Add hash operator class

The uri = uri operator was marked as HASHES, but no hash function or
hash operator class was defined.  Hash joins would therefore fail.

Bug report and original patch by Grégoire HUBERT (@chanmix51).

closes #3

Makefile

@@ -1,11 +1,12 @@
 PG_CONFIG = pg_config
 PKG_CONFIG = pkg-config
 
-extension_version = 0
+extension_version = 1
 
 EXTENSION = uri
 MODULE_big = uri
 OBJS = uri.o
+DATA = uri--0--1.sql
 DATA_built = uri--$(extension_version).sql
 
 ifeq (no,$(shell $(PKG_CONFIG) liburiparser || echo no))

test/expected/test.out

@@ -289,3 +289,14 @@ SELECT DISTINCT b FROM test ORDER BY b;
  ssh://username@review.openstack.org:29418/openstack/nova.git
 (17 rows)
 
+CREATE TABLE test2 (x text, y uri);
+INSERT INTO test2 VALUES ('foo', 'http://www.postgresql.org/');
+-- check hashing (issue petere/pguri#3)
+SET enable_nestloop = off;
+SET enable_mergejoin = off;
+SELECT * FROM test JOIN test2 ON b = y AND a = 1;
+ a |             b              |  x  |             y              
+---+----------------------------+-----+----------------------------
+ 1 | http://www.postgresql.org/ | foo | http://www.postgresql.org/
+(1 row)
+

test/sql/test.sql

@@ -48,3 +48,12 @@ SELECT b AS uri,
 \x off
 
 SELECT DISTINCT b FROM test ORDER BY b;
+
+
+CREATE TABLE test2 (x text, y uri);
+INSERT INTO test2 VALUES ('foo', 'http://www.postgresql.org/');
+
+-- check hashing (issue petere/pguri#3)
+SET enable_nestloop = off;
+SET enable_mergejoin = off;
+SELECT * FROM test JOIN test2 ON b = y AND a = 1;

uri--0--1.sql

@@ -0,0 +1,10 @@
+CREATE FUNCTION uri_hash(uri) RETURNS integer
+    IMMUTABLE
+    STRICT
+    LANGUAGE C
+    AS '$libdir/uri';
+
+CREATE OPERATOR CLASS uri_ops_hash
+    DEFAULT FOR TYPE uri USING hash AS
+        OPERATOR        1       =,
+        FUNCTION        1       uri_hash(uri);

uri.c

@@ -1,4 +1,5 @@
 #include <postgres.h>
+#include <access/hash.h>
 #include <catalog/pg_type.h>
 #include <fmgr.h>
 #include <lib/stringinfo.h>
@@ -16,9 +17,11 @@ typedef struct varlena uritype;
 
 
 #define DatumGetUriP(X)		((uritype *) PG_DETOAST_DATUM(X))
+#define DatumGetUriPP(X)	((uritype *) PG_DETOAST_DATUM_PACKED(X))
 #define UriPGetDatum(X)		PointerGetDatum(X)
 
 #define PG_GETARG_URI_P(n)	DatumGetUriP(PG_GETARG_DATUM(n))
+#define PG_GETARG_URI_PP(n)	DatumGetUriPP(PG_GETARG_DATUM(n))
 #define PG_RETURN_URI_P(x)	PG_RETURN_POINTER(x)
 
 
@@ -436,3 +439,19 @@ uri_cmp(PG_FUNCTION_ARGS)
 
 	PG_RETURN_INT32(_uri_cmp(arg1, arg2));
 }
+
+PG_FUNCTION_INFO_V1(uri_hash);
+Datum
+uri_hash(PG_FUNCTION_ARGS)
+{
+	uritype	   *key = PG_GETARG_URI_PP(0);
+	Datum		result;
+
+	result = hash_any((unsigned char *) VARDATA_ANY(key),
+					  VARSIZE_ANY_EXHDR(key));
+
+	/* Avoid leaking memory for toasted inputs */
+	PG_FREE_IF_COPY(key, 0);
+
+	return result;
+}

uri.control

@@ -1,4 +1,4 @@
 comment = 'uri type'
-default_version = 0
+default_version = 1
 module_pathname = '$libdir/uri'
 relocatable = true

uri.sql

@@ -130,6 +130,12 @@ CREATE FUNCTION uri_cmp(uri, uri) RETURNS integer
     LANGUAGE C
     AS '$libdir/uri';
 
+CREATE FUNCTION uri_hash(uri) RETURNS integer
+    IMMUTABLE
+    STRICT
+    LANGUAGE C
+    AS '$libdir/uri';
+
 CREATE OPERATOR < (
     LEFTARG = uri,
     RIGHTARG = uri,
@@ -200,3 +206,8 @@ CREATE OPERATOR CLASS uri_ops
         OPERATOR        4       >= ,
         OPERATOR        5       > ,
         FUNCTION        1       uri_cmp(uri, uri);
+
+CREATE OPERATOR CLASS uri_ops_hash
+    DEFAULT FOR TYPE uri USING hash AS
+        OPERATOR        1       =,
+        FUNCTION        1       uri_hash(uri);