mirror of
https://github.com/tursodatabase/libsql.git
synced 2024-12-15 06:19:42 +00:00
705 lines
19 KiB
Plaintext
705 lines
19 KiB
Plaintext
#
|
|
# 2001 September 15
|
|
#
|
|
# The author disclaims copyright to this source code. In place of
|
|
# a legal notice, here is a blessing:
|
|
#
|
|
# May you do good and not evil.
|
|
# May you find forgiveness for yourself and forgive others.
|
|
# May you share freely, never taking more than you give.
|
|
#
|
|
#***********************************************************************
|
|
# This file implements regression tests for SQLite library. The
|
|
# focus of this script is page cache subsystem.
|
|
#
|
|
# $Id: collate4.test,v 1.9 2008/01/05 17:39:30 danielk1977 Exp $
|
|
|
|
set testdir [file dirname $argv0]
|
|
source $testdir/tester.tcl
|
|
|
|
db collate TEXT text_collate
|
|
proc text_collate {a b} {
|
|
return [string compare $a $b]
|
|
}
|
|
|
|
# Do an SQL statement. Append the search count to the end of the result.
|
|
#
|
|
proc count sql {
|
|
set ::sqlite_search_count 0
|
|
return [concat [execsql $sql] $::sqlite_search_count]
|
|
}
|
|
|
|
# This procedure executes the SQL. Then it checks the generated program
|
|
# for the SQL and appends a "nosort" to the result if the program contains the
|
|
# SortCallback opcode. If the program does not contain the SortCallback
|
|
# opcode it appends "sort"
|
|
#
|
|
proc cksort {sql} {
|
|
set ::sqlite_sort_count 0
|
|
set data [execsql $sql]
|
|
if {$::sqlite_sort_count} {set x sort} {set x nosort}
|
|
lappend data $x
|
|
return $data
|
|
}
|
|
|
|
#
|
|
# Test cases are organized roughly as follows:
|
|
#
|
|
# collate4-1.* ORDER BY.
|
|
# collate4-2.* WHERE clauses.
|
|
# collate4-3.* constraints (primary key, unique).
|
|
# collate4-4.* simple min() or max() queries.
|
|
# collate4-5.* REINDEX command
|
|
# collate4-6.* INTEGER PRIMARY KEY indices.
|
|
#
|
|
|
|
#
|
|
# These tests - collate4-1.* - check that indices are correctly
|
|
# selected or not selected to implement ORDER BY clauses when
|
|
# user defined collation sequences are involved.
|
|
#
|
|
# Because these tests also exercise all the different ways indices
|
|
# can be created, they also serve to verify that indices are correctly
|
|
# initialized with user-defined collation sequences when they are
|
|
# created.
|
|
#
|
|
# Tests named collate4-1.1.* use indices with a single column. Tests
|
|
# collate4-1.2.* use indices with two columns.
|
|
#
|
|
do_test collate4-1.1.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a COLLATE NOCASE, b COLLATE TEXT);
|
|
INSERT INTO collate4t1 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t1 VALUES( 'b', 'b' );
|
|
INSERT INTO collate4t1 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t1 VALUES( 'B', 'B' );
|
|
INSERT INTO collate4t1 VALUES( 'A', 'A' );
|
|
CREATE INDEX collate4i1 ON collate4t1(a);
|
|
CREATE INDEX collate4i2 ON collate4t1(b);
|
|
}
|
|
} {}
|
|
do_test collate4-1.1.1 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a}
|
|
} {{} a A b B nosort}
|
|
do_test collate4-1.1.2 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a COLLATE NOCASE}
|
|
} {{} a A b B nosort}
|
|
do_test collate4-1.1.3 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a COLLATE TEXT}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.1.4 {
|
|
cksort {SELECT b FROM collate4t1 ORDER BY b}
|
|
} {{} A B a b nosort}
|
|
do_test collate4-1.1.5 {
|
|
cksort {SELECT b FROM collate4t1 ORDER BY b COLLATE TEXT}
|
|
} {{} A B a b nosort}
|
|
do_test collate4-1.1.6 {
|
|
cksort {SELECT b FROM collate4t1 ORDER BY b COLLATE NOCASE, rowid}
|
|
} {{} a A b B sort}
|
|
|
|
do_test collate4-1.1.7 {
|
|
execsql {
|
|
CREATE TABLE collate4t2(
|
|
a PRIMARY KEY COLLATE NOCASE,
|
|
b UNIQUE COLLATE TEXT
|
|
);
|
|
INSERT INTO collate4t2 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t2 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t2 VALUES( 'B', 'B' );
|
|
}
|
|
} {}
|
|
do_test collate4-1.1.8 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.1.9 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a COLLATE NOCASE}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.1.10 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a COLLATE TEXT}
|
|
} {{} B a sort}
|
|
do_test collate4-1.1.11 {
|
|
cksort {SELECT b FROM collate4t2 ORDER BY b}
|
|
} {{} B a nosort}
|
|
do_test collate4-1.1.12 {
|
|
cksort {SELECT b FROM collate4t2 ORDER BY b COLLATE TEXT}
|
|
} {{} B a nosort}
|
|
do_test collate4-1.1.13 {
|
|
cksort {SELECT b FROM collate4t2 ORDER BY b COLLATE NOCASE}
|
|
} {{} a B sort}
|
|
|
|
do_test collate4-1.1.14 {
|
|
execsql {
|
|
CREATE TABLE collate4t3(
|
|
b COLLATE TEXT,
|
|
a COLLATE NOCASE,
|
|
UNIQUE(a), PRIMARY KEY(b)
|
|
);
|
|
INSERT INTO collate4t3 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t3 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t3 VALUES( 'B', 'B' );
|
|
}
|
|
} {}
|
|
do_test collate4-1.1.15 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.1.16 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE NOCASE}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.1.17 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE TEXT}
|
|
} {{} B a sort}
|
|
do_test collate4-1.1.18 {
|
|
cksort {SELECT b FROM collate4t3 ORDER BY b}
|
|
} {{} B a nosort}
|
|
do_test collate4-1.1.19 {
|
|
cksort {SELECT b FROM collate4t3 ORDER BY b COLLATE TEXT}
|
|
} {{} B a nosort}
|
|
do_test collate4-1.1.20 {
|
|
cksort {SELECT b FROM collate4t3 ORDER BY b COLLATE NOCASE}
|
|
} {{} a B sort}
|
|
|
|
do_test collate4-1.1.21 {
|
|
execsql {
|
|
CREATE TABLE collate4t4(a COLLATE NOCASE, b COLLATE TEXT);
|
|
INSERT INTO collate4t4 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t4 VALUES( 'b', 'b' );
|
|
INSERT INTO collate4t4 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t4 VALUES( 'B', 'B' );
|
|
INSERT INTO collate4t4 VALUES( 'A', 'A' );
|
|
CREATE INDEX collate4i3 ON collate4t4(a COLLATE TEXT);
|
|
CREATE INDEX collate4i4 ON collate4t4(b COLLATE NOCASE);
|
|
}
|
|
} {}
|
|
do_test collate4-1.1.22 {
|
|
cksort {SELECT a FROM collate4t4 ORDER BY a, rowid}
|
|
} {{} a A b B sort}
|
|
do_test collate4-1.1.23 {
|
|
cksort {SELECT a FROM collate4t4 ORDER BY a COLLATE NOCASE, rowid}
|
|
} {{} a A b B sort}
|
|
do_test collate4-1.1.24 {
|
|
cksort {SELECT a FROM collate4t4 ORDER BY a COLLATE TEXT, rowid}
|
|
} {{} A B a b nosort}
|
|
do_test collate4-1.1.25 {
|
|
cksort {SELECT b FROM collate4t4 ORDER BY b}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.1.26 {
|
|
cksort {SELECT b FROM collate4t4 ORDER BY b COLLATE TEXT}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.1.27 {
|
|
cksort {SELECT b FROM collate4t4 ORDER BY b COLLATE NOCASE}
|
|
} {{} a A b B nosort}
|
|
|
|
do_test collate4-1.1.30 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
DROP TABLE collate4t2;
|
|
DROP TABLE collate4t3;
|
|
DROP TABLE collate4t4;
|
|
}
|
|
} {}
|
|
|
|
do_test collate4-1.2.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a COLLATE NOCASE, b COLLATE TEXT);
|
|
INSERT INTO collate4t1 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t1 VALUES( 'b', 'b' );
|
|
INSERT INTO collate4t1 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t1 VALUES( 'B', 'B' );
|
|
INSERT INTO collate4t1 VALUES( 'A', 'A' );
|
|
CREATE INDEX collate4i1 ON collate4t1(a, b);
|
|
}
|
|
} {}
|
|
do_test collate4-1.2.1 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a}
|
|
} {{} A a B b nosort}
|
|
do_test collate4-1.2.2 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a COLLATE nocase}
|
|
} {{} A a B b nosort}
|
|
do_test collate4-1.2.3 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a COLLATE text}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.2.4 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a, b}
|
|
} {{} A a B b nosort}
|
|
do_test collate4-1.2.5 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a, b COLLATE nocase, rowid}
|
|
} {{} a A b B sort}
|
|
do_test collate4-1.2.6 {
|
|
cksort {SELECT a FROM collate4t1 ORDER BY a, b COLLATE text}
|
|
} {{} A a B b nosort}
|
|
|
|
do_test collate4-1.2.7 {
|
|
execsql {
|
|
CREATE TABLE collate4t2(
|
|
a COLLATE NOCASE,
|
|
b COLLATE TEXT,
|
|
PRIMARY KEY(a, b)
|
|
);
|
|
INSERT INTO collate4t2 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t2 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t2 VALUES( 'B', 'B' );
|
|
}
|
|
} {}
|
|
do_test collate4-1.2.8 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.2.9 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a COLLATE nocase}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.2.10 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a COLLATE text}
|
|
} {{} B a sort}
|
|
do_test collate4-1.2.11 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a, b}
|
|
} {{} a B nosort}
|
|
do_test collate4-1.2.12 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a, b COLLATE nocase}
|
|
} {{} a B sort}
|
|
do_test collate4-1.2.13 {
|
|
cksort {SELECT a FROM collate4t2 ORDER BY a, b COLLATE text}
|
|
} {{} a B nosort}
|
|
|
|
do_test collate4-1.2.14 {
|
|
execsql {
|
|
CREATE TABLE collate4t3(a COLLATE NOCASE, b COLLATE TEXT);
|
|
INSERT INTO collate4t3 VALUES( 'a', 'a' );
|
|
INSERT INTO collate4t3 VALUES( 'b', 'b' );
|
|
INSERT INTO collate4t3 VALUES( NULL, NULL );
|
|
INSERT INTO collate4t3 VALUES( 'B', 'B' );
|
|
INSERT INTO collate4t3 VALUES( 'A', 'A' );
|
|
CREATE INDEX collate4i2 ON collate4t3(a COLLATE TEXT, b COLLATE NOCASE);
|
|
}
|
|
} {}
|
|
do_test collate4-1.2.15 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a, rowid}
|
|
} {{} a A b B sort}
|
|
do_test collate4-1.2.16 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE nocase, rowid}
|
|
} {{} a A b B sort}
|
|
do_test collate4-1.2.17 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text}
|
|
} {{} A B a b nosort}
|
|
do_test collate4-1.2.18 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text, b}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.2.19 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text, b COLLATE nocase}
|
|
} {{} A B a b nosort}
|
|
do_test collate4-1.2.20 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text, b COLLATE text}
|
|
} {{} A B a b sort}
|
|
do_test collate4-1.2.21 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text DESC}
|
|
} {b a B A {} nosort}
|
|
do_test collate4-1.2.22 {
|
|
cksort {SELECT a FROM collate4t3 ORDER BY a COLLATE text DESC, b}
|
|
} {b a B A {} sort}
|
|
do_test collate4-1.2.23 {
|
|
cksort {SELECT a FROM collate4t3
|
|
ORDER BY a COLLATE text DESC, b COLLATE nocase}
|
|
} {b a B A {} sort}
|
|
do_test collate4-1.2.24 {
|
|
cksort {SELECT a FROM collate4t3
|
|
ORDER BY a COLLATE text DESC, b COLLATE nocase DESC}
|
|
} {b a B A {} nosort}
|
|
|
|
do_test collate4-1.2.25 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
DROP TABLE collate4t2;
|
|
DROP TABLE collate4t3;
|
|
}
|
|
} {}
|
|
|
|
#
|
|
# These tests - collate4-2.* - check that indices are correctly
|
|
# selected or not selected to implement WHERE clauses when user
|
|
# defined collation sequences are involved.
|
|
#
|
|
# Indices may optimise WHERE clauses using <, >, <=, >=, = or IN
|
|
# operators.
|
|
#
|
|
do_test collate4-2.1.0 {
|
|
execsql {
|
|
PRAGMA automatic_index=OFF;
|
|
CREATE TABLE collate4t1(a COLLATE NOCASE);
|
|
CREATE TABLE collate4t2(b COLLATE TEXT);
|
|
|
|
INSERT INTO collate4t1 VALUES('a');
|
|
INSERT INTO collate4t1 VALUES('A');
|
|
INSERT INTO collate4t1 VALUES('b');
|
|
INSERT INTO collate4t1 VALUES('B');
|
|
INSERT INTO collate4t1 VALUES('c');
|
|
INSERT INTO collate4t1 VALUES('C');
|
|
INSERT INTO collate4t1 VALUES('d');
|
|
INSERT INTO collate4t1 VALUES('D');
|
|
INSERT INTO collate4t1 VALUES('e');
|
|
INSERT INTO collate4t1 VALUES('D');
|
|
|
|
INSERT INTO collate4t2 VALUES('A');
|
|
INSERT INTO collate4t2 VALUES('Z');
|
|
}
|
|
} {}
|
|
do_test collate4-2.1.1 {
|
|
count {
|
|
SELECT * FROM collate4t2, collate4t1 WHERE a = b;
|
|
}
|
|
} {A a A A 19}
|
|
do_test collate4-2.1.2 {
|
|
execsql {
|
|
CREATE INDEX collate4i1 ON collate4t1(a);
|
|
}
|
|
count {
|
|
SELECT * FROM collate4t2, collate4t1 WHERE a = b;
|
|
}
|
|
} {A a A A 4}
|
|
do_test collate4-2.1.3 {
|
|
count {
|
|
SELECT * FROM collate4t2, collate4t1 WHERE b = a;
|
|
}
|
|
} {A A 19}
|
|
do_test collate4-2.1.4 {
|
|
execsql {
|
|
DROP INDEX collate4i1;
|
|
CREATE INDEX collate4i1 ON collate4t1(a COLLATE TEXT);
|
|
}
|
|
count {
|
|
SELECT * FROM collate4t2, collate4t1 WHERE a = b
|
|
ORDER BY collate4t2.rowid, collate4t1.rowid
|
|
}
|
|
} {A a A A 19}
|
|
do_test collate4-2.1.5 {
|
|
count {
|
|
SELECT * FROM collate4t2, collate4t1 WHERE b = a;
|
|
}
|
|
} {A A 3}
|
|
ifcapable subquery {
|
|
do_test collate4-2.1.6 {
|
|
count {
|
|
SELECT a FROM collate4t1 WHERE a IN (SELECT * FROM collate4t2)
|
|
ORDER BY rowid
|
|
}
|
|
} {a A 10}
|
|
do_test collate4-2.1.7 {
|
|
execsql {
|
|
DROP INDEX collate4i1;
|
|
CREATE INDEX collate4i1 ON collate4t1(a);
|
|
}
|
|
count {
|
|
SELECT a FROM collate4t1 WHERE a IN (SELECT * FROM collate4t2)
|
|
ORDER BY rowid
|
|
}
|
|
} {a A 5}
|
|
do_test collate4-2.1.8 {
|
|
count {
|
|
SELECT a FROM collate4t1 WHERE a IN ('z', 'a');
|
|
}
|
|
} {a A 4}
|
|
do_test collate4-2.1.9 {
|
|
execsql {
|
|
DROP INDEX collate4i1;
|
|
CREATE INDEX collate4i1 ON collate4t1(a COLLATE TEXT);
|
|
}
|
|
count {
|
|
SELECT a FROM collate4t1 WHERE a IN ('z', 'a') ORDER BY rowid;
|
|
}
|
|
} {a A 9}
|
|
}
|
|
do_test collate4-2.1.10 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
DROP TABLE collate4t2;
|
|
}
|
|
} {}
|
|
|
|
do_test collate4-2.2.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a COLLATE nocase, b COLLATE text, c);
|
|
CREATE TABLE collate4t2(a COLLATE nocase, b COLLATE text, c COLLATE TEXT);
|
|
|
|
INSERT INTO collate4t1 VALUES('0', '0', '0');
|
|
INSERT INTO collate4t1 VALUES('0', '0', '1');
|
|
INSERT INTO collate4t1 VALUES('0', '1', '0');
|
|
INSERT INTO collate4t1 VALUES('0', '1', '1');
|
|
INSERT INTO collate4t1 VALUES('1', '0', '0');
|
|
INSERT INTO collate4t1 VALUES('1', '0', '1');
|
|
INSERT INTO collate4t1 VALUES('1', '1', '0');
|
|
INSERT INTO collate4t1 VALUES('1', '1', '1');
|
|
insert into collate4t2 SELECT * FROM collate4t1;
|
|
}
|
|
} {}
|
|
do_test collate4-2.2.1 {
|
|
count {
|
|
SELECT * FROM collate4t2 NOT INDEXED NATURAL JOIN collate4t1 NOT INDEXED;
|
|
}
|
|
} {0 0 0 0 0 1 0 1 0 0 1 1 1 0 0 1 0 1 1 1 0 1 1 1 63}
|
|
do_test collate4-2.2.1b {
|
|
execsql {
|
|
CREATE INDEX collate4i1 ON collate4t1(a, b, c);
|
|
}
|
|
count {
|
|
SELECT * FROM collate4t2 NATURAL JOIN collate4t1;
|
|
}
|
|
} {0 0 0 0 0 1 0 1 0 0 1 1 1 0 0 1 0 1 1 1 0 1 1 1 29}
|
|
do_test collate4-2.2.2 {
|
|
execsql {
|
|
DROP INDEX collate4i1;
|
|
CREATE INDEX collate4i1 ON collate4t1(a, b, c COLLATE text);
|
|
}
|
|
count {
|
|
SELECT * FROM collate4t2 NATURAL JOIN collate4t1;
|
|
}
|
|
} {0 0 0 0 0 1 0 1 0 0 1 1 1 0 0 1 0 1 1 1 0 1 1 1 22}
|
|
|
|
do_test collate4-2.2.10 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
DROP TABLE collate4t2;
|
|
}
|
|
} {}
|
|
|
|
#
|
|
# These tests - collate4-3.* verify that indices that implement
|
|
# UNIQUE and PRIMARY KEY constraints operate correctly with user
|
|
# defined collation sequences.
|
|
#
|
|
do_test collate4-3.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a PRIMARY KEY COLLATE NOCASE);
|
|
}
|
|
} {}
|
|
do_test collate4-3.1 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES('abc');
|
|
INSERT INTO collate4t1 VALUES('ABC');
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.2 {
|
|
execsql {
|
|
SELECT * FROM collate4t1;
|
|
}
|
|
} {abc}
|
|
do_test collate4-3.3 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 SELECT upper(a) FROM collate4t1;
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.4 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES(1);
|
|
UPDATE collate4t1 SET a = 'abc';
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.5 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
CREATE TABLE collate4t1(a COLLATE NOCASE UNIQUE);
|
|
}
|
|
} {}
|
|
do_test collate4-3.6 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES('abc');
|
|
INSERT INTO collate4t1 VALUES('ABC');
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.7 {
|
|
execsql {
|
|
SELECT * FROM collate4t1;
|
|
}
|
|
} {abc}
|
|
do_test collate4-3.8 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 SELECT upper(a) FROM collate4t1;
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.9 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES(1);
|
|
UPDATE collate4t1 SET a = 'abc';
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.10 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
CREATE TABLE collate4t1(a);
|
|
CREATE UNIQUE INDEX collate4i1 ON collate4t1(a COLLATE NOCASE);
|
|
}
|
|
} {}
|
|
do_test collate4-3.11 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES('abc');
|
|
INSERT INTO collate4t1 VALUES('ABC');
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.12 {
|
|
execsql {
|
|
SELECT * FROM collate4t1;
|
|
}
|
|
} {abc}
|
|
do_test collate4-3.13 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 SELECT upper(a) FROM collate4t1;
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
do_test collate4-3.14 {
|
|
catchsql {
|
|
INSERT INTO collate4t1 VALUES(1);
|
|
UPDATE collate4t1 SET a = 'abc';
|
|
}
|
|
} {1 {UNIQUE constraint failed: collate4t1.a}}
|
|
|
|
do_test collate4-3.15 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
}
|
|
} {}
|
|
|
|
# Mimic the SQLite 2 collation type NUMERIC.
|
|
db collate numeric numeric_collate
|
|
proc numeric_collate {lhs rhs} {
|
|
if {$lhs == $rhs} {return 0}
|
|
return [expr ($lhs>$rhs)?1:-1]
|
|
}
|
|
|
|
#
|
|
# These tests - collate4-4.* check that min() and max() only ever
|
|
# use indices constructed with built-in collation type numeric.
|
|
#
|
|
# CHANGED: min() and max() now use the collation type. If there
|
|
# is an indice that can be used, it is used.
|
|
#
|
|
do_test collate4-4.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a COLLATE TEXT);
|
|
INSERT INTO collate4t1 VALUES('2');
|
|
INSERT INTO collate4t1 VALUES('10');
|
|
INSERT INTO collate4t1 VALUES('20');
|
|
INSERT INTO collate4t1 VALUES('104');
|
|
}
|
|
} {}
|
|
do_test collate4-4.1 {
|
|
count {
|
|
SELECT max(a) FROM collate4t1
|
|
}
|
|
} {20 3}
|
|
do_test collate4-4.2 {
|
|
count {
|
|
SELECT min(a) FROM collate4t1
|
|
}
|
|
} {10 3}
|
|
do_test collate4-4.3 {
|
|
# Test that the index with collation type TEXT is used.
|
|
execsql {
|
|
CREATE INDEX collate4i1 ON collate4t1(a);
|
|
}
|
|
count {
|
|
SELECT min(a) FROM collate4t1;
|
|
}
|
|
} {10 1}
|
|
do_test collate4-4.4 {
|
|
count {
|
|
SELECT max(a) FROM collate4t1;
|
|
}
|
|
} {20 0}
|
|
do_test collate4-4.5 {
|
|
# Test that the index with collation type NUMERIC is not used.
|
|
execsql {
|
|
DROP INDEX collate4i1;
|
|
CREATE INDEX collate4i1 ON collate4t1(a COLLATE NUMERIC);
|
|
}
|
|
count {
|
|
SELECT min(a) FROM collate4t1;
|
|
}
|
|
} {10 3}
|
|
do_test collate4-4.6 {
|
|
count {
|
|
SELECT max(a) FROM collate4t1;
|
|
}
|
|
} {20 3}
|
|
do_test collate4-4.7 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
}
|
|
} {}
|
|
|
|
# Also test the scalar min() and max() functions.
|
|
#
|
|
do_test collate4-4.8 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a COLLATE TEXT, b COLLATE NUMERIC);
|
|
INSERT INTO collate4t1 VALUES('11', '101');
|
|
INSERT INTO collate4t1 VALUES('101', '11')
|
|
}
|
|
} {}
|
|
do_test collate4-4.9 {
|
|
execsql {
|
|
SELECT max(a, b) FROM collate4t1;
|
|
}
|
|
} {11 11}
|
|
do_test collate4-4.10 {
|
|
execsql {
|
|
SELECT max(b, a) FROM collate4t1;
|
|
}
|
|
} {101 101}
|
|
do_test collate4-4.11 {
|
|
execsql {
|
|
SELECT max(a, '101') FROM collate4t1;
|
|
}
|
|
} {11 101}
|
|
do_test collate4-4.12 {
|
|
execsql {
|
|
SELECT max('101', a) FROM collate4t1;
|
|
}
|
|
} {11 101}
|
|
do_test collate4-4.13 {
|
|
execsql {
|
|
SELECT max(b, '101') FROM collate4t1;
|
|
}
|
|
} {101 101}
|
|
do_test collate4-4.14 {
|
|
execsql {
|
|
SELECT max('101', b) FROM collate4t1;
|
|
}
|
|
} {101 101}
|
|
|
|
do_test collate4-4.15 {
|
|
execsql {
|
|
DROP TABLE collate4t1;
|
|
}
|
|
} {}
|
|
|
|
#
|
|
# These tests - collate4.6.* - ensure that implict INTEGER PRIMARY KEY
|
|
# indices do not confuse collation sequences.
|
|
#
|
|
# These indices are never used for sorting in SQLite. And you can't
|
|
# create another index on an INTEGER PRIMARY KEY column, so we don't have
|
|
# to test that.
|
|
# (Revised 2004-Nov-22): The ROWID can be used for sorting now.
|
|
#
|
|
do_test collate4-6.0 {
|
|
execsql {
|
|
CREATE TABLE collate4t1(a INTEGER PRIMARY KEY);
|
|
INSERT INTO collate4t1 VALUES(101);
|
|
INSERT INTO collate4t1 VALUES(10);
|
|
INSERT INTO collate4t1 VALUES(15);
|
|
}
|
|
} {}
|
|
do_test collate4-6.1 {
|
|
cksort {
|
|
SELECT * FROM collate4t1 ORDER BY 1;
|
|
}
|
|
} {10 15 101 nosort}
|
|
do_test collate4-6.2 {
|
|
cksort {
|
|
SELECT * FROM collate4t1 ORDER BY oid;
|
|
}
|
|
} {10 15 101 nosort}
|
|
do_test collate4-6.3 {
|
|
cksort {
|
|
SELECT * FROM collate4t1 ORDER BY oid||'' COLLATE TEXT;
|
|
}
|
|
} {10 101 15 sort}
|
|
|
|
finish_test
|