postgresql/contrib/intarray/_int_tool.c

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

408 lines
6.5 KiB
C
Raw Normal View History

/*
2010-09-20 22:08:53 +02:00
* contrib/intarray/_int_tool.c
*/
#include "postgres.h"
#include <limits.h>
2003-06-11 21:31:05 +02:00
#include "_int.h"
#include "catalog/pg_type.h"
#include "common/int.h"
#include "lib/qunique.h"
2003-06-11 21:31:05 +02:00
/* arguments are assumed sorted & unique-ified */
2003-06-11 21:31:05 +02:00
bool
inner_int_contains(ArrayType *a, ArrayType *b)
{
int na,
nb;
int i,
j,
n;
int *da,
*db;
na = ARRNELEMS(a);
nb = ARRNELEMS(b);
da = ARRPTR(a);
db = ARRPTR(b);
i = j = n = 0;
while (i < na && j < nb)
{
2003-06-11 21:31:05 +02:00
if (da[i] < db[j])
i++;
else if (da[i] == db[j])
{
n++;
i++;
j++;
}
else
break; /* db[j] is not in da */
}
2003-06-11 21:31:05 +02:00
return (n == nb);
2003-06-11 21:31:05 +02:00
}
/* arguments are assumed sorted */
2003-06-11 21:31:05 +02:00
bool
inner_int_overlap(ArrayType *a, ArrayType *b)
{
int na,
nb;
int i,
j;
int *da,
*db;
na = ARRNELEMS(a);
nb = ARRNELEMS(b);
da = ARRPTR(a);
db = ARRPTR(b);
i = j = 0;
while (i < na && j < nb)
{
2003-06-11 21:31:05 +02:00
if (da[i] < db[j])
i++;
else if (da[i] == db[j])
return true;
2003-06-11 21:31:05 +02:00
else
j++;
}
2003-06-11 21:31:05 +02:00
return false;
2003-06-11 21:31:05 +02:00
}
ArrayType *
inner_int_union(ArrayType *a, ArrayType *b)
{
ArrayType *r = NULL;
CHECKARRVALID(a);
CHECKARRVALID(b);
if (ARRISEMPTY(a) && ARRISEMPTY(b))
2003-06-11 21:31:05 +02:00
return new_intArrayType(0);
if (ARRISEMPTY(a))
2003-06-11 21:31:05 +02:00
r = copy_intArrayType(b);
if (ARRISEMPTY(b))
2003-06-11 21:31:05 +02:00
r = copy_intArrayType(a);
if (!r)
2003-06-11 21:31:05 +02:00
{
int na = ARRNELEMS(a),
nb = ARRNELEMS(b);
int *da = ARRPTR(a),
*db = ARRPTR(b);
int i,
j,
*dr;
2003-06-11 21:31:05 +02:00
r = new_intArrayType(na + nb);
dr = ARRPTR(r);
/* union */
i = j = 0;
while (i < na && j < nb)
{
if (da[i] == db[j])
{
*dr++ = da[i++];
j++;
}
else if (da[i] < db[j])
2003-06-11 21:31:05 +02:00
*dr++ = da[i++];
else
*dr++ = db[j++];
}
2003-06-11 21:31:05 +02:00
while (i < na)
*dr++ = da[i++];
while (j < nb)
*dr++ = db[j++];
r = resize_intArrayType(r, dr - ARRPTR(r));
2003-06-11 21:31:05 +02:00
}
if (ARRNELEMS(r) > 1)
r = _int_unique(r);
return r;
}
ArrayType *
inner_int_inter(ArrayType *a, ArrayType *b)
{
ArrayType *r;
int na,
nb;
int *da,
*db,
*dr;
int i,
j,
k;
2003-06-11 21:31:05 +02:00
if (ARRISEMPTY(a) || ARRISEMPTY(b))
2003-06-11 21:31:05 +02:00
return new_intArrayType(0);
na = ARRNELEMS(a);
nb = ARRNELEMS(b);
da = ARRPTR(a);
db = ARRPTR(b);
r = new_intArrayType(Min(na, nb));
2003-06-11 21:31:05 +02:00
dr = ARRPTR(r);
i = j = k = 0;
2003-06-11 21:31:05 +02:00
while (i < na && j < nb)
{
2003-06-11 21:31:05 +02:00
if (da[i] < db[j])
i++;
else if (da[i] == db[j])
{
if (k == 0 || dr[k - 1] != db[j])
dr[k++] = db[j];
2003-06-11 21:31:05 +02:00
i++;
j++;
}
else
j++;
}
2003-06-11 21:31:05 +02:00
if (k == 0)
2003-06-11 21:31:05 +02:00
{
pfree(r);
return new_intArrayType(0);
}
else
return resize_intArrayType(r, k);
2003-06-11 21:31:05 +02:00
}
void
rt__int_size(ArrayType *a, float *size)
{
*size = (float) ARRNELEMS(a);
}
/* qsort_arg comparison function for isort() */
static int
isort_cmp(const void *a, const void *b, void *arg)
{
int32 aval = *((const int32 *) a);
int32 bval = *((const int32 *) b);
if (aval < bval)
return -1;
if (aval > bval)
return 1;
/*
* Report if we have any duplicates. If there are equal keys, qsort must
* compare them at some point, else it wouldn't know whether one should go
* before or after the other.
*/
*((bool *) arg) = true;
return 0;
}
/* Sort the given data (len >= 2). Return true if any duplicates found */
2003-06-11 21:31:05 +02:00
bool
isort(int32 *a, int len)
2003-06-11 21:31:05 +02:00
{
bool r = false;
2003-06-11 21:31:05 +02:00
qsort_arg(a, len, sizeof(int32), isort_cmp, &r);
2003-06-11 21:31:05 +02:00
return r;
}
/* Create a new int array with room for "num" elements */
2003-06-11 21:31:05 +02:00
ArrayType *
new_intArrayType(int num)
{
ArrayType *r;
int nbytes;
/* if no elements, return a zero-dimensional array */
if (num <= 0)
{
Assert(num == 0);
r = construct_empty_array(INT4OID);
return r;
}
nbytes = ARR_OVERHEAD_NONULLS(1) + sizeof(int) * num;
2003-06-11 21:31:05 +02:00
r = (ArrayType *) palloc0(nbytes);
SET_VARSIZE(r, nbytes);
ARR_NDIM(r) = 1;
r->dataoffset = 0; /* marker for no null bitmap */
2003-06-11 21:31:05 +02:00
ARR_ELEMTYPE(r) = INT4OID;
ARR_DIMS(r)[0] = num;
ARR_LBOUND(r)[0] = 1;
2003-06-11 21:31:05 +02:00
return r;
}
ArrayType *
resize_intArrayType(ArrayType *a, int num)
{
int nbytes;
int i;
2003-06-11 21:31:05 +02:00
/* if no elements, return a zero-dimensional array */
if (num <= 0)
{
Assert(num == 0);
Fix intarray's GiST opclasses to not fail for empty arrays with <@. contrib/intarray considers "arraycol <@ constant-array" to be indexable, but its GiST opclass code fails to reliably find index entries for empty array values (which of course should trivially match such queries). This is because the test condition to see whether we should descend through a non-leaf node is wrong. Unfortunately, empty array entries could be anywhere in the index, as these index opclasses are currently designed. So there's no way to fix this except by lobotomizing <@ indexscans to scan the whole index ... which is what this patch does. That's pretty unfortunate: the performance is now actually worse than a seqscan, in most cases. We'd be better off to remove <@ from the GiST opclasses entirely, and perhaps a future non-back-patchable patch will do so. In the meantime, applications whose performance is adversely impacted have a couple of options. They could switch to a GIN index, which doesn't have this bug, or they could replace "arraycol <@ constant-array" with "arraycol <@ constant-array AND arraycol && constant-array". That will provide about the same performance as before, and it will find all non-empty subsets of the given constant-array, which is all that could reliably be expected of the query before. While at it, add some more regression test cases to improve code coverage of contrib/intarray. In passing, adjust resize_intArrayType so that when it's returning an empty array, it uses construct_empty_array for that rather than cowboy hacking on the input array. While the hack produces an array that looks valid for most purposes, it isn't bitwise equal to empty arrays produced by other code paths, which could have subtle odd effects. I don't think this code path is performance-critical enough to justify such shortcuts. (Back-patch this part only as far as v11; before commit 01783ac36 we were not careful about this in other intarray code paths either.) Back-patch the <@ fixes to all supported versions, since this was broken from day one. Patch by me; thanks to Alexander Korotkov for review. Discussion: https://postgr.es/m/458.1565114141@sss.pgh.pa.us
2019-08-07 00:04:51 +02:00
a = construct_empty_array(INT4OID);
return a;
}
2003-06-11 21:31:05 +02:00
if (num == ARRNELEMS(a))
return a;
nbytes = ARR_DATA_OFFSET(a) + sizeof(int) * num;
2003-06-11 21:31:05 +02:00
a = (ArrayType *) repalloc(a, nbytes);
SET_VARSIZE(a, nbytes);
/* usually the array should be 1-D already, but just in case ... */
for (i = 0; i < ARR_NDIM(a); i++)
{
ARR_DIMS(a)[i] = num;
num = 1;
}
2003-06-11 21:31:05 +02:00
return a;
}
ArrayType *
copy_intArrayType(ArrayType *a)
{
ArrayType *r;
int n = ARRNELEMS(a);
2003-06-11 21:31:05 +02:00
r = new_intArrayType(n);
memcpy(ARRPTR(r), ARRPTR(a), n * sizeof(int32));
2003-06-11 21:31:05 +02:00
return r;
}
/* num for compressed key */
int
internal_size(int *a, int len)
{
int i;
int64 size = 0;
2003-06-11 21:31:05 +02:00
for (i = 0; i < len; i += 2)
{
2003-06-11 21:31:05 +02:00
if (!i || a[i] != a[i - 1]) /* do not count repeated range */
size += (int64) (a[i + 1]) - (int64) (a[i]) + 1;
}
2003-06-11 21:31:05 +02:00
if (size > (int64) INT_MAX || size < (int64) INT_MIN)
return -1; /* overflow */
return (int) size;
2003-06-11 21:31:05 +02:00
}
/* unique-ify elements of r in-place ... r must be sorted already */
2003-06-11 21:31:05 +02:00
ArrayType *
_int_unique(ArrayType *r)
{
int num = ARRNELEMS(r);
bool duplicates_found; /* not used */
2003-06-11 21:31:05 +02:00
num = qunique_arg(ARRPTR(r), num, sizeof(int), isort_cmp,
&duplicates_found);
2003-06-11 21:31:05 +02:00
return resize_intArrayType(r, num);
2003-06-11 21:31:05 +02:00
}
void
Implement operator class parameters PostgreSQL provides set of template index access methods, where opclasses have much freedom in the semantics of indexing. These index AMs are GiST, GIN, SP-GiST and BRIN. There opclasses define representation of keys, operations on them and supported search strategies. So, it's natural that opclasses may be faced some tradeoffs, which require user-side decision. This commit implements opclass parameters allowing users to set some values, which tell opclass how to index the particular dataset. This commit doesn't introduce new storage in system catalog. Instead it uses pg_attribute.attoptions, which is used for table column storage options but unused for index attributes. In order to evade changing signature of each opclass support function, we implement unified way to pass options to opclass support functions. Options are set to fn_expr as the constant bytea expression. It's possible due to the fact that opclass support functions are executed outside of expressions, so fn_expr is unused for them. This commit comes with some examples of opclass options usage. We parametrize signature length in GiST. That applies to multiple opclasses: tsvector_ops, gist__intbig_ops, gist_ltree_ops, gist__ltree_ops, gist_trgm_ops and gist_hstore_ops. Also we parametrize maximum number of integer ranges for gist__int_ops. However, the main future usage of this feature is expected to be json, where users would be able to specify which way to index particular json parts. Catversion is bumped. Discussion: https://postgr.es/m/d22c3a18-31c7-1879-fc11-4c1ce2f5e5af%40postgrespro.ru Author: Nikita Glukhov, revised by me Reviwed-by: Nikolay Shaplov, Robert Haas, Tom Lane, Tomas Vondra, Alvaro Herrera
2020-03-30 18:17:11 +02:00
gensign(BITVECP sign, int *a, int len, int siglen)
2003-06-11 21:31:05 +02:00
{
int i;
/* we assume that the sign vector is previously zeroed */
for (i = 0; i < len; i++)
{
Implement operator class parameters PostgreSQL provides set of template index access methods, where opclasses have much freedom in the semantics of indexing. These index AMs are GiST, GIN, SP-GiST and BRIN. There opclasses define representation of keys, operations on them and supported search strategies. So, it's natural that opclasses may be faced some tradeoffs, which require user-side decision. This commit implements opclass parameters allowing users to set some values, which tell opclass how to index the particular dataset. This commit doesn't introduce new storage in system catalog. Instead it uses pg_attribute.attoptions, which is used for table column storage options but unused for index attributes. In order to evade changing signature of each opclass support function, we implement unified way to pass options to opclass support functions. Options are set to fn_expr as the constant bytea expression. It's possible due to the fact that opclass support functions are executed outside of expressions, so fn_expr is unused for them. This commit comes with some examples of opclass options usage. We parametrize signature length in GiST. That applies to multiple opclasses: tsvector_ops, gist__intbig_ops, gist_ltree_ops, gist__ltree_ops, gist_trgm_ops and gist_hstore_ops. Also we parametrize maximum number of integer ranges for gist__int_ops. However, the main future usage of this feature is expected to be json, where users would be able to specify which way to index particular json parts. Catversion is bumped. Discussion: https://postgr.es/m/d22c3a18-31c7-1879-fc11-4c1ce2f5e5af%40postgrespro.ru Author: Nikita Glukhov, revised by me Reviwed-by: Nikolay Shaplov, Robert Haas, Tom Lane, Tomas Vondra, Alvaro Herrera
2020-03-30 18:17:11 +02:00
HASH(sign, *a, siglen);
2003-06-11 21:31:05 +02:00
a++;
}
}
int32
intarray_match_first(ArrayType *a, int32 elem)
{
int32 *aa,
c,
i;
CHECKARRVALID(a);
c = ARRNELEMS(a);
2003-06-11 21:31:05 +02:00
aa = ARRPTR(a);
for (i = 0; i < c; i++)
if (aa[i] == elem)
return (i + 1);
return 0;
}
ArrayType *
intarray_add_elem(ArrayType *a, int32 elem)
{
ArrayType *result;
int32 *r;
int32 c;
2003-06-11 21:31:05 +02:00
CHECKARRVALID(a);
c = ARRNELEMS(a);
2003-06-11 21:31:05 +02:00
result = new_intArrayType(c + 1);
r = ARRPTR(result);
if (c > 0)
memcpy(r, ARRPTR(a), c * sizeof(int32));
r[c] = elem;
return result;
}
ArrayType *
intarray_concat_arrays(ArrayType *a, ArrayType *b)
{
ArrayType *result;
int32 ac = ARRNELEMS(a);
int32 bc = ARRNELEMS(b);
2003-06-11 21:31:05 +02:00
CHECKARRVALID(a);
CHECKARRVALID(b);
2003-06-11 21:31:05 +02:00
result = new_intArrayType(ac + bc);
if (ac)
memcpy(ARRPTR(result), ARRPTR(a), ac * sizeof(int32));
if (bc)
memcpy(ARRPTR(result) + ac, ARRPTR(b), bc * sizeof(int32));
return result;
}
ArrayType *
int_to_intset(int32 elem)
2003-06-11 21:31:05 +02:00
{
ArrayType *result;
int32 *aa;
result = new_intArrayType(1);
aa = ARRPTR(result);
aa[0] = elem;
2003-06-11 21:31:05 +02:00
return result;
}
int
compASC(const void *a, const void *b)
{
return pg_cmp_s32(*(const int32 *) a, *(const int32 *) b);
2003-06-11 21:31:05 +02:00
}
int
compDESC(const void *a, const void *b)
{
return pg_cmp_s32(*(const int32 *) b, *(const int32 *) a);
2003-06-11 21:31:05 +02:00
}