File:  [ELWIX - Embedded LightWeight unIX -] / embedaddon / sqlite3 / test / misc1.test
Revision 1.1.1.1 (vendor branch): download - view: text, annotated - select for diffs - revision graph
Tue Feb 21 17:04:16 2012 UTC (12 years, 8 months ago) by misho
Branches: sqlite3, MAIN
CVS tags: v3_7_10, HEAD
sqlite3

# 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.
#
# This file implements tests for miscellanous features that were
# left out of other test files.
#
# $Id: misc1.test,v 1.1.1.1 2012/02/21 17:04:16 misho Exp $

set testdir [file dirname $argv0]
source $testdir/tester.tcl

# 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]
}

# Mimic the SQLite 2 collation type TEXT.
db collate text text_collate
proc numeric_collate {lhs rhs} {
  return [string compare $lhs $rhs]
}

# Test the creation and use of tables that have a large number
# of columns.
#
do_test misc1-1.1 {
  set cmd "CREATE TABLE manycol(x0 text"
  for {set i 1} {$i<=99} {incr i} {
    append cmd ",x$i text"
  }
  append cmd ")";
  execsql $cmd
  set cmd "INSERT INTO manycol VALUES(0"
  for {set i 1} {$i<=99} {incr i} {
    append cmd ",$i"
  }
  append cmd ")";
  execsql $cmd
  execsql "SELECT x99 FROM manycol"
} 99
do_test misc1-1.2 {
  execsql {SELECT x0, x10, x25, x50, x75 FROM manycol}
} {0 10 25 50 75}
do_test misc1-1.3.1 {
  for {set j 100} {$j<=1000} {incr j 100} {
    set cmd "INSERT INTO manycol VALUES($j"
    for {set i 1} {$i<=99} {incr i} {
      append cmd ",[expr {$i+$j}]"
    }
    append cmd ")"
    execsql $cmd
  }
  execsql {SELECT x50 FROM manycol ORDER BY x80+0}
} {50 150 250 350 450 550 650 750 850 950 1050}
do_test misc1-1.3.2 {
  execsql {SELECT x50 FROM manycol ORDER BY x80}
} {1050 150 250 350 450 550 650 750 50 850 950}
do_test misc1-1.4 {
  execsql {SELECT x75 FROM manycol WHERE x50=350}
} 375
do_test misc1-1.5 {
  execsql {SELECT x50 FROM manycol WHERE x99=599}
} 550
do_test misc1-1.6 {
  execsql {CREATE INDEX manycol_idx1 ON manycol(x99)}
  execsql {SELECT x50 FROM manycol WHERE x99=899}
} 850
do_test misc1-1.7 {
  execsql {SELECT count(*) FROM manycol}
} 11
do_test misc1-1.8 {
  execsql {DELETE FROM manycol WHERE x98=1234}
  execsql {SELECT count(*) FROM manycol}
} 11
do_test misc1-1.9 {
  execsql {DELETE FROM manycol WHERE x98=998}
  execsql {SELECT count(*) FROM manycol}
} 10
do_test misc1-1.10 {
  execsql {DELETE FROM manycol WHERE x99=500}
  execsql {SELECT count(*) FROM manycol}
} 10
do_test misc1-1.11 {
  execsql {DELETE FROM manycol WHERE x99=599}
  execsql {SELECT count(*) FROM manycol}
} 9

# Check GROUP BY expressions that name two or more columns.
#
do_test misc1-2.1 {
  execsql {
    BEGIN TRANSACTION;
    CREATE TABLE agger(one text, two text, three text, four text);
    INSERT INTO agger VALUES(1, 'one', 'hello', 'yes');
    INSERT INTO agger VALUES(2, 'two', 'howdy', 'no');
    INSERT INTO agger VALUES(3, 'thr', 'howareya', 'yes');
    INSERT INTO agger VALUES(4, 'two', 'lothere', 'yes');
    INSERT INTO agger VALUES(5, 'one', 'atcha', 'yes');
    INSERT INTO agger VALUES(6, 'two', 'hello', 'no');
    COMMIT
  }
  execsql {SELECT count(*) FROM agger}
} 6
do_test misc1-2.2 {
  execsql {SELECT sum(one), two, four FROM agger
           GROUP BY two, four ORDER BY sum(one) desc}
} {8 two no 6 one yes 4 two yes 3 thr yes}
do_test misc1-2.3 {
  execsql {SELECT sum((one)), (two), (four) FROM agger
           GROUP BY (two), (four) ORDER BY sum(one) desc}
} {8 two no 6 one yes 4 two yes 3 thr yes}

# Here's a test for a bug found by Joel Lucsy.  The code below
# was causing an assertion failure.
#
do_test misc1-3.1 {
  set r [execsql {
    CREATE TABLE t1(a);
    INSERT INTO t1 VALUES('hi');
    PRAGMA full_column_names=on;
    SELECT rowid, * FROM t1;
  }]
  lindex $r 1
} {hi}

# Here's a test for yet another bug found by Joel Lucsy.  The code
# below was causing an assertion failure.
#
do_test misc1-4.1 {
  execsql {
    BEGIN;
    CREATE TABLE t2(a);
    INSERT INTO t2 VALUES('This is a long string to use up a lot of disk -');
    UPDATE t2 SET a=a||a||a||a;
    INSERT INTO t2 SELECT '1 - ' || a FROM t2;
    INSERT INTO t2 SELECT '2 - ' || a FROM t2;
    INSERT INTO t2 SELECT '3 - ' || a FROM t2;
    INSERT INTO t2 SELECT '4 - ' || a FROM t2;
    INSERT INTO t2 SELECT '5 - ' || a FROM t2;
    INSERT INTO t2 SELECT '6 - ' || a FROM t2;
    COMMIT;
    SELECT count(*) FROM t2;
  }
} {64}

# Make sure we actually see a semicolon or end-of-file in the SQL input
# before executing a command.  Thus if "WHERE" is misspelled on an UPDATE,
# the user won't accidently update every record.
#
do_test misc1-5.1 {
  catchsql {
    CREATE TABLE t3(a,b);
    INSERT INTO t3 VALUES(1,2);
    INSERT INTO t3 VALUES(3,4);
    UPDATE t3 SET a=0 WHEREwww b=2;
  }
} {1 {near "WHEREwww": syntax error}}
do_test misc1-5.2 {
  execsql {
    SELECT * FROM t3 ORDER BY a;
  }
} {1 2 3 4}

# Certain keywords (especially non-standard keywords like "REPLACE") can
# also be used as identifiers.  The way this works in the parser is that
# the parser first detects a syntax error, the error handling routine
# sees that the special keyword caused the error, then replaces the keyword
# with "ID" and tries again.
#
# Check the operation of this logic.
#
do_test misc1-6.1 {
  catchsql {
    CREATE TABLE t4(
      abort, asc, begin, cluster, conflict, copy, delimiters, desc, end,
      explain, fail, ignore, key, offset, pragma, replace, temp,
      vacuum, view
    );
  }
} {0 {}}
do_test misc1-6.2 {
  catchsql {
    INSERT INTO t4
       VALUES(1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19);
  }
} {0 {}}
do_test misc1-6.3 {
  execsql {
    SELECT * FROM t4
  }
} {1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19}
do_test misc1-6.4 {
  execsql {
    SELECT abort+asc,max(key,pragma,temp) FROM t4
  }
} {3 17}

# Test for multi-column primary keys, and for multiple primary keys.
#
do_test misc1-7.1 {
  catchsql {
    CREATE TABLE error1(
      a TYPE PRIMARY KEY,
      b TYPE PRIMARY KEY
    );
  }
} {1 {table "error1" has more than one primary key}}
do_test misc1-7.2 {
  catchsql {
    CREATE TABLE error1(
      a INTEGER PRIMARY KEY,
      b TYPE PRIMARY KEY
    );
  }
} {1 {table "error1" has more than one primary key}}
do_test misc1-7.3 {
  execsql {
    CREATE TABLE t5(a,b,c,PRIMARY KEY(a,b));
    INSERT INTO t5 VALUES(1,2,3);
    SELECT * FROM t5 ORDER BY a;
  }
} {1 2 3}
do_test misc1-7.4 {
  catchsql {
    INSERT INTO t5 VALUES(1,2,4);
  }
} {1 {columns a, b are not unique}}
do_test misc1-7.5 {
  catchsql {
    INSERT INTO t5 VALUES(0,2,4);
  }
} {0 {}}
do_test misc1-7.6 {
  execsql {
    SELECT * FROM t5 ORDER BY a;
  }
} {0 2 4 1 2 3}

do_test misc1-8.1 {
  catchsql {
    SELECT *;
  }
} {1 {no tables specified}}
do_test misc1-8.2 {
  catchsql {
    SELECT t1.*;
  }
} {1 {no such table: t1}}

execsql {
  DROP TABLE t1;
  DROP TABLE t2;
  DROP TABLE t3;
  DROP TABLE t4;
}

# 64-bit integers are represented exactly.
#
do_test misc1-9.1 {
  catchsql {
    CREATE TABLE t1(a unique not null, b unique not null);
    INSERT INTO t1 VALUES('a',1234567890123456789);
    INSERT INTO t1 VALUES('b',1234567891123456789);
    INSERT INTO t1 VALUES('c',1234567892123456789);
    SELECT * FROM t1;
  }
} {0 {a 1234567890123456789 b 1234567891123456789 c 1234567892123456789}}

# A WHERE clause is not allowed to contain more than 99 terms.  Check to
# make sure this limit is enforced.
#
# 2005-07-16: There is no longer a limit on the number of terms in a
# WHERE clause.  But keep these tests just so that we have some tests
# that use a large number of terms in the WHERE clause.
#
do_test misc1-10.0 {
  execsql {SELECT count(*) FROM manycol}
} {9}
do_test misc1-10.1 {
  set ::where {WHERE x0>=0}
  for {set i 1} {$i<=99} {incr i} {
    append ::where " AND x$i<>0"
  }
  catchsql "SELECT count(*) FROM manycol $::where"
} {0 9}
do_test misc1-10.2 {
  catchsql "SELECT count(*) FROM manycol $::where AND rowid>0"
} {0 9}
do_test misc1-10.3 {
  regsub "x0>=0" $::where "x0=0" ::where
  catchsql "DELETE FROM manycol $::where"
} {0 {}}
do_test misc1-10.4 {
  execsql {SELECT count(*) FROM manycol}
} {8}
do_test misc1-10.5 {
  catchsql "DELETE FROM manycol $::where AND rowid>0"
} {0 {}}
do_test misc1-10.6 {
  execsql {SELECT x1 FROM manycol WHERE x0=100}
} {101}
do_test misc1-10.7 {
  regsub "x0=0" $::where "x0=100" ::where
  catchsql "UPDATE manycol SET x1=x1+1 $::where"
} {0 {}}
do_test misc1-10.8 {
  execsql {SELECT x1 FROM manycol WHERE x0=100}
} {102}
do_test misc1-10.9 {
  catchsql "UPDATE manycol SET x1=x1+1 $::where AND rowid>0"
} {0 {}}
do_test misc1-10.10 {
  execsql {SELECT x1 FROM manycol WHERE x0=100}
} {103}

# Make sure the initialization works even if a database is opened while
# another process has the database locked.
#
# Update for v3: The BEGIN doesn't lock the database so the schema is read
# and the SELECT returns successfully.
do_test misc1-11.1 {
  execsql {BEGIN}
  execsql {UPDATE t1 SET a=0 WHERE 0}
  sqlite3 db2 test.db
  set rc [catch {db2 eval {SELECT count(*) FROM t1}} msg]
  lappend rc $msg
# v2 result: {1 {database is locked}}
} {0 3}
do_test misc1-11.2 {
  execsql {COMMIT}
  set rc [catch {db2 eval {SELECT count(*) FROM t1}} msg]
  db2 close
  lappend rc $msg
} {0 3}

# Make sure string comparisons really do compare strings in format4+.
# Similar tests in the format3.test file show that for format3 and earlier
# all comparisions where numeric if either operand looked like a number.
#
do_test misc1-12.1 {
  execsql {SELECT '0'=='0.0'}
} {0}
do_test misc1-12.2 {
  execsql {SELECT '0'==0.0}
} {0}
do_test misc1-12.3 {
  execsql {SELECT '12345678901234567890'=='12345678901234567891'}
} {0}
do_test misc1-12.4 {
  execsql {
    CREATE TABLE t6(a INT UNIQUE, b TEXT UNIQUE);
    INSERT INTO t6 VALUES('0','0.0');
    SELECT * FROM t6;
  }
} {0 0.0}
ifcapable conflict {
  do_test misc1-12.5 {
    execsql {
      INSERT OR IGNORE INTO t6 VALUES(0.0,'x');
      SELECT * FROM t6;
    }
  } {0 0.0}
  do_test misc1-12.6 {
    execsql {
      INSERT OR IGNORE INTO t6 VALUES('y',0);
      SELECT * FROM t6;
    }
  } {0 0.0 y 0}
}
do_test misc1-12.7 {
  execsql {
    CREATE TABLE t7(x INTEGER, y TEXT, z);
    INSERT INTO t7 VALUES(0,0,1);
    INSERT INTO t7 VALUES(0.0,0,2);
    INSERT INTO t7 VALUES(0,0.0,3);
    INSERT INTO t7 VALUES(0.0,0.0,4);
    SELECT DISTINCT x, y FROM t7 ORDER BY z;
  }
} {0 0 0 0.0}
do_test misc1-12.8 {
  execsql {
    SELECT min(z), max(z), count(z) FROM t7 GROUP BY x ORDER BY 1;
  }
} {1 4 4}
do_test misc1-12.9 {
  execsql {
    SELECT min(z), max(z), count(z) FROM t7 GROUP BY y ORDER BY 1;
  }
} {1 2 2 3 4 2}

# This used to be an error.  But we changed the code so that arbitrary
# identifiers can be used as a collating sequence.  Collation is by text
# if the identifier contains "text", "blob", or "clob" and is numeric
# otherwise.
#
# Update: In v3, it is an error again.
#
#do_test misc1-12.10 {
#  catchsql {
#    SELECT * FROM t6 ORDER BY a COLLATE unknown;
#  }
#} {0 {0 0 y 0}}
do_test misc1-12.11 {
  execsql {
    CREATE TABLE t8(x TEXT COLLATE numeric, y INTEGER COLLATE text, z);
    INSERT INTO t8 VALUES(0,0,1);
    INSERT INTO t8 VALUES(0.0,0,2);
    INSERT INTO t8 VALUES(0,0.0,3);
    INSERT INTO t8 VALUES(0.0,0.0,4);
    SELECT DISTINCT x, y FROM t8 ORDER BY z;
  }
} {0 0 0.0 0}
do_test misc1-12.12 {
  execsql {
    SELECT min(z), max(z), count(z) FROM t8 GROUP BY x ORDER BY 1;
  }
} {1 3 2 2 4 2}
do_test misc1-12.13 {
  execsql {
    SELECT min(z), max(z), count(z) FROM t8 GROUP BY y ORDER BY 1;
  }
} {1 4 4}

# There was a problem with realloc() in the OP_MemStore operation of
# the VDBE.  A buffer was being reallocated but some pointers into 
# the old copy of the buffer were not being moved over to the new copy.
# The following code tests for the problem.
#
ifcapable subquery {
  do_test misc1-13.1 {
     execsql {
       CREATE TABLE t9(x,y);
       INSERT INTO t9 VALUES('one',1);
       INSERT INTO t9 VALUES('two',2);
       INSERT INTO t9 VALUES('three',3);
       INSERT INTO t9 VALUES('four',4);
       INSERT INTO t9 VALUES('five',5);
       INSERT INTO t9 VALUES('six',6);
       INSERT INTO t9 VALUES('seven',7);
       INSERT INTO t9 VALUES('eight',8);
       INSERT INTO t9 VALUES('nine',9);
       INSERT INTO t9 VALUES('ten',10);
       INSERT INTO t9 VALUES('eleven',11);
       SELECT y FROM t9
       WHERE x=(SELECT x FROM t9 WHERE y=1)
          OR x=(SELECT x FROM t9 WHERE y=2)
          OR x=(SELECT x FROM t9 WHERE y=3)
          OR x=(SELECT x FROM t9 WHERE y=4)
          OR x=(SELECT x FROM t9 WHERE y=5)
          OR x=(SELECT x FROM t9 WHERE y=6)
          OR x=(SELECT x FROM t9 WHERE y=7)
          OR x=(SELECT x FROM t9 WHERE y=8)
          OR x=(SELECT x FROM t9 WHERE y=9)
          OR x=(SELECT x FROM t9 WHERE y=10)
          OR x=(SELECT x FROM t9 WHERE y=11)
          OR x=(SELECT x FROM t9 WHERE y=12)
          OR x=(SELECT x FROM t9 WHERE y=13)
          OR x=(SELECT x FROM t9 WHERE y=14)
       ;
     }
  } {1 2 3 4 5 6 7 8 9 10 11}
}

# Make sure a database connection still works after changing the
# working directory.
#
do_test misc1-14.1 {
  file mkdir tempdir
  cd tempdir
  execsql {BEGIN}
  file exists ./test.db-journal
} {0}
do_test misc1-14.2a {
  execsql {UPDATE t1 SET a=a||'x' WHERE 0}
  file exists ../test.db-journal
} {0}
do_test misc1-14.2b {
  execsql {UPDATE t1 SET a=a||'y' WHERE 1}
  file exists ../test.db-journal
} {1}
do_test misc1-14.3 {
  cd ..
  forcedelete tempdir
  execsql {COMMIT}
  file exists ./test.db-journal
} {0}

# A failed create table should not leave the table in the internal
# data structures.  Ticket #238.
#
do_test misc1-15.1.1 {
  catchsql {
    CREATE TABLE t10 AS SELECT c1;
  }
} {1 {no such column: c1}}
do_test misc1-15.1.2 {
  catchsql {
    CREATE TABLE t10 AS SELECT t9.c1;
  }
} {1 {no such column: t9.c1}}
do_test misc1-15.1.3 {
  catchsql {
    CREATE TABLE t10 AS SELECT main.t9.c1;
  }
} {1 {no such column: main.t9.c1}}
do_test misc1-15.2 {
  catchsql {
    CREATE TABLE t10 AS SELECT 1;
  }
  # The bug in ticket #238 causes the statement above to fail with
  # the error "table t10 alread exists"
} {0 {}}

# Test for memory leaks when a CREATE TABLE containing a primary key
# fails.  Ticket #249.
#
do_test misc1-16.1 {
  catchsql {SELECT name FROM sqlite_master LIMIT 1}
  catchsql {
    CREATE TABLE test(a integer, primary key(a));
  }
} {0 {}}
do_test misc1-16.2 {
  catchsql {
    CREATE TABLE test(a integer, primary key(a));
  }
} {1 {table test already exists}}
do_test misc1-16.3 {
  catchsql {
    CREATE TABLE test2(a text primary key, b text, primary key(a,b));
  }
} {1 {table "test2" has more than one primary key}}
do_test misc1-16.4 {
  execsql {
    INSERT INTO test VALUES(1);
    SELECT rowid, a FROM test;
  }
} {1 1}
do_test misc1-16.5 {
  execsql {
    INSERT INTO test VALUES(5);
    SELECT rowid, a FROM test;
  }
} {1 1 5 5}
do_test misc1-16.6 {
  execsql {
    INSERT INTO test VALUES(NULL);
    SELECT rowid, a FROM test;
  }
} {1 1 5 5 6 6}

ifcapable trigger&&tempdb {
# Ticket #333: Temp triggers that modify persistent tables.
#
do_test misc1-17.1 {
  execsql {
    BEGIN;
    CREATE TABLE RealTable(TestID INTEGER PRIMARY KEY, TestString TEXT);
    CREATE TEMP TABLE TempTable(TestID INTEGER PRIMARY KEY, TestString TEXT);
    CREATE TEMP TRIGGER trigTest_1 AFTER UPDATE ON TempTable BEGIN
      INSERT INTO RealTable(TestString) 
         SELECT new.TestString FROM TempTable LIMIT 1;
    END;
    INSERT INTO TempTable(TestString) VALUES ('1');
    INSERT INTO TempTable(TestString) VALUES ('2');
    UPDATE TempTable SET TestString = TestString + 1 WHERE TestID=1 OR TestId=2;
    COMMIT;
    SELECT TestString FROM RealTable ORDER BY 1;
  }
} {2 3}
}

do_test misc1-18.1 {
  set n [sqlite3_sleep 100]
  expr {$n>=100}
} {1}

finish_test

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