mirror/postgresql
2
0
Fork 0
mirror of https://git.postgresql.org/git/postgresql.git synced 2025-03-13 19:57:53 +08:00
Code Issues Packages Projects Releases Wiki Activity

Add support for nearest-neighbor (KNN) searches to SP-GiST

Browse Source 
Currently, KNN searches were supported only by GiST.  SP-GiST also capable to
support them.  This commit implements that support.  SP-GiST scan stack is
replaced with queue, which serves as stack if no ordering is specified.  KNN
support is provided for three SP-GIST opclasses: quad_point_ops, kd_point_ops
and poly_ops (catversion is bumped).  Some common parts between GiST and SP-GiST
KNNs are extracted into separate functions.

Discussion: https://postgr.es/m/570825e8-47d0-4732-2bf6-88d67d2d51c8%40postgrespro.ru
Author: Nikita Glukhov, Alexander Korotkov based on GSoC work by Vlad Sterzhanov
Review: Andrey Borodin, Alexander Korotkov
This commit is contained in:
Alexander Korotkov 2018-09-19 01:54:10 +03:00
parent d0cfc3d6a4
commit 2a6368343f
29 changed files with 1681 additions and 428 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
doc/src/sgml/indices.sgml
View File

@ -281,6 +281,13 @@ SELECT * FROM places ORDER BY location <-> point '(101,456)' LIMIT 10;
For more information see <xref linkend="spgist"/>.
</para>
<para>
Like GiST, SP-GiST supports <quote>nearest-neighbor</quote> searches.
For SP-GiST operator classes that support distance ordering, the
corresponding operator is specified in the <quote>Ordering Operators</quote>
column in <xref linkend="spgist-builtin-opclasses-table"/>.
</para>
<para>
<indexterm>
<primary>index</primary>

58
doc/src/sgml/spgist.sgml
View File

@ -64,12 +64,13 @@
<table id="spgist-builtin-opclasses-table">
<title>Built-in <acronym>SP-GiST</acronym> Operator Classes</title>
<tgroup cols="3">
<tgroup cols="4">
<thead>
<row>
<entry>Name</entry>
<entry>Indexed Data Type</entry>
<entry>Indexable Operators</entry>
<entry>Ordering Operators</entry>
</row>
</thead>
<tbody>
@ -84,6 +85,9 @@
<literal>&gt;^</literal>
<literal>~=</literal>
</entry>
<entry>
<literal>&lt;-&gt;</literal>
</entry>
</row>
<row>
<entry><literal>quad_point_ops</literal></entry>
@ -96,6 +100,9 @@
<literal>&gt;^</literal>
<literal>~=</literal>
</entry>
<entry>
<literal>&lt;-&gt;</literal>
</entry>
</row>
<row>
<entry><literal>range_ops</literal></entry>
@ -111,6 +118,8 @@
<literal>&gt;&gt;</literal>
<literal>@&gt;</literal>
</entry>
<entry>
</entry>
</row>
<row>
<entry><literal>box_ops</literal></entry>
@ -129,6 +138,8 @@
<literal>|&gt;&gt;</literal>
<literal>|&amp;&gt;</literal>
</entry>
<entry>
</entry>
</row>
<row>
<entry><literal>poly_ops</literal></entry>
@ -147,6 +158,9 @@
<literal>|&gt;&gt;</literal>
<literal>|&amp;&gt;</literal>
</entry>
<entry>
<literal>&lt;-&gt;</literal>
</entry>
</row>
<row>
<entry><literal>text_ops</literal></entry>
@ -163,6 +177,8 @@
<literal>~&gt;~</literal>
<literal>^@</literal>
</entry>
<entry>
</entry>
</row>
<row>
<entry><literal>inet_ops</literal></entry>
@ -180,6 +196,8 @@
<literal>&lt;=</literal>
<literal>=</literal>
</entry>
<entry>
</entry>
</row>
</tbody>
</tgroup>
@ -191,6 +209,12 @@
supports the same operators but uses a different index data structure which
may offer better performance in some applications.
</para>
<para>
The <literal>quad_point_ops</literal>, <literal>kd_point_ops</literal> and
<literal>poly_ops</literal> operator classes support the <literal>&lt;-&gt;</literal>
ordering operator, which enables the k-nearest neighbor (<literal>k-NN</literal>)
search over indexed point or polygon datasets.
</para>
</sect1>
@ -630,7 +654,10 @@ CREATE FUNCTION my_inner_consistent(internal, internal) RETURNS void ...
typedef struct spgInnerConsistentIn
{
ScanKey scankeys; /* array of operators and comparison values */
int nkeys; /* length of array */
ScanKey orderbys; /* array of ordering operators and comparison
* values */
int nkeys; /* length of scankeys array */
int norderbys; /* length of orderbys array */
Datum reconstructedValue; /* value reconstructed at parent */
void *traversalValue; /* opclass-specific traverse value */
@ -653,6 +680,7 @@ typedef struct spgInnerConsistentOut
int *levelAdds; /* increment level by this much for each */
Datum *reconstructedValues; /* associated reconstructed values */
void **traversalValues; /* opclass-specific traverse values */
double **distances; /* associated distances */
} spgInnerConsistentOut;
</programlisting>
@ -667,6 +695,8 @@ typedef struct spgInnerConsistentOut
In particular it is not necessary to check <structfield>sk_flags</structfield> to
see if the comparison value is NULL, because the SP-GiST core code
will filter out such conditions.
The array <structfield>orderbys</structfield>, of length <structfield>norderbys</structfield>,
describes ordering operators (if any) in the same manner.
<structfield>reconstructedValue</structfield> is the value reconstructed for the
parent tuple; it is <literal>(Datum) 0</literal> at the root level or if the
<function>inner_consistent</function> function did not provide a value at the
@ -709,6 +739,10 @@ typedef struct spgInnerConsistentOut
of <structname>spgConfigOut</structname>.<structfield>leafType</structfield> type
reconstructed for each child node to be visited; otherwise, leave
<structfield>reconstructedValues</structfield> as NULL.
If ordered search is performed, set <structfield>distances</structfield>
to an array of distance values according to <structfield>orderbys</structfield>
array (nodes with lowest distances will be processed first). Leave it
NULL otherwise.
If it is desired to pass down additional out-of-band information
(<quote>traverse values</quote>) to lower levels of the tree search,
set <structfield>traversalValues</structfield> to an array of the appropriate
@ -717,6 +751,7 @@ typedef struct spgInnerConsistentOut
Note that the <function>inner_consistent</function> function is
responsible for palloc'ing the
<structfield>nodeNumbers</structfield>, <structfield>levelAdds</structfield>,
<structfield>distances</structfield>,
<structfield>reconstructedValues</structfield>, and
<structfield>traversalValues</structfield> arrays in the current memory context.
However, any output traverse values pointed to by
@ -747,7 +782,10 @@ CREATE FUNCTION my_leaf_consistent(internal, internal) RETURNS bool ...
typedef struct spgLeafConsistentIn
{
ScanKey scankeys; /* array of operators and comparison values */
int nkeys; /* length of array */
ScanKey orderbys; /* array of ordering operators and comparison
* values */
int nkeys; /* length of scankeys array */
int norderbys; /* length of orderbys array */
Datum reconstructedValue; /* value reconstructed at parent */
void *traversalValue; /* opclass-specific traverse value */
@ -759,8 +797,10 @@ typedef struct spgLeafConsistentIn
typedef struct spgLeafConsistentOut
{
Datum leafValue; /* reconstructed original data, if any */
bool recheck; /* set true if operator must be rechecked */
Datum leafValue; /* reconstructed original data, if any */
bool recheck; /* set true if operator must be rechecked */
bool recheckDistances; /* set true if distances must be rechecked */
double *distances; /* associated distances */
} spgLeafConsistentOut;
</programlisting>
@ -775,6 +815,8 @@ typedef struct spgLeafConsistentOut
In particular it is not necessary to check <structfield>sk_flags</structfield> to
see if the comparison value is NULL, because the SP-GiST core code
will filter out such conditions.
The array <structfield>orderbys</structfield>, of length <structfield>norderbys</structfield>,
describes the ordering operators in the same manner.
<structfield>reconstructedValue</structfield> is the value reconstructed for the
parent tuple; it is <literal>(Datum) 0</literal> at the root level or if the
<function>inner_consistent</function> function did not provide a value at the
@ -803,6 +845,12 @@ typedef struct spgLeafConsistentOut
<structfield>recheck</structfield> may be set to <literal>true</literal> if the match
is uncertain and so the operator(s) must be re-applied to the actual
heap tuple to verify the match.
If ordered search is performed, set <structfield>distances</structfield>
to an array of distance values according to <structfield>orderbys</structfield>
array. Leave it NULL otherwise. If at least one of returned distances
is not exact, set <structfield>recheckDistances</structfield> to true.
In this case, the executor will calculate the exact distances after
fetching the tuple from the heap, and will reorder the tuples if needed.
</para>
</listitem>
</varlistentry>

2
doc/src/sgml/xindex.sgml
View File

@ -1242,7 +1242,7 @@ SELECT sum(x) OVER (ORDER BY x RANGE BETWEEN 5 PRECEDING AND 10 FOLLOWING)
<title>Ordering Operators</title>
<para>
Some index access methods (currently, only GiST) support the concept of
Some index access methods (currently, only GiST and SP-GiST) support the concept of
<firstterm>ordering operators</firstterm>. What we have been discussing so far
are <firstterm>search operators</firstterm>. A search operator is one for which
the index can be searched to find all rows satisfying

46
src/backend/access/gist/gistget.c
View File

@ -14,9 +14,9 @@
*/
#include "postgres.h"
#include "access/genam.h"
#include "access/gist_private.h"
#include "access/relscan.h"
#include "catalog/pg_type.h"
#include "miscadmin.h"
#include "storage/lmgr.h"
#include "storage/predicate.h"
@ -543,7 +543,6 @@ getNextNearest(IndexScanDesc scan)
{
GISTScanOpaque so = (GISTScanOpaque) scan->opaque;
bool res = false;
int i;
if (scan->xs_hitup)
{
@ -564,45 +563,10 @@ getNextNearest(IndexScanDesc scan)
/* found a heap item at currently minimal distance */
scan->xs_ctup.t_self = item->data.heap.heapPtr;
scan->xs_recheck = item->data.heap.recheck;
scan->xs_recheckorderby = item->data.heap.recheckDistances;
for (i = 0; i < scan->numberOfOrderBys; i++)
{
if (so->orderByTypes[i] == FLOAT8OID)
{
#ifndef USE_FLOAT8_BYVAL
/* must free any old value to avoid memory leakage */
if (!scan->xs_orderbynulls[i])
pfree(DatumGetPointer(scan->xs_orderbyvals[i]));
#endif
scan->xs_orderbyvals[i] = Float8GetDatum(item->distances[i]);
scan->xs_orderbynulls[i] = false;
}
else if (so->orderByTypes[i] == FLOAT4OID)
{
/* convert distance function's result to ORDER BY type */
#ifndef USE_FLOAT4_BYVAL
/* must free any old value to avoid memory leakage */
if (!scan->xs_orderbynulls[i])
pfree(DatumGetPointer(scan->xs_orderbyvals[i]));
#endif
scan->xs_orderbyvals[i] = Float4GetDatum((float4) item->distances[i]);
scan->xs_orderbynulls[i] = false;
}
else
{
/*
* If the ordering operator's return value is anything
* else, we don't know how to convert the float8 bound
* calculated by the distance function to that. The
* executor won't actually need the order by values we
* return here, if there are no lossy results, so only
* insist on converting if the *recheck flag is set.
*/
if (scan->xs_recheckorderby)
elog(ERROR, "GiST operator family's FOR ORDER BY operator must return float8 or float4 if the distance function is lossy");
scan->xs_orderbynulls[i] = true;
}
}
index_store_float8_orderby_distances(scan, so->orderByTypes,
item->distances,
item->data.heap.recheckDistances);
/* in an index-only scan, also return the reconstructed tuple. */
if (scan->xs_want_itup)

37
src/backend/access/gist/gistutil.c
View File

@ -23,6 +23,7 @@
#include "storage/lmgr.h"
#include "utils/float.h"
#include "utils/syscache.h"
#include "utils/lsyscache.h"
/*
@ -871,12 +872,6 @@ gistproperty(Oid index_oid, int attno,
IndexAMProperty prop, const char *propname,
bool *res, bool *isnull)
{
HeapTuple tuple;
Form_pg_index rd_index PG_USED_FOR_ASSERTS_ONLY;
Form_pg_opclass rd_opclass;
Datum datum;
bool disnull;
oidvector *indclass;
Oid opclass,
opfamily,
opcintype;
@ -910,41 +905,19 @@ gistproperty(Oid index_oid, int attno,
}
/* First we need to know the column's opclass. */
tuple = SearchSysCache1(INDEXRELID, ObjectIdGetDatum(index_oid));
if (!HeapTupleIsValid(tuple))
opclass = get_index_column_opclass(index_oid, attno);
if (!OidIsValid(opclass))
{
*isnull = true;
return true;
}
rd_index = (Form_pg_index) GETSTRUCT(tuple);
/* caller is supposed to guarantee this */
Assert(attno > 0 && attno <= rd_index->indnatts);
datum = SysCacheGetAttr(INDEXRELID, tuple,
Anum_pg_index_indclass, &disnull);
Assert(!disnull);
indclass = ((oidvector *) DatumGetPointer(datum));
opclass = indclass->values[attno - 1];
ReleaseSysCache(tuple);
/* Now look up the opclass family and input datatype. */
tuple = SearchSysCache1(CLAOID, ObjectIdGetDatum(opclass));
if (!HeapTupleIsValid(tuple))
if (!get_opclass_opfamily_and_input_type(opclass, &opfamily, &opcintype))
{
*isnull = true;
return true;
}
rd_opclass = (Form_pg_opclass) GETSTRUCT(tuple);
opfamily = rd_opclass->opcfamily;
opcintype = rd_opclass->opcintype;
ReleaseSysCache(tuple);
/* And now we can check whether the function is provided. */
@ -967,6 +940,8 @@ gistproperty(Oid index_oid, int attno,
Int16GetDatum(GIST_COMPRESS_PROC));
}
*isnull = false;
return true;
}

70
src/backend/access/index/indexam.c
View File

@ -74,6 +74,7 @@
#include "access/transam.h"
#include "access/xlog.h"
#include "catalog/index.h"
#include "catalog/pg_type.h"
#include "pgstat.h"
#include "storage/bufmgr.h"
#include "storage/lmgr.h"
@ -897,3 +898,72 @@ index_getprocinfo(Relation irel,
return locinfo;
}
/* ----------------
* index_store_float8_orderby_distances
*
* Convert AM distance function's results (that can be inexact)
* to ORDER BY types and save them into xs_orderbyvals/xs_orderbynulls
* for a possible recheck.
* ----------------
*/
void
index_store_float8_orderby_distances(IndexScanDesc scan, Oid *orderByTypes,
double *distances, bool recheckOrderBy)
{
int i;
scan->xs_recheckorderby = recheckOrderBy;
if (!distances)
{
Assert(!scan->xs_recheckorderby);
for (i = 0; i < scan->numberOfOrderBys; i++)
{
scan->xs_orderbyvals[i] = (Datum) 0;
scan->xs_orderbynulls[i] = true;
}
return;
}
for (i = 0; i < scan->numberOfOrderBys; i++)
{
if (orderByTypes[i] == FLOAT8OID)
{
#ifndef USE_FLOAT8_BYVAL
/* must free any old value to avoid memory leakage */
if (!scan->xs_orderbynulls[i])
pfree(DatumGetPointer(scan->xs_orderbyvals[i]));
#endif
scan->xs_orderbyvals[i] = Float8GetDatum(distances[i]);
scan->xs_orderbynulls[i] = false;
}
else if (orderByTypes[i] == FLOAT4OID)
{
/* convert distance function's result to ORDER BY type */
#ifndef USE_FLOAT4_BYVAL
/* must free any old value to avoid memory leakage */
if (!scan->xs_orderbynulls[i])
pfree(DatumGetPointer(scan->xs_orderbyvals[i]));
#endif
scan->xs_orderbyvals[i] = Float4GetDatum((float4) distances[i]);
scan->xs_orderbynulls[i] = false;
}
else
{
/*
* If the ordering operator's return value is anything else, we
* don't know how to convert the float8 bound calculated by the
* distance function to that. The executor won't actually need
* the order by values we return here, if there are no lossy
* results, so only insist on converting if the *recheck flag is
* set.
*/
if (scan->xs_recheckorderby)
elog(ERROR, "ORDER BY operator must return float8 or float4 if the distance function is lossy");
scan->xs_orderbynulls[i] = true;
}
}
}

3
src/backend/access/spgist/Makefile
View File

@ -14,6 +14,7 @@ include $(top_builddir)/src/Makefile.global
OBJS = spgutils.o spginsert.o spgscan.o spgvacuum.o spgvalidate.o \
spgdoinsert.o spgxlog.o \
spgtextproc.o spgquadtreeproc.o spgkdtreeproc.o
spgtextproc.o spgquadtreeproc.o spgkdtreeproc.o \
spgproc.o
include $(top_srcdir)/src/backend/common.mk

6
src/backend/access/spgist/README
View File

@ -41,7 +41,11 @@ contain exactly one inner tuple.
When the search traversal algorithm reaches an inner tuple, it chooses a set
of nodes to continue tree traverse in depth. If it reaches a leaf page it
scans a list of leaf tuples to find the ones that match the query.
scans a list of leaf tuples to find the ones that match the query. SP-GiST
also supports ordered (nearest-neighbor) searches - that is during scan pending
nodes are put into priority queue, so traversal is performed by the
closest-first model.
The insertion algorithm descends the tree similarly, except it must choose
just one node to descend to from each inner tuple. Insertion might also have

82
src/backend/access/spgist/spgkdtreeproc.c
View File

@ -16,9 +16,11 @@
#include "postgres.h"
#include "access/spgist.h"
#include "access/spgist_private.h"
#include "access/stratnum.h"
#include "catalog/pg_type.h"
#include "utils/builtins.h"
#include "utils/float.h"
#include "utils/geo_decls.h"
@ -162,6 +164,7 @@ spg_kd_inner_consistent(PG_FUNCTION_ARGS)
double coord;
int which;
int i;
BOX bboxes[2];
Assert(in->hasPrefix);
coord = DatumGetFloat8(in->prefixDatum);
@ -248,12 +251,87 @@ spg_kd_inner_consistent(PG_FUNCTION_ARGS)
}
/* We must descend into the children identified by which */
out->nodeNumbers = (int *) palloc(sizeof(int) * 2);
out->nNodes = 0;
/* Fast-path for no matching children */
if (!which)
PG_RETURN_VOID();
out->nodeNumbers = (int *) palloc(sizeof(int) * 2);
/*
* When ordering scan keys are specified, we've to calculate distance for
* them. In order to do that, we need calculate bounding boxes for both
* children nodes. Calculation of those bounding boxes on non-zero level
* require knowledge of bounding box of upper node. So, we save bounding
* boxes to traversalValues.
*/
if (in->norderbys > 0)
{
BOX infArea;
BOX *area;
out->distances = (double **) palloc(sizeof(double *) * in->nNodes);
out->traversalValues = (void **) palloc(sizeof(void *) * in->nNodes);
if (in->level == 0)
{
float8 inf = get_float8_infinity();
infArea.high.x = inf;
infArea.high.y = inf;
infArea.low.x = -inf;
infArea.low.y = -inf;
area = &infArea;
}
else
{
area = (BOX *) in->traversalValue;
Assert(area);
}
bboxes[0].low = area->low;
bboxes[1].high = area->high;
if (in->level % 2)
{
/* split box by x */
bboxes[0].high.x = bboxes[1].low.x = coord;
bboxes[0].high.y = area->high.y;
bboxes[1].low.y = area->low.y;
}
else
{
/* split box by y */
bboxes[0].high.y = bboxes[1].low.y = coord;
bboxes[0].high.x = area->high.x;
bboxes[1].low.x = area->low.x;
}
}
for (i = 1; i <= 2; i++)
{
if (which & (1 << i))
out->nodeNumbers[out->nNodes++] = i - 1;
{
out->nodeNumbers[out->nNodes] = i - 1;
if (in->norderbys > 0)
{
MemoryContext oldCtx = MemoryContextSwitchTo(
in->traversalMemoryContext);
BOX *box = box_copy(&bboxes[i - 1]);
MemoryContextSwitchTo(oldCtx);
out->traversalValues[out->nNodes] = box;
out->distances[out->nNodes] = spg_key_orderbys_distances(
BoxPGetDatum(box), false,
in->orderbys, in->norderbys);
}
out->nNodes++;
}
}
/* Set up level increments, too */

88
src/backend/access/spgist/spgproc.c Normal file
View File

@ -0,0 +1,88 @@
/*-------------------------------------------------------------------------
*
* spgproc.c
* Common supporting procedures for SP-GiST opclasses.
*
*
* Portions Copyright (c) 1996-2018, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
* src/backend/access/spgist/spgproc.c
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#include <math.h>
#include "access/spgist_private.h"
#include "utils/builtins.h"
#include "utils/float.h"
#include "utils/geo_decls.h"
#define point_point_distance(p1,p2) \
DatumGetFloat8(DirectFunctionCall2(point_distance, \
PointPGetDatum(p1), PointPGetDatum(p2)))
/* Point-box distance in the assumption that box is aligned by axis */
static double
point_box_distance(Point *point, BOX *box)
{
double dx,
dy;
if (isnan(point->x) || isnan(box->low.x) ||
isnan(point->y) || isnan(box->low.y))
return get_float8_nan();
if (point->x < box->low.x)
dx = box->low.x - point->x;
else if (point->x > box->high.x)
dx = point->x - box->high.x;
else
dx = 0.0;
if (point->y < box->low.y)
dy = box->low.y - point->y;
else if (point->y > box->high.y)
dy = point->y - box->high.y;
else
dy = 0.0;
return HYPOT(dx, dy);
}
/*
* Returns distances from given key to array of ordering scan keys. Leaf key
* is expected to be point, non-leaf key is expected to be box. Scan key
* arguments are expected to be points.
*/
double *
spg_key_orderbys_distances(Datum key, bool isLeaf,
ScanKey orderbys, int norderbys)
{
int sk_num;
double *distances = (double *) palloc(norderbys * sizeof(double)),
*distance = distances;
for (sk_num = 0; sk_num < norderbys; ++sk_num, ++orderbys, ++distance)
{
Point *point = DatumGetPointP(orderbys->sk_argument);
*distance = isLeaf ? point_point_distance(point, DatumGetPointP(key))
: point_box_distance(point, DatumGetBoxP(key));
}
return distances;
}
BOX *
box_copy(BOX *orig)
{
BOX *result = palloc(sizeof(BOX));
*result = *orig;
return result;
}

115
src/backend/access/spgist/spgquadtreeproc.c
View File

@ -17,8 +17,10 @@
#include "access/spgist.h"
#include "access/stratnum.h"
#include "access/spgist_private.h"
#include "catalog/pg_type.h"
#include "utils/builtins.h"
#include "utils/float.h"
#include "utils/geo_decls.h"
@ -77,6 +79,38 @@ getQuadrant(Point *centroid, Point *tst)
return 0;
}
/* Returns bounding box of a given quadrant inside given bounding box */
static BOX *
getQuadrantArea(BOX *bbox, Point *centroid, int quadrant)
{
BOX *result = (BOX *) palloc(sizeof(BOX));
switch (quadrant)
{
case 1:
result->high = bbox->high;
result->low = *centroid;
break;
case 2:
result->high.x = bbox->high.x;
result->high.y = centroid->y;
result->low.x = centroid->x;
result->low.y = bbox->low.y;
break;
case 3:
result->high = *centroid;
result->low = bbox->low;
break;
case 4:
result->high.x = centroid->x;
result->high.y = bbox->high.y;
result->low.x = bbox->low.x;
result->low.y = centroid->y;
break;
}
return result;
}
Datum
spg_quad_choose(PG_FUNCTION_ARGS)
@ -196,19 +230,68 @@ spg_quad_inner_consistent(PG_FUNCTION_ARGS)
spgInnerConsistentIn *in = (spgInnerConsistentIn *) PG_GETARG_POINTER(0);
spgInnerConsistentOut *out = (spgInnerConsistentOut *) PG_GETARG_POINTER(1);
Point *centroid;
BOX infbbox;
BOX *bbox = NULL;
int which;
int i;
Assert(in->hasPrefix);
centroid = DatumGetPointP(in->prefixDatum);
/*
* When ordering scan keys are specified, we've to calculate distance for
* them. In order to do that, we need calculate bounding boxes for all
* children nodes. Calculation of those bounding boxes on non-zero level
* require knowledge of bounding box of upper node. So, we save bounding
* boxes to traversalValues.
*/
if (in->norderbys > 0)
{
out->distances = (double **) palloc(sizeof(double *) * in->nNodes);
out->traversalValues = (void **) palloc(sizeof(void *) * in->nNodes);
if (in->level == 0)
{
double inf = get_float8_infinity();
infbbox.high.x = inf;
infbbox.high.y = inf;
infbbox.low.x = -inf;
infbbox.low.y = -inf;
bbox = &infbbox;
}
else
{
bbox = in->traversalValue;
Assert(bbox);
}
}
if (in->allTheSame)
{
/* Report that all nodes should be visited */
out->nNodes = in->nNodes;
out->nodeNumbers = (int *) palloc(sizeof(int) * in->nNodes);
for (i = 0; i < in->nNodes; i++)
{
out->nodeNumbers[i] = i;
if (in->norderbys > 0)
{
MemoryContext oldCtx = MemoryContextSwitchTo(
in->traversalMemoryContext);
/* Use parent quadrant box as traversalValue */
BOX *quadrant = box_copy(bbox);
MemoryContextSwitchTo(oldCtx);
out->traversalValues[i] = quadrant;
out->distances[i] = spg_key_orderbys_distances(
BoxPGetDatum(quadrant), false,
in->orderbys, in->norderbys);
}
}
PG_RETURN_VOID();
}
@ -286,13 +369,37 @@ spg_quad_inner_consistent(PG_FUNCTION_ARGS)
break; /* no need to consider remaining conditions */
}
out->levelAdds = palloc(sizeof(int) * 4);
for (i = 0; i < 4; ++i)
out->levelAdds[i] = 1;
/* We must descend into the quadrant(s) identified by which */
out->nodeNumbers = (int *) palloc(sizeof(int) * 4);
out->nNodes = 0;
for (i = 1; i <= 4; i++)
{
if (which & (1 << i))
out->nodeNumbers[out->nNodes++] = i - 1;
{
out->nodeNumbers[out->nNodes] = i - 1;
if (in->norderbys > 0)
{
MemoryContext oldCtx = MemoryContextSwitchTo(
in->traversalMemoryContext);
BOX *quadrant = getQuadrantArea(bbox, centroid, i);
MemoryContextSwitchTo(oldCtx);
out->traversalValues[out->nNodes] = quadrant;
out->distances[out->nNodes] = spg_key_orderbys_distances(
BoxPGetDatum(quadrant), false,
in->orderbys, in->norderbys);
}
out->nNodes++;
}
}
PG_RETURN_VOID();
@ -356,5 +463,11 @@ spg_quad_leaf_consistent(PG_FUNCTION_ARGS)
break;
}
if (res && in->norderbys > 0)
/* ok, it passes -> let's compute the distances */
out->distances = spg_key_orderbys_distances(
BoxPGetDatum(in->leafDatum), true,
in->orderbys, in->norderbys);
PG_RETURN_BOOL(res);
}

895
src/backend/access/spgist/spgscan.c
View File

File diff suppressed because it is too large Load Diff

92
src/backend/access/spgist/spgutils.c
View File

@ -15,17 +15,26 @@
#include "postgres.h"
#include "access/amvalidate.h"
#include "access/htup_details.h"
#include "access/reloptions.h"
#include "access/spgist_private.h"
#include "access/transam.h"
#include "access/xact.h"
#include "catalog/pg_amop.h"
#include "optimizer/paths.h"
#include "storage/bufmgr.h"
#include "storage/indexfsm.h"
#include "storage/lmgr.h"
#include "utils/builtins.h"
#include "utils/catcache.h"
#include "utils/index_selfuncs.h"
#include "utils/lsyscache.h"
#include "utils/syscache.h"
extern Expr *spgcanorderbyop(IndexOptInfo *index,
PathKey *pathkey, int pathkeyno,
Expr *orderby_clause, int *indexcol_p);
/*
* SP-GiST handler function: return IndexAmRoutine with access method parameters
@ -39,7 +48,7 @@ spghandler(PG_FUNCTION_ARGS)
amroutine->amstrategies = 0;
amroutine->amsupport = SPGISTNProc;
amroutine->amcanorder = false;
amroutine->amcanorderbyop = false;
amroutine->amcanorderbyop = true;
amroutine->amcanbackward = false;
amroutine->amcanunique = false;
amroutine->amcanmulticol = false;
@ -61,7 +70,7 @@ spghandler(PG_FUNCTION_ARGS)
amroutine->amcanreturn = spgcanreturn;
amroutine->amcostestimate = spgcostestimate;
amroutine->amoptions = spgoptions;
amroutine->amproperty = NULL;
amroutine->amproperty = spgproperty;
amroutine->amvalidate = spgvalidate;
amroutine->ambeginscan = spgbeginscan;
amroutine->amrescan = spgrescan;
@ -949,3 +958,82 @@ SpGistPageAddNewItem(SpGistState *state, Page page, Item item, Size size,
return offnum;
}
/*
* spgproperty() -- Check boolean properties of indexes.
*
* This is optional for most AMs, but is required for SP-GiST because the core
* property code doesn't support AMPROP_DISTANCE_ORDERABLE.
*/
bool
spgproperty(Oid index_oid, int attno,
IndexAMProperty prop, const char *propname,
bool *res, bool *isnull)
{
Oid opclass,
opfamily,
opcintype;
CatCList *catlist;
int i;
/* Only answer column-level inquiries */
if (attno == 0)
return false;
switch (prop)
{
case AMPROP_DISTANCE_ORDERABLE:
break;
default:
return false;
}
/*
* Currently, SP-GiST distance-ordered scans require that there be a
* distance operator in the opclass with the default types. So we assume
* that if such a operator exists, then there's a reason for it.
*/
/* First we need to know the column's opclass. */
opclass = get_index_column_opclass(index_oid, attno);
if (!OidIsValid(opclass))
{
*isnull = true;
return true;
}
/* Now look up the opclass family and input datatype. */
if (!get_opclass_opfamily_and_input_type(opclass, &opfamily, &opcintype))
{
*isnull = true;
return true;
}
/* And now we can check whether the operator is provided. */
catlist = SearchSysCacheList1(AMOPSTRATEGY,
ObjectIdGetDatum(opfamily));
*res = false;
for (i = 0; i < catlist->n_members; i++)
{
HeapTuple amoptup = &catlist->members[i]->tuple;
Form_pg_amop amopform = (Form_pg_amop) GETSTRUCT(amoptup);
if (amopform->amoppurpose == AMOP_ORDER &&
(amopform->amoplefttype == opcintype ||
amopform->amoprighttype == opcintype) &&
opfamily_can_sort_type(amopform->amopsortfamily,
get_op_rettype(amopform->amopopr)))
{
*res = true;
break;
}
}
ReleaseSysCacheList(catlist);
*isnull = false;
return true;
}

27
src/backend/access/spgist/spgvalidate.c
View File

@ -187,6 +187,7 @@ spgvalidate(Oid opclassoid)
{
HeapTuple oprtup = &oprlist->members[i]->tuple;
Form_pg_amop oprform = (Form_pg_amop) GETSTRUCT(oprtup);
Oid op_rettype;
/* TODO: Check that only allowed strategy numbers exist */
if (oprform->amopstrategy < 1 || oprform->amopstrategy > 63)
@ -200,20 +201,26 @@ spgvalidate(Oid opclassoid)
result = false;
}
/* spgist doesn't support ORDER BY operators */
if (oprform->amoppurpose != AMOP_SEARCH ||
OidIsValid(oprform->amopsortfamily))
/* spgist supports ORDER BY operators */
if (oprform->amoppurpose != AMOP_SEARCH)
{
ereport(INFO,
(errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
errmsg("operator family \"%s\" of access method %s contains invalid ORDER BY specification for operator %s",
opfamilyname, "spgist",
format_operator(oprform->amopopr))));
result = false;
/* ... and operator result must match the claimed btree opfamily */
op_rettype = get_op_rettype(oprform->amopopr);
if (!opfamily_can_sort_type(oprform->amopsortfamily, op_rettype))
{
ereport(INFO,
(errcode(ERRCODE_INVALID_OBJECT_DEFINITION),
errmsg("operator family \"%s\" of access method %s contains invalid ORDER BY specification for operator %s",
opfamilyname, "spgist",
format_operator(oprform->amopopr))));
result = false;
}
}
else
op_rettype = BOOLOID;
/* Check operator signature --- same for all spgist strategies */
if (!check_amop_signature(oprform->amopopr, BOOLOID,
if (!check_amop_signature(oprform->amopopr, op_rettype,
oprform->amoplefttype,
oprform->amoprighttype))
{

101
src/backend/utils/adt/geo_spgist.c
View File

@ -74,9 +74,11 @@
#include "postgres.h"
#include "access/spgist.h"
#include "access/spgist_private.h"
#include "access/stratnum.h"
#include "catalog/pg_type.h"
#include "utils/float.h"
#include "utils/fmgroids.h"
#include "utils/fmgrprotos.h"
#include "utils/geo_decls.h"
@ -367,6 +369,31 @@ overAbove4D(RectBox *rect_box, RangeBox *query)
return overHigher2D(&rect_box->range_box_y, &query->right);
}
/* Lower bound for the distance between point and rect_box */
static double
pointToRectBoxDistance(Point *point, RectBox *rect_box)
{
double dx;
double dy;
if (point->x < rect_box->range_box_x.left.low)
dx = rect_box->range_box_x.left.low - point->x;
else if (point->x > rect_box->range_box_x.right.high)
dx = point->x - rect_box->range_box_x.right.high;
else
dx = 0;
if (point->y < rect_box->range_box_y.left.low)
dy = rect_box->range_box_y.left.low - point->y;
else if (point->y > rect_box->range_box_y.right.high)
dy = point->y - rect_box->range_box_y.right.high;
else
dy = 0;
return HYPOT(dx, dy);
}
/*
* SP-GiST config function
*/
@ -534,17 +561,6 @@ spg_box_quad_inner_consistent(PG_FUNCTION_ARGS)
RangeBox *centroid,
**queries;
if (in->allTheSame)
{
/* Report that all nodes should be visited */
out->nNodes = in->nNodes;
out->nodeNumbers = (int *) palloc(sizeof(int) * in->nNodes);
for (i = 0; i < in->nNodes; i++)
out->nodeNumbers[i] = i;
PG_RETURN_VOID();
}
/*
* We are saving the traversal value or initialize it an unbounded one, if
* we have just begun to walk the tree.
@ -554,6 +570,40 @@ spg_box_quad_inner_consistent(PG_FUNCTION_ARGS)
else
rect_box = initRectBox();
if (in->allTheSame)
{
/* Report that all nodes should be visited */
out->nNodes = in->nNodes;
out->nodeNumbers = (int *) palloc(sizeof(int) * in->nNodes);
for (i = 0; i < in->nNodes; i++)
out->nodeNumbers[i] = i;
if (in->norderbys > 0 && in->nNodes > 0)
{
double *distances = palloc(sizeof(double) * in->norderbys);
int j;
for (j = 0; j < in->norderbys; j++)
{
Point *pt = DatumGetPointP(in->orderbys[j].sk_argument);
distances[j] = pointToRectBoxDistance(pt, rect_box);
}
out->distances = (double **) palloc(sizeof(double *) * in->nNodes);
out->distances[0] = distances;
for (i = 1; i < in->nNodes; i++)
{
out->distances[i] = palloc(sizeof(double) * in->norderbys);
memcpy(out->distances[i], distances,
sizeof(double) * in->norderbys);
}
}
PG_RETURN_VOID();
}
/*
* We are casting the prefix and queries to RangeBoxes for ease of the
* following operations.
@ -571,6 +621,8 @@ spg_box_quad_inner_consistent(PG_FUNCTION_ARGS)
out->nNodes = 0;
out->nodeNumbers = (int *) palloc(sizeof(int) * in->nNodes);
out->traversalValues = (void **) palloc(sizeof(void *) * in->nNodes);
if (in->norderbys > 0)
out->distances = (double **) palloc(sizeof(double *) * in->nNodes);
/*
* We switch memory context, because we want to allocate memory for new
@ -648,6 +700,22 @@ spg_box_quad_inner_consistent(PG_FUNCTION_ARGS)
{
out->traversalValues[out->nNodes] = next_rect_box;
out->nodeNumbers[out->nNodes] = quadrant;
if (in->norderbys > 0)
{
double *distances = palloc(sizeof(double) * in->norderbys);
int j;
out->distances[out->nNodes] = distances;
for (j = 0; j < in->norderbys; j++)
{
Point *pt = DatumGetPointP(in->orderbys[j].sk_argument);
distances[j] = pointToRectBoxDistance(pt, next_rect_box);
}
}
out->nNodes++;
}
else
@ -763,6 +831,17 @@ spg_box_quad_leaf_consistent(PG_FUNCTION_ARGS)
break;
}
if (flag && in->norderbys > 0)
{
Oid distfnoid = in->orderbys[0].sk_func.fn_oid;
out->distances = spg_key_orderbys_distances(leaf, false,
in->orderbys, in->norderbys);
/* Recheck is necessary when computing distance to polygon */
out->recheckDistances = distfnoid == F_DIST_POLYP;
}
PG_RETURN_BOOL(flag);
}

68
src/backend/utils/cache/lsyscache.c vendored
View File

@ -1067,6 +1067,32 @@ get_opclass_input_type(Oid opclass)
return result;
}
/*
* get_opclass_family_and_input_type
*
* Returns the OID of the operator family the opclass belongs to,
* the OID of the datatype the opclass indexes
*/
bool
get_opclass_opfamily_and_input_type(Oid opclass, Oid *opfamily, Oid *opcintype)
{
HeapTuple tp;
Form_pg_opclass cla_tup;
tp = SearchSysCache1(CLAOID, ObjectIdGetDatum(opclass));
if (!HeapTupleIsValid(tp))
return false;
cla_tup = (Form_pg_opclass) GETSTRUCT(tp);
*opfamily = cla_tup->opcfamily;
*opcintype = cla_tup->opcintype;
ReleaseSysCache(tp);
return true;
}
/* ---------- OPERATOR CACHE ---------- */
/*
@ -3106,3 +3132,45 @@ get_range_subtype(Oid rangeOid)
else
return InvalidOid;
}
/* ---------- PG_INDEX CACHE ---------- */
/*
* get_index_column_opclass
*
* Given the index OID and column number,
* return opclass of the index column
* or InvalidOid if the index was not found.
*/
Oid
get_index_column_opclass(Oid index_oid, int attno)
{
HeapTuple tuple;
Form_pg_index rd_index PG_USED_FOR_ASSERTS_ONLY;
Datum datum;
bool isnull;
oidvector *indclass;
Oid opclass;
/* First we need to know the column's opclass. */
tuple = SearchSysCache1(INDEXRELID, ObjectIdGetDatum(index_oid));
if (!HeapTupleIsValid(tuple))
return InvalidOid;
rd_index = (Form_pg_index) GETSTRUCT(tuple);
/* caller is supposed to guarantee this */
Assert(attno > 0 && attno <= rd_index->indnatts);
datum = SysCacheGetAttr(INDEXRELID, tuple,
Anum_pg_index_indclass, &isnull);
Assert(!isnull);
indclass = ((oidvector *) DatumGetPointer(datum));
opclass = indclass->values[attno - 1];
ReleaseSysCache(tuple);
return opclass;
}

3
src/include/access/genam.h
View File

@ -174,6 +174,9 @@ extern RegProcedure index_getprocid(Relation irel, AttrNumber attnum,
uint16 procnum);
extern FmgrInfo *index_getprocinfo(Relation irel, AttrNumber attnum,
uint16 procnum);
extern void index_store_float8_orderby_distances(IndexScanDesc scan,
Oid *orderByTypes, double *distances,
bool recheckOrderBy);
/*
* index access method support routines (in genam.c)

13
src/include/access/spgist.h
View File

@ -136,7 +136,10 @@ typedef struct spgPickSplitOut
typedef struct spgInnerConsistentIn
{
ScanKey scankeys; /* array of operators and comparison values */
int nkeys; /* length of array */
ScanKey orderbys; /* array of ordering operators and comparison
* values */
int nkeys; /* length of scankeys array */
int norderbys; /* length of orderbys array */
Datum reconstructedValue; /* value reconstructed at parent */
void *traversalValue; /* opclass-specific traverse value */
@ -159,6 +162,7 @@ typedef struct spgInnerConsistentOut
int *levelAdds; /* increment level by this much for each */
Datum *reconstructedValues; /* associated reconstructed values */
void **traversalValues; /* opclass-specific traverse values */
double **distances; /* associated distances */
} spgInnerConsistentOut;
/*
@ -167,7 +171,10 @@ typedef struct spgInnerConsistentOut
typedef struct spgLeafConsistentIn
{
ScanKey scankeys; /* array of operators and comparison values */
int nkeys; /* length of array */
ScanKey orderbys; /* array of ordering operators and comparison
* values */
int nkeys; /* length of scankeys array */
int norderbys; /* length of orderbys array */
Datum reconstructedValue; /* value reconstructed at parent */
void *traversalValue; /* opclass-specific traverse value */
@ -181,6 +188,8 @@ typedef struct spgLeafConsistentOut
{
Datum leafValue; /* reconstructed original data, if any */
bool recheck; /* set true if operator must be rechecked */
bool recheckDistances; /* set true if distances must be rechecked */
double *distances; /* associated distances */
} spgLeafConsistentOut;

48
src/include/access/spgist_private.h
View File

@ -18,6 +18,7 @@
#include "access/spgist.h"
#include "nodes/tidbitmap.h"
#include "storage/buf.h"
#include "utils/geo_decls.h"
#include "utils/relcache.h"
@ -130,14 +131,35 @@ typedef struct SpGistState
bool isBuild; /* true if doing index build */
} SpGistState;
typedef struct SpGistSearchItem
{
pairingheap_node phNode; /* pairing heap node */
Datum value; /* value reconstructed from parent or
* leafValue if heaptuple */
void *traversalValue; /* opclass-specific traverse value */
int level; /* level of items on this page */
ItemPointerData heapPtr; /* heap info, if heap tuple */
bool isNull; /* SearchItem is NULL item */
bool isLeaf; /* SearchItem is heap item */
bool recheck; /* qual recheck is needed */
bool recheckDistances; /* distance recheck is needed */
/* array with numberOfOrderBys entries */
double distances[FLEXIBLE_ARRAY_MEMBER];
} SpGistSearchItem;
#define SizeOfSpGistSearchItem(n_distances) \
(offsetof(SpGistSearchItem, distances) + sizeof(double) * (n_distances))
/*
* Private state of an index scan
*/
typedef struct SpGistScanOpaqueData
{
SpGistState state; /* see above */
pairingheap *scanQueue; /* queue of to be visited items */
MemoryContext tempCxt; /* short-lived memory context */
MemoryContext traversalCxt; /* memory context for traversalValues */
MemoryContext traversalCxt; /* single scan lifetime memory context */
/* Control flags showing whether to search nulls and/or non-nulls */
bool searchNulls; /* scan matches (all) null entries */
@ -146,9 +168,18 @@ typedef struct SpGistScanOpaqueData
/* Index quals to be passed to opclass (null-related quals removed) */
int numberOfKeys; /* number of index qualifier conditions */
ScanKey keyData; /* array of index qualifier descriptors */
int numberOfOrderBys; /* number of ordering operators */
ScanKey orderByData; /* array of ordering op descriptors */
Oid *orderByTypes; /* array of ordering op return types */
Oid indexCollation; /* collation of index column */
/* Stack of yet-to-be-visited pages */
List *scanStack; /* List of ScanStackEntrys */
/* Opclass defined functions: */
FmgrInfo innerConsistentFn;
FmgrInfo leafConsistentFn;
/* Pre-allocated workspace arrays: */
double *zeroDistances;
double *infDistances;
/* These fields are only used in amgetbitmap scans: */
TIDBitmap *tbm; /* bitmap being filled */
@ -161,7 +192,10 @@ typedef struct SpGistScanOpaqueData
int iPtr; /* index for scanning through same */
ItemPointerData heapPtrs[MaxIndexTuplesPerPage]; /* TIDs from cur page */
bool recheck[MaxIndexTuplesPerPage]; /* their recheck flags */
bool recheckDistances[MaxIndexTuplesPerPage]; /* distance recheck
* flags */
HeapTuple reconTups[MaxIndexTuplesPerPage]; /* reconstructed tuples */
double *distances[MaxIndexTuplesPerPage]; /* distances (for recheck) */
/*
* Note: using MaxIndexTuplesPerPage above is a bit hokey since
@ -410,6 +444,9 @@ extern OffsetNumber SpGistPageAddNewItem(SpGistState *state, Page page,
Item item, Size size,
OffsetNumber *startOffset,
bool errorOK);
extern bool spgproperty(Oid index_oid, int attno,
IndexAMProperty prop, const char *propname,
bool *res, bool *isnull);
/* spgdoinsert.c */
extern void spgUpdateNodeLink(SpGistInnerTuple tup, int nodeN,
@ -421,4 +458,9 @@ extern void spgPageIndexMultiDelete(SpGistState *state, Page page,
extern bool spgdoinsert(Relation index, SpGistState *state,
ItemPointer heapPtr, Datum datum, bool isnull);
/* spgproc.c */
extern double *spg_key_orderbys_distances(Datum key, bool isLeaf,
ScanKey orderbys, int norderbys);
extern BOX *box_copy(BOX *orig);
#endif /* SPGIST_PRIVATE_H */

2
src/include/catalog/catversion.h
View File

@ -53,6 +53,6 @@
*/
/* yyyymmddN */
#define CATALOG_VERSION_NO 201809181
#define CATALOG_VERSION_NO 201809191
#endif

12
src/include/catalog/pg_amop.dat
View File

@ -1401,6 +1401,10 @@
{ amopfamily => 'spgist/quad_point_ops', amoplefttype => 'point',
amoprighttype => 'box', amopstrategy => '8', amopopr => '<@(point,box)',
amopmethod => 'spgist' },
{ amopfamily => 'spgist/quad_point_ops', amoplefttype => 'point',
amoprighttype => 'point', amopstrategy => '15', amopopr => '<->(point,point)',
amopmethod => 'spgist', amoppurpose => 'o',
amopsortfamily => 'btree/float_ops' },
# SP-GiST kd_point_ops
{ amopfamily => 'spgist/kd_point_ops', amoplefttype => 'point',
@ -1421,6 +1425,10 @@
{ amopfamily => 'spgist/kd_point_ops', amoplefttype => 'point',
amoprighttype => 'box', amopstrategy => '8', amopopr => '<@(point,box)',
amopmethod => 'spgist' },
{ amopfamily => 'spgist/kd_point_ops', amoplefttype => 'point',
amoprighttype => 'point', amopstrategy => '15', amopopr => '<->(point,point)',
amopmethod => 'spgist', amoppurpose => 'o',
amopsortfamily => 'btree/float_ops' },
# SP-GiST text_ops
{ amopfamily => 'spgist/text_ops', amoplefttype => 'text',
@ -1590,6 +1598,10 @@
{ amopfamily => 'spgist/poly_ops', amoplefttype => 'polygon',
amoprighttype => 'polygon', amopstrategy => '12',
amopopr => '|&>(polygon,polygon)', amopmethod => 'spgist' },
{ amopfamily => 'spgist/poly_ops', amoplefttype => 'polygon',
amoprighttype => 'point', amopstrategy => '15',
amopopr => '<->(polygon,point)', amoppurpose => 'o', amopmethod => 'spgist',
amopsortfamily => 'btree/float_ops' },
# GiST inet_ops
{ amopfamily => 'gist/network_ops', amoplefttype => 'inet',

3
src/include/utils/lsyscache.h
View File

@ -95,6 +95,8 @@ extern char *get_constraint_name(Oid conoid);
extern char *get_language_name(Oid langoid, bool missing_ok);
extern Oid get_opclass_family(Oid opclass);
extern Oid get_opclass_input_type(Oid opclass);
extern bool get_opclass_opfamily_and_input_type(Oid opclass,
Oid *opfamily, Oid *opcintype);
extern RegProcedure get_opcode(Oid opno);
extern char *get_opname(Oid opno);
extern Oid get_op_rettype(Oid opno);
@ -176,6 +178,7 @@ extern void free_attstatsslot(AttStatsSlot *sslot);
extern char *get_namespace_name(Oid nspid);
extern char *get_namespace_name_or_temp(Oid nspid);
extern Oid get_range_subtype(Oid rangeOid);
extern Oid get_index_column_opclass(Oid index_oid, int attno);
#define type_is_array(typid) (get_element_type(typid) != InvalidOid)
/* type_is_array_domain accepts both plain arrays and domains over arrays */

27
src/test/regress/expected/amutils.out
View File

@ -83,7 +83,8 @@ select prop,
pg_index_column_has_property('onek_hundred'::regclass, 1, prop) as btree,
pg_index_column_has_property('hash_i4_index'::regclass, 1, prop) as hash,
pg_index_column_has_property('gcircleind'::regclass, 1, prop) as gist,
pg_index_column_has_property('sp_radix_ind'::regclass, 1, prop) as spgist,
pg_index_column_has_property('sp_radix_ind'::regclass, 1, prop) as spgist_radix,
pg_index_column_has_property('sp_quad_ind'::regclass, 1, prop) as spgist_quad,
pg_index_column_has_property('botharrayidx'::regclass, 1, prop) as gin,
pg_index_column_has_property('brinidx'::regclass, 1, prop) as brin
from unnest(array['asc', 'desc', 'nulls_first', 'nulls_last',
@ -92,18 +93,18 @@ select prop,
'bogus']::text[])
with ordinality as u(prop,ord)
order by ord;
prop | btree | hash | gist | spgist | gin | brin
--------------------+-------+------+------+--------+-----+------
asc | t | f | f | f | f | f
desc | f | f | f | f | f | f
nulls_first | f | f | f | f | f | f
nulls_last | t | f | f | f | f | f
orderable | t | f | f | f | f | f
distance_orderable | f | f | t | f | f | f
returnable | t | f | f | t | f | f
search_array | t | f | f | f | f | f
search_nulls | t | f | t | t | f | t
bogus | | | | | |
prop | btree | hash | gist | spgist_radix | spgist_quad | gin | brin
--------------------+-------+------+------+--------------+-------------+-----+------
asc | t | f | f | f | f | f | f
desc | f | f | f | f | f | f | f
nulls_first | f | f | f | f | f | f | f
nulls_last | t | f | f | f | f | f | f
orderable | t | f | f | f | f | f | f
distance_orderable | f | f | t | f | t | f | f
returnable | t | f | f | t | t | f | f
search_array | t | f | f | f | f | f | f
search_nulls | t | f | t | t | t | f | t
bogus | | | | | | |
(10 rows)
select prop,

139
src/test/regress/expected/create_index.out
View File

@ -294,6 +294,15 @@ SELECT count(*) FROM quad_point_tbl WHERE p ~= '(4585, 365)';
1
(1 row)
CREATE TEMP TABLE quad_point_tbl_ord_seq1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
CREATE TEMP TABLE quad_point_tbl_ord_seq2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
CREATE TEMP TABLE quad_point_tbl_ord_seq3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcdef';
count
-------
@ -888,6 +897,71 @@ SELECT count(*) FROM quad_point_tbl WHERE p ~= '(4585, 365)';
1
(1 row)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
QUERY PLAN
-----------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_quad_ind on quad_point_tbl
Order By: (p <-> '(0,0)'::point)
(3 rows)
CREATE TEMP TABLE quad_point_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
SELECT * FROM quad_point_tbl_ord_seq1 seq FULL JOIN quad_point_tbl_ord_idx1 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
QUERY PLAN
-----------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_quad_ind on quad_point_tbl
Index Cond: (p <@ '(1000,1000),(200,200)'::box)
Order By: (p <-> '(0,0)'::point)
(4 rows)
CREATE TEMP TABLE quad_point_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
SELECT * FROM quad_point_tbl_ord_seq2 seq FULL JOIN quad_point_tbl_ord_idx2 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
QUERY PLAN
-----------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_quad_ind on quad_point_tbl
Index Cond: (p IS NOT NULL)
Order By: (p <-> '(333,400)'::point)
(4 rows)
CREATE TEMP TABLE quad_point_tbl_ord_idx3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
SELECT * FROM quad_point_tbl_ord_seq3 seq FULL JOIN quad_point_tbl_ord_idx3 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT count(*) FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
QUERY PLAN
@ -993,6 +1067,71 @@ SELECT count(*) FROM kd_point_tbl WHERE p ~= '(4585, 365)';
1
(1 row)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl;
QUERY PLAN
-------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_kd_ind on kd_point_tbl
Order By: (p <-> '(0,0)'::point)
(3 rows)
CREATE TEMP TABLE kd_point_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl;
SELECT * FROM quad_point_tbl_ord_seq1 seq FULL JOIN kd_point_tbl_ord_idx1 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
QUERY PLAN
---------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_kd_ind on kd_point_tbl
Index Cond: (p <@ '(1000,1000),(200,200)'::box)
Order By: (p <-> '(0,0)'::point)
(4 rows)
CREATE TEMP TABLE kd_point_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
SELECT * FROM quad_point_tbl_ord_seq2 seq FULL JOIN kd_point_tbl_ord_idx2 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM kd_point_tbl WHERE p IS NOT NULL;
QUERY PLAN
-------------------------------------------------------
WindowAgg
-> Index Only Scan using sp_kd_ind on kd_point_tbl
Index Cond: (p IS NOT NULL)
Order By: (p <-> '(333,400)'::point)
(4 rows)
CREATE TEMP TABLE kd_point_tbl_ord_idx3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM kd_point_tbl WHERE p IS NOT NULL;
SELECT * FROM quad_point_tbl_ord_seq3 seq FULL JOIN kd_point_tbl_ord_idx3 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
n | dist | p | n | dist | p
---+------+---+---+------+---
(0 rows)
EXPLAIN (COSTS OFF)
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcdef';
QUERY PLAN

3
src/test/regress/expected/opr_sanity.out
View File

@ -1911,6 +1911,7 @@ ORDER BY 1, 2, 3;
4000 | 12 | <=
4000 | 12 | |&>
4000 | 14 | >=
4000 | 15 | <->
4000 | 15 | >
4000 | 16 | @>
4000 | 18 | =
@ -1924,7 +1925,7 @@ ORDER BY 1, 2, 3;
4000 | 26 | >>
4000 | 27 | >>=
4000 | 28 | ^@
(122 rows)
(123 rows)
-- Check that all opclass search operators have selectivity estimators.
-- This is not absolutely required, but it seems a reasonable thing

48
src/test/regress/expected/polygon.out
View File

@ -462,6 +462,54 @@ SELECT count(*) FROM quad_poly_tbl WHERE p ~= polygon '((200, 300),(210, 310),(2
1000
(1 row)
-- test ORDER BY distance
SET enable_indexscan = ON;
SET enable_bitmapscan = OFF;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl;
QUERY PLAN
-----------------------------------------------------------
WindowAgg
-> Index Scan using quad_poly_tbl_idx on quad_poly_tbl
Order By: (p <-> '(123,456)'::point)
(3 rows)
CREATE TEMP TABLE quad_poly_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl;
SELECT *
FROM quad_poly_tbl_ord_seq1 seq FULL JOIN quad_poly_tbl_ord_idx1 idx
ON seq.n = idx.n AND seq.id = idx.id AND
(seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
WHERE seq.id IS NULL OR idx.id IS NULL;
n | dist | id | n | dist | id
---+------+----+---+------+----
(0 rows)
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl WHERE p <@ polygon '((300,300),(400,600),(600,500),(700,200))';
QUERY PLAN
---------------------------------------------------------------------------------
WindowAgg
-> Index Scan using quad_poly_tbl_idx on quad_poly_tbl
Index Cond: (p <@ '((300,300),(400,600),(600,500),(700,200))'::polygon)
Order By: (p <-> '(123,456)'::point)
(4 rows)
CREATE TEMP TABLE quad_poly_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl WHERE p <@ polygon '((300,300),(400,600),(600,500),(700,200))';
SELECT *
FROM quad_poly_tbl_ord_seq2 seq FULL JOIN quad_poly_tbl_ord_idx2 idx
ON seq.n = idx.n AND seq.id = idx.id AND
(seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
WHERE seq.id IS NULL OR idx.id IS NULL;
n | dist | id | n | dist | id
---+------+----+---+------+----
(0 rows)
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;

3
src/test/regress/sql/amutils.sql
View File

@ -40,7 +40,8 @@ select prop,
pg_index_column_has_property('onek_hundred'::regclass, 1, prop) as btree,
pg_index_column_has_property('hash_i4_index'::regclass, 1, prop) as hash,
pg_index_column_has_property('gcircleind'::regclass, 1, prop) as gist,
pg_index_column_has_property('sp_radix_ind'::regclass, 1, prop) as spgist,
pg_index_column_has_property('sp_radix_ind'::regclass, 1, prop) as spgist_radix,
pg_index_column_has_property('sp_quad_ind'::regclass, 1, prop) as spgist_quad,
pg_index_column_has_property('botharrayidx'::regclass, 1, prop) as gin,
pg_index_column_has_property('brinidx'::regclass, 1, prop) as brin
from unnest(array['asc', 'desc', 'nulls_first', 'nulls_last',

78
src/test/regress/sql/create_index.sql
View File

@ -198,6 +198,18 @@ SELECT count(*) FROM quad_point_tbl WHERE p >^ '(5000, 4000)';
SELECT count(*) FROM quad_point_tbl WHERE p ~= '(4585, 365)';
CREATE TEMP TABLE quad_point_tbl_ord_seq1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
CREATE TEMP TABLE quad_point_tbl_ord_seq2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
CREATE TEMP TABLE quad_point_tbl_ord_seq3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcdef';
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcde';
@ -363,6 +375,39 @@ EXPLAIN (COSTS OFF)
SELECT count(*) FROM quad_point_tbl WHERE p ~= '(4585, 365)';
SELECT count(*) FROM quad_point_tbl WHERE p ~= '(4585, 365)';
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
CREATE TEMP TABLE quad_point_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl;
SELECT * FROM quad_point_tbl_ord_seq1 seq FULL JOIN quad_point_tbl_ord_idx1 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
CREATE TEMP TABLE quad_point_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM quad_point_tbl WHERE p <@ box '(200,200,1000,1000)';
SELECT * FROM quad_point_tbl_ord_seq2 seq FULL JOIN quad_point_tbl_ord_idx2 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
CREATE TEMP TABLE quad_point_tbl_ord_idx3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM quad_point_tbl WHERE p IS NOT NULL;
SELECT * FROM quad_point_tbl_ord_seq3 seq FULL JOIN quad_point_tbl_ord_idx3 idx
ON seq.n = idx.n
AND (seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT count(*) FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
SELECT count(*) FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
@ -391,6 +436,39 @@ EXPLAIN (COSTS OFF)
SELECT count(*) FROM kd_point_tbl WHERE p ~= '(4585, 365)';
SELECT count(*) FROM kd_point_tbl WHERE p ~= '(4585, 365)';
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl;
CREATE TEMP TABLE kd_point_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl;
SELECT * FROM quad_point_tbl_ord_seq1 seq FULL JOIN kd_point_tbl_ord_idx1 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
CREATE TEMP TABLE kd_point_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> '0,0') n, p <-> '0,0' dist, p
FROM kd_point_tbl WHERE p <@ box '(200,200,1000,1000)';
SELECT * FROM quad_point_tbl_ord_seq2 seq FULL JOIN kd_point_tbl_ord_idx2 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM kd_point_tbl WHERE p IS NOT NULL;
CREATE TEMP TABLE kd_point_tbl_ord_idx3 AS
SELECT rank() OVER (ORDER BY p <-> '333,400') n, p <-> '333,400' dist, p
FROM kd_point_tbl WHERE p IS NOT NULL;
SELECT * FROM quad_point_tbl_ord_seq3 seq FULL JOIN kd_point_tbl_ord_idx3 idx
ON seq.n = idx.n AND
(seq.dist = idx.dist AND seq.p ~= idx.p OR seq.p IS NULL AND idx.p IS NULL)
WHERE seq.n IS NULL OR idx.n IS NULL;
EXPLAIN (COSTS OFF)
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcdef';
SELECT count(*) FROM radix_text_tbl WHERE t = 'P0123456789abcdef';

33
src/test/regress/sql/polygon.sql
View File

@ -206,6 +206,39 @@ EXPLAIN (COSTS OFF)
SELECT count(*) FROM quad_poly_tbl WHERE p ~= polygon '((200, 300),(210, 310),(230, 290))';
SELECT count(*) FROM quad_poly_tbl WHERE p ~= polygon '((200, 300),(210, 310),(230, 290))';
-- test ORDER BY distance
SET enable_indexscan = ON;
SET enable_bitmapscan = OFF;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl;
CREATE TEMP TABLE quad_poly_tbl_ord_idx1 AS
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl;
SELECT *
FROM quad_poly_tbl_ord_seq1 seq FULL JOIN quad_poly_tbl_ord_idx1 idx
ON seq.n = idx.n AND seq.id = idx.id AND
(seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
WHERE seq.id IS NULL OR idx.id IS NULL;
EXPLAIN (COSTS OFF)
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl WHERE p <@ polygon '((300,300),(400,600),(600,500),(700,200))';
CREATE TEMP TABLE quad_poly_tbl_ord_idx2 AS
SELECT rank() OVER (ORDER BY p <-> point '123,456') n, p <-> point '123,456' dist, id
FROM quad_poly_tbl WHERE p <@ polygon '((300,300),(400,600),(600,500),(700,200))';
SELECT *
FROM quad_poly_tbl_ord_seq2 seq FULL JOIN quad_poly_tbl_ord_idx2 idx
ON seq.n = idx.n AND seq.id = idx.id AND
(seq.dist = idx.dist OR seq.dist IS NULL AND idx.dist IS NULL)
WHERE seq.id IS NULL OR idx.id IS NULL;
RESET enable_seqscan;
RESET enable_indexscan;
RESET enable_bitmapscan;