Added new function to hash any PostgreSQL data type

hll--1.0.sql

@@ -319,6 +319,13 @@ CREATE FUNCTION hll_hash_text(text, integer default 0)
      AS 'MODULE_PATHNAME', 'hll_hash_varlena'
      LANGUAGE C STRICT IMMUTABLE;
 
+-- Hash any scalar data type.
+--
+CREATE FUNCTION hll_hash_any(anyelement, integer default 0)
+     RETURNS hll_hashval
+     AS 'MODULE_PATHNAME', 'hll_hash_any'
+     LANGUAGE C STRICT IMMUTABLE;
+
 
 -- ----------------------------------------------------------------
 -- Operators

hll.c

@@ -30,6 +30,7 @@
 #include "utils/array.h"
 #include "utils/bytea.h"
 #include "utils/int8.h"
+#include "utils/lsyscache.h"
 #include "utils/memutils.h"
 #include "catalog/pg_type.h"
 
@@ -2571,6 +2572,70 @@ hll_hash_varlena(PG_FUNCTION_ARGS)
     PG_RETURN_INT64(out[0]);
 }
 
+
+// Hash any scalar data type.
+//
+PG_FUNCTION_INFO_V1(hll_hash_any);
+Datum		hll_hash_any(PG_FUNCTION_ARGS);
+Datum
+hll_hash_any(PG_FUNCTION_ARGS)
+{
+    Datum keyDatum = PG_GETARG_DATUM(0);
+    Datum seedDatum = PG_GETARG_DATUM(1);
+    Datum hashResultDatum = 0;
+
+    Oid keyTypeId = get_fn_expr_argtype(fcinfo->flinfo, 0);
+    int16 keyTypeLength = get_typlen(keyTypeId);
+
+    /* dispatch to corresponding hash function for key type */
+    switch (keyTypeLength)
+    {
+    case 1:
+        hashResultDatum = DirectFunctionCall2(hll_hash_1byte, keyDatum, seedDatum);
+        break;
+
+    case 2:
+        hashResultDatum = DirectFunctionCall2(hll_hash_2byte, keyDatum, seedDatum);
+        break;
+
+    case 4:
+        hashResultDatum = DirectFunctionCall2(hll_hash_4byte, keyDatum, seedDatum);
+        break;
+
+    case 8:
+        hashResultDatum = DirectFunctionCall2(hll_hash_8byte, keyDatum, seedDatum);
+        break;
+
+    case -1:
+    case -2:
+        hashResultDatum = DirectFunctionCall2(hll_hash_varlena, keyDatum, seedDatum);
+        break;
+
+    default:
+    {
+        /*
+         * We have a fixed-size type such as char(10), macaddr, circle, etc. We
+         * first convert this type to its variable-length binary representation
+         * and then dispatch to the variable-length hashing function.
+         */
+        Oid keyTypeSendFunction = InvalidOid;
+        bool keyTypeVarlena = false;
+        Datum keyBinaryDatum = 0;
+
+        /* no need to worry about SPI for these types' output functions */
+        getTypeBinaryOutputInfo(keyTypeId, &keyTypeSendFunction, &keyTypeVarlena);
+        keyBinaryDatum = OidFunctionCall1(keyTypeSendFunction, keyDatum);
+
+        hashResultDatum = DirectFunctionCall2(hll_hash_varlena, keyBinaryDatum,
+                                              seedDatum);
+        break;
+    }
+    }
+
+    PG_RETURN_INT64(hashResultDatum);
+}
+
+
 PG_FUNCTION_INFO_V1(hll_eq);
 Datum		hll_eq(PG_FUNCTION_ARGS);
 Datum

regress/Makefile

@@ -30,6 +30,7 @@ SQL =	\
 		typmod.sql \
 		typmod_insert.sql \
 		hash.sql \
+		hash_any.sql \
 		murmur_bigint.sql \
 		murmur_bytea.sql \
 		equal.sql \

regress/hash_any.ref

@@ -0,0 +1,175 @@
+-- ----------------------------------------------------------------
+-- Misc Tests on hash_any Function
+-- ----------------------------------------------------------------
+SELECT hll_set_output_version(1);
+ hll_set_output_version 
+------------------------
+                      1
+(1 row)
+
+-- ---------------- Check hash and hash_any function results match
+SELECT hll_hash_boolean(FALSE) = hll_hash_any(FALSE);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_boolean(TRUE) = hll_hash_any(TRUE);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_smallint(0::smallint) = hll_hash_any(0::smallint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_smallint(100::smallint) = hll_hash_any(100::smallint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_smallint(-100::smallint) = hll_hash_any(-100::smallint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_integer(0) = hll_hash_any(0);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_integer(100) = hll_hash_any(100);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_integer(-100) = hll_hash_any(-100);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bigint(0) = hll_hash_any(0::bigint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bigint(100) = hll_hash_any(100::bigint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bigint(-100) = hll_hash_any(-100::bigint);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bytea(E'\\x') = hll_hash_any(E'\\x'::bytea);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bytea(E'\\x41') = hll_hash_any(E'\\x41'::bytea);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bytea(E'\\x42') = hll_hash_any(E'\\x42'::bytea);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_bytea(E'\\x4142') = hll_hash_any(E'\\x4142'::bytea);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_text('') = hll_hash_any(''::text);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_text('A') = hll_hash_any('A'::text);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_text('B') = hll_hash_any('B'::text);
+ ?column? 
+----------
+ t
+(1 row)
+
+SELECT hll_hash_text('AB') = hll_hash_any('AB'::text);
+ ?column? 
+----------
+ t
+(1 row)
+
+-- ---------------- Check several types not handled by default hash functions
+-- ---------------- macaddr
+SELECT hll_hash_any('08:00:2b:01:02:03'::macaddr);
+     hll_hash_any     
+----------------------
+ -4883882473551067169
+(1 row)
+
+SELECT hll_hash_any('08002b010203'::macaddr);
+     hll_hash_any     
+----------------------
+ -4883882473551067169
+(1 row)
+
+SELECT hll_hash_any('01-23-45-67-89-ab'::macaddr);
+    hll_hash_any     
+---------------------
+ 3974616115244794976
+(1 row)
+
+SELECT hll_hash_any('012345-6789ab'::macaddr);
+    hll_hash_any     
+---------------------
+ 3974616115244794976
+(1 row)
+
+-- ---------------- interval
+SELECT hll_hash_any('1 year 2 months 3 days 4 hours 5 minutes 6seconds'::interval);
+    hll_hash_any     
+---------------------
+ 1647734813508782007
+(1 row)
+
+SELECT hll_hash_any('P1Y2M3DT4H5M6S'::interval);
+    hll_hash_any     
+---------------------
+ 1647734813508782007
+(1 row)
+
+SELECT hll_hash_any('1997-06 20 12:00:00'::interval);
+    hll_hash_any     
+---------------------
+ 3706410791461549552
+(1 row)
+
+SELECT hll_hash_any('P1997-06-20T12:00:00'::interval);
+    hll_hash_any     
+---------------------
+ 3706410791461549552
+(1 row)
+

regress/hash_any.sql

@@ -0,0 +1,48 @@
+-- ----------------------------------------------------------------
+-- Misc Tests on hash_any Function
+-- ----------------------------------------------------------------
+
+SELECT hll_set_output_version(1);
+
+-- ---------------- Check hash and hash_any function results match
+
+SELECT hll_hash_boolean(FALSE) = hll_hash_any(FALSE);
+SELECT hll_hash_boolean(TRUE) = hll_hash_any(TRUE);
+
+SELECT hll_hash_smallint(0::smallint) = hll_hash_any(0::smallint);
+SELECT hll_hash_smallint(100::smallint) = hll_hash_any(100::smallint);
+SELECT hll_hash_smallint(-100::smallint) = hll_hash_any(-100::smallint);
+
+SELECT hll_hash_integer(0) = hll_hash_any(0);
+SELECT hll_hash_integer(100) = hll_hash_any(100);
+SELECT hll_hash_integer(-100) = hll_hash_any(-100);
+
+SELECT hll_hash_bigint(0) = hll_hash_any(0::bigint);
+SELECT hll_hash_bigint(100) = hll_hash_any(100::bigint);
+SELECT hll_hash_bigint(-100) = hll_hash_any(-100::bigint);
+
+SELECT hll_hash_bytea(E'\\x') = hll_hash_any(E'\\x'::bytea);
+SELECT hll_hash_bytea(E'\\x41') = hll_hash_any(E'\\x41'::bytea);
+SELECT hll_hash_bytea(E'\\x42') = hll_hash_any(E'\\x42'::bytea);
+SELECT hll_hash_bytea(E'\\x4142') = hll_hash_any(E'\\x4142'::bytea);
+
+SELECT hll_hash_text('') = hll_hash_any(''::text);
+SELECT hll_hash_text('A') = hll_hash_any('A'::text);
+SELECT hll_hash_text('B') = hll_hash_any('B'::text);
+SELECT hll_hash_text('AB') = hll_hash_any('AB'::text);
+
+-- ---------------- Check several types not handled by default hash functions
+
+-- ---------------- macaddr
+
+SELECT hll_hash_any('08:00:2b:01:02:03'::macaddr);
+SELECT hll_hash_any('08002b010203'::macaddr);
+SELECT hll_hash_any('01-23-45-67-89-ab'::macaddr);
+SELECT hll_hash_any('012345-6789ab'::macaddr);
+
+-- ---------------- interval
+
+SELECT hll_hash_any('1 year 2 months 3 days 4 hours 5 minutes 6seconds'::interval);
+SELECT hll_hash_any('P1Y2M3DT4H5M6S'::interval);
+SELECT hll_hash_any('1997-06 20 12:00:00'::interval);
+SELECT hll_hash_any('P1997-06-20T12:00:00'::interval);