Add support for CREATE TABLE AS

include/pgduckdb/pgduckdb_planner.hpp

@@ -5,4 +5,4 @@ extern "C" {
 #include "optimizer/planner.h"
 }
 
-PlannedStmt *DuckdbPlanNode(Query *parse, const char *query_string, int cursor_options, ParamListInfo bound_params);
+PlannedStmt *DuckdbPlanNode(Query *parse, int cursor_options, ParamListInfo bound_params);

src/pgduckdb_hooks.cpp

@@ -46,20 +46,37 @@ IsAllowedStatement(Query *query) {
 	if (query->hasModifyingCTE) {
 		return false;
 	}
-	/* For `SELECT ..` ActivePortal doesn't exist */
-	if (!ActivePortal)
-		return true;
-	/* `EXPLAIN ...` should be allowed */
-	if (ActivePortal->commandTag == CMDTAG_EXPLAIN)
-		return true;
-	return false;
+
+	/* We don't support modifying statements yet */
+	if (query->commandType != CMD_SELECT) {
+		return false;
+	}
+
+	/*
+	 * If there's no rtable, we're only selecting constants. There's no point
+	 * in using DuckDB for that.
+	 */
+	if (!query->rtable) {
+		return false;
+	}
+
+	/*
+	 * If any table is from pg_catalog, we don't want to use DuckDB. This is
+	 * because DuckDB has its own pg_catalog tables that contain different data
+	 * then Postgres its pg_catalog tables.
+	 */
+	if (IsCatalogTable(query->rtable)) {
+		return false;
+	}
+
+	/* Anything else is hopefully fine... */
+	return true;
 }
 
 static PlannedStmt *
 DuckdbPlannerHook(Query *parse, const char *query_string, int cursor_options, ParamListInfo bound_params) {
-	if (duckdb_execution && IsAllowedStatement(parse) && pgduckdb::IsExtensionRegistered() && parse->rtable &&
-	    !IsCatalogTable(parse->rtable) && parse->commandType == CMD_SELECT) {
-		PlannedStmt *duckdb_plan = DuckdbPlanNode(parse, query_string, cursor_options, bound_params);
+	if (duckdb_execution && IsAllowedStatement(parse) && pgduckdb::IsExtensionRegistered()) {
+		PlannedStmt *duckdb_plan = DuckdbPlanNode(parse, cursor_options, bound_params);
 		if (duckdb_plan) {
 			return duckdb_plan;
 		}

src/pgduckdb_planner.cpp

@@ -5,6 +5,8 @@ extern "C" {
 #include "catalog/pg_type.h"
 #include "nodes/makefuncs.h"
 #include "optimizer/optimizer.h"
+#include "tcop/pquery.h"
+#include "utils/ruleutils.h"
 #include "utils/syscache.h"
 }
 
@@ -100,7 +102,12 @@ CreatePlan(Query *query, const char *query_string, ParamListInfo bound_params) {
 }
 
 PlannedStmt *
-DuckdbPlanNode(Query *parse, const char *query_string, int cursor_options, ParamListInfo bound_params) {
+DuckdbPlanNode(Query *parse, int cursor_options, ParamListInfo bound_params) {
+	const char *query_string = pg_get_querydef(parse, false);
+
+	if (ActivePortal && ActivePortal->commandTag == CMDTAG_EXPLAIN) {
+		query_string = psprintf("EXPLAIN %s", query_string);
+	}
 	/* We need to check can we DuckDB create plan */
 	Plan *duckdb_plan = (Plan *)castNode(CustomScan, CreatePlan(parse, query_string, bound_params));
 

test/regression/data/web_page.csv

@@ -0,0 +1,60 @@
+1|AAAAAAAABAAAAAAA|1997-09-03||2450810|2452620|Y|98539|http://www.foo.com|welcome|2531|8|3|4
+2|AAAAAAAACAAAAAAA|1997-09-03|2000-09-02|2450814|2452580|N||http://www.foo.com|protected|1564|4|3|1
+3|AAAAAAAACAAAAAAA|2000-09-03||2450814|2452611|N||http://www.foo.com|feedback|1564|4|3|4
+4|AAAAAAAAEAAAAAAA|1997-09-03|1999-09-03|2450812|2452579|N||http://www.foo.com|general|3732|18|7|1
+5|AAAAAAAAEAAAAAAA|1999-09-04|2001-09-02|2450812|2452597|N||http://www.foo.com|welcome|3732|18|3|1
+6|AAAAAAAAEAAAAAAA|2001-09-03||2450814|2452597|N||http://www.foo.com|ad|3732|18|7|4
+7|AAAAAAAAHAAAAAAA|1997-09-03||2450815|2452574|N||http://www.foo.com|feedback|3034|18|7|4
+8|AAAAAAAAIAAAAAAA|1997-09-03|2000-09-02|2450815|2452646|Y|1898|http://www.foo.com|protected|3128|12|2|4
+9|AAAAAAAAIAAAAAAA|2000-09-03||2450807|2452579|Y|84146|http://www.foo.com|welcome|3128|13|5|3
+10|AAAAAAAAKAAAAAAA|1997-09-03|1999-09-03||2452623|N||http://www.foo.com|||||
+11|AAAAAAAAKAAAAAAA|1999-09-04|2001-09-02|2450814|2452611|N||http://www.foo.com|welcome|7046|23|4|4
+12|AAAAAAAAKAAAAAAA|2001-09-03||2450815|2452611|N||http://www.foo.com|protected|7046|17|4|4
+13|AAAAAAAANAAAAAAA|1997-09-03||2450807|2452629|N||http://www.foo.com|protected|2281|6|4|1
+14|AAAAAAAAOAAAAAAA|1997-09-03|2000-09-02|2450810|2452639|N||http://www.foo.com|dynamic|5676|19|6|0
+15|AAAAAAAAOAAAAAAA|2000-09-03||2450810|2452639|N||http://www.foo.com|dynamic|2469|10|5|2
+16|AAAAAAAAABAAAAAA|1997-09-03|1999-09-03|2450814|2452601|Y|33463|http://www.foo.com|feedback|701|2|1|4
+17|AAAAAAAAABAAAAAA|1999-09-04|2001-09-02|2450812|2452645|N||http://www.foo.com|general|701|11|1|3
+18|AAAAAAAAABAAAAAA|2001-09-03||2450812|2452608|N||http://www.foo.com|ad|4080|11|6|3
+19|AAAAAAAADBAAAAAA|1997-09-03||2450808|2452648|Y|57610|http://www.foo.com|general|2347|9|7|4
+20|AAAAAAAAEBAAAAAA|1997-09-03|2000-09-02|2450809|2452555|Y|46487|http://www.foo.com|ad|1147|3|6|0
+21|AAAAAAAAEBAAAAAA|2000-09-03||2450809|2452555|Y|10897|http://www.foo.com|general|1147|3|6|4
+22|AAAAAAAAGBAAAAAA|1997-09-03|1999-09-03|2450812|2452565|Y|20213|http://www.foo.com|general|5663|25|3|4
+23|AAAAAAAAGBAAAAAA|1999-09-04|2001-09-02|2450812|2452623|Y|20213|http://www.foo.com|order|4729|23|6|4
+24|AAAAAAAAGBAAAAAA|2001-09-03||2450812|2452646|Y|20213|http://www.foo.com|dynamic|5918|23|6|1
+25|AAAAAAAAJBAAAAAA|1997-09-03||2450811|2452620|N||http://www.foo.com|feedback|1526|9|4|2
+26|AAAAAAAAKBAAAAAA|1997-09-03|2000-09-02|2450812|2452636|Y|98376|http://www.foo.com|ad|1826|9|3|1
+27|AAAAAAAAKBAAAAAA|2000-09-03||2450812|2452607|Y|98376|http://www.foo.com|protected|1553|9|1|1
+28|AAAAAAAAMBAAAAAA|1997-09-03|1999-09-03|2450807|2452572|N||http://www.foo.com|protected|1308|4|1|2
+29|AAAAAAAAMBAAAAAA|1999-09-04|2001-09-02|2450808|2452611|N||http://www.foo.com|order|1308|4|1|2
+30|AAAAAAAAMBAAAAAA|2001-09-03||2450808|2452611|N||http://www.foo.com|general|3872|18|1|4
+31|AAAAAAAAPBAAAAAA|1997-09-03||2450810|2452596|N||http://www.foo.com|general|1732|3|6|0
+32|AAAAAAAAACAAAAAA|1997-09-03|2000-09-02|2450808|2452585|N||http://www.foo.com|welcome|5104|20|7|4
+33|AAAAAAAAACAAAAAA|2000-09-03||2450808|2452585|N||http://www.foo.com|protected|2129|7|1|0
+34|AAAAAAAACCAAAAAA|1997-09-03|1999-09-03|2450808|2452616|N||http://www.foo.com|welcome|2726|12|5|2
+35|AAAAAAAACCAAAAAA|1999-09-04|2001-09-02|2450808|2452591|N||http://www.foo.com|protected|2726|12|1|2
+36|AAAAAAAACCAAAAAA|2001-09-03||2450812|2452613|N||http://www.foo.com|dynamic|2726|3|1|2
+37|AAAAAAAAFCAAAAAA|1997-09-03||2450809|2452556|N||http://www.foo.com|ad|3076|15|3|0
+38|AAAAAAAAGCAAAAAA|1997-09-03|2000-09-02|2450811|2452583|Y|37285|http://www.foo.com|general|3096|18|3|0
+39|AAAAAAAAGCAAAAAA|2000-09-03||2450815|2452583|N||http://www.foo.com|general|3096|18|3|0
+40|AAAAAAAAICAAAAAA|1997-09-03|1999-09-03|2450813|2452576|N||http://www.foo.com|general|4402|18|4|2
+41|AAAAAAAAICAAAAAA|1999-09-04|2001-09-02|2450813|2452579|Y|16769|http://www.foo.com|welcome|784|3|4|4
+42|AAAAAAAAICAAAAAA|2001-09-03||2450813|2452579|Y|60150|http://www.foo.com|dynamic|1451|3|4|4
+43|AAAAAAAALCAAAAAA|1997-09-03||2450814|2452580|Y|64793|http://www.foo.com|ad|3760|12|3|2
+44|AAAAAAAAMCAAAAAA|1997-09-03|2000-09-02|2450811|2452602|Y|92078|http://www.foo.com|ad|4179|19|7|1
+45|AAAAAAAAMCAAAAAA|2000-09-03||2450811|2452575|Y|98633|http://www.foo.com|feedback|4584|19|7|4
+46|AAAAAAAAOCAAAAAA|1997-09-03|1999-09-03|2450809|2452574|N||http://www.foo.com|protected|1711|4|5|1
+47|AAAAAAAAOCAAAAAA|1999-09-04|2001-09-02|2450815|2452574|N||http://www.foo.com|welcome|1711|4|5|1
+48|AAAAAAAAOCAAAAAA|2001-09-03||2450815|2452622|N||http://www.foo.com|ad|1732|9|5|1
+49|AAAAAAAABDAAAAAA|1997-09-03||2450809|2452618|N||http://www.foo.com|order|4894|20|3|2
+50|AAAAAAAACDAAAAAA|1997-09-03|2000-09-02|2450808|2452615|N||http://www.foo.com|welcome|5262|16|5|2
+51|AAAAAAAACDAAAAAA|2000-09-03||2450811|2452564|N||http://www.foo.com|general|3423|19|7|1
+52|AAAAAAAAEDAAAAAA|1997-09-03|1999-09-03|2450815|2452606|N||http://www.foo.com|welcome|3306|21|7|1
+53|AAAAAAAAEDAAAAAA|1999-09-04|2001-09-02|2450808|2452636|N||http://www.foo.com|dynamic|3306|21|7|1
+54|AAAAAAAAEDAAAAAA|2001-09-03||2450808|2452629|N||http://www.foo.com|protected|1931|7|2|2
+55|AAAAAAAAHDAAAAAA|1997-09-03||2450811|2452549|N||http://www.foo.com|order|3788|19|1|0
+56|AAAAAAAAIDAAAAAA|1997-09-03|2000-09-02|2450815|2452554|N||http://www.foo.com|protected|5733|24|2|2
+57|AAAAAAAAIDAAAAAA|2000-09-03||2450811|2452568|N||http://www.foo.com|ad|5733|16|2|2
+58|AAAAAAAAKDAAAAAA|1997-09-03|1999-09-03|2450813|2452619|Y|7625|http://www.foo.com|ad|6577|24|4|3
+59|AAAAAAAAKDAAAAAA|1999-09-04|2001-09-02|2450813|2452624|Y|80555|http://www.foo.com|general|6577|24|2|3
+60|AAAAAAAAKDAAAAAA|2001-09-03||2450813|2452566|Y|80555|http://www.foo.com|welcome|6577|24|2|3

test/regression/expected/basic.out

@@ -2,41 +2,41 @@ CREATE TABLE t(a INT);
 INSERT INTO t SELECT g % 10 from generate_series(1,1000000) g;
 SET client_min_messages to 'DEBUG1';
 SELECT COUNT(*) FROM t;
- count_star() 
---------------
-      1000000
+  count  
+---------
+ 1000000
 (1 row)
 
 SELECT a, COUNT(*) FROM t WHERE a > 5 GROUP BY a ORDER BY a;
- a | count_star() 
----+--------------
- 6 |       100000
- 7 |       100000
- 8 |       100000
- 9 |       100000
+ a | count  
+---+--------
+ 6 | 100000
+ 7 | 100000
+ 8 | 100000
+ 9 | 100000
 (4 rows)
 
 SET duckdb.max_threads_per_query to 4;
 SELECT COUNT(*) FROM t;
- count_star() 
---------------
-      1000000
+  count  
+---------
+ 1000000
 (1 row)
 
 SELECT a, COUNT(*) FROM t WHERE a > 5 GROUP BY a ORDER BY a;
- a | count_star() 
----+--------------
- 6 |       100000
- 7 |       100000
- 8 |       100000
- 9 |       100000
+ a | count  
+---+--------
+ 6 | 100000
+ 7 | 100000
+ 8 | 100000
+ 9 | 100000
 (4 rows)
 
 CREATE TABLE empty(a INT);
 SELECT COUNT(*) FROM empty;
- count_star() 
---------------
-            0
+ count 
+-------
+     0
 (1 row)
 
 SET duckdb.max_threads_per_query TO default;

test/regression/expected/create_table_as.out

@@ -0,0 +1,16 @@
+\getenv pwd PWD
+CREATE TABLE webpages AS SELECT * FROM read_csv(:'pwd' || '/data/web_page.csv') as (column00 int, column01 text, column02 date);
+select * from webpages order by column00 limit 2;
+ column00 |     column01     |  column02  
+----------+------------------+------------
+        1 | AAAAAAAABAAAAAAA | 09-03-1997
+        2 | AAAAAAAACAAAAAAA | 09-03-1997
+(2 rows)
+
+select count(*) from webpages;
+ count 
+-------
+    60
+(1 row)
+
+DROP TABLE webpages;

test/regression/expected/execution_error.out

@@ -5,5 +5,5 @@ insert into int_as_varchar SELECT * from (
 ) t(a);
 select a::INTEGER from int_as_varchar;
 ERROR:  Duckdb execute returned an error: Conversion Error: Could not convert string 'abc' to INT32
-LINE 1: select a::INTEGER from int_as_varchar;
-                ^
+LINE 1:  SELECT (a)::integer AS a
+                   ^

test/regression/expected/hugeint_conversion.out

@@ -1,8 +1,8 @@
 create table hugeint_sum(a int);
 insert into hugeint_sum select g from generate_series(1,100) g;
 select pg_typeof(sum(a)) from hugeint_sum;
- pg_typeof(sum(a)) 
--------------------
+ pg_typeof 
+-----------
  hugeint
 (1 row)
 

test/regression/expected/materialized_view.out

@@ -1,42 +1,42 @@
-CREATE TABLE t(a INT, b TEXT);
-INSERT INTO t SELECT g % 1000, MD5(g::TEXT) FROM generate_series(1,20000) g;
+CREATE TABLE t(a INT, b VARCHAR);
+INSERT INTO t SELECT g % 1000, MD5(g::VARCHAR) FROM generate_series(1,20000) g;
 SELECT COUNT(*) FROM t WHERE a % 100 = 0;
- count_star() 
---------------
-          200
+ count 
+-------
+   200
 (1 row)
 
 CREATE MATERIALIZED VIEW tv AS SELECT * FROM t WHERE a % 100 = 0;
 SELECT COUNT(*) FROM tv;
- count_star() 
---------------
-          200
+ count 
+-------
+   200
 (1 row)
 
 INSERT INTO t SELECT g % 1000, MD5(g::TEXT) FROM generate_series(1,20000) g;
 SELECT COUNT(*) FROM t WHERE a % 100 = 0;
- count_star() 
---------------
-          400
+ count 
+-------
+   400
 (1 row)
 
 REFRESH MATERIALIZED VIEW tv;
 SELECT COUNT(*) FROM tv;
- count_star() 
---------------
-          400
+ count 
+-------
+   400
 (1 row)
 
 SELECT COUNT(*) FROM t WHERE (a % 100 = 0) AND (a < 30);
- count_star() 
---------------
-           40
+ count 
+-------
+    40
 (1 row)
 
 SELECT COUNT(*) FROM tv WHERE a < 30;
- count_star() 
---------------
-           40
+ count 
+-------
+    40
 (1 row)
 
 DROP MATERIALIZED VIEW tv;

test/regression/expected/search_path.out

@@ -1,54 +1,54 @@
 CREATE TABLE t(a INT);
 INSERT INTO t SELECT g from generate_series(1,10) g;
 SELECT count(*) FROM t;
- count_star() 
---------------
-           10
+ count 
+-------
+    10
 (1 row)
 
 SELECT count(*) FROM public.t;
- count_star() 
---------------
-           10
+ count 
+-------
+    10
 (1 row)
 
 -- Create schema `other`
 CREATE SCHEMA other;
 CREATE TABLE other.t(a INT);
 INSERT INTO other.t SELECT g from generate_series(1,100) g;
 SELECT count(*) FROM other.t;
- count_star() 
---------------
-          100
+ count 
+-------
+   100
 (1 row)
 
 -- Test fully qualified table name combinations
 SELECT count(*) FROM public.t, other.t;
- count_star() 
---------------
-         1000
+ count 
+-------
+  1000
 (1 row)
 
 SELECT count(*) FROM t, other.t;
- count_star() 
---------------
-         1000
+ count 
+-------
+  1000
 (1 row)
 
 SELECT count(*) FROM t,t;
 ERROR:  table name "t" specified more than once
 -- search_path ORDER matters.
 SET search_path TO other, public;
 SELECT count(*) FROM t;
- count_star() 
---------------
-          100
+ count 
+-------
+   100
 (1 row)
 
 SELECT count(*) FROM t, public.t;
- count_star() 
---------------
-         1000
+ count 
+-------
+  1000
 (1 row)
 
 -- No search_path
@@ -58,9 +58,9 @@ ERROR:  relation "t" does not exist
 LINE 1: SELECT count(*) FROM t, other.t;
                              ^
 SELECT count(*) FROM public.t, other.t;
- count_star() 
---------------
-         1000
+ count 
+-------
+  1000
 (1 row)
 
 -- Cleanup