2008-05-12 02:00:54 +02:00
|
|
|
/*
|
2010-09-20 22:08:53 +02:00
|
|
|
* contrib/hstore/hstore_gist.c
|
2008-05-12 02:00:54 +02:00
|
|
|
*/
|
|
|
|
#include "postgres.h"
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
#include "access/gist.h"
|
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
|
|
|
#include "access/reloptions.h"
|
2015-05-15 22:03:16 +02:00
|
|
|
#include "access/stratnum.h"
|
2009-09-30 21:50:22 +02:00
|
|
|
#include "catalog/pg_type.h"
|
2008-05-12 02:00:54 +02:00
|
|
|
#include "hstore.h"
|
2019-10-23 05:56:22 +02:00
|
|
|
#include "utils/pg_crc.h"
|
2008-05-12 02:00:54 +02:00
|
|
|
|
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
|
|
|
/* gist_hstore_ops opclass options */
|
|
|
|
typedef struct
|
|
|
|
{
|
|
|
|
int32 vl_len_; /* varlena header (do not touch directly!) */
|
|
|
|
int siglen; /* signature length in bytes */
|
|
|
|
} GistHstoreOptions;
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
/* bigint defines */
|
|
|
|
#define BITBYTE 8
|
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
|
|
|
#define SIGLEN_DEFAULT (sizeof(int32) * 4)
|
|
|
|
#define SIGLEN_MAX GISTMaxIndexKeySize
|
|
|
|
#define SIGLENBIT(siglen) ((siglen) * BITBYTE)
|
|
|
|
#define GET_SIGLEN() (PG_HAS_OPCLASS_OPTIONS() ? \
|
|
|
|
((GistHstoreOptions *) PG_GET_OPCLASS_OPTIONS())->siglen : \
|
|
|
|
SIGLEN_DEFAULT)
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
typedef char *BITVECP;
|
|
|
|
|
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
|
|
|
#define LOOPBYTE(siglen) \
|
|
|
|
for (i = 0; i < (siglen); i++)
|
2006-09-05 20:00:58 +02:00
|
|
|
|
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
|
|
|
#define LOOPBIT(siglen) \
|
|
|
|
for (i = 0; i < SIGLENBIT(siglen); i++)
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
/* beware of multiple evaluation of arguments to these macros! */
|
|
|
|
#define GETBYTE(x,i) ( *( (BITVECP)(x) + (int)( (i) / BITBYTE ) ) )
|
|
|
|
#define GETBITBYTE(x,i) ( (*((char*)(x)) >> (i)) & 0x01 )
|
|
|
|
#define CLRBIT(x,i) GETBYTE(x,i) &= ~( 0x01 << ( (i) % BITBYTE ) )
|
|
|
|
#define SETBIT(x,i) GETBYTE(x,i) |= ( 0x01 << ( (i) % BITBYTE ) )
|
|
|
|
#define GETBIT(x,i) ( (GETBYTE(x,i) >> ( (i) % BITBYTE )) & 0x01 )
|
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
|
|
|
#define HASHVAL(val, siglen) (((unsigned int)(val)) % SIGLENBIT(siglen))
|
|
|
|
#define HASH(sign, val, siglen) SETBIT((sign), HASHVAL(val, siglen))
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
typedef struct
|
|
|
|
{
|
2007-02-28 23:44:38 +01:00
|
|
|
int32 vl_len_; /* varlena header (do not touch directly!) */
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 flag;
|
2015-02-21 22:12:14 +01:00
|
|
|
char data[FLEXIBLE_ARRAY_MEMBER];
|
2009-06-11 16:49:15 +02:00
|
|
|
} GISTTYPE;
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
#define ALLISTRUE 0x04
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
#define ISALLTRUE(x) ( ((GISTTYPE*)x)->flag & ALLISTRUE )
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2012-06-25 00:51:46 +02:00
|
|
|
#define GTHDRSIZE (VARHDRSZ + sizeof(int32))
|
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
|
|
|
#define CALCGTSIZE(flag, siglen) ( GTHDRSIZE+(((flag) & ALLISTRUE) ? 0 : (siglen)) )
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
#define GETSIGN(x) ( (BITVECP)( (char*)x+GTHDRSIZE ) )
|
|
|
|
|
|
|
|
#define SUMBIT(val) ( \
|
2006-09-05 20:00:58 +02:00
|
|
|
GETBITBYTE((val),0) + \
|
|
|
|
GETBITBYTE((val),1) + \
|
|
|
|
GETBITBYTE((val),2) + \
|
|
|
|
GETBITBYTE((val),3) + \
|
|
|
|
GETBITBYTE((val),4) + \
|
|
|
|
GETBITBYTE((val),5) + \
|
|
|
|
GETBITBYTE((val),6) + \
|
|
|
|
GETBITBYTE((val),7) \
|
|
|
|
)
|
|
|
|
|
|
|
|
#define GETENTRY(vec,pos) ((GISTTYPE *) DatumGetPointer((vec)->vector[(pos)].key))
|
|
|
|
|
|
|
|
#define WISH_F(a,b,c) (double)( -(double)(((a)-(b))*((a)-(b))*((a)-(b)))*(c) )
|
|
|
|
|
Switch to CRC-32C in WAL and other places.
The old algorithm was found to not be the usual CRC-32 algorithm, used by
Ethernet et al. We were using a non-reflected lookup table with code meant
for a reflected lookup table. That's a strange combination that AFAICS does
not correspond to any bit-wise CRC calculation, which makes it difficult to
reason about its properties. Although it has worked well in practice, seems
safer to use a well-known algorithm.
Since we're changing the algorithm anyway, we might as well choose a
different polynomial. The Castagnoli polynomial has better error-correcting
properties than the traditional CRC-32 polynomial, even if we had
implemented it correctly. Another reason for picking that is that some new
CPUs have hardware support for calculating CRC-32C, but not CRC-32, let
alone our strange variant of it. This patch doesn't add any support for such
hardware, but a future patch could now do that.
The old algorithm is kept around for tsquery and pg_trgm, which use the
values in indexes that need to remain compatible so that pg_upgrade works.
While we're at it, share the old lookup table for CRC-32 calculation
between hstore, ltree and core. They all use the same table, so might as
well.
2014-11-04 10:35:15 +01:00
|
|
|
/* shorthand for calculating CRC-32 of a single chunk of data. */
|
|
|
|
static pg_crc32
|
|
|
|
crc32_sz(char *buf, int size)
|
|
|
|
{
|
2015-05-24 03:35:49 +02:00
|
|
|
pg_crc32 crc;
|
Switch to CRC-32C in WAL and other places.
The old algorithm was found to not be the usual CRC-32 algorithm, used by
Ethernet et al. We were using a non-reflected lookup table with code meant
for a reflected lookup table. That's a strange combination that AFAICS does
not correspond to any bit-wise CRC calculation, which makes it difficult to
reason about its properties. Although it has worked well in practice, seems
safer to use a well-known algorithm.
Since we're changing the algorithm anyway, we might as well choose a
different polynomial. The Castagnoli polynomial has better error-correcting
properties than the traditional CRC-32 polynomial, even if we had
implemented it correctly. Another reason for picking that is that some new
CPUs have hardware support for calculating CRC-32C, but not CRC-32, let
alone our strange variant of it. This patch doesn't add any support for such
hardware, but a future patch could now do that.
The old algorithm is kept around for tsquery and pg_trgm, which use the
values in indexes that need to remain compatible so that pg_upgrade works.
While we're at it, share the old lookup table for CRC-32 calculation
between hstore, ltree and core. They all use the same table, so might as
well.
2014-11-04 10:35:15 +01:00
|
|
|
|
|
|
|
INIT_TRADITIONAL_CRC32(crc);
|
|
|
|
COMP_TRADITIONAL_CRC32(crc, buf, size);
|
|
|
|
FIN_TRADITIONAL_CRC32(crc);
|
|
|
|
|
|
|
|
return crc;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
PG_FUNCTION_INFO_V1(ghstore_in);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_out);
|
2006-10-04 02:30:14 +02:00
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_in(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
elog(ERROR, "Not implemented");
|
|
|
|
PG_RETURN_DATUM(0);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_out(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
elog(ERROR, "Not implemented");
|
|
|
|
PG_RETURN_DATUM(0);
|
|
|
|
}
|
|
|
|
|
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
|
|
|
static GISTTYPE *
|
|
|
|
ghstore_alloc(bool allistrue, int siglen, BITVECP sign)
|
|
|
|
{
|
|
|
|
int flag = allistrue ? ALLISTRUE : 0;
|
|
|
|
int size = CALCGTSIZE(flag, siglen);
|
|
|
|
GISTTYPE *res = palloc(size);
|
|
|
|
|
|
|
|
SET_VARSIZE(res, size);
|
|
|
|
res->flag = flag;
|
|
|
|
|
|
|
|
if (!allistrue)
|
|
|
|
{
|
|
|
|
if (sign)
|
|
|
|
memcpy(GETSIGN(res), sign, siglen);
|
|
|
|
else
|
|
|
|
memset(GETSIGN(res), 0, siglen);
|
|
|
|
}
|
|
|
|
|
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
PG_FUNCTION_INFO_V1(ghstore_consistent);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_compress);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_decompress);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_penalty);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_picksplit);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_union);
|
|
|
|
PG_FUNCTION_INFO_V1(ghstore_same);
|
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
|
|
|
PG_FUNCTION_INFO_V1(ghstore_options);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_compress(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTENTRY *retval = entry;
|
2006-10-04 02:30:14 +02:00
|
|
|
|
|
|
|
if (entry->leafkey)
|
|
|
|
{
|
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
|
|
|
GISTTYPE *res = ghstore_alloc(false, siglen, NULL);
|
2009-09-30 21:50:22 +02:00
|
|
|
HStore *val = DatumGetHStoreP(entry->key);
|
|
|
|
HEntry *hsent = ARRPTR(val);
|
|
|
|
char *ptr = STRPTR(val);
|
2010-02-26 03:01:40 +01:00
|
|
|
int count = HS_COUNT(val);
|
|
|
|
int i;
|
2006-10-04 02:30:14 +02:00
|
|
|
|
2009-09-30 21:50:22 +02:00
|
|
|
for (i = 0; i < count; ++i)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2010-02-26 03:01:40 +01:00
|
|
|
int h;
|
2006-10-04 02:30:14 +02:00
|
|
|
|
2015-11-19 20:54:05 +01:00
|
|
|
h = crc32_sz((char *) HSTORE_KEY(hsent, ptr, i),
|
|
|
|
HSTORE_KEYLEN(hsent, 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(GETSIGN(res), h, siglen);
|
2015-11-19 20:54:05 +01:00
|
|
|
if (!HSTORE_VALISNULL(hsent, i))
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2015-11-19 20:54:05 +01:00
|
|
|
h = crc32_sz((char *) HSTORE_VAL(hsent, ptr, i),
|
|
|
|
HSTORE_VALLEN(hsent, 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(GETSIGN(res), h, siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
|
|
|
|
gistentryinit(*retval, PointerGetDatum(res),
|
|
|
|
entry->rel, entry->page,
|
|
|
|
entry->offset,
|
2017-08-16 06:22:32 +02:00
|
|
|
false);
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else if (!ISALLTRUE(DatumGetPointer(entry->key)))
|
|
|
|
{
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 i;
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTTYPE *res;
|
2006-10-04 02:30:14 +02:00
|
|
|
BITVECP sign = GETSIGN(DatumGetPointer(entry->key));
|
|
|
|
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 01:13:02 +01:00
|
|
|
{
|
2007-11-16 02:12:24 +01:00
|
|
|
if ((sign[i] & 0xff) != 0xff)
|
|
|
|
PG_RETURN_POINTER(retval);
|
2007-11-16 01:13:02 +01:00
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
|
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
|
|
|
res = ghstore_alloc(true, siglen, NULL);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
retval = (GISTENTRY *) palloc(sizeof(GISTENTRY));
|
|
|
|
gistentryinit(*retval, PointerGetDatum(res),
|
|
|
|
entry->rel, entry->page,
|
2006-10-04 02:30:14 +02:00
|
|
|
entry->offset,
|
2017-08-16 06:22:32 +02:00
|
|
|
false);
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(retval);
|
|
|
|
}
|
|
|
|
|
2010-09-16 04:54:01 +02:00
|
|
|
/*
|
|
|
|
* Since type ghstore isn't toastable (and doesn't need to be),
|
|
|
|
* this function can be a no-op.
|
|
|
|
*/
|
2006-09-05 20:00:58 +02:00
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_decompress(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2010-09-16 04:54:01 +02:00
|
|
|
PG_RETURN_POINTER(PG_GETARG_POINTER(0));
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_same(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTTYPE *a = (GISTTYPE *) PG_GETARG_POINTER(0);
|
|
|
|
GISTTYPE *b = (GISTTYPE *) PG_GETARG_POINTER(1);
|
2006-10-04 02:30:14 +02:00
|
|
|
bool *result = (bool *) PG_GETARG_POINTER(2);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
if (ISALLTRUE(a) && ISALLTRUE(b))
|
|
|
|
*result = true;
|
|
|
|
else if (ISALLTRUE(a))
|
|
|
|
*result = false;
|
|
|
|
else if (ISALLTRUE(b))
|
|
|
|
*result = false;
|
2006-10-04 02:30:14 +02:00
|
|
|
else
|
|
|
|
{
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 i;
|
2006-10-04 02:30:14 +02:00
|
|
|
BITVECP sa = GETSIGN(a),
|
|
|
|
sb = GETSIGN(b);
|
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
*result = true;
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 01:13:02 +01:00
|
|
|
{
|
|
|
|
if (sa[i] != sb[i])
|
|
|
|
{
|
|
|
|
*result = false;
|
|
|
|
break;
|
|
|
|
}
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
PG_RETURN_POINTER(result);
|
|
|
|
}
|
|
|
|
|
2012-06-25 00:51:46 +02:00
|
|
|
static int32
|
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
|
|
|
sizebitvec(BITVECP sign, int siglen)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 size = 0,
|
2006-10-04 02:30:14 +02:00
|
|
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 01:13:02 +01:00
|
|
|
{
|
|
|
|
size += SUMBIT(sign);
|
|
|
|
sign = (BITVECP) (((char *) sign) + 1);
|
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
return size;
|
|
|
|
}
|
2006-10-04 02:30:14 +02:00
|
|
|
|
2006-09-05 20:00:58 +02:00
|
|
|
static int
|
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
|
|
|
hemdistsign(BITVECP a, BITVECP b, int siglen)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
|
|
|
int i,
|
|
|
|
dist = 0;
|
2006-09-05 20:00:58 +02:00
|
|
|
|
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
|
|
|
LOOPBIT(siglen)
|
2007-11-16 01:13:02 +01:00
|
|
|
{
|
|
|
|
if (GETBIT(a, i) != GETBIT(b, i))
|
2006-09-05 20:00:58 +02:00
|
|
|
dist++;
|
2007-11-16 01:13:02 +01:00
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
return dist;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
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
|
|
|
hemdist(GISTTYPE *a, GISTTYPE *b, int siglen)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
|
|
|
if (ISALLTRUE(a))
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
if (ISALLTRUE(b))
|
|
|
|
return 0;
|
|
|
|
else
|
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
|
|
|
return SIGLENBIT(siglen) - sizebitvec(GETSIGN(b), siglen);
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else if (ISALLTRUE(b))
|
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
|
|
|
return SIGLENBIT(siglen) - sizebitvec(GETSIGN(a), siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
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
|
|
|
return hemdistsign(GETSIGN(a), GETSIGN(b), siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
|
2012-06-25 00:51:46 +02:00
|
|
|
static int32
|
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
|
|
|
unionkey(BITVECP sbase, GISTTYPE *add, int siglen)
|
2006-09-05 20:00:58 +02:00
|
|
|
{
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 i;
|
2006-10-04 02:30:14 +02:00
|
|
|
BITVECP sadd = GETSIGN(add);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
if (ISALLTRUE(add))
|
|
|
|
return 1;
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 02:12:24 +01:00
|
|
|
sbase[i] |= sadd[i];
|
2006-09-05 20:00:58 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_union(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0);
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 len = entryvec->n;
|
2006-10-04 02:30:14 +02:00
|
|
|
|
|
|
|
int *size = (int *) PG_GETARG_POINTER(1);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 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
|
|
|
GISTTYPE *result = ghstore_alloc(false, siglen, NULL);
|
|
|
|
BITVECP base = GETSIGN(result);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
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
|
|
|
if (unionkey(base, GETENTRY(entryvec, i), siglen))
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
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
|
|
|
result->flag |= ALLISTRUE;
|
|
|
|
SET_VARSIZE(result, CALCGTSIZE(ALLISTRUE, siglen));
|
2006-09-05 20:00:58 +02:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
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
|
|
|
*size = VARSIZE(result);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
PG_RETURN_POINTER(result);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_penalty(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTENTRY *origentry = (GISTENTRY *) PG_GETARG_POINTER(0); /* always ISSIGNKEY */
|
|
|
|
GISTENTRY *newentry = (GISTENTRY *) PG_GETARG_POINTER(1);
|
2006-10-04 02:30:14 +02:00
|
|
|
float *penalty = (float *) PG_GETARG_POINTER(2);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
2006-09-05 20:00:58 +02:00
|
|
|
GISTTYPE *origval = (GISTTYPE *) DatumGetPointer(origentry->key);
|
|
|
|
GISTTYPE *newval = (GISTTYPE *) DatumGetPointer(newentry->key);
|
|
|
|
|
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
|
|
|
*penalty = hemdist(origval, newval, siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
PG_RETURN_POINTER(penalty);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
typedef struct
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
OffsetNumber pos;
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 cost;
|
2007-11-16 02:12:24 +01:00
|
|
|
} SPLITCOST;
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
static int
|
2006-10-04 02:30:14 +02:00
|
|
|
comparecost(const void *a, const void *b)
|
|
|
|
{
|
2011-09-11 20:54:32 +02:00
|
|
|
return ((const SPLITCOST *) a)->cost - ((const SPLITCOST *) b)->cost;
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
Datum
|
2006-10-04 02:30:14 +02:00
|
|
|
ghstore_picksplit(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0);
|
2006-09-05 20:00:58 +02:00
|
|
|
OffsetNumber maxoff = entryvec->n - 2;
|
|
|
|
|
|
|
|
GIST_SPLITVEC *v = (GIST_SPLITVEC *) PG_GETARG_POINTER(1);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
2006-09-05 20:00:58 +02:00
|
|
|
OffsetNumber k,
|
|
|
|
j;
|
2006-10-04 02:30:14 +02:00
|
|
|
GISTTYPE *datum_l,
|
2006-09-05 20:00:58 +02:00
|
|
|
*datum_r;
|
2006-10-04 02:30:14 +02:00
|
|
|
BITVECP union_l,
|
2006-09-05 20:00:58 +02:00
|
|
|
union_r;
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 size_alpha,
|
2006-10-04 02:30:14 +02:00
|
|
|
size_beta;
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 size_waste,
|
2006-09-05 20:00:58 +02:00
|
|
|
waste = -1;
|
2012-06-25 00:51:46 +02:00
|
|
|
int32 nbytes;
|
2006-09-05 20:00:58 +02:00
|
|
|
OffsetNumber seed_1 = 0,
|
|
|
|
seed_2 = 0;
|
|
|
|
OffsetNumber *left,
|
|
|
|
*right;
|
2006-10-04 02:30:14 +02:00
|
|
|
BITVECP ptr;
|
|
|
|
int i;
|
2006-09-05 20:00:58 +02:00
|
|
|
SPLITCOST *costvector;
|
2006-10-04 02:30:14 +02:00
|
|
|
GISTTYPE *_k,
|
2006-09-05 20:00:58 +02:00
|
|
|
*_j;
|
|
|
|
|
|
|
|
nbytes = (maxoff + 2) * sizeof(OffsetNumber);
|
|
|
|
v->spl_left = (OffsetNumber *) palloc(nbytes);
|
|
|
|
v->spl_right = (OffsetNumber *) palloc(nbytes);
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
for (k = FirstOffsetNumber; k < maxoff; k = OffsetNumberNext(k))
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
_k = GETENTRY(entryvec, k);
|
2006-10-04 02:30:14 +02:00
|
|
|
for (j = OffsetNumberNext(k); j <= maxoff; j = OffsetNumberNext(j))
|
|
|
|
{
|
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
|
|
|
size_waste = hemdist(_k, GETENTRY(entryvec, j), siglen);
|
2006-10-04 02:30:14 +02:00
|
|
|
if (size_waste > waste)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
waste = size_waste;
|
|
|
|
seed_1 = k;
|
|
|
|
seed_2 = j;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
left = v->spl_left;
|
|
|
|
v->spl_nleft = 0;
|
|
|
|
right = v->spl_right;
|
|
|
|
v->spl_nright = 0;
|
|
|
|
|
|
|
|
if (seed_1 == 0 || seed_2 == 0)
|
|
|
|
{
|
|
|
|
seed_1 = 1;
|
|
|
|
seed_2 = 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* form initial .. */
|
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
|
|
|
datum_l = ghstore_alloc(ISALLTRUE(GETENTRY(entryvec, seed_1)), siglen,
|
|
|
|
GETSIGN(GETENTRY(entryvec, seed_1)));
|
|
|
|
datum_r = ghstore_alloc(ISALLTRUE(GETENTRY(entryvec, seed_2)), siglen,
|
|
|
|
GETSIGN(GETENTRY(entryvec, seed_2)));
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
maxoff = OffsetNumberNext(maxoff);
|
|
|
|
/* sort before ... */
|
|
|
|
costvector = (SPLITCOST *) palloc(sizeof(SPLITCOST) * maxoff);
|
|
|
|
for (j = FirstOffsetNumber; j <= maxoff; j = OffsetNumberNext(j))
|
|
|
|
{
|
|
|
|
costvector[j - 1].pos = j;
|
|
|
|
_j = GETENTRY(entryvec, j);
|
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
|
|
|
size_alpha = hemdist(datum_l, _j, siglen);
|
|
|
|
size_beta = hemdist(datum_r, _j, siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
costvector[j - 1].cost = abs(size_alpha - size_beta);
|
|
|
|
}
|
|
|
|
qsort((void *) costvector, maxoff, sizeof(SPLITCOST), comparecost);
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
union_l = GETSIGN(datum_l);
|
|
|
|
union_r = GETSIGN(datum_r);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
for (k = 0; k < maxoff; k++)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
j = costvector[k].pos;
|
2006-10-04 02:30:14 +02:00
|
|
|
if (j == seed_1)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
*left++ = j;
|
|
|
|
v->spl_nleft++;
|
|
|
|
continue;
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else if (j == seed_2)
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
*right++ = j;
|
|
|
|
v->spl_nright++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
_j = GETENTRY(entryvec, j);
|
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
|
|
|
size_alpha = hemdist(datum_l, _j, siglen);
|
|
|
|
size_beta = hemdist(datum_r, _j, siglen);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
if (size_alpha < size_beta + WISH_F(v->spl_nleft, v->spl_nright, 0.0001))
|
|
|
|
{
|
|
|
|
if (ISALLTRUE(datum_l) || ISALLTRUE(_j))
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
if (!ISALLTRUE(datum_l))
|
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
|
|
|
MemSet((void *) union_l, 0xff, siglen);
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
ptr = GETSIGN(_j);
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 02:12:24 +01:00
|
|
|
union_l[i] |= ptr[i];
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
*left++ = j;
|
|
|
|
v->spl_nleft++;
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
if (ISALLTRUE(datum_r) || ISALLTRUE(_j))
|
|
|
|
{
|
2006-09-05 20:00:58 +02:00
|
|
|
if (!ISALLTRUE(datum_r))
|
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
|
|
|
MemSet((void *) union_r, 0xff, siglen);
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
ptr = GETSIGN(_j);
|
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
|
|
|
LOOPBYTE(siglen)
|
2007-11-16 02:12:24 +01:00
|
|
|
union_r[i] |= ptr[i];
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
|
|
|
*right++ = j;
|
|
|
|
v->spl_nright++;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*right = *left = FirstOffsetNumber;
|
|
|
|
|
|
|
|
v->spl_ldatum = PointerGetDatum(datum_l);
|
|
|
|
v->spl_rdatum = PointerGetDatum(datum_r);
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(v);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
Datum
|
|
|
|
ghstore_consistent(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
GISTTYPE *entry = (GISTTYPE *) DatumGetPointer(((GISTENTRY *) PG_GETARG_POINTER(0))->key);
|
2007-03-14 15:21:53 +01:00
|
|
|
StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2);
|
2010-02-26 03:01:40 +01:00
|
|
|
|
2008-04-14 19:05:34 +02:00
|
|
|
/* Oid subtype = PG_GETARG_OID(3); */
|
|
|
|
bool *recheck = (bool *) PG_GETARG_POINTER(4);
|
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
|
|
|
int siglen = GET_SIGLEN();
|
2006-10-04 02:30:14 +02:00
|
|
|
bool res = true;
|
|
|
|
BITVECP sign;
|
|
|
|
|
2008-04-14 19:05:34 +02:00
|
|
|
/* All cases served by this function are inexact */
|
|
|
|
*recheck = true;
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
if (ISALLTRUE(entry))
|
2006-09-05 20:00:58 +02:00
|
|
|
PG_RETURN_BOOL(true);
|
|
|
|
|
2006-10-04 02:30:14 +02:00
|
|
|
sign = GETSIGN(entry);
|
2007-03-14 15:21:53 +01:00
|
|
|
|
2009-09-30 21:50:22 +02:00
|
|
|
if (strategy == HStoreContainsStrategyNumber ||
|
|
|
|
strategy == HStoreOldContainsStrategyNumber)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2017-09-18 21:21:23 +02:00
|
|
|
HStore *query = PG_GETARG_HSTORE_P(1);
|
2007-03-14 15:21:53 +01:00
|
|
|
HEntry *qe = ARRPTR(query);
|
|
|
|
char *qv = STRPTR(query);
|
2010-02-26 03:01:40 +01:00
|
|
|
int count = HS_COUNT(query);
|
|
|
|
int i;
|
2006-10-04 02:30:14 +02:00
|
|
|
|
2009-09-30 21:50:22 +02:00
|
|
|
for (i = 0; res && i < count; ++i)
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2015-11-19 20:54:05 +01:00
|
|
|
int crc = crc32_sz((char *) HSTORE_KEY(qe, qv, i),
|
|
|
|
HSTORE_KEYLEN(qe, i));
|
2007-03-14 15:21:53 +01:00
|
|
|
|
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
|
|
|
if (GETBIT(sign, HASHVAL(crc, siglen)))
|
2006-10-04 02:30:14 +02:00
|
|
|
{
|
2015-11-19 20:54:05 +01:00
|
|
|
if (!HSTORE_VALISNULL(qe, i))
|
2007-03-14 15:21:53 +01:00
|
|
|
{
|
2015-11-19 20:54:05 +01:00
|
|
|
crc = crc32_sz((char *) HSTORE_VAL(qe, qv, i),
|
|
|
|
HSTORE_VALLEN(qe, 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
|
|
|
if (!GETBIT(sign, HASHVAL(crc, siglen)))
|
2007-03-14 15:21:53 +01:00
|
|
|
res = false;
|
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
2007-03-14 15:21:53 +01:00
|
|
|
else
|
|
|
|
res = false;
|
2006-10-04 02:30:14 +02:00
|
|
|
}
|
2006-09-05 20:00:58 +02:00
|
|
|
}
|
2007-11-16 02:12:24 +01:00
|
|
|
else if (strategy == HStoreExistsStrategyNumber)
|
2007-03-14 15:21:53 +01:00
|
|
|
{
|
2009-09-30 21:50:22 +02:00
|
|
|
text *query = PG_GETARG_TEXT_PP(1);
|
|
|
|
int crc = crc32_sz(VARDATA_ANY(query), VARSIZE_ANY_EXHDR(query));
|
2007-03-14 15:21:53 +01:00
|
|
|
|
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
|
|
|
res = (GETBIT(sign, HASHVAL(crc, siglen))) ? true : false;
|
2007-03-14 15:21:53 +01:00
|
|
|
}
|
2009-09-30 21:50:22 +02:00
|
|
|
else if (strategy == HStoreExistsAllStrategyNumber)
|
|
|
|
{
|
2010-02-26 03:01:40 +01:00
|
|
|
ArrayType *query = PG_GETARG_ARRAYTYPE_P(1);
|
|
|
|
Datum *key_datums;
|
|
|
|
bool *key_nulls;
|
|
|
|
int key_count;
|
|
|
|
int i;
|
2009-09-30 21:50:22 +02:00
|
|
|
|
|
|
|
deconstruct_array(query,
|
2020-03-04 16:34:25 +01:00
|
|
|
TEXTOID, -1, false, TYPALIGN_INT,
|
2009-09-30 21:50:22 +02:00
|
|
|
&key_datums, &key_nulls, &key_count);
|
|
|
|
|
|
|
|
for (i = 0; res && i < key_count; ++i)
|
|
|
|
{
|
2010-02-26 03:01:40 +01:00
|
|
|
int crc;
|
|
|
|
|
2009-09-30 21:50:22 +02:00
|
|
|
if (key_nulls[i])
|
|
|
|
continue;
|
|
|
|
crc = crc32_sz(VARDATA(key_datums[i]), VARSIZE(key_datums[i]) - VARHDRSZ);
|
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
|
|
|
if (!(GETBIT(sign, HASHVAL(crc, siglen))))
|
2017-08-16 06:22:32 +02:00
|
|
|
res = false;
|
2009-09-30 21:50:22 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
else if (strategy == HStoreExistsAnyStrategyNumber)
|
|
|
|
{
|
2010-02-26 03:01:40 +01:00
|
|
|
ArrayType *query = PG_GETARG_ARRAYTYPE_P(1);
|
|
|
|
Datum *key_datums;
|
|
|
|
bool *key_nulls;
|
|
|
|
int key_count;
|
|
|
|
int i;
|
2009-09-30 21:50:22 +02:00
|
|
|
|
|
|
|
deconstruct_array(query,
|
2020-03-04 16:34:25 +01:00
|
|
|
TEXTOID, -1, false, TYPALIGN_INT,
|
2009-09-30 21:50:22 +02:00
|
|
|
&key_datums, &key_nulls, &key_count);
|
|
|
|
|
2017-08-16 06:22:32 +02:00
|
|
|
res = false;
|
2009-09-30 21:50:22 +02:00
|
|
|
|
|
|
|
for (i = 0; !res && i < key_count; ++i)
|
|
|
|
{
|
2010-02-26 03:01:40 +01:00
|
|
|
int crc;
|
|
|
|
|
2009-09-30 21:50:22 +02:00
|
|
|
if (key_nulls[i])
|
|
|
|
continue;
|
|
|
|
crc = crc32_sz(VARDATA(key_datums[i]), VARSIZE(key_datums[i]) - VARHDRSZ);
|
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
|
|
|
if (GETBIT(sign, HASHVAL(crc, siglen)))
|
2017-08-16 06:22:32 +02:00
|
|
|
res = true;
|
2009-09-30 21:50:22 +02:00
|
|
|
}
|
|
|
|
}
|
2007-03-14 15:21:53 +01:00
|
|
|
else
|
|
|
|
elog(ERROR, "Unsupported strategy number: %d", strategy);
|
2006-09-05 20:00:58 +02:00
|
|
|
|
|
|
|
PG_RETURN_BOOL(res);
|
|
|
|
}
|
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
|
|
|
|
|
|
|
Datum
|
|
|
|
ghstore_options(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
local_relopts *relopts = (local_relopts *) PG_GETARG_POINTER(0);
|
|
|
|
|
|
|
|
init_local_reloptions(relopts, sizeof(GistHstoreOptions));
|
|
|
|
add_local_int_reloption(relopts, "siglen",
|
|
|
|
"signature length in bytes",
|
|
|
|
SIGLEN_DEFAULT, 1, SIGLEN_MAX,
|
|
|
|
offsetof(GistHstoreOptions, siglen));
|
|
|
|
|
|
|
|
PG_RETURN_VOID();
|
|
|
|
}
|