sl@0: # 2008 June 26
sl@0: #
sl@0: # The author disclaims copyright to this source code.  In place of
sl@0: # a legal notice, here is a blessing:
sl@0: #
sl@0: #    May you do good and not evil.
sl@0: #    May you find forgiveness for yourself and forgive others.
sl@0: #    May you share freely, never taking more than you give.
sl@0: #
sl@0: #*************************************************************************
sl@0: # This file exercises some new testing functions in the FTS2 module,
sl@0: # and then uses them to do some basic tests that FTS2 is internally
sl@0: # working as expected.
sl@0: #
sl@0: # $Id: fts2p.test,v 1.1 2008/07/22 23:32:28 shess Exp $
sl@0: #
sl@0: 
sl@0: set testdir [file dirname $argv0]
sl@0: source $testdir/tester.tcl
sl@0: 
sl@0: # If SQLITE_ENABLE_FTS2 is not defined, omit this file.
sl@0: ifcapable !fts2 {
sl@0:   finish_test
sl@0:   return
sl@0: }
sl@0: 
sl@0: #*************************************************************************
sl@0: # Probe to see if support for these functions is compiled in.
sl@0: # TODO(shess): Change main.mk to do the right thing and remove this test.
sl@0: db eval {
sl@0:   DROP TABLE IF EXISTS t1;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) VALUES (1, 'x');
sl@0: }
sl@0: 
sl@0: set s {SELECT dump_terms(t1, 1) FROM t1 LIMIT 1}
sl@0: set r {1 {unable to use function dump_terms in the requested context}}
sl@0: if {[catchsql $s]==$r} {
sl@0:   finish_test
sl@0:   return
sl@0: }
sl@0: 
sl@0: #*************************************************************************
sl@0: # Test that the new functions give appropriate errors.
sl@0: do_test fts2p-0.0 {
sl@0:   catchsql {
sl@0:     SELECT dump_terms(t1, 1) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_terms: incorrect arguments}}
sl@0: 
sl@0: do_test fts2p-0.1 {
sl@0:   catchsql {
sl@0:     SELECT dump_terms(t1, 0, 0, 0) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_terms: incorrect arguments}}
sl@0: 
sl@0: do_test fts2p-0.2 {
sl@0:   catchsql {
sl@0:     SELECT dump_terms(1, t1) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {unable to use function dump_terms in the requested context}}
sl@0: 
sl@0: do_test fts2p-0.3 {
sl@0:   catchsql {
sl@0:     SELECT dump_terms(t1, 16, 16) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_terms: segment not found}}
sl@0: 
sl@0: do_test fts2p-0.4 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: incorrect arguments}}
sl@0: 
sl@0: do_test fts2p-0.5 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1, NULL) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: empty second argument}}
sl@0: 
sl@0: do_test fts2p-0.6 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1, '') FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: empty second argument}}
sl@0: 
sl@0: do_test fts2p-0.7 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1, 'a', 0) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: incorrect arguments}}
sl@0: 
sl@0: do_test fts2p-0.8 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1, 'a', 0, 0, 0) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: incorrect arguments}}
sl@0: 
sl@0: do_test fts2p-0.9 {
sl@0:   catchsql {
sl@0:     SELECT dump_doclist(t1, 'a', 16, 16) FROM t1 LIMIT 1;
sl@0:   }
sl@0: } {1 {dump_doclist: segment not found}}
sl@0: 
sl@0: #*************************************************************************
sl@0: # Utility function to check for the expected terms in the segment
sl@0: # level/index.  _all version does same but for entire index.
sl@0: proc check_terms {test level index terms} {
sl@0:   # TODO(shess): Figure out why uplevel in do_test can't catch
sl@0:   # $level and $index directly.
sl@0:   set ::level $level
sl@0:   set ::index $index
sl@0:   do_test $test.terms {
sl@0:     execsql {
sl@0:       SELECT dump_terms(t1, $::level, $::index) FROM t1 LIMIT 1;
sl@0:     }
sl@0:   } [list $terms]
sl@0: }
sl@0: proc check_terms_all {test terms} {
sl@0:   do_test $test.terms {
sl@0:     execsql {
sl@0:       SELECT dump_terms(t1) FROM t1 LIMIT 1;
sl@0:     }
sl@0:   } [list $terms]
sl@0: }
sl@0: 
sl@0: # Utility function to check for the expected doclist for the term in
sl@0: # segment level/index.  _all version does same for entire index.
sl@0: proc check_doclist {test level index term doclist} {
sl@0:   # TODO(shess): Again, why can't the non-:: versions work?
sl@0:   set ::term $term
sl@0:   set ::level $level
sl@0:   set ::index $index
sl@0:   do_test $test {
sl@0:     execsql {
sl@0:       SELECT dump_doclist(t1, $::term, $::level, $::index) FROM t1 LIMIT 1;
sl@0:     }
sl@0:   } [list $doclist]
sl@0: }
sl@0: proc check_doclist_all {test term doclist} {
sl@0:   set ::term $term
sl@0:   do_test $test {
sl@0:     execsql {
sl@0:       SELECT dump_doclist(t1, $::term) FROM t1 LIMIT 1;
sl@0:     }
sl@0:   } [list $doclist]
sl@0: }
sl@0: 
sl@0: #*************************************************************************
sl@0: # Test the segments resulting from straight-forward inserts.
sl@0: db eval {
sl@0:   DROP TABLE IF EXISTS t1;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) VALUES (1, 'This is a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (2, 'That was a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (3, 'This is a test');
sl@0: }
sl@0: 
sl@0: # Check for expected segments and expected matches.
sl@0: do_test fts2p-1.0.segments {
sl@0:   execsql {
sl@0:     SELECT level, idx FROM t1_segdir ORDER BY level, idx;
sl@0:   }
sl@0: } {0 0 0 1 0 2}
sl@0: do_test fts2p-1.0.matches {
sl@0:   execsql {
sl@0:     SELECT OFFSETS(t1) FROM t1
sl@0:      WHERE t1 MATCH 'this OR that OR was OR a OR is OR test' ORDER BY rowid;
sl@0:   }
sl@0: } [list {0 0 0 4 0 4 5 2 0 3 8 1 0 5 10 4} \
sl@0:         {0 1 0 4 0 2 5 3 0 3 9 1 0 5 11 4} \
sl@0:         {0 0 0 4 0 4 5 2 0 3 8 1 0 5 10 4}]
sl@0: 
sl@0: # Check the specifics of the segments constructed.
sl@0: # Logical view of entire index.
sl@0: check_terms_all   fts2p-1.0.1   {a is test that this was}
sl@0: check_doclist_all fts2p-1.0.1.1 a {[1 0[2]] [2 0[2]] [3 0[2]]}
sl@0: check_doclist_all fts2p-1.0.1.2 is {[1 0[1]] [3 0[1]]}
sl@0: check_doclist_all fts2p-1.0.1.3 test {[1 0[3]] [2 0[3]] [3 0[3]]}
sl@0: check_doclist_all fts2p-1.0.1.4 that {[2 0[0]]}
sl@0: check_doclist_all fts2p-1.0.1.5 this {[1 0[0]] [3 0[0]]}
sl@0: check_doclist_all fts2p-1.0.1.6 was {[2 0[1]]}
sl@0: 
sl@0: # Segment 0,0
sl@0: check_terms   fts2p-1.0.2   0 0 {a is test this}
sl@0: check_doclist fts2p-1.0.2.1 0 0 a {[1 0[2]]}
sl@0: check_doclist fts2p-1.0.2.2 0 0 is {[1 0[1]]}
sl@0: check_doclist fts2p-1.0.2.3 0 0 test {[1 0[3]]}
sl@0: check_doclist fts2p-1.0.2.4 0 0 this {[1 0[0]]}
sl@0: 
sl@0: # Segment 0,1
sl@0: check_terms   fts2p-1.0.3   0 1 {a test that was}
sl@0: check_doclist fts2p-1.0.3.1 0 1 a {[2 0[2]]}
sl@0: check_doclist fts2p-1.0.3.2 0 1 test {[2 0[3]]}
sl@0: check_doclist fts2p-1.0.3.3 0 1 that {[2 0[0]]}
sl@0: check_doclist fts2p-1.0.3.4 0 1 was {[2 0[1]]}
sl@0: 
sl@0: # Segment 0,2
sl@0: check_terms   fts2p-1.0.4   0 2 {a is test this}
sl@0: check_doclist fts2p-1.0.4.1 0 2 a {[3 0[2]]}
sl@0: check_doclist fts2p-1.0.4.2 0 2 is {[3 0[1]]}
sl@0: check_doclist fts2p-1.0.4.3 0 2 test {[3 0[3]]}
sl@0: check_doclist fts2p-1.0.4.4 0 2 this {[3 0[0]]}
sl@0: 
sl@0: #*************************************************************************
sl@0: # Test the segments resulting from inserts followed by a delete.
sl@0: db eval {
sl@0:   DROP TABLE IF EXISTS t1;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) VALUES (1, 'This is a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (2, 'That was a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (3, 'This is a test');
sl@0:   DELETE FROM t1 WHERE rowid = 1;
sl@0: }
sl@0: 
sl@0: do_test fts2p-1.1.segments {
sl@0:   execsql {
sl@0:     SELECT level, idx FROM t1_segdir ORDER BY level, idx;
sl@0:   }
sl@0: } {0 0 0 1 0 2 0 3}
sl@0: do_test fts2p-1.1.matches {
sl@0:   execsql {
sl@0:     SELECT OFFSETS(t1) FROM t1
sl@0:      WHERE t1 MATCH 'this OR that OR was OR a OR is OR test' ORDER BY rowid;
sl@0:   }
sl@0: } {{0 1 0 4 0 2 5 3 0 3 9 1 0 5 11 4} {0 0 0 4 0 4 5 2 0 3 8 1 0 5 10 4}}
sl@0: 
sl@0: check_terms_all fts2p-1.1.1 {a is test that this was}
sl@0: check_doclist_all fts2p-1.1.1.1 a {[2 0[2]] [3 0[2]]}
sl@0: check_doclist_all fts2p-1.1.1.2 is {[3 0[1]]}
sl@0: check_doclist_all fts2p-1.1.1.3 test {[2 0[3]] [3 0[3]]}
sl@0: check_doclist_all fts2p-1.1.1.4 that {[2 0[0]]}
sl@0: check_doclist_all fts2p-1.1.1.5 this {[3 0[0]]}
sl@0: check_doclist_all fts2p-1.1.1.6 was {[2 0[1]]}
sl@0: 
sl@0: check_terms fts2p-1.1.2 0 0 {a is test this}
sl@0: check_doclist fts2p-1.1.2.1 0 0 a {[1 0[2]]}
sl@0: check_doclist fts2p-1.1.2.2 0 0 is {[1 0[1]]}
sl@0: check_doclist fts2p-1.1.2.3 0 0 test {[1 0[3]]}
sl@0: check_doclist fts2p-1.1.2.4 0 0 this {[1 0[0]]}
sl@0: 
sl@0: check_terms fts2p-1.1.3 0 1 {a test that was}
sl@0: check_doclist fts2p-1.1.3.1 0 1 a {[2 0[2]]}
sl@0: check_doclist fts2p-1.1.3.2 0 1 test {[2 0[3]]}
sl@0: check_doclist fts2p-1.1.3.3 0 1 that {[2 0[0]]}
sl@0: check_doclist fts2p-1.1.3.4 0 1 was {[2 0[1]]}
sl@0: 
sl@0: check_terms fts2p-1.1.4 0 2 {a is test this}
sl@0: check_doclist fts2p-1.1.4.1 0 2 a {[3 0[2]]}
sl@0: check_doclist fts2p-1.1.4.2 0 2 is {[3 0[1]]}
sl@0: check_doclist fts2p-1.1.4.3 0 2 test {[3 0[3]]}
sl@0: check_doclist fts2p-1.1.4.4 0 2 this {[3 0[0]]}
sl@0: 
sl@0: check_terms fts2p-1.1.5 0 3 {a is test this}
sl@0: check_doclist fts2p-1.1.5.1 0 3 a {[1]}
sl@0: check_doclist fts2p-1.1.5.2 0 3 is {[1]}
sl@0: check_doclist fts2p-1.1.5.3 0 3 test {[1]}
sl@0: check_doclist fts2p-1.1.5.4 0 3 this {[1]}
sl@0: 
sl@0: #*************************************************************************
sl@0: # Test results when all references to certain tokens are deleted.
sl@0: db eval {
sl@0:   DROP TABLE IF EXISTS t1;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) VALUES (1, 'This is a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (2, 'That was a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (3, 'This is a test');
sl@0:   DELETE FROM t1 WHERE rowid IN (1,3);
sl@0: }
sl@0: 
sl@0: # Still 4 segments because 0,3 will contain deletes for rowid 1 and 3.
sl@0: do_test fts2p-1.2.segments {
sl@0:   execsql {
sl@0:     SELECT level, idx FROM t1_segdir ORDER BY level, idx;
sl@0:   }
sl@0: } {0 0 0 1 0 2 0 3}
sl@0: do_test fts2p-1.2.matches {
sl@0:   execsql {
sl@0:     SELECT OFFSETS(t1) FROM t1
sl@0:      WHERE t1 MATCH 'this OR that OR was OR a OR is OR test' ORDER BY rowid;
sl@0:   }
sl@0: } {{0 1 0 4 0 2 5 3 0 3 9 1 0 5 11 4}}
sl@0: 
sl@0: check_terms_all fts2p-1.2.1 {a is test that this was}
sl@0: check_doclist_all fts2p-1.2.1.1 a {[2 0[2]]}
sl@0: check_doclist_all fts2p-1.2.1.2 is {}
sl@0: check_doclist_all fts2p-1.2.1.3 test {[2 0[3]]}
sl@0: check_doclist_all fts2p-1.2.1.4 that {[2 0[0]]}
sl@0: check_doclist_all fts2p-1.2.1.5 this {}
sl@0: check_doclist_all fts2p-1.2.1.6 was {[2 0[1]]}
sl@0: 
sl@0: check_terms fts2p-1.2.2 0 0 {a is test this}
sl@0: check_doclist fts2p-1.2.2.1 0 0 a {[1 0[2]]}
sl@0: check_doclist fts2p-1.2.2.2 0 0 is {[1 0[1]]}
sl@0: check_doclist fts2p-1.2.2.3 0 0 test {[1 0[3]]}
sl@0: check_doclist fts2p-1.2.2.4 0 0 this {[1 0[0]]}
sl@0: 
sl@0: check_terms fts2p-1.2.3 0 1 {a test that was}
sl@0: check_doclist fts2p-1.2.3.1 0 1 a {[2 0[2]]}
sl@0: check_doclist fts2p-1.2.3.2 0 1 test {[2 0[3]]}
sl@0: check_doclist fts2p-1.2.3.3 0 1 that {[2 0[0]]}
sl@0: check_doclist fts2p-1.2.3.4 0 1 was {[2 0[1]]}
sl@0: 
sl@0: check_terms fts2p-1.2.4 0 2 {a is test this}
sl@0: check_doclist fts2p-1.2.4.1 0 2 a {[3 0[2]]}
sl@0: check_doclist fts2p-1.2.4.2 0 2 is {[3 0[1]]}
sl@0: check_doclist fts2p-1.2.4.3 0 2 test {[3 0[3]]}
sl@0: check_doclist fts2p-1.2.4.4 0 2 this {[3 0[0]]}
sl@0: 
sl@0: check_terms fts2p-1.2.5 0 3 {a is test this}
sl@0: check_doclist fts2p-1.2.5.1 0 3 a {[1] [3]}
sl@0: check_doclist fts2p-1.2.5.2 0 3 is {[1] [3]}
sl@0: check_doclist fts2p-1.2.5.3 0 3 test {[1] [3]}
sl@0: check_doclist fts2p-1.2.5.4 0 3 this {[1] [3]}
sl@0: 
sl@0: #*************************************************************************
sl@0: # Test results when everything is optimized manually.
sl@0: db eval {
sl@0:   DROP TABLE IF EXISTS t1;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) VALUES (1, 'This is a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (2, 'That was a test');
sl@0:   INSERT INTO t1 (rowid, c) VALUES (3, 'This is a test');
sl@0:   DELETE FROM t1 WHERE rowid IN (1,3);
sl@0:   DROP TABLE IF EXISTS t1old;
sl@0:   ALTER TABLE t1 RENAME TO t1old;
sl@0:   CREATE VIRTUAL TABLE t1 USING fts2(c);
sl@0:   INSERT INTO t1 (rowid, c) SELECT rowid, c FROM t1old;
sl@0:   DROP TABLE t1old;
sl@0: }
sl@0: 
sl@0: # Should be a single optimal segment with the same logical results.
sl@0: do_test fts2p-1.3.segments {
sl@0:   execsql {
sl@0:     SELECT level, idx FROM t1_segdir ORDER BY level, idx;
sl@0:   }
sl@0: } {0 0}
sl@0: do_test fts2p-1.3.matches {
sl@0:   execsql {
sl@0:     SELECT OFFSETS(t1) FROM t1
sl@0:      WHERE t1 MATCH 'this OR that OR was OR a OR is OR test' ORDER BY rowid;
sl@0:   }
sl@0: } {{0 1 0 4 0 2 5 3 0 3 9 1 0 5 11 4}}
sl@0: 
sl@0: check_terms_all fts2p-1.3.1 {a test that was}
sl@0: check_doclist_all fts2p-1.3.1.1 a {[2 0[2]]}
sl@0: check_doclist_all fts2p-1.3.1.2 test {[2 0[3]]}
sl@0: check_doclist_all fts2p-1.3.1.3 that {[2 0[0]]}
sl@0: check_doclist_all fts2p-1.3.1.4 was {[2 0[1]]}
sl@0: 
sl@0: check_terms fts2p-1.3.2 0 0 {a test that was}
sl@0: check_doclist fts2p-1.3.2.1 0 0 a {[2 0[2]]}
sl@0: check_doclist fts2p-1.3.2.2 0 0 test {[2 0[3]]}
sl@0: check_doclist fts2p-1.3.2.3 0 0 that {[2 0[0]]}
sl@0: check_doclist fts2p-1.3.2.4 0 0 was {[2 0[1]]}
sl@0: 
sl@0: finish_test