postgresql/contrib/ltree/ltree_op.c

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

591 lines
11 KiB
C
Raw Normal View History

/*
* op function for ltree
* Teodor Sigaev <teodor@stack.net>
2010-09-21 04:08:53 +08:00
* contrib/ltree/ltree_op.c
*/
#include "postgres.h"
#include <ctype.h>
#include "access/htup_details.h"
#include "catalog/pg_statistic.h"
#include "ltree.h"
#include "utils/builtins.h"
#include "utils/lsyscache.h"
#include "utils/selfuncs.h"
PG_MODULE_MAGIC;
/* compare functions */
PG_FUNCTION_INFO_V1(ltree_cmp);
PG_FUNCTION_INFO_V1(ltree_lt);
PG_FUNCTION_INFO_V1(ltree_le);
PG_FUNCTION_INFO_V1(ltree_eq);
PG_FUNCTION_INFO_V1(ltree_ne);
PG_FUNCTION_INFO_V1(ltree_ge);
PG_FUNCTION_INFO_V1(ltree_gt);
PG_FUNCTION_INFO_V1(nlevel);
PG_FUNCTION_INFO_V1(ltree_isparent);
PG_FUNCTION_INFO_V1(ltree_risparent);
PG_FUNCTION_INFO_V1(subltree);
PG_FUNCTION_INFO_V1(subpath);
PG_FUNCTION_INFO_V1(ltree_index);
PG_FUNCTION_INFO_V1(ltree_addltree);
PG_FUNCTION_INFO_V1(ltree_addtext);
PG_FUNCTION_INFO_V1(ltree_textadd);
PG_FUNCTION_INFO_V1(lca);
PG_FUNCTION_INFO_V1(ltree2text);
PG_FUNCTION_INFO_V1(text2ltree);
PG_FUNCTION_INFO_V1(ltreeparentsel);
int
ltree_compare(const ltree *a, const ltree *b)
{
ltree_level *al = LTREE_FIRST(a);
ltree_level *bl = LTREE_FIRST(b);
int an = a->numlevel;
int bn = b->numlevel;
2002-09-05 04:31:48 +08:00
while (an > 0 && bn > 0)
{
int res;
if ((res = memcmp(al->name, bl->name, Min(al->len, bl->len))) == 0)
{
if (al->len != bl->len)
return (al->len - bl->len) * 10 * (an + 1);
}
else
{
if (res < 0)
res = -1;
else
res = 1;
return res * 10 * (an + 1);
}
2002-09-05 04:31:48 +08:00
an--;
bn--;
al = LEVEL_NEXT(al);
bl = LEVEL_NEXT(bl);
}
return (a->numlevel - b->numlevel) * 10 * (an + 1);
}
#define RUNCMP \
ltree *a = PG_GETARG_LTREE_P(0); \
ltree *b = PG_GETARG_LTREE_P(1); \
int res = ltree_compare(a,b); \
PG_FREE_IF_COPY(a,0); \
PG_FREE_IF_COPY(b,1)
Datum
ltree_cmp(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_INT32(res);
}
Datum
ltree_lt(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res < 0);
}
Datum
ltree_le(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res <= 0);
}
Datum
ltree_eq(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res == 0);
}
Datum
ltree_ge(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res >= 0);
}
Datum
ltree_gt(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res > 0);
}
Datum
ltree_ne(PG_FUNCTION_ARGS)
{
RUNCMP;
PG_RETURN_BOOL(res != 0);
}
Datum
nlevel(PG_FUNCTION_ARGS)
{
ltree *a = PG_GETARG_LTREE_P(0);
int res = a->numlevel;
2002-09-05 04:31:48 +08:00
PG_FREE_IF_COPY(a, 0);
PG_RETURN_INT32(res);
}
bool
inner_isparent(const ltree *c, const ltree *p)
{
ltree_level *cl = LTREE_FIRST(c);
ltree_level *pl = LTREE_FIRST(p);
int pn = p->numlevel;
if (pn > c->numlevel)
return false;
while (pn > 0)
{
if (cl->len != pl->len)
return false;
if (memcmp(cl->name, pl->name, cl->len) != 0)
return false;
pn--;
cl = LEVEL_NEXT(cl);
pl = LEVEL_NEXT(pl);
}
return true;
}
Datum
ltree_isparent(PG_FUNCTION_ARGS)
{
ltree *c = PG_GETARG_LTREE_P(1);
ltree *p = PG_GETARG_LTREE_P(0);
bool res = inner_isparent(c, p);
2002-09-05 04:31:48 +08:00
PG_FREE_IF_COPY(c, 1);
PG_FREE_IF_COPY(p, 0);
PG_RETURN_BOOL(res);
}
Datum
ltree_risparent(PG_FUNCTION_ARGS)
{
ltree *c = PG_GETARG_LTREE_P(0);
ltree *p = PG_GETARG_LTREE_P(1);
bool res = inner_isparent(c, p);
2002-09-05 04:31:48 +08:00
PG_FREE_IF_COPY(c, 0);
PG_FREE_IF_COPY(p, 1);
PG_RETURN_BOOL(res);
}
static ltree *
inner_subltree(ltree *t, int32 startpos, int32 endpos)
{
char *start = NULL,
*end = NULL;
ltree_level *ptr = LTREE_FIRST(t);
ltree *res;
int i;
if (startpos < 0 || endpos < 0 || startpos >= t->numlevel || startpos > endpos)
ereport(ERROR,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("invalid positions")));
if (endpos > t->numlevel)
endpos = t->numlevel;
start = end = (char *) ptr;
for (i = 0; i < endpos; i++)
{
if (i == startpos)
start = (char *) ptr;
if (i == endpos - 1)
{
end = (char *) LEVEL_NEXT(ptr);
break;
}
ptr = LEVEL_NEXT(ptr);
}
res = (ltree *) palloc0(LTREE_HDRSIZE + (end - start));
SET_VARSIZE(res, LTREE_HDRSIZE + (end - start));
res->numlevel = endpos - startpos;
2002-09-05 04:31:48 +08:00
memcpy(LTREE_FIRST(res), start, end - start);
return res;
}
Datum
subltree(PG_FUNCTION_ARGS)
{
ltree *t = PG_GETARG_LTREE_P(0);
ltree *res = inner_subltree(t, PG_GETARG_INT32(1), PG_GETARG_INT32(2));
PG_FREE_IF_COPY(t, 0);
PG_RETURN_POINTER(res);
}
Datum
subpath(PG_FUNCTION_ARGS)
{
ltree *t = PG_GETARG_LTREE_P(0);
int32 start = PG_GETARG_INT32(1);
int32 len = (fcinfo->nargs == 3) ? PG_GETARG_INT32(2) : 0;
int32 end;
ltree *res;
2002-09-05 04:31:48 +08:00
end = start + len;
2002-09-05 04:31:48 +08:00
if (start < 0)
{
start = t->numlevel + start;
end = start + len;
}
if (start < 0)
{ /* start > t->numlevel */
start = t->numlevel + start;
end = start + len;
}
if (len < 0)
end = t->numlevel + len;
else if (len == 0)
end = (fcinfo->nargs == 3) ? start : 0xffff;
res = inner_subltree(t, start, end);
PG_FREE_IF_COPY(t, 0);
PG_RETURN_POINTER(res);
}
static ltree *
ltree_concat(ltree *a, ltree *b)
2002-09-05 04:31:48 +08:00
{
ltree *r;
int numlevel = (int) a->numlevel + b->numlevel;
if (numlevel > LTREE_MAX_LEVELS)
ereport(ERROR,
(errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED),
errmsg("number of ltree levels (%d) exceeds the maximum allowed (%d)",
numlevel, LTREE_MAX_LEVELS)));
2002-09-05 04:31:48 +08:00
r = (ltree *) palloc0(VARSIZE(a) + VARSIZE(b) - LTREE_HDRSIZE);
SET_VARSIZE(r, VARSIZE(a) + VARSIZE(b) - LTREE_HDRSIZE);
r->numlevel = (uint16) numlevel;
2002-09-05 04:31:48 +08:00
memcpy(LTREE_FIRST(r), LTREE_FIRST(a), VARSIZE(a) - LTREE_HDRSIZE);
memcpy(((char *) LTREE_FIRST(r)) + VARSIZE(a) - LTREE_HDRSIZE,
LTREE_FIRST(b),
VARSIZE(b) - LTREE_HDRSIZE);
return r;
2002-09-05 04:31:48 +08:00
}
Datum
ltree_addltree(PG_FUNCTION_ARGS)
{
ltree *a = PG_GETARG_LTREE_P(0);
ltree *b = PG_GETARG_LTREE_P(1);
ltree *r;
r = ltree_concat(a, b);
PG_FREE_IF_COPY(a, 0);
PG_FREE_IF_COPY(b, 1);
PG_RETURN_POINTER(r);
}
Datum
ltree_addtext(PG_FUNCTION_ARGS)
{
ltree *a = PG_GETARG_LTREE_P(0);
text *b = PG_GETARG_TEXT_PP(1);
char *s;
ltree *r,
*tmp;
2002-09-05 04:31:48 +08:00
s = text_to_cstring(b);
tmp = (ltree *) DatumGetPointer(DirectFunctionCall1(ltree_in,
PointerGetDatum(s)));
pfree(s);
r = ltree_concat(a, tmp);
pfree(tmp);
2002-09-05 04:31:48 +08:00
PG_FREE_IF_COPY(a, 0);
PG_FREE_IF_COPY(b, 1);
PG_RETURN_POINTER(r);
}
Datum
ltree_index(PG_FUNCTION_ARGS)
{
ltree *a = PG_GETARG_LTREE_P(0);
ltree *b = PG_GETARG_LTREE_P(1);
int start = (fcinfo->nargs == 3) ? PG_GETARG_INT32(2) : 0;
int i,
j;
ltree_level *startptr,
*aptr,
*bptr;
bool found = false;
2003-08-04 08:43:34 +08:00
if (start < 0)
{
if (-start >= a->numlevel)
start = 0;
else
start = (int) (a->numlevel) + start;
}
if (a->numlevel - start < b->numlevel || a->numlevel == 0 || b->numlevel == 0)
{
PG_FREE_IF_COPY(a, 0);
PG_FREE_IF_COPY(b, 1);
PG_RETURN_INT32(-1);
}
startptr = LTREE_FIRST(a);
for (i = 0; i <= a->numlevel - b->numlevel; i++)
{
if (i >= start)
{
aptr = startptr;
bptr = LTREE_FIRST(b);
for (j = 0; j < b->numlevel; j++)
{
if (!(aptr->len == bptr->len && memcmp(aptr->name, bptr->name, aptr->len) == 0))
break;
aptr = LEVEL_NEXT(aptr);
bptr = LEVEL_NEXT(bptr);
}
2003-08-04 08:43:34 +08:00
if (j == b->numlevel)
{
found = true;
break;
}
}
startptr = LEVEL_NEXT(startptr);
}
2003-08-04 08:43:34 +08:00
if (!found)
i = -1;
PG_FREE_IF_COPY(a, 0);
PG_FREE_IF_COPY(b, 1);
PG_RETURN_INT32(i);
}
Datum
ltree_textadd(PG_FUNCTION_ARGS)
{
ltree *a = PG_GETARG_LTREE_P(1);
text *b = PG_GETARG_TEXT_PP(0);
char *s;
ltree *r,
*tmp;
2002-09-05 04:31:48 +08:00
s = text_to_cstring(b);
2002-09-05 04:31:48 +08:00
tmp = (ltree *) DatumGetPointer(DirectFunctionCall1(ltree_in,
PointerGetDatum(s)));
2002-09-05 04:31:48 +08:00
pfree(s);
2002-09-05 04:31:48 +08:00
r = ltree_concat(tmp, a);
2002-09-05 04:31:48 +08:00
pfree(tmp);
2002-09-05 04:31:48 +08:00
PG_FREE_IF_COPY(a, 1);
PG_FREE_IF_COPY(b, 0);
PG_RETURN_POINTER(r);
}
/*
* Common code for variants of lca(), find longest common ancestor of inputs
*
* Returns NULL if there is no common ancestor, ie, the longest common
* prefix is empty.
*/
ltree *
lca_inner(ltree **a, int len)
{
int tmp,
num,
i,
reslen;
ltree **ptr;
ltree_level *l1,
*l2;
ltree *res;
2002-09-05 04:31:48 +08:00
if (len <= 0)
return NULL; /* no inputs? */
if ((*a)->numlevel == 0)
return NULL; /* any empty input means NULL result */
/* num is the length of the longest common ancestor so far */
num = (*a)->numlevel - 1;
/* Compare each additional input to *a */
ptr = a + 1;
while (ptr - a < len)
{
if ((*ptr)->numlevel == 0)
return NULL;
else if ((*ptr)->numlevel == 1)
num = 0;
else
{
l1 = LTREE_FIRST(*a);
l2 = LTREE_FIRST(*ptr);
tmp = Min(num, (*ptr)->numlevel - 1);
num = 0;
for (i = 0; i < tmp; i++)
{
if (l1->len == l2->len &&
memcmp(l1->name, l2->name, l1->len) == 0)
num = i + 1;
else
break;
l1 = LEVEL_NEXT(l1);
l2 = LEVEL_NEXT(l2);
}
}
ptr++;
}
/* Now compute size of result ... */
reslen = LTREE_HDRSIZE;
l1 = LTREE_FIRST(*a);
for (i = 0; i < num; i++)
{
reslen += MAXALIGN(l1->len + LEVEL_HDRSIZE);
l1 = LEVEL_NEXT(l1);
}
/* ... and construct it by copying from *a */
res = (ltree *) palloc0(reslen);
SET_VARSIZE(res, reslen);
res->numlevel = num;
l1 = LTREE_FIRST(*a);
l2 = LTREE_FIRST(res);
for (i = 0; i < num; i++)
{
memcpy(l2, l1, MAXALIGN(l1->len + LEVEL_HDRSIZE));
l1 = LEVEL_NEXT(l1);
l2 = LEVEL_NEXT(l2);
}
return res;
}
Datum
lca(PG_FUNCTION_ARGS)
{
int i;
ltree **a,
*res;
2002-09-05 04:31:48 +08:00
a = (ltree **) palloc(sizeof(ltree *) * fcinfo->nargs);
for (i = 0; i < fcinfo->nargs; i++)
a[i] = PG_GETARG_LTREE_P(i);
res = lca_inner(a, (int) fcinfo->nargs);
for (i = 0; i < fcinfo->nargs; i++)
PG_FREE_IF_COPY(a[i], i);
pfree(a);
2002-09-05 04:31:48 +08:00
if (res)
PG_RETURN_POINTER(res);
else
PG_RETURN_NULL();
}
Datum
text2ltree(PG_FUNCTION_ARGS)
{
text *in = PG_GETARG_TEXT_PP(0);
char *s;
ltree *out;
s = text_to_cstring(in);
out = (ltree *) DatumGetPointer(DirectFunctionCall1(ltree_in,
PointerGetDatum(s)));
pfree(s);
PG_FREE_IF_COPY(in, 0);
PG_RETURN_POINTER(out);
}
Datum
ltree2text(PG_FUNCTION_ARGS)
{
ltree *in = PG_GETARG_LTREE_P(0);
char *ptr;
int i;
ltree_level *curlevel;
text *out;
2003-08-04 08:43:34 +08:00
out = (text *) palloc(VARSIZE(in) + VARHDRSZ);
ptr = VARDATA(out);
curlevel = LTREE_FIRST(in);
for (i = 0; i < in->numlevel; i++)
{
if (i != 0)
{
*ptr = '.';
ptr++;
}
memcpy(ptr, curlevel->name, curlevel->len);
ptr += curlevel->len;
curlevel = LEVEL_NEXT(curlevel);
}
2003-08-04 08:43:34 +08:00
SET_VARSIZE(out, ptr - ((char *) out));
PG_FREE_IF_COPY(in, 0);
2003-08-04 08:43:34 +08:00
PG_RETURN_POINTER(out);
}
/*
* ltreeparentsel - Selectivity of parent relationship for ltree data types.
Improve selectivity estimation for assorted match-style operators. Quite a few matching operators such as JSONB's @> used "contsel" and "contjoinsel" as their selectivity estimators. That was a bad idea, because (a) contsel is only a stub, yielding a fixed default estimate, and (b) that default is 0.001, meaning we estimate these operators as five times more selective than equality, which is surely pretty silly. There's a good model for improving this in ltree's ltreeparentsel(): for any "var OP constant" query, we can try applying the operator to all of the column's MCV and histogram values, taking the latter as being a random sample of the non-MCV values. That code is actually 100% generic, except for the question of exactly what default selectivity ought to be plugged in when we don't have stats. Hence, migrate the guts of ltreeparentsel() into the core code, provide wrappers "matchingsel" and "matchingjoinsel" with a more-appropriate default estimate, and use those for the non-geometric operators that formerly used contsel (mostly JSONB containment operators and tsquery matching). Also apply this code to some match-like operators in hstore, ltree, and pg_trgm, including the former users of ltreeparentsel as well as ones that improperly used contsel. Since commit 911e70207 just created new versions of those extensions that we haven't released yet, we can sneak this change into those new versions instead of having to create an additional generation of update scripts. Patch by me, reviewed by Alexey Bashtanov Discussion: https://postgr.es/m/12237.1582833074@sss.pgh.pa.us
2020-04-01 22:32:33 +08:00
*
* This function is not used anymore, if the ltree extension has been
* updated to 1.2 or later.
*/
Datum
ltreeparentsel(PG_FUNCTION_ARGS)
{
PlannerInfo *root = (PlannerInfo *) PG_GETARG_POINTER(0);
Oid operator = PG_GETARG_OID(1);
List *args = (List *) PG_GETARG_POINTER(2);
int varRelid = PG_GETARG_INT32(3);
double selec;
Improve selectivity estimation for assorted match-style operators. Quite a few matching operators such as JSONB's @> used "contsel" and "contjoinsel" as their selectivity estimators. That was a bad idea, because (a) contsel is only a stub, yielding a fixed default estimate, and (b) that default is 0.001, meaning we estimate these operators as five times more selective than equality, which is surely pretty silly. There's a good model for improving this in ltree's ltreeparentsel(): for any "var OP constant" query, we can try applying the operator to all of the column's MCV and histogram values, taking the latter as being a random sample of the non-MCV values. That code is actually 100% generic, except for the question of exactly what default selectivity ought to be plugged in when we don't have stats. Hence, migrate the guts of ltreeparentsel() into the core code, provide wrappers "matchingsel" and "matchingjoinsel" with a more-appropriate default estimate, and use those for the non-geometric operators that formerly used contsel (mostly JSONB containment operators and tsquery matching). Also apply this code to some match-like operators in hstore, ltree, and pg_trgm, including the former users of ltreeparentsel as well as ones that improperly used contsel. Since commit 911e70207 just created new versions of those extensions that we haven't released yet, we can sneak this change into those new versions instead of having to create an additional generation of update scripts. Patch by me, reviewed by Alexey Bashtanov Discussion: https://postgr.es/m/12237.1582833074@sss.pgh.pa.us
2020-04-01 22:32:33 +08:00
/* Use generic restriction selectivity logic, with default 0.001. */
Use query collation, not column's collation, while examining statistics. Commit 5e0928005 changed the planner so that, instead of blindly using DEFAULT_COLLATION_OID when invoking operators for selectivity estimation, it would use the collation of the column whose statistics we're considering. This was recognized as still being not quite the right thing, but it seemed like a good incremental improvement. However, shortly thereafter we introduced nondeterministic collations, and that creates cases where operators can fail if they're passed the wrong collation. We don't want planning to fail in cases where the query itself would work, so this means that we *must* use the query's collation when invoking operators for estimation purposes. The only real problem this creates is in ineq_histogram_selectivity, where the binary search might produce a garbage answer if we perform comparisons using a different collation than the column's histogram is ordered with. However, when the query's collation is significantly different from the column's default collation, the estimate we previously generated would be pretty irrelevant anyway; so it's not clear that this will result in noticeably worse estimates in practice. (A follow-on patch will improve this situation in HEAD, but it seems too invasive for back-patch.) The patch requires changing the signatures of mcv_selectivity and allied functions, which are exported and very possibly are used by extensions. In HEAD, I just did that, but an API/ABI break of this sort isn't acceptable in stable branches. Therefore, in v12 the patch introduces "mcv_selectivity_ext" and so on, with signatures matching HEAD, and makes the old functions into wrappers that assume DEFAULT_COLLATION_OID should be used. That does not match the prior behavior, but it should avoid risk of failure in most cases. (In practice, I think most extension datatypes aren't collation-aware, so the change probably doesn't matter to them.) Per report from James Lucas. Back-patch to v12 where the problem was introduced. Discussion: https://postgr.es/m/CAAFmbbOvfi=wMM=3qRsPunBSLb8BFREno2oOzSBS=mzfLPKABw@mail.gmail.com
2020-06-06 04:18:50 +08:00
selec = generic_restriction_selectivity(root, operator, InvalidOid,
Improve selectivity estimation for assorted match-style operators. Quite a few matching operators such as JSONB's @> used "contsel" and "contjoinsel" as their selectivity estimators. That was a bad idea, because (a) contsel is only a stub, yielding a fixed default estimate, and (b) that default is 0.001, meaning we estimate these operators as five times more selective than equality, which is surely pretty silly. There's a good model for improving this in ltree's ltreeparentsel(): for any "var OP constant" query, we can try applying the operator to all of the column's MCV and histogram values, taking the latter as being a random sample of the non-MCV values. That code is actually 100% generic, except for the question of exactly what default selectivity ought to be plugged in when we don't have stats. Hence, migrate the guts of ltreeparentsel() into the core code, provide wrappers "matchingsel" and "matchingjoinsel" with a more-appropriate default estimate, and use those for the non-geometric operators that formerly used contsel (mostly JSONB containment operators and tsquery matching). Also apply this code to some match-like operators in hstore, ltree, and pg_trgm, including the former users of ltreeparentsel as well as ones that improperly used contsel. Since commit 911e70207 just created new versions of those extensions that we haven't released yet, we can sneak this change into those new versions instead of having to create an additional generation of update scripts. Patch by me, reviewed by Alexey Bashtanov Discussion: https://postgr.es/m/12237.1582833074@sss.pgh.pa.us
2020-04-01 22:32:33 +08:00
args, varRelid,
0.001);
PG_RETURN_FLOAT8((float8) selec);
}