pushdown for new functions in pg_re2 0.3

Author: serprex

src/custom_types.c

@@ -310,13 +310,17 @@ static char *re2_func_map[][2] = {
 	{"countmatches", "countMatches"},
 	{"countmatchescaseinsensitive", "countMatchesCaseInsensitive"},
 	{"extractall", "extractAll"},
+	{"extractallgroupshorizontal", "extractAllGroupsHorizontal"},
+	{"extractallgroupsvertical", "extractAllGroupsVertical"},
 	{"extractgroups", "extractGroups"},
 	{"multimatchallindices", "multiMatchAllIndices"},
 	{"multimatchany", "multiMatchAny"},
 	{"multimatchanyindex", "multiMatchAnyIndex"},
 	{"regexpextract", "regexpExtract"},
+	{"regexpquotemeta", "regexpQuoteMeta"},
 	{"replaceregexpall", "replaceRegexpAll"},
 	{"replaceregexpone", "replaceRegexpOne"},
+	{"splitbyregexp", "splitByRegexp"},
 	{NULL, NULL},
 };
 

test/expected/re2_functions.out

@@ -275,6 +275,84 @@ SELECT * FROM t1 WHERE re2multimatchallindices(val, VARIADIC ARRAY['POSIX','PCRE
   1 | POSIX uses BRE and ERE
 (1 row)
 
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2regexpquotemeta(val) <> val;
+                                   QUERY PLAN                                   
+--------------------------------------------------------------------------------
+ Foreign Scan on re2_test.t1
+   Output: id
+   Remote SQL: SELECT id FROM re2_test.t1 WHERE ((regexpQuoteMeta(val) <> val))
+(3 rows)
+
+SELECT id FROM t1 WHERE re2regexpquotemeta(val) <> val;
+ id 
+----
+(0 rows)
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2splitbyregexp(val, ' ') = ARRAY['re2','uses','finite','automata'];
+                                                     QUERY PLAN                                                     
+--------------------------------------------------------------------------------------------------------------------
+ Foreign Scan on re2_test.t1
+   Output: id
+   Remote SQL: SELECT id FROM re2_test.t1 WHERE ((splitByRegexp(' ', val, 0) = ['re2','uses','finite','automata']))
+(3 rows)
+
+SELECT id FROM t1 WHERE re2splitbyregexp(val, ' ') = ARRAY['re2','uses','finite','automata'];
+ id 
+----
+  2
+(1 row)
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2splitbyregexp(val, ' ', 2) = ARRAY['re2','uses'];
+                                           QUERY PLAN                                           
+------------------------------------------------------------------------------------------------
+ Foreign Scan on re2_test.t1
+   Output: id
+   Remote SQL: SELECT id FROM re2_test.t1 WHERE ((splitByRegexp(' ', val, 2) = ['re2','uses']))
+(3 rows)
+
+SELECT id FROM t1 WHERE re2splitbyregexp(val, ' ', 2) = ARRAY['re2','uses'];
+ id 
+----
+  2
+(1 row)
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE array_length(re2extractallgroupsvertical(val, '(\w+) (\w+)'), 1) > 0;
+                                                  QUERY PLAN                                                   
+---------------------------------------------------------------------------------------------------------------
+ Foreign Scan on re2_test.t1
+   Output: id
+   Remote SQL: SELECT id FROM re2_test.t1 WHERE ((length(extractAllGroupsVertical(val, '(\\w+) (\\w+)')) > 0))
+(3 rows)
+
+SELECT id FROM t1 WHERE array_length(re2extractallgroupsvertical(val, '(\w+) (\w+)'), 1) > 0;
+ id 
+----
+  1
+  2
+  3
+(3 rows)
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE array_length(re2extractallgroupshorizontal(val, '(\w+) (\w+)'), 1) = 2;
+                                                   QUERY PLAN                                                    
+-----------------------------------------------------------------------------------------------------------------
+ Foreign Scan on re2_test.t1
+   Output: id
+   Remote SQL: SELECT id FROM re2_test.t1 WHERE ((length(extractAllGroupsHorizontal(val, '(\\w+) (\\w+)')) = 2))
+(3 rows)
+
+SELECT id FROM t1 WHERE array_length(re2extractallgroupshorizontal(val, '(\w+) (\w+)'), 1) = 2;
+ id 
+----
+  1
+  2
+  3
+(3 rows)
+
 DROP EXTENSION re2;
 DROP USER MAPPING FOR CURRENT_USER SERVER re2_svr;
 SELECT clickhouse_raw_query('DROP DATABASE re2_test');

test/sql/re2_functions.sql

@@ -85,6 +85,26 @@ EXPLAIN (VERBOSE, COSTS OFF)
 SELECT * FROM t1 WHERE re2multimatchallindices(val, VARIADIC ARRAY['POSIX','PCRE']) = ARRAY[1];
 SELECT * FROM t1 WHERE re2multimatchallindices(val, VARIADIC ARRAY['POSIX','PCRE']) = ARRAY[1];
 
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2regexpquotemeta(val) <> val;
+SELECT id FROM t1 WHERE re2regexpquotemeta(val) <> val;
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2splitbyregexp(' ', val) = ARRAY['re2','uses','finite','automata'];
+SELECT id FROM t1 WHERE re2splitbyregexp(' ', val) = ARRAY['re2','uses','finite','automata'];
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE re2splitbyregexp(' ', val, 2) = ARRAY['re2','uses'];
+SELECT id FROM t1 WHERE re2splitbyregexp(' ', val, 2) = ARRAY['re2','uses'];
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE array_length(re2extractallgroupsvertical(val, '(\w+) (\w+)'), 1) > 0;
+SELECT id FROM t1 WHERE array_length(re2extractallgroupsvertical(val, '(\w+) (\w+)'), 1) > 0;
+
+EXPLAIN (VERBOSE, COSTS OFF)
+SELECT id FROM t1 WHERE array_length(re2extractallgroupshorizontal(val, '(\w+) (\w+)'), 1) = 2;
+SELECT id FROM t1 WHERE array_length(re2extractallgroupshorizontal(val, '(\w+) (\w+)'), 1) = 2;
+
 DROP EXTENSION re2;
 DROP USER MAPPING FOR CURRENT_USER SERVER re2_svr;
 SELECT clickhouse_raw_query('DROP DATABASE re2_test');