path: root/ext/fts5/test/fts5prefix.test

# 2015 Jan 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 contains tests focused on prefix indexes.
#

source [file join [file dirname [info script]] fts5_common.tcl]
set testprefix fts5prefix

# If SQLITE_ENABLE_FTS5 is defined, omit this file.
ifcapable !fts5 {
  finish_test
  return
}

do_execsql_test 1.0 {
  CREATE VIRTUAL TABLE xx USING fts5(x, prefix=1);
  INSERT INTO xx VALUES('one two three');
  INSERT INTO xx VALUES('four five six');
  INSERT INTO xx VALUES('seven eight nine ten');
}

do_execsql_test 1.1 {
  SELECT rowid FROM xx WHERE xx MATCH 't*'
} {1 3}


#-------------------------------------------------------------------------
# Check that prefix indexes really do index n-character prefixes, not 
# n-byte prefixes. Use the ascii tokenizer so as not to be confused by
# diacritic removal.
#
do_execsql_test 2.0 { 
  CREATE VIRTUAL TABLE t1 USING fts5(x, tokenize = ascii, prefix = 2) 
}

do_test 2.1 {
  foreach {rowid string} {
    1 "\xCA\xCB\xCC\xCD"
    2 "\u1234\u5678\u4321\u8765"
  } {
    execsql { INSERT INTO t1(rowid, x) VALUES($rowid, $string) }
  }
} {}

do_execsql_test 2.2 {
  INSERT INTO t1(t1) VALUES('integrity-check');
}

foreach {tn q res} {
  1 "SELECT rowid FROM t1 WHERE t1 MATCH '\xCA\xCB*'" 1
  2 "SELECT rowid FROM t1 WHERE t1 MATCH '\u1234\u5678*'" 2
} {
  do_execsql_test 2.3.$tn $q $res
}

#-------------------------------------------------------------------------
# Check that prefix queries with:
#
#   * a column filter, and
#   * no prefix index.
#
# work Ok.
#
do_execsql_test 3.0 {
  CREATE VIRTUAL TABLE t3 USING fts5(a, b, c);
  INSERT INTO t3(t3, rank) VALUES('pgsz', 32);
  BEGIN;
    INSERT INTO t3 VALUES('acb ccc bba', 'cca bba bca', 'bbc ccc bca'); -- 1
    INSERT INTO t3 VALUES('cbb cac cab', 'abb aac bba', 'aab ccc cac'); -- 2
    INSERT INTO t3 VALUES('aac bcb aac', 'acb bcb caa', 'aca bab bca'); -- 3
    INSERT INTO t3 VALUES('aab ccb ccc', 'aca cba cca', 'aca aac cbb'); -- 4
    INSERT INTO t3 VALUES('bac aab bab', 'ccb bac cba', 'acb aba abb'); -- 5
    INSERT INTO t3 VALUES('bab abc ccb', 'acb cba abb', 'cbb aaa cab'); -- 6
    INSERT INTO t3 VALUES('cbb bbc baa', 'aab aca baa', 'bcc cca aca'); -- 7
    INSERT INTO t3 VALUES('abc bba abb', 'cac abc cba', 'acc aac cac'); -- 8
    INSERT INTO t3 VALUES('bbc bbc cab', 'bcb ccb cba', 'bcc cac acb'); -- 9
  COMMIT;
}

foreach {tn match res} {
  1 "a : c*" {1 2 4 6 7 9}
  2 "b : c*" {1 3 4 5 6 8 9}
  3 "c : c*" {1 2 4 6 7 8 9}
  4 "a : b*" {1 3 5 6 7 8 9}
  5 "b : b*" {1 2 3 5 7 9}
  6 "c : b*" {1 3 7 9}
  7 "a : a*" {1 3 4 5 6 8}
  8 "b : a*" {2 3 4 6 7 8}
  9 "c : a*" {2 3 4 5 6 7 8 9}
} {
  do_execsql_test 3.1.$tn {
    SELECT rowid FROM t3($match)
  } $res
}

do_test 3.2 {
  expr srand(0)
  execsql { DELETE FROM t3 }
  for {set i 0} {$i < 1000} {incr i} {
    set a [fts5_rnddoc 3]
    set b [fts5_rnddoc 8]
    set c [fts5_rnddoc 20]
    execsql { INSERT INTO t3 VALUES($a, $b, $c) }
  }
  execsql { INSERT INTO t3(t3) VALUES('integrity-check') }
} {}

proc gmatch {col pattern} {
  expr {[lsearch -glob $col $pattern]>=0}
}
db func gmatch gmatch

proc ghl {col pattern} {
  foreach t $col {
    if {[string match $pattern $t]} {
      lappend res "*$t*"
    } else {
      lappend res $t
    }
  }
  set res
}
db func ghl ghl

set COLS(a) 0
set COLS(b) 1
set COLS(c) 2

for {set x 0} {$x<2} {incr x} {
  foreach {tn pattern} {
    1  {xa*}
    2  {xb*}
    3  {xc*}
    4  {xd*}
    5  {xe*}
    6  {xf*}
    7  {xg*}
    8  {xh*}
    9  {xi*}
    10 {xj*}
  } {
    foreach col {a b c} {

      # Check that the list of returned rowids is correct.
      #
      set res [db eval "SELECT rowid FROM t3 WHERE gmatch($col, '$pattern')"]
      set query "$col : $pattern"
      do_execsql_test 3.3.$x.$tn.$col.rowid {
        SELECT rowid FROM t3($query);
      } $res

      # Check that the highlight() function works.
      #
      set res [db eval \
        "SELECT ghl($col, '$pattern') FROM t3 WHERE gmatch($col, '$pattern')"
      ]
      set idx $COLS($col)
      do_execsql_test 3.3.$x.$tn.$col.highlight {
        SELECT highlight(t3, $idx, '*', '*') FROM t3($query);
      } $res
    }

    foreach colset {{a b} {b c} {c a} {a c} {b a}} {
      # Check that the list of returned rowids is correct.
      #
      foreach {col1 col2} $colset {}
      set expr "gmatch($col1, '$pattern') OR gmatch($col2, '$pattern')"
      set res [db eval "SELECT rowid FROM t3 WHERE $expr"]
      set query "{$colset} : $pattern"
      do_execsql_test 3.3.$x.$tn.{$colset}.rowid {
        SELECT rowid FROM t3($query);
      } $res

      set resq    "SELECT ghl($col1, '$pattern'), ghl($col2, '$pattern')"
      append resq " FROM t3 WHERE $expr"
      set res [db eval $resq]
      set idx1 $COLS($col1)
      set idx2 $COLS($col2)
      do_execsql_test 3.3.$x.$tn.{$colset}.highlight {
        SELECT highlight(t3, $idx1, '*', '*'), highlight(t3, $idx2, '*', '*')
          FROM t3($query)
      } $res
    }
  }
  execsql { INSERT INTO t3(t3) VALUES('optimize') }
  execsql { INSERT INTO t3(t3) VALUES('integrity-check') }
}

#-------------------------------------------------------------------------
#
reset_db
do_execsql_test 4.0 {
  CREATE VIRTUAL TABLE t2 USING fts5(c1, c2);
  INSERT INTO t2 VALUES('xa xb', 'xb xa');

  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 2
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 4
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 8
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 16
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 32
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 64
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 128
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 256
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 512
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 1024
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 2048
  INSERT INTO t2 SELECT c1||' '||c1, c2||' '||c2 FROM t2; -- 4096

  SELECT count(*) FROM t2('x*');
} {4096}

do_execsql_test 4.1 {
  UPDATE t2 SET c2 = 'ya yb';
  SELECT count(*) FROM t2('c1:x*');
  SELECT count(*) FROM t2('c2:x*');
} {4096 0}

do_execsql_test 4.2 {
  UPDATE t2 SET c2 = 'xa';
  SELECT count(*) FROM t2('c1:x*');
  SELECT count(*) FROM t2('c2:x*');
} {4096 4096}

#-------------------------------------------------------------------------
#
reset_db
proc rnddoc {n} {
  set map [list a b c d]
  set doc [list]
  for {set i 0} {$i < $n} {incr i} {
    lappend doc "x[lindex $map [expr int(rand()*4)]]"
  }
  set doc
}
set cols [list]
for {set i 1} {$i<250} {incr i} { 
  lappend cols "c$i" 
  lappend vals "'[rnddoc 10]'"
}

do_test 5.0 {
  execsql "CREATE VIRTUAL TABLE t4 USING fts5([join $cols ,])"
  execsql {INSERT INTO t4(t4, rank) VALUES('pgsz', 32)}
  execsql "INSERT INTO t4 VALUES([join $vals ,])"
  execsql "INSERT INTO t4 VALUES([join $vals ,])"
  execsql "INSERT INTO t4 VALUES([join $vals ,])"
  execsql "INSERT INTO t4 VALUES([join $vals ,])"
} {}

proc gmatch {col pattern} {
  expr {[lsearch -glob $col $pattern]>=0}
}
db func gmatch gmatch
foreach {tn col pattern} {
  1 c100 {xa*}
  2 c200 {xb*}
} {
  set res [db eval "SELECT rowid FROM t4 WHERE gmatch($col, \$pattern)"]
  set query "$col : $pattern"
  do_execsql_test 5.$tn { SELECT rowid FROM t4($query) } $res
}

reset_db
db func fts5_rnddoc fts5_rnddoc
do_test 6.0 {
  execsql {
    CREATE VIRTUAL TABLE t5 USING fts5(x, y);
    INSERT INTO t5 VALUES( fts5_rnddoc(10000), fts5_rnddoc(10000) );
    INSERT INTO t5 VALUES( fts5_rnddoc(10000), fts5_rnddoc(10000) );
    INSERT INTO t5 VALUES( fts5_rnddoc(10000), fts5_rnddoc(10000) );
    INSERT INTO t5 VALUES( fts5_rnddoc(10000), fts5_rnddoc(10000) );
  }
} {}

proc gmatch {col pattern} {
  expr {[lsearch -glob $col $pattern]>=0}
}
db func gmatch gmatch
foreach {tn col pattern} {
  1 y {xa*}
  2 y {xb*}
  3 y {xc*}
  4 x {xa*}
  5 x {xb*}
  6 x {xc*}
} {
  set res [db eval "SELECT rowid FROM t5 WHERE gmatch($col, \$pattern)"]
  set query "$col : $pattern"
  do_execsql_test 6.$tn { SELECT rowid FROM t5($query) } $res
}

#-------------------------------------------------------------------------
# Check that the various ways of creating prefix indexes produce the
# same database on disk.
#
save_prng_state
foreach {tn create} {
  1 { CREATE VIRTUAL TABLE tt USING fts5(x, y, prefix="1,2,3") }
  2 { CREATE VIRTUAL TABLE tt USING fts5(x, y, prefix="1 2 3") }
  3 { CREATE VIRTUAL TABLE tt USING fts5(x, y, prefix=1, prefix=2, prefix=3) }
  4 { CREATE VIRTUAL TABLE tt USING fts5(x, y, prefix="1 2", prefix=3) }
} {
  execsql { DROP TABLE IF EXISTS tt }
  restore_prng_state
  execsql $create
  execsql {
    INSERT INTO tt VALUES('cc b ggg ccc aa eee hh', 'aa g b hh a e');
    INSERT INTO tt VALUES('cc bb cc gg j g cc', 'ii jjj ggg jjj cc cc');
    INSERT INTO tt VALUES('h eee cc h iii', 'aaa iii dd iii dd');
    INSERT INTO tt VALUES('jjj hh eee c e b gg', 'j bbb jj ddd jj');
    INSERT INTO tt VALUES('ii hhh aaa ff c hhh iii', 'j cc hh bb e');
    INSERT INTO tt VALUES('e fff hhh i aaa', 'g b aa gg c aa dd');
    INSERT INTO tt VALUES('i aaa ccc gg hhh aa h', 'j bbb bbb d ff');
    INSERT INTO tt VALUES('g f gg ff ff jjj d', 'jjj d j fff fff ee j');
    INSERT INTO tt VALUES('a cc e ccc jjj c', 'ccc iii d bb a eee g');
    INSERT INTO tt VALUES('jj hh hh bb bbb gg', 'j c jjj bb iii f');
    INSERT INTO tt VALUES('a ggg g cc ccc aa', 'jjj j j aaa c');
    INSERT INTO tt VALUES('ddd j dd b i', 'aaa bbb iii ggg ff ccc ddd');
    INSERT INTO tt VALUES('jj ii hh c ii h gg', 'hhh bbb ddd bbb hh g ggg');
    INSERT INTO tt VALUES('aa hhh ccc h ggg ccc', 'iii d jj a ff ii');
  }

  #db eval {SELECT rowid, fts5_decode(rowid, block) aS r FROM tt_data} {puts $r}

  if {$tn==1} {
    set ::checksum [execsql {SELECT md5sum(id, block) FROM tt_data}]
  } else {
    do_execsql_test 7.$tn {
      SELECT md5sum(id, block) FROM tt_data
    } [list $::checksum]
  }
}

finish_test