add GUCs for array similarity & function

funbringer · funbringer · commit 9e873ece04f3 · 2017-10-18T15:56:35.000+03:00
diff --git a/src/rum.h b/src/rum.h
@@ -801,9 +801,31 @@ extern Datum rum_ts_distance_td(PG_FUNCTION_ARGS);
 
 extern Datum tsquery_to_distance_query(PG_FUNCTION_ARGS);
 
+/* rum_arr_utils.c */
+typedef enum SimilarityType
+{
+	SMT_COSINE  = 1,
+	SMT_JACCARD = 2,
+	SMT_OVERLAP = 3
+} SimilarityType;
+
+#define RUM_SIMILARITY_FUNCTION_DEFAULT		SMT_COSINE
+#define RUM_SIMILARITY_THRESHOLD_DEFAULT	0.5
+
+extern Datum rum_anyarray_config(PG_FUNCTION_ARGS);
+extern Datum rum_extract_anyarray(PG_FUNCTION_ARGS);
+extern Datum rum_extract_anyarray_query(PG_FUNCTION_ARGS);
+extern Datum rum_anyarray_consistent(PG_FUNCTION_ARGS);
+extern Datum rum_anyarray_ordering(PG_FUNCTION_ARGS);
+extern Datum rum_anyarray_similar(PG_FUNCTION_ARGS);
+extern Datum rum_anyarray_distance(PG_FUNCTION_ARGS);
+
 
 /* GUC parameters */
-extern PGDLLIMPORT int RumFuzzySearchLimit;
+extern PGDLLIMPORT int		RumFuzzySearchLimit;
+extern PGDLLIMPORT float8	RumArraySimilarityThreshold;
+extern PGDLLIMPORT int		RumArraySimilarityFunction;
+
 
 /*
  * Functions for reading ItemPointers with additional information. Used in
diff --git a/src/rum_arr_utils.c b/src/rum_arr_utils.c
@@ -97,12 +97,9 @@ typedef struct SimpleArray
 	AnyArrayTypeInfo   *info;
 } SimpleArray;
 
-typedef enum SimilarityType
-{
-	AA_Cosine,
-	AA_Jaccard,
-	AA_Overlap
-} SimilarityType;
+
+float8	RumArraySimilarityThreshold	= RUM_SIMILARITY_THRESHOLD_DEFAULT;
+int		RumArraySimilarityFunction	= RUM_SIMILARITY_FUNCTION_DEFAULT;
 
 
 PG_FUNCTION_INFO_V1(rum_anyarray_config);
@@ -117,10 +114,6 @@ PG_FUNCTION_INFO_V1(rum_anyarray_similar);
 PG_FUNCTION_INFO_V1(rum_anyarray_distance);
 
 
-static SimilarityType	SmlType = AA_Cosine;
-static float8			SmlLimit = 0.5;
-
-
 static Oid getAMProc(Oid amOid, Oid typid);
 
 static AnyArrayTypeInfo *getAnyArrayTypeInfo(MemoryContext ctx, Oid typid);
@@ -139,7 +132,6 @@ static int32 getNumOfIntersect(SimpleArray *sa, SimpleArray *sb);
 static float8 getSimilarity(SimpleArray *sa, SimpleArray *sb, int32 intersection);
 
 
-
 /*
  * Specifies additional information type for operator class.
  */
@@ -390,7 +382,8 @@ rum_anyarray_consistent(PG_FUNCTION_ARGS)
 
 					INIT_DUMMY_SIMPLE_ARRAY(&sa, nentries);
 					INIT_DUMMY_SIMPLE_ARRAY(&sb, nkeys);
-					res = getSimilarity(&sa, &sb, intersection) >= SmlLimit;
+					res = getSimilarity(&sa, &sb, intersection) >=
+								RumArraySimilarityThreshold;
 				}
 				else
 					res = false;
@@ -491,7 +484,7 @@ rum_anyarray_similar(PG_FUNCTION_ARGS)
 	PG_FREE_IF_COPY(b, 1);
 	PG_FREE_IF_COPY(a, 0);
 
-	PG_RETURN_BOOL(result >= SmlLimit);
+	PG_RETURN_BOOL(result >= RumArraySimilarityThreshold);
 }
 
 Datum
@@ -851,19 +844,19 @@ getSimilarity(SimpleArray *sa, SimpleArray *sb, int32 intersection)
 {
 	float8 result = 0.0;
 
-	switch (SmlType)
+	switch (RumArraySimilarityFunction)
 	{
-		case AA_Cosine:
+		case SMT_COSINE:
 			result = ((float8) intersection) /
 						sqrt(((float8) sa->nelems) * ((float8) sb->nelems));
 			break;
-		case AA_Jaccard:
+		case SMT_JACCARD:
 			result = ((float8) intersection) /
 						(((float8) sa->nelems) +
 						 ((float8) sb->nelems) -
 						 ((float8) intersection));
 			break;
-		case AA_Overlap:
+		case SMT_OVERLAP:
 			result = intersection;
 			break;
 		default:
diff --git a/src/rumutil.c b/src/rumutil.c
@@ -36,6 +36,14 @@ PG_FUNCTION_INFO_V1(rumhandler);
 /* Kind of relation optioms for rum index */
 static relopt_kind rum_relopt_kind;
 
+static const struct config_enum_entry rum_array_similarity_function_opts[] =
+{
+	{ "cosine",		SMT_COSINE,		false },
+	{ "jaccard",	SMT_JACCARD,	false },
+	{ "overlap",	SMT_OVERLAP,	false },
+	{ NULL,			0,				false }
+};
+
 /*
  * Module load callback
  */
@@ -51,6 +59,23 @@ _PG_init(void)
 							PGC_USERSET, 0,
 							NULL, NULL, NULL);
 
+	DefineCustomRealVariable("rum_array_similarity_threshold",
+							 "Sets the array similarity threshold.",
+							 NULL,
+							 &RumArraySimilarityThreshold,
+							 RUM_SIMILARITY_THRESHOLD_DEFAULT, 0.0, 1.0,
+							 PGC_USERSET, 0,
+							 NULL, NULL, NULL);
+
+	DefineCustomEnumVariable("rum_array_similarity_function",
+							 "Sets the array similarity function.",
+							 NULL,
+							 &RumArraySimilarityFunction,
+							 RUM_SIMILARITY_FUNCTION_DEFAULT,
+							 rum_array_similarity_function_opts,
+							 PGC_USERSET, 0,
+							 NULL, NULL, NULL);
+
 	rum_relopt_kind = add_reloption_kind();
 
 	add_string_reloption(rum_relopt_kind, "attach",
