Skip to content

Commit b7117a3

Browse files
author
Vladlen Popolitov
committed
preConsistent function for int8 implemented
1 parent c047a94 commit b7117a3

File tree

7 files changed

+225
-57
lines changed

7 files changed

+225
-57
lines changed

‎Makefile‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -8,7 +8,7 @@ EXTENSION = anyarray
88
DATA = anyarray--1.0.sql anyarray--1.2.sql
99
PGFILEDESC = "anyarray - functions and operators for one-dimensional arrays"
1010

11-
REGRESS = init anyarray \
11+
REGRESS = security init anyarray \
1212
int2 int4 int8 float4 float8 numeric \
1313
text varchar char varbit bit bytea \
1414
interval money oid \

‎anyarray--1.2.sql‎

Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1025,6 +1025,11 @@ RETURNS bool
10251025
AS 'MODULE_PATHNAME'
10261026
LANGUAGE C IMMUTABLE STRICT;
10271027

1028+
CREATE FUNCTION rumanyarray_preconsistent(internal, smallint, anyarray, integer, internal, internal, internal, internal)
1029+
RETURNS bool
1030+
AS 'MODULE_PATHNAME'
1031+
LANGUAGE C IMMUTABLE STRICT;
1032+
10281033
CREATE FUNCTION rumanyarray_ordering(internal,smallint,anyarray,int,internal,internal,internal,internal,internal)
10291034
RETURNS float8
10301035
AS 'MODULE_PATHNAME'
@@ -1114,5 +1119,25 @@ AS
11141119
FUNCTION 3 rumextract_anyarray_query(anyarray,internal,smallint,internal,internal,internal,internal),
11151120
FUNCTION 4 rumanyarray_consistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
11161121
FUNCTION 6 rumanyarray_config(internal),
1122+
--FUNCTION 7 rumanyarray_preconsistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
1123+
FUNCTION 8 rumanyarray_ordering(internal,smallint,anyarray,int,internal,internal,internal,internal,internal),
1124+
STORAGE int8;
1125+
1126+
CREATE OPERATOR CLASS _int8_aa_ops_beta
1127+
FOR TYPE _int8 USING rum
1128+
AS
1129+
OPERATOR 1 && (anyarray, anyarray),
1130+
OPERATOR 2 @> (anyarray, anyarray),
1131+
OPERATOR 3 <@ (anyarray, anyarray),
1132+
OPERATOR 4 = (anyarray, anyarray),
1133+
OPERATOR 5 <%%> (anyarray, anyarray),
1134+
OPERATOR 20 <==> (anyarray, anyarray) FOR ORDER BY pg_catalog.float_ops,
1135+
--dispatch function 1 for concrete type
1136+
FUNCTION 1 btint8cmp(int8,int8),
1137+
FUNCTION 2 rumextract_anyarray(anyarray,internal,internal,internal,internal),
1138+
FUNCTION 3 rumextract_anyarray_query(anyarray,internal,smallint,internal,internal,internal,internal),
1139+
FUNCTION 4 rumanyarray_consistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
1140+
FUNCTION 6 rumanyarray_config(internal),
1141+
FUNCTION 7 rumanyarray_preconsistent(internal,smallint,anyarray,integer,internal,internal,internal,internal),
11171142
FUNCTION 8 rumanyarray_ordering(internal,smallint,anyarray,int,internal,internal,internal,internal,internal),
11181143
STORAGE int8;

‎anyarray_rum.c‎

Lines changed: 169 additions & 37 deletions
Original file line numberDiff line numberDiff line change
@@ -40,10 +40,27 @@ RUM support functions
4040
Datum *rum_extract_value(Datum itemValue, int32 *nkeys, bool **nullFlags, Datum **addInfo, bool **nullFlagsAddInfo)
4141
Datum *rum_extract_query(Datum query, int32 *nkeys, StrategyNumber n, bool **pmatch, Pointer **extra_data,
4242
bool **nullFlags, int32 *searchMode)
43-
bool consistent(bool check[], StrategyNumber n, Datum query, int32 nkeys, Pointer extra_data[],
43+
bool rum_consistent(bool check[], StrategyNumber n, Datum query, int32 nkeys, Pointer extra_data[],
4444
bool *recheck, Datum queryKeys[], bool nullFlags[], Datum **addInfo, bool **nullFlagsAddInfo)
4545
int32 rum_compare_prefix(Datum a, Datum b,StrategyNumber n,void *addInfo)
4646
void rum_config(RumConfig *config)
47+
48+
bool rum_pre_consistent(bool check[], StrategyNumber n, Datum query, int32 nkeys, Pointer extra_data[],
49+
bool *recheck, Datum queryKeys[], bool nullFlags[])
50+
Return value:
51+
false if index key value is not consistent with query
52+
true if key value MAYBE consistent and will be rechecked in rum_consistent function
53+
Parameters:
54+
check[] contains false, if element is not match any query element.
55+
if all elements in check[] equal to true, this function is not called
56+
if all elements in check[] equal to false, this function is not called
57+
StrategyNumber - the number of the strategy from the class operator
58+
query
59+
nkeys quantity of elements in check[]. If nkeys==0, this function is not called
60+
recheck - parameter is not used
61+
queryKeys[] - returned by rum_extract_query
62+
nullFlags[] - returned by rum_extract_query (or all false, if rum_extract_query did not return nullFlags)
63+
4764
double rum_ordering_distance(bool check[], ?StrategyNumber n, ?Datum query, int32 nkeys, ?Pointer extra_data[],
4865
?bool *recheck, ?Datum queryKeys[], ?bool nullFlags[], Datum **addInfo, bool **nullFlagsAddInfo)
4966
Datum rum_join_pos(Datum, Datum)
@@ -228,39 +245,45 @@ rumanyarray_consistent(PG_FUNCTION_ARGS)
228245
bool *recheck = (bool *) PG_GETARG_POINTER(5);
229246

230247
/* Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6); */
231-
bool *nullFlags = (bool *) PG_GETARG_POINTER(7);
248+
249+
/* rumextract_anyarray_query does not return nullFlags, so RUM initializes all them by false */
250+
/* bool *nullFlags = (bool *) PG_GETARG_POINTER(7); */
232251

233252
Datum *addInfo = (Datum *) PG_GETARG_POINTER(8);
234-
bool *addInfoIsNull = (bool *) PG_GETARG_POINTER(9);
253+
/* rumextract_anyarray_query initializes all addInfoIsNull elements by false */
254+
/* bool *addInfoIsNull = (bool *) PG_GETARG_POINTER(9); */
235255

236256
bool res;
237257
int32 i;
238258

239259
switch (strategy)
240260
{
241261
case RUM_OVERLAP_STRATEGY:
262+
/* at least one element in check[] is true, so result = true */
263+
*recheck = false;
264+
res = true;
242265
/* result is not lossy */
243-
*recheck = false;
266+
/* *recheck = false; */
244267
/* must have a match for at least one non-null element */
245-
res = false;
268+
/* res = false;
246269
for (i = 0; i < nkeys; i++)
247270
{
248271
if (check[i] && !nullFlags[i])
249272
{
250273
res = true;
251274
break;
252275
}
253-
}
276+
}*/
254277
break;
255278
case RUM_CONTAINS_STRATEGY:
256-
/* result is not lossy */
279+
/* result is not lossy */
257280
*recheck = false;
258281

259282
/* must have all elements in check[] true, and no nulls */
260283
res = true;
261284
for (i = 0; i < nkeys; i++)
262285
{
263-
if (!check[i] || nullFlags[i])
286+
if (!check[i] /* || nullFlags[i] */)
264287
{
265288
res = false;
266289
break;
@@ -272,19 +295,28 @@ rumanyarray_consistent(PG_FUNCTION_ARGS)
272295
*recheck = true;
273296

274297
/* query must have <= amount of elements than array */
298+
if(nkeys > 0 )
299+
{
300+
res = ( DatumGetInt32(addInfo[0]) <= nkeys);
301+
} else {
302+
/* empty arrays in query and index */
303+
*recheck = false;
304+
res = true;
305+
}
306+
/*
275307
res = true;
276-
for (i = 0; i < nkeys; i++)
308+
for (i = 0; i < nkeys; i++)
277309
{
278-
if (!addInfoIsNull[i] && DatumGetInt32(addInfo[i]) > nkeys)
310+
if ( !addInfoIsNull[i] && DatumGetInt32(addInfo[i]) > nkeys)
279311
{
280312
res = false;
281313
break;
282314
}
283-
}
315+
} */
284316
break;
285317
case RUM_EQUAL_STRATEGY:
286318
/* we will need recheck */
287-
*recheck = true;
319+
/* *recheck = true; */
288320

289321
/*
290322
* Must have all elements in check[] true; no discrimination
@@ -293,6 +325,25 @@ rumanyarray_consistent(PG_FUNCTION_ARGS)
293325
*
294326
* Also, query and array must have equal amount of elements.
295327
*/
328+
*recheck = true;
329+
res = true;
330+
if(nkeys > 0)
331+
{
332+
if (DatumGetInt32(addInfo[0]) != nkeys)
333+
{
334+
res = false;
335+
} else {
336+
for (i = 0; i < nkeys; i++)
337+
{
338+
if (!check[i])
339+
{
340+
res = false;
341+
break;
342+
}
343+
}
344+
}
345+
}
346+
/*
296347
res = true;
297348
for (i = 0; i < nkeys; i++)
298349
{
@@ -307,7 +358,7 @@ rumanyarray_consistent(PG_FUNCTION_ARGS)
307358
res = false;
308359
break;
309360
}
310-
}
361+
}*/
311362
break;
312363
case RUM_SIMILAR_STRATEGY:
313364
/* we won't need recheck */
@@ -317,48 +368,129 @@ rumanyarray_consistent(PG_FUNCTION_ARGS)
317368
int32 intersection = 0,
318369
nentries = -1;
319370
SimpleArray sa, sb;
371+
float8 sml;
320372

321373
for (i = 0; i < nkeys; i++)
322374
if (check[i])
323375
intersection++;
376+
if(intersection > 0 )
377+
{
378+
/* extract array's length from addInfo */
379+
nentries = DatumGetInt32(addInfo[0]);
380+
/*for (i = 0; i < nkeys; i++)
381+
{
382+
if (!addInfoIsNull[i])
383+
{
384+
nentries = DatumGetInt32(addInfo[i]);
385+
break;
386+
}
387+
} */
388+
389+
/* there must be addInfo */
390+
Assert(nentries >= 0);
391+
392+
INIT_DUMMY_SIMPLE_ARRAY(&sa, nentries);
393+
INIT_DUMMY_SIMPLE_ARRAY(&sb, nkeys);
394+
sml = getSimilarityValue(&sa, &sb, intersection);
395+
}
396+
else
397+
{
398+
sml = 0.0 ;
399+
}
400+
res = (sml >= SmlLimit);
401+
}
402+
break;
403+
default:
404+
elog(ERROR, "rum_anyarray_consistent: unknown strategy number: %d",
405+
strategy);
406+
res = false;
407+
}
324408

325-
if (intersection > 0)
326-
{
327-
float8 sml;
409+
PG_RETURN_BOOL(res);
410+
}
328411

329-
/* extract array's length from addInfo */
330-
for (i = 0; i < nkeys; i++)
331-
{
332-
if (!addInfoIsNull[i])
333-
{
334-
nentries = DatumGetInt32(addInfo[i]);
335-
break;
336-
}
337-
}
412+
PG_FUNCTION_INFO_V1(rumanyarray_preconsistent);
413+
Datum
414+
rumanyarray_preconsistent(PG_FUNCTION_ARGS)
415+
{
416+
bool *check = (bool *) PG_GETARG_POINTER(0);
417+
StrategyNumber strategy = PG_GETARG_UINT16(1);
418+
/* ArrayType *query = PG_GETARG_ARRAYTYPE_P(2); */
419+
int32 nkeys = PG_GETARG_INT32(3);
420+
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
421+
/* bool *recheck = (bool *) PG_GETARG_POINTER(5); */
422+
/* Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6); */
423+
/* bool *nullFlags = (bool *) PG_GETARG_POINTER(7); */
338424

339-
/* there must be addInfo */
340-
Assert(nentries >= 0);
425+
bool res;
426+
int32 i;
341427

342-
INIT_DUMMY_SIMPLE_ARRAY(&sa, nentries);
343-
INIT_DUMMY_SIMPLE_ARRAY(&sb, nkeys);
344-
sml = getSimilarityValue(&sa, &sb, intersection);
428+
switch (strategy)
429+
{
430+
case RUM_OVERLAP_STRATEGY:
431+
/*
432+
* at least one check[i]==true
433+
* preConsistent function is not called, if all check[i]==false
434+
*/
435+
res = true;
436+
break;
437+
case RUM_CONTAINS_STRATEGY:
438+
/*
439+
* at least one check[i]==false
440+
* preConsistent function is not called, if all check[i]==true
441+
*/
442+
res = false;
443+
break;
444+
case RUM_CONTAINED_STRATEGY:
445+
/* we will need recheck */
446+
res = true;
447+
break;
448+
case RUM_EQUAL_STRATEGY:
449+
/*
450+
* at least one check[i]==false
451+
* preConsistent function is not called, if all check[i]==true
452+
*/
453+
res = false;
454+
break;
455+
case RUM_SIMILAR_STRATEGY:
456+
{
457+
int32 intersection = 0;
458+
459+
for (i = 0; i < nkeys; i++)
460+
if (check[i])
461+
intersection++;
345462

346-
res = (sml >= SmlLimit);
463+
switch(SmlType)
464+
{
465+
case AA_Cosine:
466+
/* intersection / sqrt(nkeys * intersection) */
467+
res = (sqrt(((double)intersection) / (double)nkeys) >= SmlLimit);
468+
break;
469+
case AA_Jaccard:
470+
res = ((((double)intersection) / (double)nkeys) >= SmlLimit);
471+
break;
472+
case AA_Overlap:
473+
/*
474+
* if intersection >= SmlLimit, so result = true
475+
* if intersection < SmlLimit, so result = false
476+
*/
477+
res = (((double)intersection) >= SmlLimit);
478+
break;
479+
default:
480+
elog(ERROR, "unknown similarity type");
347481
}
348-
else
349-
res = false;
350482
}
351483
break;
352484
default:
353-
elog(ERROR, "rum_anyarray_consistent: unknown strategy number: %d",
354-
strategy);
485+
elog(ERROR, "rum_anyarray_preconsistent: unknown strategy number: %d", strategy);
355486
res = false;
356487
}
357488

358489
PG_RETURN_BOOL(res);
359490
}
360491

361492

493+
362494
PG_FUNCTION_INFO_V1(rumanyarray_ordering);
363495
/*
364496
* Similarity and distance
@@ -689,10 +821,10 @@ rumanyarray_consistent_with_position(PG_FUNCTION_ARGS)
689821
/* ArrayType *query = PG_GETARG_ARRAYTYPE_P(2); */
690822
int32 nkeys = PG_GETARG_INT32(3);
691823

692-
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */ Datum *extra_data = (Datum *) PG_GETARG_POINTER(4);
824+
/* Pointer *extra_data = (Pointer *) PG_GETARG_POINTER(4); */
693825
bool *recheck = (bool *) PG_GETARG_POINTER(5);
694826

695-
/* Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6); */ Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6);
827+
/* Datum *queryKeys = (Datum *) PG_GETARG_POINTER(6); */
696828
bool *nullFlags = (bool *) PG_GETARG_POINTER(7);
697829

698830
Datum *addInfo = (Datum *) PG_GETARG_POINTER(8);

0 commit comments

Comments
 (0)