRPTR(out);
if (nkeys == 0)
{
/* return a copy of the input, unchanged */
memcpy(out, hs, VARSIZE(hs));
HS_FIXSIZE(out, hs_count);
HS_SETCOUNT(out, hs_count);
PG_RETURN_POINTER(out);
}
/*
* this is in effect a merge between hs and key_pairs, both of which are
* already sorted by (keylen,key); we take keys from hs only
*/
for (i = j = 0; i < hs_count;)
{
int difference;
2010-02-26 05:01:40 +03:00
if (j >= nkeys)
difference = -1;
else
{
int skeylen = HSTORE_KEYLEN(es, i);
2010-02-26 05:01:40 +03:00
if (skeylen == key_pairs[j].keylen)
difference = memcmp(HSTORE_KEY(es, ps, i),
key_pairs[j].key,
key_pairs[j].keylen);
else
difference = (skeylen > key_pairs[j].keylen) ? 1 : -1;
}
if (difference > 0)
++j;
else if (difference == 0)
++i, ++j;
else
{
HS_COPYITEM(ed, bufd, pd,
HSTORE_KEY(es, ps, i), HSTORE_KEYLEN(es, i),
HSTORE_VALLEN(es, i), HSTORE_VALISNULL(es, i));
++outcount;
++i;
}
}
HS_FINALIZE(out, outcount, bufd, pd);
PG_RETURN_POINTER(out);
}
2006-10-04 04:30:14 +04:00
PG_FUNCTION_INFO_V1(hstore_delete_hstore);
Datum
hstore_delete_hstore(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
HStore *hs2 = PG_GETARG_HSTORE_P(1);
HStore *out = palloc(VARSIZE(hs));
int hs_count = HS_COUNT(hs);
int hs2_count = HS_COUNT(hs2);
char *ps,
*ps2,
*bufd,
*pd;
HEntry *es,
*es2,
*ed;
int i,
j;
int outcount = 0;
SET_VARSIZE(out, VARSIZE(hs));
HS_SETCOUNT(out, hs_count); /* temporary! */
ps = STRPTR(hs);
es = ARRPTR(hs);
ps2 = STRPTR(hs2);
es2 = ARRPTR(hs2);
bufd = pd = STRPTR(out);
ed = ARRPTR(out);
if (hs2_count == 0)
{
/* return a copy of the input, unchanged */
memcpy(out, hs, VARSIZE(hs));
HS_FIXSIZE(out, hs_count);
HS_SETCOUNT(out, hs_count);
PG_RETURN_POINTER(out);
}
/*
* this is in effect a merge between hs and hs2, both of which are already
* sorted by (keylen,key); we take keys from hs only; for equal keys, we
* take the value from hs unless the values are equal
*/
for (i = j = 0; i < hs_count;)
{
int difference;
2010-02-26 05:01:40 +03:00
if (j >= hs2_count)
difference = -1;
else
{
int skeylen = HSTORE_KEYLEN(es, i);
int s2keylen = HSTORE_KEYLEN(es2, j);
2010-02-26 05:01:40 +03:00
if (skeylen == s2keylen)
difference = memcmp(HSTORE_KEY(es, ps, i),
HSTORE_KEY(es2, ps2, j),
skeylen);
else
difference = (skeylen > s2keylen) ? 1 : -1;
}
if (difference > 0)
++j;
else if (difference == 0)
{
int svallen = HSTORE_VALLEN(es, i);
int snullval = HSTORE_VALISNULL(es, i);
if (snullval != HSTORE_VALISNULL(es2, j) ||
(!snullval && (svallen != HSTORE_VALLEN(es2, j) ||
memcmp(HSTORE_VAL(es, ps, i),
HSTORE_VAL(es2, ps2, j),
svallen) != 0)))
{
HS_COPYITEM(ed, bufd, pd,
HSTORE_KEY(es, ps, i), HSTORE_KEYLEN(es, i),
svallen, snullval);
++outcount;
}
++i, ++j;
}
else
{
HS_COPYITEM(ed, bufd, pd,
HSTORE_KEY(es, ps, i), HSTORE_KEYLEN(es, i),
HSTORE_VALLEN(es, i), HSTORE_VALISNULL(es, i));
++outcount;
++i;
}
}
2006-10-04 04:30:14 +04:00
HS_FINALIZE(out, outcount, bufd, pd);
PG_RETURN_POINTER(out);
}
PG_FUNCTION_INFO_V1(hstore_concat);
Datum
hstore_concat(PG_FUNCTION_ARGS)
{
HStore *s1 = PG_GETARG_HSTORE_P(0);
HStore *s2 = PG_GETARG_HSTORE_P(1);
HStore *out = palloc(VARSIZE(s1) + VARSIZE(s2));
char *ps1,
*ps2,
*bufd,
*pd;
HEntry *es1,
*es2,
*ed;
int s1idx;
int s2idx;
int s1count = HS_COUNT(s1);
int s2count = HS_COUNT(s2);
int outcount = 0;
SET_VARSIZE(out, VARSIZE(s1) + VARSIZE(s2) - HSHRDSIZE);
HS_SETCOUNT(out, s1count + s2count);
if (s1count == 0)
{
/* return a copy of the input, unchanged */
memcpy(out, s2, VARSIZE(s2));
HS_FIXSIZE(out, s2count);
HS_SETCOUNT(out, s2count);
PG_RETURN_POINTER(out);
}
if (s2count == 0)
{
/* return a copy of the input, unchanged */
memcpy(out, s1, VARSIZE(s1));
HS_FIXSIZE(out, s1count);
HS_SETCOUNT(out, s1count);
PG_RETURN_POINTER(out);
}
ps1 = STRPTR(s1);
ps2 = STRPTR(s2);
bufd = pd = STRPTR(out);
es1 = ARRPTR(s1);
es2 = ARRPTR(s2);
ed = ARRPTR(out);
2006-10-04 04:30:14 +04:00
/*
* this is in effect a merge between s1 and s2, both of which are already
* sorted by (keylen,key); we take s2 for equal keys
*/
2006-10-04 04:30:14 +04:00
for (s1idx = s2idx = 0; s1idx < s1count || s2idx < s2count; ++outcount)
{
int difference;
2010-02-26 05:01:40 +03:00
if (s1idx >= s1count)
difference = 1;
else if (s2idx >= s2count)
difference = -1;
else
{
int s1keylen = HSTORE_KEYLEN(es1, s1idx);
int s2keylen = HSTORE_KEYLEN(es2, s2idx);
2010-02-26 05:01:40 +03:00
if (s1keylen == s2keylen)
difference = memcmp(HSTORE_KEY(es1, ps1, s1idx),
HSTORE_KEY(es2, ps2, s2idx),
s1keylen);
else
difference = (s1keylen > s2keylen) ? 1 : -1;
}
if (difference >= 0)
{
HS_COPYITEM(ed, bufd, pd,
HSTORE_KEY(es2, ps2, s2idx), HSTORE_KEYLEN(es2, s2idx),
HSTORE_VALLEN(es2, s2idx), HSTORE_VALISNULL(es2, s2idx));
++s2idx;
if (difference == 0)
++s1idx;
}
else
{
HS_COPYITEM(ed, bufd, pd,
HSTORE_KEY(es1, ps1, s1idx), HSTORE_KEYLEN(es1, s1idx),
HSTORE_VALLEN(es1, s1idx), HSTORE_VALISNULL(es1, s1idx));
++s1idx;
}
}
HS_FINALIZE(out, outcount, bufd, pd);
2006-10-04 04:30:14 +04:00
PG_RETURN_POINTER(out);
}
2006-10-04 04:30:14 +04:00
PG_FUNCTION_INFO_V1(hstore_slice_to_array);
Datum
hstore_slice_to_array(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
HEntry *entries = ARRPTR(hs);
char *ptr = STRPTR(hs);
ArrayType *key_array = PG_GETARG_ARRAYTYPE_P(1);
ArrayType *aout;
Datum *key_datums;
bool *key_nulls;
Datum *out_datums;
bool *out_nulls;
int key_count;
int i;
deconstruct_array_builtin(key_array, TEXTOID, &key_datums, &key_nulls, &key_count);
if (key_count == 0)
{
aout = construct_empty_array(TEXTOID);
PG_RETURN_POINTER(aout);
}
out_datums = palloc(sizeof(Datum) * key_count);
out_nulls = palloc(sizeof(bool) * key_count);
2006-10-04 04:30:14 +04:00
for (i = 0; i < key_count; ++i)
{
text *key = (text *) DatumGetPointer(key_datums[i]);
int idx;
if (key_nulls[i])
idx = -1;
else
idx = hstoreFindKey(hs, NULL, VARDATA(key), VARSIZE(key) - VARHDRSZ);
if (idx < 0 || HSTORE_VALISNULL(entries, idx))
{
out_nulls[i] = true;
out_datums[i] = (Datum) 0;
}
else
{
out_datums[i] =
PointerGetDatum(cstring_to_text_with_len(HSTORE_VAL(entries, ptr, idx),
HSTORE_VALLEN(entries, idx)));
out_nulls[i] = false;
}
}
2006-10-04 04:30:14 +04:00
aout = construct_md_array(out_datums, out_nulls,
ARR_NDIM(key_array),
ARR_DIMS(key_array),
ARR_LBOUND(key_array),
TEXTOID, -1, false, TYPALIGN_INT);
PG_RETURN_POINTER(aout);
}
PG_FUNCTION_INFO_V1(hstore_slice_to_hstore);
Datum
hstore_slice_to_hstore(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
HEntry *entries = ARRPTR(hs);
char *ptr = STRPTR(hs);
ArrayType *key_array = PG_GETARG_ARRAYTYPE_P(1);
HStore *out;
int nkeys;
Pairs *key_pairs = hstoreArrayToPairs(key_array, &nkeys);
Pairs *out_pairs;
int bufsiz;
int lastidx = 0;
int i;
int out_count = 0;
if (nkeys == 0)
{
out = hstorePairs(NULL, 0, 0);
PG_RETURN_POINTER(out);
}
2006-10-04 04:30:14 +04:00
/* hstoreArrayToPairs() checked overflow */
out_pairs = palloc(sizeof(Pairs) * nkeys);
bufsiz = 0;
/*
* we exploit the fact that the pairs list is already sorted into strictly
* increasing order to narrow the hstoreFindKey search; each search can
* start one entry past the previous "found" entry, or at the lower bound
* of the last search.
*/
for (i = 0; i < nkeys; ++i)
{
int idx = hstoreFindKey(hs, &lastidx,
key_pairs[i].key, key_pairs[i].keylen);
if (idx >= 0)
{
out_pairs[out_count].key = key_pairs[i].key;
bufsiz += (out_pairs[out_count].keylen = key_pairs[i].keylen);
out_pairs[out_count].val = HSTORE_VAL(entries, ptr, idx);
bufsiz += (out_pairs[out_count].vallen = HSTORE_VALLEN(entries, idx));
out_pairs[out_count].isnull = HSTORE_VALISNULL(entries, idx);
out_pairs[out_count].needfree = false;
++out_count;
}
}
/*
* we don't use hstoreUniquePairs here because we know that the pairs list
* is already sorted and uniq'ed.
*/
2006-10-04 04:30:14 +04:00
out = hstorePairs(out_pairs, out_count, bufsiz);
PG_RETURN_POINTER(out);
}
PG_FUNCTION_INFO_V1(hstore_akeys);
Datum
hstore_akeys(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
Datum *d;
ArrayType *a;
HEntry *entries = ARRPTR(hs);
char *base = STRPTR(hs);
int count = HS_COUNT(hs);
int i;
2006-10-04 04:30:14 +04:00
if (count == 0)
{
a = construct_empty_array(TEXTOID);
PG_RETURN_POINTER(a);
}
2006-10-04 04:30:14 +04:00
d = (Datum *) palloc(sizeof(Datum) * count);
2006-10-04 04:30:14 +04:00
for (i = 0; i < count; ++i)
{
text *t = cstring_to_text_with_len(HSTORE_KEY(entries, base, i),
HSTORE_KEYLEN(entries, i));
2010-02-26 05:01:40 +03:00
d[i] = PointerGetDatum(t);
}
a = construct_array_builtin(d, count, TEXTOID);
PG_RETURN_POINTER(a);
}
PG_FUNCTION_INFO_V1(hstore_avals);
Datum
hstore_avals(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
Datum *d;
bool *nulls;
ArrayType *a;
HEntry *entries = ARRPTR(hs);
char *base = STRPTR(hs);
int count = HS_COUNT(hs);
int lb = 1;
int i;
2006-10-04 04:30:14 +04:00
if (count == 0)
{
a = construct_empty_array(TEXTOID);
PG_RETURN_POINTER(a);
}
2006-10-04 04:30:14 +04:00
d = (Datum *) palloc(sizeof(Datum) * count);
nulls = (bool *) palloc(sizeof(bool) * count);
2006-10-04 04:30:14 +04:00
for (i = 0; i < count; ++i)
{
if (HSTORE_VALISNULL(entries, i))
{
d[i] = (Datum) 0;
nulls[i] = true;
}
else
{
text *item = cstring_to_text_with_len(HSTORE_VAL(entries, base, i),
HSTORE_VALLEN(entries, i));
2010-02-26 05:01:40 +03:00
d[i] = PointerGetDatum(item);
nulls[i] = false;
}
}
a = construct_md_array(d, nulls, 1, &count, &lb,
TEXTOID, -1, false, TYPALIGN_INT);
PG_RETURN_POINTER(a);
}
static ArrayType *
hstore_to_array_internal(HStore *hs, int ndims)
{
HEntry *entries = ARRPTR(hs);
char *base = STRPTR(hs);
int count = HS_COUNT(hs);
int out_size[2] = {0, 2};
int lb[2] = {1, 1};
Datum *out_datums;
bool *out_nulls;
int i;
Assert(ndims < 3);
if (count == 0 || ndims == 0)
return construct_empty_array(TEXTOID);
out_size[0] = count * 2 / ndims;
out_datums = palloc(sizeof(Datum) * count * 2);
out_nulls = palloc(sizeof(bool) * count * 2);
for (i = 0; i < count; ++i)
{
text *key = cstring_to_text_with_len(HSTORE_KEY(entries, base, i),
HSTORE_KEYLEN(entries, i));
2010-02-26 05:01:40 +03:00
out_datums[i * 2] = PointerGetDatum(key);
out_nulls[i * 2] = false;
if (HSTORE_VALISNULL(entries, i))
{
out_datums[i * 2 + 1] = (Datum) 0;
out_nulls[i * 2 + 1] = true;
}
else
{
text *item = cstring_to_text_with_len(HSTORE_VAL(entries, base, i),
HSTORE_VALLEN(entries, i));
2010-02-26 05:01:40 +03:00
out_datums[i * 2 + 1] = PointerGetDatum(item);
out_nulls[i * 2 + 1] = false;
}
}
return construct_md_array(out_datums, out_nulls,
ndims, out_size, lb,
TEXTOID, -1, false, TYPALIGN_INT);
}
PG_FUNCTION_INFO_V1(hstore_to_array);
Datum
hstore_to_array(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
ArrayType *out = hstore_to_array_internal(hs, 1);
PG_RETURN_POINTER(out);
}
PG_FUNCTION_INFO_V1(hstore_to_matrix);
Datum
hstore_to_matrix(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
ArrayType *out = hstore_to_array_internal(hs, 2);
PG_RETURN_POINTER(out);
}
/*
* Common initialization function for the various set-returning
* funcs. fcinfo is only passed if the function is to return a
* composite; it will be used to look up the return tupledesc.
* we stash a copy of the hstore in the multi-call context in
* case it was originally toasted. (At least I assume that's why;
* there was no explanatory comment in the original code. --AG)
*/
static void
setup_firstcall(FuncCallContext *funcctx, HStore *hs,
Change function call information to be variable length. Before this change FunctionCallInfoData, the struct arguments etc for V1 function calls are stored in, always had space for FUNC_MAX_ARGS/100 arguments, storing datums and their nullness in two arrays. For nearly every function call 100 arguments is far more than needed, therefore wasting memory. Arg and argnull being two separate arrays also guarantees that to access a single argument, two cachelines have to be touched. Change the layout so there's a single variable-length array with pairs of value / isnull. That drastically reduces memory consumption for most function calls (on x86-64 a two argument function now uses 64bytes, previously 936 bytes), and makes it very likely that argument value and its nullness are on the same cacheline. Arguments are stored in a new NullableDatum struct, which, due to padding, needs more memory per argument than before. But as usually far fewer arguments are stored, and individual arguments are cheaper to access, that's still a clear win. It's likely that there's other places where conversion to NullableDatum arrays would make sense, e.g. TupleTableSlots, but that's for another commit. Because the function call information is now variable-length allocations have to take the number of arguments into account. For heap allocations that can be done with SizeForFunctionCallInfoData(), for on-stack allocations there's a new LOCAL_FCINFO(name, nargs) macro that helps to allocate an appropriately sized and aligned variable. Some places with stack allocation function call information don't know the number of arguments at compile time, and currently variably sized stack allocations aren't allowed in postgres. Therefore allow for FUNC_MAX_ARGS space in these cases. They're not that common, so for now that seems acceptable. Because of the need to allocate FunctionCallInfo of the appropriate size, older extensions may need to update their code. To avoid subtle breakages, the FunctionCallInfoData struct has been renamed to FunctionCallInfoBaseData. Most code only references FunctionCallInfo, so that shouldn't cause much collateral damage. This change is also a prerequisite for more efficient expression JIT compilation (by allocating the function call information on the stack, allowing LLVM to optimize it away); previously the size of the call information caused problems inside LLVM's optimizer. Author: Andres Freund Reviewed-By: Tom Lane Discussion: https://postgr.es/m/20180605172952.x34m5uz6ju6enaem@alap3.anarazel.de
2019-01-27 01:17:52 +03:00
FunctionCallInfo fcinfo)
{
MemoryContext oldcontext;
HStore *st;
oldcontext = MemoryContextSwitchTo(funcctx->multi_call_memory_ctx);
st = (HStore *) palloc(VARSIZE(hs));
memcpy(st, hs, VARSIZE(hs));
funcctx->user_fctx = (void *) st;
if (fcinfo)
{
TupleDesc tupdesc;
/* Build a tuple descriptor for our result type */
if (get_call_result_type(fcinfo, NULL, &tupdesc) != TYPEFUNC_COMPOSITE)
elog(ERROR, "return type must be a row type");
2010-02-26 05:01:40 +03:00
funcctx->tuple_desc = BlessTupleDesc(tupdesc);
}
MemoryContextSwitchTo(oldcontext);
}
PG_FUNCTION_INFO_V1(hstore_skeys);
Datum
hstore_skeys(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
HStore *hs;
int i;
2006-10-04 04:30:14 +04:00
if (SRF_IS_FIRSTCALL())
2006-10-04 04:30:14 +04:00
{
hs = PG_GETARG_HSTORE_P(0);
funcctx = SRF_FIRSTCALL_INIT();
setup_firstcall(funcctx, hs, NULL);
}
funcctx = SRF_PERCALL_SETUP();
hs = (HStore *) funcctx->user_fctx;
i = funcctx->call_cntr;
2006-10-04 04:30:14 +04:00
if (i < HS_COUNT(hs))
{
HEntry *entries = ARRPTR(hs);
text *item;
item = cstring_to_text_with_len(HSTORE_KEY(entries, STRPTR(hs), i),
HSTORE_KEYLEN(entries, i));
SRF_RETURN_NEXT(funcctx, PointerGetDatum(item));
}
SRF_RETURN_DONE(funcctx);
}
PG_FUNCTION_INFO_V1(hstore_svals);
Datum
hstore_svals(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
HStore *hs;
int i;
2006-10-04 04:30:14 +04:00
if (SRF_IS_FIRSTCALL())
2006-10-04 04:30:14 +04:00
{
hs = PG_GETARG_HSTORE_P(0);
funcctx = SRF_FIRSTCALL_INIT();
setup_firstcall(funcctx, hs, NULL);
}
funcctx = SRF_PERCALL_SETUP();
hs = (HStore *) funcctx->user_fctx;
i = funcctx->call_cntr;
if (i < HS_COUNT(hs))
{
HEntry *entries = ARRPTR(hs);
2006-10-04 04:30:14 +04:00
if (HSTORE_VALISNULL(entries, i))
2006-10-04 04:30:14 +04:00
{
ReturnSetInfo *rsi;
/* ugly ugly ugly. why no macro for this? */
(funcctx)->call_cntr++;
rsi = (ReturnSetInfo *) fcinfo->resultinfo;
rsi->isDone = ExprMultipleResult;
PG_RETURN_NULL();
}
else
{
text *item;
item = cstring_to_text_with_len(HSTORE_VAL(entries, STRPTR(hs), i),
HSTORE_VALLEN(entries, i));
SRF_RETURN_NEXT(funcctx, PointerGetDatum(item));
}
}
SRF_RETURN_DONE(funcctx);
}
PG_FUNCTION_INFO_V1(hstore_contains);
Datum
hstore_contains(PG_FUNCTION_ARGS)
{
HStore *val = PG_GETARG_HSTORE_P(0);
HStore *tmpl = PG_GETARG_HSTORE_P(1);
bool res = true;
HEntry *te = ARRPTR(tmpl);
char *tstr = STRPTR(tmpl);
HEntry *ve = ARRPTR(val);
char *vstr = STRPTR(val);
int tcount = HS_COUNT(tmpl);
int lastidx = 0;
int i;
/*
* we exploit the fact that keys in "tmpl" are in strictly increasing
* order to narrow the hstoreFindKey search; each search can start one
* entry past the previous "found" entry, or at the lower bound of the
* search
*/
for (i = 0; res && i < tcount; ++i)
{
int idx = hstoreFindKey(val, &lastidx,
HSTORE_KEY(te, tstr, i),
HSTORE_KEYLEN(te, i));
2006-10-04 04:30:14 +04:00
if (idx >= 0)
{
bool nullval = HSTORE_VALISNULL(te, i);
int vallen = HSTORE_VALLEN(te, i);
if (nullval != HSTORE_VALISNULL(ve, idx) ||
(!nullval && (vallen != HSTORE_VALLEN(ve, idx) ||
memcmp(HSTORE_VAL(te, tstr, i),
HSTORE_VAL(ve, vstr, idx),
vallen) != 0)))
res = false;
}
else
res = false;
}
PG_RETURN_BOOL(res);
}
2006-10-04 04:30:14 +04:00
PG_FUNCTION_INFO_V1(hstore_contained);
Datum
hstore_contained(PG_FUNCTION_ARGS)
{
PG_RETURN_DATUM(DirectFunctionCall2(hstore_contains,
PG_GETARG_DATUM(1),
PG_GETARG_DATUM(0)
));
}
PG_FUNCTION_INFO_V1(hstore_each);
Datum
hstore_each(PG_FUNCTION_ARGS)
{
FuncCallContext *funcctx;
HStore *hs;
int i;
if (SRF_IS_FIRSTCALL())
{
hs = PG_GETARG_HSTORE_P(0);
funcctx = SRF_FIRSTCALL_INIT();
setup_firstcall(funcctx, hs, fcinfo);
}
funcctx = SRF_PERCALL_SETUP();
hs = (HStore *) funcctx->user_fctx;
i = funcctx->call_cntr;
2006-10-04 04:30:14 +04:00
if (i < HS_COUNT(hs))
{
HEntry *entries = ARRPTR(hs);
char *ptr = STRPTR(hs);
Datum res,
dvalues[2];
bool nulls[2] = {false, false};
text *item;
HeapTuple tuple;
2006-10-04 04:30:14 +04:00
item = cstring_to_text_with_len(HSTORE_KEY(entries, ptr, i),
HSTORE_KEYLEN(entries, i));
dvalues[0] = PointerGetDatum(item);
if (HSTORE_VALISNULL(entries, i))
{
dvalues[1] = (Datum) 0;
nulls[1] = true;
}
else
{
item = cstring_to_text_with_len(HSTORE_VAL(entries, ptr, i),
HSTORE_VALLEN(entries, i));
dvalues[1] = PointerGetDatum(item);
}
tuple = heap_form_tuple(funcctx->tuple_desc, dvalues, nulls);
res = HeapTupleGetDatum(tuple);
SRF_RETURN_NEXT(funcctx, res);
}
SRF_RETURN_DONE(funcctx);
2006-10-04 04:30:14 +04:00
}
/*
* btree sort order for hstores isn't intended to be useful; we really only
* care about equality versus non-equality. we compare the entire string
* buffer first, then the entry pos array.
*/
PG_FUNCTION_INFO_V1(hstore_cmp);
Datum
hstore_cmp(PG_FUNCTION_ARGS)
{
HStore *hs1 = PG_GETARG_HSTORE_P(0);
HStore *hs2 = PG_GETARG_HSTORE_P(1);
int hcount1 = HS_COUNT(hs1);
int hcount2 = HS_COUNT(hs2);
int res = 0;
if (hcount1 == 0 || hcount2 == 0)
{
/*
* if either operand is empty, and the other is nonempty, the nonempty
* one is larger. If both are empty they are equal.
*/
if (hcount1 > 0)
res = 1;
else if (hcount2 > 0)
res = -1;
}
else
{
/* here we know both operands are nonempty */
char *str1 = STRPTR(hs1);
char *str2 = STRPTR(hs2);
HEntry *ent1 = ARRPTR(hs1);
HEntry *ent2 = ARRPTR(hs2);
size_t len1 = HSE_ENDPOS(ent1[2 * hcount1 - 1]);
size_t len2 = HSE_ENDPOS(ent2[2 * hcount2 - 1]);
res = memcmp(str1, str2, Min(len1, len2));
if (res == 0)
{
if (len1 > len2)
res = 1;
else if (len1 < len2)
res = -1;
else if (hcount1 > hcount2)
res = 1;
else if (hcount2 > hcount1)
res = -1;
else
{
int count = hcount1 * 2;
int i;
for (i = 0; i < count; ++i)
if (HSE_ENDPOS(ent1[i]) != HSE_ENDPOS(ent2[i]) ||
HSE_ISNULL(ent1[i]) != HSE_ISNULL(ent2[i]))
break;
if (i < count)
{
if (HSE_ENDPOS(ent1[i]) < HSE_ENDPOS(ent2[i]))
res = -1;
else if (HSE_ENDPOS(ent1[i]) > HSE_ENDPOS(ent2[i]))
res = 1;
else if (HSE_ISNULL(ent1[i]))
res = 1;
else if (HSE_ISNULL(ent2[i]))
res = -1;
}
}
}
else
{
res = (res > 0) ? 1 : -1;
}
}
/*
* this is a btree support function; this is one of the few places where
* memory needs to be explicitly freed.
*/
PG_FREE_IF_COPY(hs1, 0);
PG_FREE_IF_COPY(hs2, 1);
PG_RETURN_INT32(res);
}
PG_FUNCTION_INFO_V1(hstore_eq);
Datum
hstore_eq(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res == 0);
}
PG_FUNCTION_INFO_V1(hstore_ne);
Datum
hstore_ne(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res != 0);
}
PG_FUNCTION_INFO_V1(hstore_gt);
Datum
hstore_gt(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res > 0);
}
PG_FUNCTION_INFO_V1(hstore_ge);
Datum
hstore_ge(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res >= 0);
}
PG_FUNCTION_INFO_V1(hstore_lt);
Datum
hstore_lt(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res < 0);
}
PG_FUNCTION_INFO_V1(hstore_le);
Datum
hstore_le(PG_FUNCTION_ARGS)
{
int res = DatumGetInt32(DirectFunctionCall2(hstore_cmp,
PG_GETARG_DATUM(0),
PG_GETARG_DATUM(1)));
2010-02-26 05:01:40 +03:00
PG_RETURN_BOOL(res <= 0);
}
PG_FUNCTION_INFO_V1(hstore_hash);
Datum
hstore_hash(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
Datum hval = hash_any((unsigned char *) VARDATA(hs),
VARSIZE(hs) - VARHDRSZ);
/*
* This (along with hstore_hash_extended) is the only place in the code
* that cares whether the overall varlena size exactly matches the true
* data size; this assertion should be maintained by all the other code,
* but we make it explicit here.
*/
Assert(VARSIZE(hs) ==
(HS_COUNT(hs) != 0 ?
CALCDATASIZE(HS_COUNT(hs),
HSE_ENDPOS(ARRPTR(hs)[2 * HS_COUNT(hs) - 1])) :
HSHRDSIZE));
PG_FREE_IF_COPY(hs, 0);
PG_RETURN_DATUM(hval);
}
PG_FUNCTION_INFO_V1(hstore_hash_extended);
Datum
hstore_hash_extended(PG_FUNCTION_ARGS)
{
HStore *hs = PG_GETARG_HSTORE_P(0);
uint64 seed = PG_GETARG_INT64(1);
Datum hval;
hval = hash_any_extended((unsigned char *) VARDATA(hs),
VARSIZE(hs) - VARHDRSZ,
seed);
/* See comment in hstore_hash */
Assert(VARSIZE(hs) ==
(HS_COUNT(hs) != 0 ?
CALCDATASIZE(HS_COUNT(hs),
HSE_ENDPOS(ARRPTR(hs)[2 * HS_COUNT(hs) - 1])) :
HSHRDSIZE));
PG_FREE_IF_COPY(hs, 0);
PG_RETURN_DATUM(hval);
}