Annotation of embedaddon/sqlite3/test/in3.test, revision 1.1.1.1

1.1       misho       1: # 2007 November 29
                      2: #
                      3: # The author disclaims copyright to this source code.  In place of
                      4: # a legal notice, here is a blessing:
                      5: #
                      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.
                      9: #
                     10: #***********************************************************************
                     11: # This file tests the optimisations made in November 2007 of expressions 
                     12: # of the following form:
                     13: #
                     14: #     <value> IN (SELECT <column> FROM <table>)
                     15: #
                     16: # $Id: in3.test,v 1.5 2008/08/04 03:51:24 danielk1977 Exp $
                     17: 
                     18: set testdir [file dirname $argv0]
                     19: source $testdir/tester.tcl
                     20: 
                     21: ifcapable !subquery {
                     22:   finish_test
                     23:   return
                     24: }
                     25: 
                     26: # Return the number of OpenEphemeral instructions used in the
                     27: # implementation of the sql statement passed as a an argument.
                     28: #
                     29: proc nEphemeral {sql} {
                     30:   set nEph 0
                     31:   foreach op [execsql "EXPLAIN $sql"] {
                     32:     if {$op eq "OpenEphemeral"} {incr nEph}
                     33:   }
                     34:   set nEph
                     35: }
                     36: 
                     37: # This proc works the same way as execsql, except that the number
                     38: # of OpenEphemeral instructions used in the implementation of the
                     39: # statement is inserted into the start of the returned list.
                     40: #
                     41: proc exec_neph {sql} {
                     42:   return [concat [nEphemeral $sql] [execsql $sql]]
                     43: }
                     44: 
                     45: do_test in3-1.1 {
                     46:   execsql {
                     47:     CREATE TABLE t1(a PRIMARY KEY, b);
                     48:     INSERT INTO t1 VALUES(1, 2);
                     49:     INSERT INTO t1 VALUES(3, 4);
                     50:     INSERT INTO t1 VALUES(5, 6);
                     51:   }
                     52: } {}
                     53: 
                     54: # All of these queries should avoid using a temp-table:
                     55: #
                     56: do_test in3-1.2 {
                     57:   exec_neph { SELECT rowid FROM t1 WHERE rowid IN (SELECT rowid FROM t1); }
                     58: } {0 1 2 3}
                     59: do_test in3-1.3 {
                     60:   exec_neph { SELECT a FROM t1 WHERE a IN (SELECT a FROM t1); }
                     61: } {0 1 3 5}
                     62: do_test in3-1.4 {
                     63:   exec_neph { SELECT rowid FROM t1 WHERE rowid+0 IN (SELECT rowid FROM t1); }
                     64: } {0 1 2 3}
                     65: do_test in3-1.5 {
                     66:   exec_neph { SELECT a FROM t1 WHERE a+0 IN (SELECT a FROM t1); }
                     67: } {0 1 3 5}
                     68: 
                     69: # Because none of the sub-select queries in the following statements
                     70: # match the pattern ("SELECT <column> FROM <table>"), the following do 
                     71: # require a temp table.
                     72: #
                     73: do_test in3-1.6 {
                     74:   exec_neph { SELECT rowid FROM t1 WHERE rowid IN (SELECT rowid+0 FROM t1); }
                     75: } {1 1 2 3}
                     76: do_test in3-1.7 {
                     77:   exec_neph { SELECT a FROM t1 WHERE a IN (SELECT a+0 FROM t1); }
                     78: } {1 1 3 5}
                     79: do_test in3-1.8 {
                     80:   exec_neph { SELECT a FROM t1 WHERE a IN (SELECT a FROM t1 WHERE 1); }
                     81: } {1 1 3 5}
                     82: do_test in3-1.9 {
                     83:   exec_neph { SELECT a FROM t1 WHERE a IN (SELECT a FROM t1 GROUP BY a); }
                     84: } {1 1 3 5}
                     85: 
                     86: # This should not use a temp-table. Even though the sub-select does
                     87: # not exactly match the pattern "SELECT <column> FROM <table>", in
                     88: # this case the ORDER BY is a no-op and can be ignored.
                     89: do_test in3-1.10 {
                     90:   exec_neph { SELECT a FROM t1 WHERE a IN (SELECT a FROM t1 ORDER BY a); }
                     91: } {0 1 3 5}
                     92: 
                     93: # These do use the temp-table. Adding the LIMIT clause means the 
                     94: # ORDER BY cannot be ignored.
                     95: do_test in3-1.11 {
                     96:   exec_neph {SELECT a FROM t1 WHERE a IN (SELECT a FROM t1 ORDER BY a LIMIT 1)}
                     97: } {1 1}
                     98: do_test in3-1.12 {
                     99:   exec_neph {
                    100:     SELECT a FROM t1 WHERE a IN (SELECT a FROM t1 ORDER BY a LIMIT 1 OFFSET 1)
                    101:   }
                    102: } {1 3}
                    103: 
                    104: # Has to use a temp-table because of the compound sub-select.
                    105: #
                    106: ifcapable compound {
                    107:   do_test in3-1.13 {
                    108:     exec_neph {
                    109:       SELECT a FROM t1 WHERE a IN (
                    110:         SELECT a FROM t1 UNION ALL SELECT a FROM t1
                    111:       )
                    112:     }
                    113:   } {1 1 3 5}
                    114: }
                    115: 
                    116: # The first of these queries has to use the temp-table, because the 
                    117: # collation sequence used for the index on "t1.a" does not match the
                    118: # collation sequence used by the "IN" comparison. The second does not
                    119: # require a temp-table, because the collation sequences match.
                    120: #
                    121: do_test in3-1.14 {
                    122:   exec_neph { SELECT a FROM t1 WHERE a COLLATE nocase IN (SELECT a FROM t1) }
                    123: } {1 1 3 5}
                    124: do_test in3-1.15 {
                    125:   exec_neph { SELECT a FROM t1 WHERE a COLLATE binary IN (SELECT a FROM t1) }
                    126: } {0 1 3 5}
                    127: 
                    128: # Neither of these queries require a temp-table. The collation sequence
                    129: # makes no difference when using a rowid.
                    130: #
                    131: do_test in3-1.16 {
                    132:   exec_neph {SELECT a FROM t1 WHERE a COLLATE nocase IN (SELECT rowid FROM t1)}
                    133: } {0 1 3}
                    134: do_test in3-1.17 {
                    135:   exec_neph {SELECT a FROM t1 WHERE a COLLATE binary IN (SELECT rowid FROM t1)}
                    136: } {0 1 3}
                    137: 
                    138: # The following tests - in3.2.* - test a bug that was difficult to track
                    139: # down during development. They are not particularly well focused.
                    140: #
                    141: do_test in3-2.1 {
                    142:   execsql {
                    143:     DROP TABLE IF EXISTS t1;
                    144:     CREATE TABLE t1(w int, x int, y int);
                    145:     CREATE TABLE t2(p int, q int, r int, s int);
                    146:   }
                    147:   for {set i 1} {$i<=100} {incr i} {
                    148:     set w $i
                    149:     set x [expr {int(log($i)/log(2))}]
                    150:     set y [expr {$i*$i + 2*$i + 1}]
                    151:     execsql "INSERT INTO t1 VALUES($w,$x,$y)"
                    152:   }
                    153:   set maxy [execsql {select max(y) from t1}]
                    154:   db eval { INSERT INTO t2 SELECT 101-w, x, $maxy+1-y, y FROM t1 }
                    155: } {}
                    156: do_test in3-2.2 {
                    157:   execsql {
                    158:     SELECT rowid 
                    159:     FROM t1 
                    160:     WHERE rowid IN (SELECT rowid FROM t1 WHERE rowid IN (1, 2));
                    161:   }
                    162: } {1 2}
                    163: do_test in3-2.3 {
                    164:   execsql {
                    165:     select rowid from t1 where rowid IN (-1,2,4)
                    166:   }
                    167: } {2 4}
                    168: do_test in3-2.4 {
                    169:   execsql {
                    170:     SELECT rowid FROM t1 WHERE rowid IN 
                    171:        (select rowid from t1 where rowid IN (-1,2,4))
                    172:   }
                    173: } {2 4}
                    174: 
                    175: #-------------------------------------------------------------------------
                    176: # This next block of tests - in3-3.* - verify that column affinity is
                    177: # correctly handled in cases where an index might be used to optimise
                    178: # an IN (SELECT) expression.
                    179: #
                    180: do_test in3-3.1 {
                    181:   catch {execsql {
                    182:     DROP TABLE t1;
                    183:     DROP TABLE t2;
                    184:   }}
                    185: 
                    186:   execsql {
                    187: 
                    188:     CREATE TABLE t1(a BLOB, b NUMBER ,c TEXT);
                    189:     CREATE UNIQUE INDEX t1_i1 ON t1(a);        /* no affinity */
                    190:     CREATE UNIQUE INDEX t1_i2 ON t1(b);        /* numeric affinity */
                    191:     CREATE UNIQUE INDEX t1_i3 ON t1(c);        /* text affinity */
                    192: 
                    193:     CREATE TABLE t2(x BLOB, y NUMBER, z TEXT);
                    194:     CREATE UNIQUE INDEX t2_i1 ON t2(x);        /* no affinity */
                    195:     CREATE UNIQUE INDEX t2_i2 ON t2(y);        /* numeric affinity */
                    196:     CREATE UNIQUE INDEX t2_i3 ON t2(z);        /* text affinity */
                    197: 
                    198:     INSERT INTO t1 VALUES(1, 1, 1);
                    199:     INSERT INTO t2 VALUES('1', '1', '1');
                    200:   }
                    201: } {}
                    202: 
                    203: do_test in3-3.2 {
                    204:   # No affinity is applied before comparing "x" and "a". Therefore
                    205:   # the index can be used (the comparison is false, text!=number).
                    206:   exec_neph { SELECT x IN (SELECT a FROM t1) FROM t2 }
                    207: } {0 0}
                    208: do_test in3-3.3 {
                    209:   # Logically, numeric affinity is applied to both sides before 
                    210:   # the comparison.  Therefore it is possible to use index t1_i2.
                    211:   exec_neph { SELECT x IN (SELECT b FROM t1) FROM t2 }
                    212: } {0 1}
                    213: do_test in3-3.4 {
                    214:   # No affinity is applied before the comparison takes place. Making
                    215:   # it possible to use index t1_i3.
                    216:   exec_neph { SELECT x IN (SELECT c FROM t1) FROM t2 }
                    217: } {0 1}
                    218: 
                    219: do_test in3-3.5 {
                    220:   # Numeric affinity should be applied to each side before the comparison
                    221:   # takes place. Therefore we cannot use index t1_i1, which has no affinity.
                    222:   exec_neph { SELECT y IN (SELECT a FROM t1) FROM t2 }
                    223: } {1 1}
                    224: do_test in3-3.6 {
                    225:   # Numeric affinity is applied to both sides before 
                    226:   # the comparison.  Therefore it is possible to use index t1_i2.
                    227:   exec_neph { SELECT y IN (SELECT b FROM t1) FROM t2 }
                    228: } {0 1}
                    229: do_test in3-3.7 {
                    230:   # Numeric affinity is applied before the comparison takes place. 
                    231:   # Making it impossible to use index t1_i3.
                    232:   exec_neph { SELECT y IN (SELECT c FROM t1) FROM t2 }
                    233: } {1 1}
                    234: 
                    235: #---------------------------------------------------------------------
                    236: #
                    237: # Test using a multi-column index.
                    238: #
                    239: do_test in3-4.1 {
                    240:   execsql {
                    241:     CREATE TABLE t3(a, b, c);
                    242:     CREATE UNIQUE INDEX t3_i ON t3(b, a);
                    243:   }
                    244: 
                    245:   execsql {
                    246:     INSERT INTO t3 VALUES(1, 'numeric', 2);
                    247:     INSERT INTO t3 VALUES(2, 'text', 2);
                    248:     INSERT INTO t3 VALUES(3, 'real', 2);
                    249:     INSERT INTO t3 VALUES(4, 'none', 2);
                    250:   }
                    251: } {}
                    252: do_test in3-4.2 {
                    253:   exec_neph { SELECT 'text' IN (SELECT b FROM t3) }
                    254: } {0 1}
                    255: do_test in3-4.3 {
                    256:   exec_neph { SELECT 'TEXT' COLLATE nocase IN (SELECT b FROM t3) }
                    257: } {1 1}
                    258: do_test in3-4.4 {
                    259:   # A temp table must be used because t3_i.b is not guaranteed to be unique.
                    260:   exec_neph { SELECT b FROM t3 WHERE b IN (SELECT b FROM t3) }
                    261: } {1 none numeric real text}
                    262: do_test in3-4.5 {
                    263:   execsql { CREATE UNIQUE INDEX t3_i2 ON t3(b) }
                    264:   exec_neph { SELECT b FROM t3 WHERE b IN (SELECT b FROM t3) }
                    265: } {0 none numeric real text}
                    266: do_test in3-4.6 {
                    267:   execsql { DROP INDEX t3_i2 }
                    268: } {}
                    269: 
                    270: # The following two test cases verify that ticket #2991 has been fixed.
                    271: #
                    272: do_test in3-5.1 {
                    273:   execsql {
                    274:     CREATE TABLE Folders(
                    275:       folderid INTEGER PRIMARY KEY, 
                    276:       parentid INTEGER, 
                    277:       rootid INTEGER, 
                    278:       path VARCHAR(255)
                    279:     );
                    280:   }
                    281: } {}
                    282: do_test in3-5.2 {
                    283:   catchsql {
                    284:     DELETE FROM Folders WHERE folderid IN
                    285:     (SELECT folderid FROM Folder WHERE path LIKE 'C:\MP3\Albums\' || '%');
                    286:   }
                    287: } {1 {no such table: Folder}}
                    288: 
                    289: finish_test

FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>