Skip to content

Commit 544fec7

Browse files
funbringerfunbringer
committed
add similarity and distance operators for arrays, improve opclass' definitions, use addInfo for rum_anyarray_consistent()
1 parent 4326d8a commit 544fec7

File tree

2 files changed

+309
-9
lines changed

2 files changed

+309
-9
lines changed

‎gen_rum_sql--1.1--1.2.pl‎

Lines changed: 57 additions & 9 deletions
Original file line numberDiff line numberDiff line change
@@ -134,19 +134,67 @@
134134
AS 'MODULE_PATHNAME'
135135
LANGUAGE C IMMUTABLE STRICT;
136136
137+
138+
CREATE OR REPLACE FUNCTION rum_anyarray_similar(anyarray,anyarray)
139+
RETURNS bool
140+
AS 'MODULE_PATHNAME'
141+
LANGUAGE C STRICT STABLE;
142+
143+
CREATE OPERATOR % (
144+
PROCEDURE = rum_anyarray_similar,
145+
LEFTARG = anyarray,
146+
RIGHTARG = anyarray,
147+
COMMUTATOR = '%',
148+
RESTRICT = contsel,
149+
JOIN = contjoinsel
150+
);
151+
152+
153+
CREATE OR REPLACE FUNCTION rum_anyarray_distance(anyarray,anyarray)
154+
RETURNS float8
155+
AS 'MODULE_PATHNAME'
156+
LANGUAGE C STRICT STABLE;
157+
158+
CREATE OPERATOR <=> (
159+
PROCEDURE = rum_anyarray_distance,
160+
LEFTARG = anyarray,
161+
RIGHTARG = anyarray,
162+
COMMUTATOR = '<=>'
163+
);
164+
165+
166+
CREATE FUNCTION rum_extract_anyarray(anyarray,internal,internal,internal,internal)
167+
RETURNS internal
168+
AS 'MODULE_PATHNAME'
169+
LANGUAGE C IMMUTABLE STRICT;
170+
171+
CREATE FUNCTION rum_extract_anyarray_query(anyarray,internal,smallint,internal,internal,internal,internal)
172+
RETURNS internal
173+
AS 'MODULE_PATHNAME'
174+
LANGUAGE C IMMUTABLE STRICT;
175+
176+
CREATE FUNCTION rum_anyarray_consistent(internal, smallint, anyarray, integer, internal, internal, internal, internal)
177+
RETURNS bool
178+
AS 'MODULE_PATHNAME'
179+
LANGUAGE C IMMUTABLE STRICT;
180+
181+
182+
137183
CREATE OPERATOR CLASS rum_anyarray_ops
138184
DEFAULT FOR TYPE anyarray USING rum
139185
AS
140-
OPERATOR 1 && (anyarray, anyarray),
141-
OPERATOR 2 @> (anyarray, anyarray),
142-
OPERATOR 3 <@ (anyarray, anyarray),
143-
OPERATOR 4 = (anyarray, anyarray),
186+
OPERATOR 1 && (anyarray, anyarray),
187+
OPERATOR 2 @> (anyarray, anyarray),
188+
OPERATOR 3 <@ (anyarray, anyarray),
189+
OPERATOR 4 = (anyarray, anyarray),
190+
OPERATOR 5 % (anyarray, anyarray),
191+
OPERATOR 20 <=> (anyarray, anyarray) FOR ORDER BY pg_catalog.float_ops,
144192
--dispatch function 1 for concrete type
145-
FUNCTION 2 ginarrayextract(anyarray,internal,internal),
146-
FUNCTION 3 ginqueryarrayextract(anyarray,internal,smallint,internal,internal,internal,internal),
147-
FUNCTION 4 ginarrayconsistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
193+
FUNCTION 2 rum_extract_anyarray(anyarray,internal,internal,internal,internal),
194+
FUNCTION 3 rum_extract_anyarray_query(anyarray,internal,smallint,internal,internal,internal,internal),
195+
FUNCTION 4 rum_anyarray_consistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
148196
FUNCTION 6 rum_anyarray_config(internal),
149-
STORAGE anyelement;
197+
STORAGE anyelement;
150198
151199
CREATE OPERATOR CLASS rum_anyarray_addon_ops
152200
FOR TYPE anyarray USING rum
@@ -159,7 +207,7 @@
159207
FUNCTION 2 ginarrayextract(anyarray,internal,internal),
160208
FUNCTION 3 ginqueryarrayextract(anyarray,internal,smallint,internal,internal,internal,internal),
161209
FUNCTION 4 ginarrayconsistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
162-
STORAGE anyelement;
210+
STORAGE anyelement;
163211
164212
EOT
165213

‎src/rum_arr_utils.c‎

Lines changed: 252 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -18,15 +18,31 @@
1818
#include "miscadmin.h"
1919
#include "utils/array.h"
2020
#include "utils/builtins.h"
21+
#include "utils/lsyscache.h"
2122
#include "utils/typcache.h"
2223

2324
#include "rum.h"
2425

2526
#include <math.h>
2627

2728

29+
#define RumOverlapStrategy 1
30+
#define RumContainsStrategy 2
31+
#define RumContainedStrategy 3
32+
#define RumEqualStrategy 4
33+
#define RumSimilarStrategy 5
34+
35+
2836
PG_FUNCTION_INFO_V1(rum_anyarray_config);
2937

38+
PG_FUNCTION_INFO_V1(rum_extract_anyarray);
39+
PG_FUNCTION_INFO_V1(rum_extract_anyarray_query);
40+
41+
PG_FUNCTION_INFO_V1(rum_anyarray_consistent);
42+
43+
PG_FUNCTION_INFO_V1(rum_anyarray_similar);
44+
PG_FUNCTION_INFO_V1(rum_anyarray_distance);
45+
3046

3147
/*
3248
* Specifies additional information type for operator class.
@@ -41,3 +57,239 @@ rum_anyarray_config(PG_FUNCTION_ARGS)
4157

4258
PG_RETURN_VOID();
4359
}
60+
61+
62+
/* ginarrayextract() */
63+
Datum
64+
rum_extract_anyarray(PG_FUNCTION_ARGS)
65+
{
66+
/* Make copy of array input to ensure it doesn't disappear while in use */
67+
ArrayType *array = PG_GETARG_ARRAYTYPE_P_COPY(0);
68+
69+
Datum *entries;
70+
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
71+
bool **entries_isnull = (bool **) PG_GETARG_POINTER(2);
72+
73+
Datum **addInfo = (Datum **) PG_GETARG_POINTER(3);
74+
bool **addInfoIsNull = (bool **) PG_GETARG_POINTER(4);
75+
76+
int16 elmlen;
77+
bool elmbyval;
78+
char elmalign;
79+
int i;
80+
81+
get_typlenbyvalalign(ARR_ELEMTYPE(array),
82+
&elmlen, &elmbyval, &elmalign);
83+
84+
deconstruct_array(array,
85+
ARR_ELEMTYPE(array),
86+
elmlen, elmbyval, elmalign,
87+
&entries, entries_isnull, nentries);
88+
89+
*addInfo = (Datum *) palloc(*nentries * sizeof(Datum));
90+
*addInfoIsNull = (bool *) palloc(*nentries * sizeof(bool));
91+
92+
for (i = 0; i < *nentries; i++)
93+
{
94+
/* Use array's size as additional info */
95+
(*addInfo)[i] = Int32GetDatum(*nentries);
96+
(*addInfoIsNull)[i] = BoolGetDatum(false);
97+
}
98+
99+
/* we should not free array, entries[i] points into it */
100+
PG_RETURN_POINTER(entries);
101+
}
102+
103+
/* ginqueryarrayextract() */
104+
Datum
105+
rum_extract_anyarray_query(PG_FUNCTION_ARGS)
106+
{
107+
/* Make copy of array input to ensure it doesn't disappear while in use */
108+
ArrayType *array = PG_GETARG_ARRAYTYPE_P_COPY(0);
109+
110+
Datum *entries;
111+
int32 *nentries = (int32 *) PG_GETARG_POINTER(1);
112+
bool **entries_isnull = (bool **) PG_GETARG_POINTER(5);
113+
114+
StrategyNumber strategy = PG_GETARG_UINT16(2);
115+
116+
/* bool **pmatch = (bool **) PG_GETARG_POINTER(3); */
117+
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
118+
119+
int32 *searchMode = (int32 *) PG_GETARG_POINTER(6);
120+
121+
int16 elmlen;
122+
bool elmbyval;
123+
char elmalign;
124+
125+
get_typlenbyvalalign(ARR_ELEMTYPE(array),
126+
&elmlen, &elmbyval, &elmalign);
127+
128+
deconstruct_array(array,
129+
ARR_ELEMTYPE(array),
130+
elmlen, elmbyval, elmalign,
131+
&entries, entries_isnull, nentries);
132+
133+
switch (strategy)
134+
{
135+
case RumOverlapStrategy:
136+
*searchMode = GIN_SEARCH_MODE_DEFAULT;
137+
break;
138+
case RumContainsStrategy:
139+
if (*nentries > 0)
140+
*searchMode = GIN_SEARCH_MODE_DEFAULT;
141+
else /* everything contains the empty set */
142+
*searchMode = GIN_SEARCH_MODE_ALL;
143+
break;
144+
case RumContainedStrategy:
145+
/* empty set is contained in everything */
146+
*searchMode = GIN_SEARCH_MODE_INCLUDE_EMPTY;
147+
break;
148+
case RumEqualStrategy:
149+
if (*nentries > 0)
150+
*searchMode = GIN_SEARCH_MODE_DEFAULT;
151+
else
152+
*searchMode = GIN_SEARCH_MODE_INCLUDE_EMPTY;
153+
break;
154+
case RumSimilarStrategy:
155+
if (*nentries > 0)
156+
*searchMode = GIN_SEARCH_MODE_DEFAULT;
157+
else
158+
*searchMode = GIN_SEARCH_MODE_INCLUDE_EMPTY;
159+
break;
160+
default:
161+
elog(ERROR, "rum_extract_anyarray_query: unknown strategy number: %d",
162+
strategy);
163+
}
164+
165+
/* we should not free array, elems[i] points into it */
166+
PG_RETURN_POINTER(entries);
167+
}
168+
169+
170+
171+
/* ginarrayconsistent() */
172+
Datum
173+
rum_anyarray_consistent(PG_FUNCTION_ARGS)
174+
{
175+
bool *check = (bool *) PG_GETARG_POINTER(0);
176+
177+
StrategyNumber strategy = PG_GETARG_UINT16(1);
178+
179+
/* ArrayType *query = PG_GETARG_ARRAYTYPE_P(2); */
180+
int32 nkeys = PG_GETARG_INT32(3);
181+
182+
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
183+
bool *recheck = (bool *) PG_GETARG_POINTER(5);
184+
185+
/* Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6); */
186+
bool *nullFlags = (bool *) PG_GETARG_POINTER(7);
187+
188+
Datum *addInfo = (Datum *) PG_GETARG_POINTER(8);
189+
bool *addInfoIsNull = (bool *) PG_GETARG_POINTER(9);
190+
191+
bool res;
192+
int32 i;
193+
194+
switch (strategy)
195+
{
196+
case RumOverlapStrategy:
197+
/* result is not lossy */
198+
*recheck = false;
199+
/* must have a match for at least one non-null element */
200+
res = false;
201+
for (i = 0; i < nkeys; i++)
202+
{
203+
if (check[i] && !nullFlags[i])
204+
{
205+
res = true;
206+
break;
207+
}
208+
}
209+
break;
210+
case RumContainsStrategy:
211+
/* result is not lossy */
212+
*recheck = false;
213+
214+
/* must have all elements in check[] true, and no nulls */
215+
res = true;
216+
for (i = 0; i < nkeys; i++)
217+
{
218+
if (!check[i] || nullFlags[i])
219+
{
220+
res = false;
221+
break;
222+
}
223+
}
224+
break;
225+
case RumContainedStrategy:
226+
/* we will need recheck */
227+
*recheck = true;
228+
229+
/* query must have <= amount of elements than array */
230+
res = true;
231+
for (i = 0; i < nkeys; i++)
232+
{
233+
if (!addInfoIsNull[i] && DatumGetInt32(addInfo[i]) > nkeys)
234+
{
235+
res = false;
236+
break;
237+
}
238+
}
239+
break;
240+
case RumEqualStrategy:
241+
/* we will need recheck */
242+
*recheck = true;
243+
244+
/*
245+
* Must have all elements in check[] true; no discrimination
246+
* against nulls here. This is because array_contain_compare and
247+
* array_eq handle nulls differently ...
248+
*
249+
* Also, query and array must have equal amount of elements.
250+
*/
251+
res = true;
252+
for (i = 0; i < nkeys; i++)
253+
{
254+
if (!check[i])
255+
{
256+
res = false;
257+
break;
258+
}
259+
260+
if (!addInfoIsNull[i] && DatumGetInt32(addInfo[i]) != nkeys)
261+
{
262+
res = false;
263+
break;
264+
}
265+
}
266+
break;
267+
case RumSimilarStrategy:
268+
/* we will need recheck */
269+
*recheck = true;
270+
271+
/* can't do anything else useful here */
272+
res = true;
273+
break;
274+
default:
275+
elog(ERROR, "rum_anyarray_consistent: unknown strategy number: %d",
276+
strategy);
277+
res = false;
278+
}
279+
280+
PG_RETURN_BOOL(res);
281+
}
282+
283+
284+
285+
Datum
286+
rum_anyarray_similar(PG_FUNCTION_ARGS)
287+
{
288+
PG_RETURN_BOOL(true);
289+
}
290+
291+
Datum
292+
rum_anyarray_distance(PG_FUNCTION_ARGS)
293+
{
294+
PG_RETURN_FLOAT8(0.0);
295+
}

0 commit comments

Comments
 (0)