Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
/*-------------------------------------------------------------------------
|
|
|
|
*
|
|
|
|
* jsonb_gin.c
|
|
|
|
* GIN support functions for jsonb
|
|
|
|
*
|
2015-01-06 17:43:47 +01:00
|
|
|
* Copyright (c) 2014-2015, PostgreSQL Global Development Group
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*
|
|
|
|
*
|
|
|
|
* IDENTIFICATION
|
|
|
|
* src/backend/utils/adt/jsonb_gin.c
|
|
|
|
*
|
|
|
|
*-------------------------------------------------------------------------
|
|
|
|
*/
|
|
|
|
#include "postgres.h"
|
|
|
|
|
|
|
|
#include "access/gin.h"
|
2014-05-09 14:41:26 +02:00
|
|
|
#include "access/hash.h"
|
2015-05-15 22:03:16 +02:00
|
|
|
#include "access/stratnum.h"
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
#include "catalog/pg_collation.h"
|
|
|
|
#include "catalog/pg_type.h"
|
|
|
|
#include "utils/builtins.h"
|
|
|
|
#include "utils/jsonb.h"
|
|
|
|
|
|
|
|
typedef struct PathHashStack
|
|
|
|
{
|
2014-05-06 18:12:18 +02:00
|
|
|
uint32 hash;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
struct PathHashStack *parent;
|
2014-05-06 18:12:18 +02:00
|
|
|
} PathHashStack;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
static Datum make_text_key(char flag, const char *str, int len);
|
|
|
|
static Datum make_scalar_key(const JsonbValue *scalarVal, bool is_key);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
|
|
|
/*
|
|
|
|
*
|
|
|
|
* jsonb_ops GIN opclass support functions
|
|
|
|
*
|
|
|
|
*/
|
2014-05-09 14:41:26 +02:00
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
Datum
|
|
|
|
gin_compare_jsonb(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
text *arg1 = PG_GETARG_TEXT_PP(0);
|
|
|
|
text *arg2 = PG_GETARG_TEXT_PP(1);
|
|
|
|
int32 result;
|
|
|
|
char *a1p,
|
|
|
|
*a2p;
|
|
|
|
int len1,
|
|
|
|
len2;
|
|
|
|
|
|
|
|
a1p = VARDATA_ANY(arg1);
|
|
|
|
a2p = VARDATA_ANY(arg2);
|
|
|
|
|
|
|
|
len1 = VARSIZE_ANY_EXHDR(arg1);
|
|
|
|
len2 = VARSIZE_ANY_EXHDR(arg2);
|
|
|
|
|
|
|
|
/* Compare text as bttextcmp does, but always using C collation */
|
|
|
|
result = varstr_cmp(a1p, len1, a2p, len2, C_COLLATION_OID);
|
|
|
|
|
|
|
|
PG_FREE_IF_COPY(arg1, 0);
|
|
|
|
PG_FREE_IF_COPY(arg2, 1);
|
|
|
|
|
|
|
|
PG_RETURN_INT32(result);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
|
|
|
gin_extract_jsonb(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
Jsonb *jb = (Jsonb *) PG_GETARG_JSONB(0);
|
|
|
|
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
|
|
|
|
int total = 2 * JB_ROOT_COUNT(jb);
|
|
|
|
JsonbIterator *it;
|
|
|
|
JsonbValue v;
|
2014-05-09 14:41:26 +02:00
|
|
|
int i = 0,
|
|
|
|
r;
|
|
|
|
Datum *entries;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* If the root level is empty, we certainly have no keys */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
if (total == 0)
|
|
|
|
{
|
|
|
|
*nentries = 0;
|
|
|
|
PG_RETURN_POINTER(NULL);
|
|
|
|
}
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Otherwise, use 2 * root count as initial estimate of result size */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
entries = (Datum *) palloc(sizeof(Datum) * total);
|
|
|
|
|
2014-05-07 22:16:19 +02:00
|
|
|
it = JsonbIteratorInit(&jb->root);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
|
|
|
while ((r = JsonbIteratorNext(&it, &v, false)) != WJB_DONE)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Since we recurse into the object, we might need more space */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
if (i >= total)
|
|
|
|
{
|
|
|
|
total *= 2;
|
|
|
|
entries = (Datum *) repalloc(entries, sizeof(Datum) * total);
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (r)
|
|
|
|
{
|
|
|
|
case WJB_KEY:
|
2014-05-09 14:41:26 +02:00
|
|
|
entries[i++] = make_scalar_key(&v, true);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case WJB_ELEM:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Pretend string array elements are keys, see jsonb.h */
|
|
|
|
entries[i++] = make_scalar_key(&v, (v.type == jbvString));
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case WJB_VALUE:
|
2014-05-09 14:41:26 +02:00
|
|
|
entries[i++] = make_scalar_key(&v, false);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
default:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* we can ignore structural items */
|
|
|
|
break;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*nentries = i;
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(entries);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
|
|
|
gin_extract_jsonb_query(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
|
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(2);
|
|
|
|
int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
|
|
|
|
Datum *entries;
|
|
|
|
|
|
|
|
if (strategy == JsonbContainsStrategyNumber)
|
|
|
|
{
|
|
|
|
/* Query is a jsonb, so just apply gin_extract_jsonb... */
|
|
|
|
entries = (Datum *)
|
|
|
|
DatumGetPointer(DirectFunctionCall2(gin_extract_jsonb,
|
|
|
|
PG_GETARG_DATUM(0),
|
|
|
|
PointerGetDatum(nentries)));
|
|
|
|
/* ...although "contains {}" requires a full index scan */
|
2014-05-09 14:41:26 +02:00
|
|
|
if (*nentries == 0)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*searchMode = GIN_SEARCH_MODE_ALL;
|
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Query is a text string, which we treat as a key */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
text *query = PG_GETARG_TEXT_PP(0);
|
|
|
|
|
|
|
|
*nentries = 1;
|
|
|
|
entries = (Datum *) palloc(sizeof(Datum));
|
2014-05-09 14:41:26 +02:00
|
|
|
entries[0] = make_text_key(JGINFLAG_KEY,
|
|
|
|
VARDATA_ANY(query),
|
|
|
|
VARSIZE_ANY_EXHDR(query));
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsAnyStrategyNumber ||
|
|
|
|
strategy == JsonbExistsAllStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Query is a text array; each element is treated as a key */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
ArrayType *query = PG_GETARG_ARRAYTYPE_P(0);
|
|
|
|
Datum *key_datums;
|
|
|
|
bool *key_nulls;
|
|
|
|
int key_count;
|
|
|
|
int i,
|
|
|
|
j;
|
|
|
|
|
|
|
|
deconstruct_array(query,
|
|
|
|
TEXTOID, -1, false, 'i',
|
|
|
|
&key_datums, &key_nulls, &key_count);
|
|
|
|
|
|
|
|
entries = (Datum *) palloc(sizeof(Datum) * key_count);
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
for (i = 0, j = 0; i < key_count; i++)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
/* Nulls in the array are ignored */
|
|
|
|
if (key_nulls[i])
|
|
|
|
continue;
|
2014-05-09 14:41:26 +02:00
|
|
|
entries[j++] = make_text_key(JGINFLAG_KEY,
|
|
|
|
VARDATA_ANY(key_datums[i]),
|
|
|
|
VARSIZE_ANY_EXHDR(key_datums[i]));
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
*nentries = j;
|
|
|
|
/* ExistsAll with no keys should match everything */
|
|
|
|
if (j == 0 && strategy == JsonbExistsAllStrategyNumber)
|
|
|
|
*searchMode = GIN_SEARCH_MODE_ALL;
|
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
entries = NULL; /* keep compiler quiet */
|
|
|
|
}
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(entries);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
|
|
|
gin_consistent_jsonb(PG_FUNCTION_ARGS)
|
|
|
|
{
|
|
|
|
bool *check = (bool *) PG_GETARG_POINTER(0);
|
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(1);
|
|
|
|
|
|
|
|
/* Jsonb *query = PG_GETARG_JSONB(2); */
|
|
|
|
int32 nkeys = PG_GETARG_INT32(3);
|
|
|
|
|
|
|
|
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
|
|
|
|
bool *recheck = (bool *) PG_GETARG_POINTER(5);
|
|
|
|
bool res = true;
|
|
|
|
int32 i;
|
|
|
|
|
|
|
|
if (strategy == JsonbContainsStrategyNumber)
|
|
|
|
{
|
|
|
|
/*
|
2014-05-09 14:41:26 +02:00
|
|
|
* We must always recheck, since we can't tell from the index whether
|
|
|
|
* the positions of the matched items match the structure of the query
|
|
|
|
* object. (Even if we could, we'd also have to worry about hashed
|
|
|
|
* keys and the index's failure to distinguish keys from string array
|
|
|
|
* elements.) However, the tuple certainly doesn't match unless it
|
|
|
|
* contains all the query keys.
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
|
|
|
*recheck = true;
|
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
|
|
|
if (!check[i])
|
|
|
|
{
|
|
|
|
res = false;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/*
|
|
|
|
* Although the key is certainly present in the index, we must recheck
|
|
|
|
* because (1) the key might be hashed, and (2) the index match might
|
|
|
|
* be for a key that's not at top level of the JSON object. For (1),
|
|
|
|
* we could look at the query key to see if it's hashed and not
|
|
|
|
* recheck if not, but the index lacks enough info to tell about (2).
|
|
|
|
*/
|
|
|
|
*recheck = true;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
res = true;
|
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsAnyStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* As for plain exists, we must recheck */
|
|
|
|
*recheck = true;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
res = true;
|
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsAllStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* As for plain exists, we must recheck */
|
|
|
|
*recheck = true;
|
|
|
|
/* ... but unless all the keys are present, we can say "false" */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
|
|
|
if (!check[i])
|
|
|
|
{
|
|
|
|
res = false;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
|
|
|
|
PG_RETURN_BOOL(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
|
|
|
gin_triconsistent_jsonb(PG_FUNCTION_ARGS)
|
|
|
|
{
|
2014-03-31 09:26:38 +02:00
|
|
|
GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(1);
|
2014-05-06 18:12:18 +02:00
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
/* Jsonb *query = PG_GETARG_JSONB(2); */
|
|
|
|
int32 nkeys = PG_GETARG_INT32(3);
|
2014-05-06 18:12:18 +02:00
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
|
2014-05-09 14:41:26 +02:00
|
|
|
GinTernaryValue res = GIN_MAYBE;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
int32 i;
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/*
|
|
|
|
* Note that we never return GIN_TRUE, only GIN_MAYBE or GIN_FALSE; this
|
|
|
|
* corresponds to always forcing recheck in the regular consistent
|
|
|
|
* function, for the reasons listed there.
|
|
|
|
*/
|
|
|
|
if (strategy == JsonbContainsStrategyNumber ||
|
|
|
|
strategy == JsonbExistsAllStrategyNumber)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* All extracted keys must be present */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
|
|
|
if (check[i] == GIN_FALSE)
|
|
|
|
{
|
|
|
|
res = GIN_FALSE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else if (strategy == JsonbExistsStrategyNumber ||
|
|
|
|
strategy == JsonbExistsAnyStrategyNumber)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
/* At least one extracted key must be present */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
res = GIN_FALSE;
|
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
if (check[i] == GIN_TRUE ||
|
|
|
|
check[i] == GIN_MAYBE)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
res = GIN_MAYBE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
else
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
|
2014-03-31 09:26:38 +02:00
|
|
|
PG_RETURN_GIN_TERNARY_VALUE(res);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
*
|
2014-05-11 18:06:04 +02:00
|
|
|
* jsonb_path_ops GIN opclass support functions
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*
|
2014-05-11 18:06:04 +02:00
|
|
|
* In a jsonb_path_ops index, the GIN keys are uint32 hashes, one per JSON
|
2014-05-09 14:41:26 +02:00
|
|
|
* value; but the JSON key(s) leading to each value are also included in its
|
|
|
|
* hash computation. This means we can only support containment queries,
|
|
|
|
* but the index can distinguish, for example, {"foo": 42} from {"bar": 42}
|
|
|
|
* since different hashes will be generated.
|
|
|
|
*
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
|
|
|
|
|
|
|
Datum
|
2014-05-11 18:06:04 +02:00
|
|
|
gin_extract_jsonb_path(PG_FUNCTION_ARGS)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
Jsonb *jb = PG_GETARG_JSONB(0);
|
|
|
|
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
|
|
|
|
int total = 2 * JB_ROOT_COUNT(jb);
|
|
|
|
JsonbIterator *it;
|
|
|
|
JsonbValue v;
|
|
|
|
PathHashStack tail;
|
|
|
|
PathHashStack *stack;
|
|
|
|
int i = 0,
|
|
|
|
r;
|
2014-05-09 14:41:26 +02:00
|
|
|
Datum *entries;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* If the root level is empty, we certainly have no keys */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
if (total == 0)
|
|
|
|
{
|
|
|
|
*nentries = 0;
|
|
|
|
PG_RETURN_POINTER(NULL);
|
|
|
|
}
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Otherwise, use 2 * root count as initial estimate of result size */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
entries = (Datum *) palloc(sizeof(Datum) * total);
|
|
|
|
|
2014-05-11 18:06:04 +02:00
|
|
|
/* We keep a stack of partial hashes corresponding to parent key levels */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
tail.parent = NULL;
|
|
|
|
tail.hash = 0;
|
|
|
|
stack = &tail;
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
it = JsonbIteratorInit(&jb->root);
|
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
while ((r = JsonbIteratorNext(&it, &v, false)) != WJB_DONE)
|
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
PathHashStack *parent;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Since we recurse into the object, we might need more space */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
if (i >= total)
|
|
|
|
{
|
|
|
|
total *= 2;
|
|
|
|
entries = (Datum *) repalloc(entries, sizeof(Datum) * total);
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (r)
|
|
|
|
{
|
|
|
|
case WJB_BEGIN_ARRAY:
|
|
|
|
case WJB_BEGIN_OBJECT:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Push a stack level for this object */
|
|
|
|
parent = stack;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
stack = (PathHashStack *) palloc(sizeof(PathHashStack));
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
if (parent->parent)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* We pass forward hashes from previous container nesting
|
|
|
|
* levels so that nested arrays with an outermost nested
|
2014-05-06 18:12:18 +02:00
|
|
|
* object will have element hashes mixed with the
|
|
|
|
* outermost key. It's also somewhat useful to have
|
2014-05-09 14:41:26 +02:00
|
|
|
* nested objects' innermost values have hashes that are a
|
2014-05-06 18:12:18 +02:00
|
|
|
* function of not just their own key, but outer keys too.
|
2014-05-09 14:41:26 +02:00
|
|
|
*
|
|
|
|
* Nesting an array within another array will not alter
|
|
|
|
* innermost scalar element hash values, but that seems
|
|
|
|
* inconsequential.
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
2014-05-09 14:41:26 +02:00
|
|
|
stack->hash = parent->hash;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
else
|
|
|
|
{
|
|
|
|
/*
|
2014-05-09 14:41:26 +02:00
|
|
|
* At the outermost level, initialize hash with container
|
|
|
|
* type proxy value. Note that this makes JB_FARRAY and
|
|
|
|
* JB_FOBJECT part of the on-disk representation, but they
|
|
|
|
* are that in the base jsonb object storage already.
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
2014-05-06 18:12:18 +02:00
|
|
|
stack->hash = (r == WJB_BEGIN_ARRAY) ? JB_FARRAY : JB_FOBJECT;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
2014-05-09 14:41:26 +02:00
|
|
|
stack->parent = parent;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case WJB_KEY:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* initialize hash from parent */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
stack->hash = stack->parent->hash;
|
2014-05-09 14:41:26 +02:00
|
|
|
/* and mix in this key */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
JsonbHashScalarValue(&v, &stack->hash);
|
2014-05-09 14:41:26 +02:00
|
|
|
/* hash is now ready to incorporate the value */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case WJB_ELEM:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* array elements use parent hash mixed with element's hash */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
stack->hash = stack->parent->hash;
|
2014-05-09 14:41:26 +02:00
|
|
|
/* FALL THRU */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
case WJB_VALUE:
|
2014-05-09 14:41:26 +02:00
|
|
|
/* mix the element or value's hash into the prepared hash */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
JsonbHashScalarValue(&v, &stack->hash);
|
2014-05-09 14:41:26 +02:00
|
|
|
/* and emit an index entry */
|
2014-05-07 05:52:30 +02:00
|
|
|
entries[i++] = UInt32GetDatum(stack->hash);
|
2014-05-09 14:41:26 +02:00
|
|
|
/* Note: we assume we'll see KEY before another VALUE */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case WJB_END_ARRAY:
|
|
|
|
case WJB_END_OBJECT:
|
|
|
|
/* Pop the stack */
|
2014-05-09 14:41:26 +02:00
|
|
|
parent = stack->parent;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
pfree(stack);
|
2014-05-09 14:41:26 +02:00
|
|
|
stack = parent;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
default:
|
|
|
|
elog(ERROR, "invalid JsonbIteratorNext rc: %d", r);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
*nentries = i;
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(entries);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2014-05-11 18:06:04 +02:00
|
|
|
gin_extract_jsonb_query_path(PG_FUNCTION_ARGS)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
|
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(2);
|
|
|
|
int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
|
|
|
|
Datum *entries;
|
|
|
|
|
|
|
|
if (strategy != JsonbContainsStrategyNumber)
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
|
2014-05-11 18:06:04 +02:00
|
|
|
/* Query is a jsonb, so just apply gin_extract_jsonb_path ... */
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
entries = (Datum *)
|
2014-05-11 18:06:04 +02:00
|
|
|
DatumGetPointer(DirectFunctionCall2(gin_extract_jsonb_path,
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
PG_GETARG_DATUM(0),
|
|
|
|
PointerGetDatum(nentries)));
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/* ... although "contains {}" requires a full index scan */
|
|
|
|
if (*nentries == 0)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*searchMode = GIN_SEARCH_MODE_ALL;
|
|
|
|
|
|
|
|
PG_RETURN_POINTER(entries);
|
|
|
|
}
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
Datum
|
2014-05-11 18:06:04 +02:00
|
|
|
gin_consistent_jsonb_path(PG_FUNCTION_ARGS)
|
2014-05-09 14:41:26 +02:00
|
|
|
{
|
|
|
|
bool *check = (bool *) PG_GETARG_POINTER(0);
|
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(1);
|
|
|
|
|
|
|
|
/* Jsonb *query = PG_GETARG_JSONB(2); */
|
|
|
|
int32 nkeys = PG_GETARG_INT32(3);
|
|
|
|
|
|
|
|
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
|
|
|
|
bool *recheck = (bool *) PG_GETARG_POINTER(5);
|
|
|
|
bool res = true;
|
|
|
|
int32 i;
|
|
|
|
|
|
|
|
if (strategy != JsonbContainsStrategyNumber)
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
|
|
|
|
/*
|
2014-05-11 18:06:04 +02:00
|
|
|
* jsonb_path_ops is necessarily lossy, not only because of hash
|
2014-05-09 14:41:26 +02:00
|
|
|
* collisions but also because it doesn't preserve complete information
|
|
|
|
* about the structure of the JSON object. Besides, there are some
|
2014-05-11 18:06:04 +02:00
|
|
|
* special rules around the containment of raw scalars in arrays that are
|
|
|
|
* not handled here. So we must always recheck a match. However, if not
|
|
|
|
* all of the keys are present, the tuple certainly doesn't match.
|
2014-05-09 14:41:26 +02:00
|
|
|
*/
|
|
|
|
*recheck = true;
|
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
|
|
|
if (!check[i])
|
|
|
|
{
|
|
|
|
res = false;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
PG_RETURN_BOOL(res);
|
|
|
|
}
|
|
|
|
|
|
|
|
Datum
|
2014-05-11 18:06:04 +02:00
|
|
|
gin_triconsistent_jsonb_path(PG_FUNCTION_ARGS)
|
2014-05-09 14:41:26 +02:00
|
|
|
{
|
|
|
|
GinTernaryValue *check = (GinTernaryValue *) PG_GETARG_POINTER(0);
|
|
|
|
StrategyNumber strategy = PG_GETARG_UINT16(1);
|
|
|
|
|
|
|
|
/* Jsonb *query = PG_GETARG_JSONB(2); */
|
|
|
|
int32 nkeys = PG_GETARG_INT32(3);
|
|
|
|
|
|
|
|
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
|
|
|
|
GinTernaryValue res = GIN_MAYBE;
|
|
|
|
int32 i;
|
|
|
|
|
|
|
|
if (strategy != JsonbContainsStrategyNumber)
|
|
|
|
elog(ERROR, "unrecognized strategy number: %d", strategy);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Note that we never return GIN_TRUE, only GIN_MAYBE or GIN_FALSE; this
|
|
|
|
* corresponds to always forcing recheck in the regular consistent
|
|
|
|
* function, for the reasons listed there.
|
|
|
|
*/
|
|
|
|
for (i = 0; i < nkeys; i++)
|
|
|
|
{
|
|
|
|
if (check[i] == GIN_FALSE)
|
|
|
|
{
|
|
|
|
res = GIN_FALSE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
PG_RETURN_GIN_TERNARY_VALUE(res);
|
|
|
|
}
|
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
/*
|
2014-05-09 14:41:26 +02:00
|
|
|
* Construct a jsonb_ops GIN key from a flag byte and a textual representation
|
|
|
|
* (which need not be null-terminated). This function is responsible
|
|
|
|
* for hashing overlength text representations; it will add the
|
|
|
|
* JGINFLAG_HASHED bit to the flag value if it does that.
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
2014-05-09 14:41:26 +02:00
|
|
|
static Datum
|
|
|
|
make_text_key(char flag, const char *str, int len)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
|
|
|
text *item;
|
2014-05-09 14:41:26 +02:00
|
|
|
char hashbuf[10];
|
|
|
|
|
|
|
|
if (len > JGIN_MAXLENGTH)
|
|
|
|
{
|
|
|
|
uint32 hashval;
|
|
|
|
|
|
|
|
hashval = DatumGetUInt32(hash_any((const unsigned char *) str, len));
|
|
|
|
snprintf(hashbuf, sizeof(hashbuf), "%08x", hashval);
|
|
|
|
str = hashbuf;
|
|
|
|
len = 8;
|
|
|
|
flag |= JGINFLAG_HASHED;
|
|
|
|
}
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
/*
|
|
|
|
* Now build the text Datum. For simplicity we build a 4-byte-header
|
|
|
|
* varlena text Datum here, but we expect it will get converted to short
|
|
|
|
* header format when stored in the index.
|
|
|
|
*/
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
item = (text *) palloc(VARHDRSZ + len + 1);
|
|
|
|
SET_VARSIZE(item, VARHDRSZ + len + 1);
|
|
|
|
|
|
|
|
*VARDATA(item) = flag;
|
|
|
|
|
|
|
|
memcpy(VARDATA(item) + 1, str, len);
|
|
|
|
|
2014-05-09 14:41:26 +02:00
|
|
|
return PointerGetDatum(item);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2014-05-09 14:41:26 +02:00
|
|
|
* Create a textual representation of a JsonbValue that will serve as a GIN
|
|
|
|
* key in a jsonb_ops index. is_key is true if the JsonbValue is a key,
|
|
|
|
* or if it is a string array element (since we pretend those are keys,
|
|
|
|
* see jsonb.h).
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*/
|
2014-05-09 14:41:26 +02:00
|
|
|
static Datum
|
|
|
|
make_scalar_key(const JsonbValue *scalarVal, bool is_key)
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
{
|
2014-05-09 14:41:26 +02:00
|
|
|
Datum item;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
char *cstr;
|
|
|
|
|
|
|
|
switch (scalarVal->type)
|
|
|
|
{
|
|
|
|
case jbvNull:
|
2014-05-09 14:41:26 +02:00
|
|
|
Assert(!is_key);
|
|
|
|
item = make_text_key(JGINFLAG_NULL, "", 0);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case jbvBool:
|
2014-05-09 14:41:26 +02:00
|
|
|
Assert(!is_key);
|
|
|
|
item = make_text_key(JGINFLAG_BOOL,
|
|
|
|
scalarVal->val.boolean ? "t" : "f", 1);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
case jbvNumeric:
|
2014-05-09 14:41:26 +02:00
|
|
|
Assert(!is_key);
|
2014-05-06 18:12:18 +02:00
|
|
|
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
/*
|
2014-05-09 14:41:26 +02:00
|
|
|
* A normalized textual representation, free of trailing zeroes,
|
|
|
|
* is required so that numerically equal values will produce equal
|
|
|
|
* strings.
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
*
|
|
|
|
* It isn't ideal that numerics are stored in a relatively bulky
|
|
|
|
* textual format. However, it's a notationally convenient way of
|
|
|
|
* storing a "union" type in the GIN B-Tree, and indexing Jsonb
|
|
|
|
* strings takes precedence.
|
|
|
|
*/
|
2014-04-02 20:30:08 +02:00
|
|
|
cstr = numeric_normalize(scalarVal->val.numeric);
|
2014-05-09 14:41:26 +02:00
|
|
|
item = make_text_key(JGINFLAG_NUM, cstr, strlen(cstr));
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
pfree(cstr);
|
|
|
|
break;
|
|
|
|
case jbvString:
|
2014-05-09 14:41:26 +02:00
|
|
|
item = make_text_key(is_key ? JGINFLAG_KEY : JGINFLAG_STR,
|
|
|
|
scalarVal->val.string.val,
|
|
|
|
scalarVal->val.string.len);
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
break;
|
|
|
|
default:
|
2014-05-09 14:41:26 +02:00
|
|
|
elog(ERROR, "unrecognized jsonb scalar type: %d", scalarVal->type);
|
|
|
|
item = 0; /* keep compiler quiet */
|
|
|
|
break;
|
Introduce jsonb, a structured format for storing json.
The new format accepts exactly the same data as the json type. However, it is
stored in a format that does not require reparsing the orgiginal text in order
to process it, making it much more suitable for indexing and other operations.
Insignificant whitespace is discarded, and the order of object keys is not
preserved. Neither are duplicate object keys kept - the later value for a given
key is the only one stored.
The new type has all the functions and operators that the json type has,
with the exception of the json generation functions (to_json, json_agg etc.)
and with identical semantics. In addition, there are operator classes for
hash and btree indexing, and two classes for GIN indexing, that have no
equivalent in the json type.
This feature grew out of previous work by Oleg Bartunov and Teodor Sigaev, which
was intended to provide similar facilities to a nested hstore type, but which
in the end proved to have some significant compatibility issues.
Authors: Oleg Bartunov, Teodor Sigaev, Peter Geoghegan and Andrew Dunstan.
Review: Andres Freund
2014-03-23 21:40:19 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
return item;
|
|
|
|
}
|