3 # The author disclaims copyright to this source code. In place of
4 # a legal notice, here is a blessing:
6 # May you do good and not evil.
7 # May you find forgiveness for yourself and forgive others.
8 # May you share freely, never taking more than you give.
10 #***********************************************************************
12 # This file tests features of the name resolver (the component that
13 # figures out what identifiers in the SQL statement refer to) that
14 # were fixed by ticket [2500cdb9be].
16 # See also tickets [1c69be2daf] and [f617ea3125] from 2013-08-14.
18 # Also a fuzzer-discovered problem on 2015-04-23.
21 set testdir [file dirname $argv0]
22 source $testdir/tester.tcl
24 # "ORDER BY y" binds to the output result-set column named "y"
25 # if available. If no output column is named "y", then try to
26 # bind against an input column named "y".
28 # This is classical SQL92 behavior.
30 do_test resolver01-1.1 {
32 CREATE TABLE t1(x, y); INSERT INTO t1 VALUES(11,22);
33 CREATE TABLE t2(y, z); INSERT INTO t2 VALUES(33,44);
34 SELECT 1 AS y FROM t1, t2 ORDER BY y;
37 do_test resolver01-1.2 {
39 SELECT 1 AS yy FROM t1, t2 ORDER BY y;
41 } {1 {ambiguous column name: y}}
42 do_test resolver01-1.3 {
44 CREATE TABLE t3(x,y); INSERT INTO t3 VALUES(11,44),(33,22);
45 SELECT x AS y FROM t3 ORDER BY y;
48 do_test resolver01-1.4 {
50 SELECT x AS yy FROM t3 ORDER BY y;
54 # SQLite allows the WHERE clause to reference output columns if there is
55 # no other way to resolve the name.
57 do_test resolver01-1.5 {
59 SELECT x AS yy FROM t3 ORDER BY yy;
62 do_test resolver01-1.6 {
64 SELECT x AS yy FROM t3 ORDER BY 1;
68 # The "ORDER BY y COLLATE nocase" form works the same as "ORDER BY y".
69 # The "y" binds more tightly to output columns than to input columns.
71 # This is for compatibility with SQL92 and with historical SQLite behavior.
72 # Note that PostgreSQL considers "y COLLATE nocase" to be an expression
73 # and thus PostgreSQL treats this case as if it where the 3.x case below.
75 do_test resolver01-2.1 {
77 SELECT 2 AS y FROM t1, t2 ORDER BY y COLLATE nocase;
80 do_test resolver01-2.2 {
82 SELECT 2 AS yy FROM t1, t2 ORDER BY y COLLATE nocase;
84 } {1 {ambiguous column name: y}}
85 do_test resolver01-2.3 {
87 SELECT x AS y FROM t3 ORDER BY y COLLATE nocase;
90 do_test resolver01-2.4 {
92 SELECT x AS yy FROM t3 ORDER BY y COLLATE nocase;
95 do_test resolver01-2.5 {
97 SELECT x AS yy FROM t3 ORDER BY yy COLLATE nocase;
100 do_test resolver01-2.6 {
102 SELECT x AS yy FROM t3 ORDER BY 1 COLLATE nocase;
106 # But if the form is "ORDER BY expr" then bind more tightly to the
107 # the input column names and only use the output column names if no
108 # input column name matches.
110 # This is SQL99 behavior, as implemented by PostgreSQL and MS-SQL.
111 # Note that Oracle works differently.
113 do_test resolver01-3.1 {
115 SELECT 3 AS y FROM t1, t2 ORDER BY +y;
117 } {1 {ambiguous column name: y}}
118 do_test resolver01-3.2 {
120 SELECT 2 AS yy FROM t1, t2 ORDER BY +y;
122 } {1 {ambiguous column name: y}}
123 do_test resolver01-3.3 {
125 SELECT x AS y FROM t3 ORDER BY +y;
128 do_test resolver01-3.4 {
130 SELECT x AS yy FROM t3 ORDER BY +y;
133 do_test resolver01-3.5 {
135 SELECT x AS yy FROM t3 ORDER BY +yy
139 # This is the test case given in ticket [f617ea3125e9] (with table name
140 # changed from "t1" to "t4". The behavior of (1) and (3) match with
141 # PostgreSQL, but we intentionally break with PostgreSQL to provide
142 # SQL92 behavior for case (2).
144 do_execsql_test resolver01-4.1 {
145 CREATE TABLE t4(m CHAR(2));
146 INSERT INTO t4 VALUES('az');
147 INSERT INTO t4 VALUES('by');
148 INSERT INTO t4 VALUES('cx');
149 SELECT '1', substr(m,2) AS m FROM t4 ORDER BY m;
150 SELECT '2', substr(m,2) AS m FROM t4 ORDER BY m COLLATE binary;
151 SELECT '3', substr(m,2) AS m FROM t4 ORDER BY lower(m);
152 } {1 x 1 y 1 z 2 x 2 y 2 z 3 z 3 y 3 x}
154 ##########################################################################
155 # Test cases for ticket [1c69be2dafc28]: Make sure the GROUP BY binds
156 # more tightly to the input tables in all cases.
158 # This first case case has been wrong in SQLite for time out of mind.
159 # For SQLite version 3.7.17 the answer was two rows, which is wrong.
161 do_execsql_test resolver01-5.1 {
162 CREATE TABLE t5(m CHAR(2));
163 INSERT INTO t5 VALUES('ax');
164 INSERT INTO t5 VALUES('bx');
165 INSERT INTO t5 VALUES('cy');
166 SELECT count(*), substr(m,2,1) AS m FROM t5 GROUP BY m ORDER BY 1, 2;
169 # This case is unambiguous and has always been correct.
171 do_execsql_test resolver01-5.2 {
172 SELECT count(*), substr(m,2,1) AS mx FROM t5 GROUP BY m ORDER BY 1, 2;
175 # This case is not allowed in standard SQL, but SQLite allows and does
176 # the sensible thing.
178 do_execsql_test resolver01-5.3 {
179 SELECT count(*), substr(m,2,1) AS mx FROM t5 GROUP BY mx ORDER BY 1, 2;
181 do_execsql_test resolver01-5.4 {
182 SELECT count(*), substr(m,2,1) AS mx FROM t5
183 GROUP BY substr(m,2,1) ORDER BY 1, 2;
186 # These test case weere provided in the 2013-08-14 email from Rob Golsteijn
187 # that originally reported the problem of ticket [1c69be2dafc28].
189 do_execsql_test resolver01-6.1 {
190 CREATE TABLE t61(name);
191 SELECT min(name) FROM t61 GROUP BY lower(name);
193 do_execsql_test resolver01-6.2 {
194 SELECT min(name) AS name FROM t61 GROUP BY lower(name);
196 do_execsql_test resolver01-6.3 {
197 CREATE TABLE t63(name);
198 INSERT INTO t63 VALUES (NULL);
199 INSERT INTO t63 VALUES ('abc');
201 NULLIF(name,'abc') AS name
203 GROUP BY lower(name);
206 do_execsql_test resolver01-7.1 {
207 SELECT 2 AS x WHERE (SELECT x AS y WHERE 3>y);
209 do_execsql_test resolver01-7.2 {
210 SELECT 2 AS x WHERE (SELECT x AS y WHERE 1>y);