sqlite/test/like.test
drh 9f504eab49 Fix a bug in the LIKE optimizer that occurs when the last character
before the wildcard is an upper-case 'Z'.  Ticket #2959. (CVS 4807)

FossilOrigin-Name: 610574b23b5e73b71be71df66e084c5bf37f6ccd
2008-02-23 21:55:39 +00:00

515 lines
11 KiB
Plaintext

# 2005 August 13
#
# 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 file is testing the LIKE and GLOB operators and
# in particular the optimizations that occur to help those operators
# run faster.
#
# $Id: like.test,v 1.9 2008/02/23 21:55:40 drh Exp $
set testdir [file dirname $argv0]
source $testdir/tester.tcl
# Create some sample data to work with.
#
do_test like-1.0 {
execsql {
CREATE TABLE t1(x TEXT);
}
foreach str {
a
ab
abc
abcd
acd
abd
bc
bcd
xyz
ABC
CDE
{ABC abc xyz}
} {
db eval {INSERT INTO t1 VALUES(:str)}
}
execsql {
SELECT count(*) FROM t1;
}
} {12}
# Test that both case sensitive and insensitive version of LIKE work.
#
do_test like-1.1 {
execsql {
SELECT x FROM t1 WHERE x LIKE 'abc' ORDER BY 1;
}
} {ABC abc}
do_test like-1.2 {
execsql {
SELECT x FROM t1 WHERE x GLOB 'abc' ORDER BY 1;
}
} {abc}
do_test like-1.3 {
execsql {
SELECT x FROM t1 WHERE x LIKE 'ABC' ORDER BY 1;
}
} {ABC abc}
do_test like-1.4 {
execsql {
SELECT x FROM t1 WHERE x LIKE 'aBc' ORDER BY 1;
}
} {ABC abc}
do_test like-1.5 {
execsql {
PRAGMA case_sensitive_like=on;
SELECT x FROM t1 WHERE x LIKE 'abc' ORDER BY 1;
}
} {abc}
do_test like-1.6 {
execsql {
SELECT x FROM t1 WHERE x GLOB 'abc' ORDER BY 1;
}
} {abc}
do_test like-1.7 {
execsql {
SELECT x FROM t1 WHERE x LIKE 'ABC' ORDER BY 1;
}
} {ABC}
do_test like-1.8 {
execsql {
SELECT x FROM t1 WHERE x LIKE 'aBc' ORDER BY 1;
}
} {}
do_test like-1.9 {
execsql {
PRAGMA case_sensitive_like=off;
SELECT x FROM t1 WHERE x LIKE 'abc' ORDER BY 1;
}
} {ABC abc}
# Tests of the REGEXP operator
#
do_test like-2.1 {
proc test_regexp {a b} {
return [regexp $a $b]
}
db function regexp test_regexp
execsql {
SELECT x FROM t1 WHERE x REGEXP 'abc' ORDER BY 1;
}
} {{ABC abc xyz} abc abcd}
do_test like-2.2 {
execsql {
SELECT x FROM t1 WHERE x REGEXP '^abc' ORDER BY 1;
}
} {abc abcd}
# Tests of the MATCH operator
#
do_test like-2.3 {
proc test_match {a b} {
return [string match $a $b]
}
db function match test_match
execsql {
SELECT x FROM t1 WHERE x MATCH '*abc*' ORDER BY 1;
}
} {{ABC abc xyz} abc abcd}
do_test like-2.4 {
execsql {
SELECT x FROM t1 WHERE x MATCH 'abc*' ORDER BY 1;
}
} {abc abcd}
# For the remaining tests, we need to have the like optimizations
# enabled.
#
ifcapable !like_opt {
finish_test
return
}
# This procedure executes the SQL. Then it appends to the result the
# "sort" or "nosort" keyword (as in the cksort procedure above) then
# it appends the ::sqlite_query_plan variable.
#
proc queryplan {sql} {
set ::sqlite_sort_count 0
set data [execsql $sql]
if {$::sqlite_sort_count} {set x sort} {set x nosort}
lappend data $x
return [concat $data $::sqlite_query_plan]
}
# Perform tests on the like optimization.
#
# With no index on t1.x and with case sensitivity turned off, no optimization
# is performed.
#
do_test like-3.1 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1;
}
} {ABC {ABC abc xyz} abc abcd sort t1 {}}
do_test like-3.2 {
set sqlite_like_count
} {12}
# With an index on t1.x and case sensitivity on, optimize completely.
#
do_test like-3.3 {
set sqlite_like_count 0
execsql {
PRAGMA case_sensitive_like=on;
CREATE INDEX i1 ON t1(x);
}
queryplan {
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1;
}
} {abc abcd nosort {} i1}
do_test like-3.4 {
set sqlite_like_count
} 0
# Partial optimization when the pattern does not end in '%'
#
do_test like-3.5 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'a_c' ORDER BY 1;
}
} {abc nosort {} i1}
do_test like-3.6 {
set sqlite_like_count
} 6
do_test like-3.7 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'ab%d' ORDER BY 1;
}
} {abcd abd nosort {} i1}
do_test like-3.8 {
set sqlite_like_count
} 4
do_test like-3.9 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'a_c%' ORDER BY 1;
}
} {abc abcd nosort {} i1}
do_test like-3.10 {
set sqlite_like_count
} 6
# No optimization when the pattern begins with a wildcard.
# Note that the index is still used but only for sorting.
#
do_test like-3.11 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE '%bcd' ORDER BY 1;
}
} {abcd bcd nosort {} i1}
do_test like-3.12 {
set sqlite_like_count
} 12
# No optimization for case insensitive LIKE
#
do_test like-3.13 {
set sqlite_like_count 0
queryplan {
PRAGMA case_sensitive_like=off;
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1;
}
} {ABC {ABC abc xyz} abc abcd nosort {} i1}
do_test like-3.14 {
set sqlite_like_count
} 12
# No optimization without an index.
#
do_test like-3.15 {
set sqlite_like_count 0
queryplan {
PRAGMA case_sensitive_like=on;
DROP INDEX i1;
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1;
}
} {abc abcd sort t1 {}}
do_test like-3.16 {
set sqlite_like_count
} 12
# No GLOB optimization without an index.
#
do_test like-3.17 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x GLOB 'abc*' ORDER BY 1;
}
} {abc abcd sort t1 {}}
do_test like-3.18 {
set sqlite_like_count
} 12
# GLOB is optimized regardless of the case_sensitive_like setting.
#
do_test like-3.19 {
set sqlite_like_count 0
queryplan {
CREATE INDEX i1 ON t1(x);
SELECT x FROM t1 WHERE x GLOB 'abc*' ORDER BY 1;
}
} {abc abcd nosort {} i1}
do_test like-3.20 {
set sqlite_like_count
} 0
do_test like-3.21 {
set sqlite_like_count 0
queryplan {
PRAGMA case_sensitive_like=on;
SELECT x FROM t1 WHERE x GLOB 'abc*' ORDER BY 1;
}
} {abc abcd nosort {} i1}
do_test like-3.22 {
set sqlite_like_count
} 0
do_test like-3.23 {
set sqlite_like_count 0
queryplan {
PRAGMA case_sensitive_like=off;
SELECT x FROM t1 WHERE x GLOB 'a[bc]d' ORDER BY 1;
}
} {abd acd nosort {} i1}
do_test like-3.24 {
set sqlite_like_count
} 6
# No optimization if the LHS of the LIKE is not a column name or
# if the RHS is not a string.
#
do_test like-4.1 {
execsql {PRAGMA case_sensitive_like=on}
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1
}
} {abc abcd nosort {} i1}
do_test like-4.2 {
set sqlite_like_count
} 0
do_test like-4.3 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE +x LIKE 'abc%' ORDER BY 1
}
} {abc abcd nosort {} i1}
do_test like-4.4 {
set sqlite_like_count
} 12
do_test like-4.5 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE ('ab' || 'c%') ORDER BY 1
}
} {abc abcd nosort {} i1}
do_test like-4.6 {
set sqlite_like_count
} 12
# Collating sequences on the index disable the LIKE optimization.
# Or if the NOCASE collating sequence is used, the LIKE optimization
# is enabled when case_sensitive_like is OFF.
#
do_test like-5.1 {
execsql {PRAGMA case_sensitive_like=off}
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'abc%' ORDER BY 1
}
} {ABC {ABC abc xyz} abc abcd nosort {} i1}
do_test like-5.2 {
set sqlite_like_count
} 12
do_test like-5.3 {
execsql {
CREATE TABLE t2(x COLLATE NOCASE);
INSERT INTO t2 SELECT * FROM t1;
CREATE INDEX i2 ON t2(x COLLATE NOCASE);
}
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x LIKE 'abc%' ORDER BY 1
}
} {abc ABC {ABC abc xyz} abcd nosort {} i2}
do_test like-5.4 {
set sqlite_like_count
} 0
do_test like-5.5 {
execsql {
PRAGMA case_sensitive_like=on;
}
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x LIKE 'abc%' ORDER BY 1
}
} {abc abcd nosort {} i2}
do_test like-5.6 {
set sqlite_like_count
} 12
do_test like-5.7 {
execsql {
PRAGMA case_sensitive_like=off;
}
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x GLOB 'abc*' ORDER BY 1
}
} {abc abcd nosort {} i2}
do_test like-5.8 {
set sqlite_like_count
} 12
do_test like-5.11 {
execsql {PRAGMA case_sensitive_like=off}
set sqlite_like_count 0
queryplan {
SELECT x FROM t1 WHERE x LIKE 'ABC%' ORDER BY 1
}
} {ABC {ABC abc xyz} abc abcd nosort {} i1}
do_test like-5.12 {
set sqlite_like_count
} 12
do_test like-5.13 {
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x LIKE 'ABC%' ORDER BY 1
}
} {abc ABC {ABC abc xyz} abcd nosort {} i2}
do_test like-5.14 {
set sqlite_like_count
} 0
do_test like-5.15 {
execsql {
PRAGMA case_sensitive_like=on;
}
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x LIKE 'ABC%' ORDER BY 1
}
} {ABC {ABC abc xyz} nosort {} i2}
do_test like-5.16 {
set sqlite_like_count
} 12
do_test like-5.17 {
execsql {
PRAGMA case_sensitive_like=off;
}
set sqlite_like_count 0
queryplan {
SELECT x FROM t2 WHERE x GLOB 'ABC*' ORDER BY 1
}
} {ABC {ABC abc xyz} nosort {} i2}
do_test like-5.18 {
set sqlite_like_count
} 12
# Boundary case. The prefix for a LIKE comparison is rounded up
# when constructing the comparison. Example: "ab" becomes "ac".
# In other words, the last character is increased by one.
#
# Make sure this happens correctly when the last character is a
# "z" and we are doing case-insensitive comparisons.
#
# Ticket #2959
#
do_test like-5.21 {
execsql {
PRAGMA case_sensitive_like=off;
INSERT INTO t2 VALUES('ZZ-upper-upper');
INSERT INTO t2 VALUES('zZ-lower-upper');
INSERT INTO t2 VALUES('Zz-upper-lower');
INSERT INTO t2 VALUES('zz-lower-lower');
}
queryplan {
SELECT x FROM t2 WHERE x LIKE 'zz%';
}
} {zz-lower-lower zZ-lower-upper Zz-upper-lower ZZ-upper-upper nosort {} i2}
do_test like-5.22 {
queryplan {
SELECT x FROM t2 WHERE x LIKE 'zZ%';
}
} {zz-lower-lower zZ-lower-upper Zz-upper-lower ZZ-upper-upper nosort {} i2}
do_test like-5.23 {
queryplan {
SELECT x FROM t2 WHERE x LIKE 'Zz%';
}
} {zz-lower-lower zZ-lower-upper Zz-upper-lower ZZ-upper-upper nosort {} i2}
do_test like-5.24 {
queryplan {
SELECT x FROM t2 WHERE x LIKE 'ZZ%';
}
} {zz-lower-lower zZ-lower-upper Zz-upper-lower ZZ-upper-upper nosort {} i2}
do_test like-5.25 {
queryplan {
PRAGMA case_sensitive_like=on;
CREATE TABLE t3(x);
CREATE INDEX i3 ON t3(x);
INSERT INTO t3 VALUES('ZZ-upper-upper');
INSERT INTO t3 VALUES('zZ-lower-upper');
INSERT INTO t3 VALUES('Zz-upper-lower');
INSERT INTO t3 VALUES('zz-lower-lower');
SELECT x FROM t3 WHERE x LIKE 'zz%';
}
} {zz-lower-lower nosort {} i3}
do_test like-5.26 {
queryplan {
SELECT x FROM t3 WHERE x LIKE 'zZ%';
}
} {zZ-lower-upper nosort {} i3}
do_test like-5.27 {
queryplan {
SELECT x FROM t3 WHERE x LIKE 'Zz%';
}
} {Zz-upper-lower nosort {} i3}
do_test like-5.28 {
queryplan {
SELECT x FROM t3 WHERE x LIKE 'ZZ%';
}
} {ZZ-upper-upper nosort {} i3}
# ticket #2407
#
# Make sure the LIKE prefix optimization does not strip off leading
# characters of the like pattern that happen to be quote characters.
#
do_test like-6.1 {
foreach x { 'abc 'bcd 'def 'ax } {
set x2 '[string map {' ''} $x]'
db eval "INSERT INTO t2 VALUES($x2)"
}
execsql {
SELECT * FROM t2 WHERE x LIKE '''a%'
}
} {'abc 'ax}
do_test like-7.1 {
execsql {
SELECT * FROM t1 WHERE rowid GLOB '1*';
}
} {a}
finish_test