modernc.org/cc@v1.0.1/v2/testdata/_sqlite/ext/fts5/test/fts5corrupt3.test

# 2015 Apr 24
#
# 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 tests that FTS5 handles corrupt databases (i.e. internal
# inconsistencies in the backing tables) correctly. In this case 
# "correctly" means without crashing.
#

source [file join [file dirname [info script]] fts5_common.tcl]
set testprefix fts5corrupt3

# If SQLITE_ENABLE_FTS5 is defined, omit this file.
ifcapable !fts5 {
  finish_test
  return
}
sqlite3_fts5_may_be_corrupt 1

proc create_t1 {} {
  expr srand(0)
  db func rnddoc fts5_rnddoc
  db eval {
    CREATE VIRTUAL TABLE t1 USING fts5(x);
    INSERT INTO t1(t1, rank) VALUES('pgsz', 64);
    WITH ii(i) AS (SELECT 1 UNION SELECT i+1 FROM ii WHERE i<100)
      INSERT INTO t1 SELECT rnddoc(10) FROM ii;
  }
}

if 1 {

# Create a simple FTS5 table containing 100 documents. Each document 
# contains 10 terms, each of which start with the character "x".
#
do_test 1.0 { create_t1 } {}

do_test 1.1 {
  # Pick out the rowid of the right-most b-tree leaf in the new segment.
  set rowid [db one {
    SELECT max(rowid) FROM t1_data WHERE ((rowid>>31) & 0x0F)==1
  }]
  set L [db one {SELECT length(block) FROM t1_data WHERE rowid = $rowid}]
  set {} {}
} {} 

for {set i 0} {$i < $L} {incr i} {
  do_test 1.2.$i {
    catchsql {
      BEGIN;
      UPDATE t1_data SET block = substr(block, 1, $i) WHERE id = $rowid;
      INSERT INTO t1(t1) VALUES('integrity-check');
    }
  } {1 {database disk image is malformed}}
  catchsql ROLLBACK
}
 
#-------------------------------------------------------------------------
# Test that trailing bytes appended to the averages record are ignored.
#
do_execsql_test 2.1 {
  CREATE VIRTUAL TABLE t2 USING fts5(x);
  INSERT INTO t2 VALUES(rnddoc(10));
  INSERT INTO t2 VALUES(rnddoc(10));
  SELECT length(block) FROM t2_data WHERE id=1;
} {2}
do_execsql_test 2.2 {
  UPDATE t2_data SET block = block || 'abcd' WHERE id=1;
  SELECT length(block) FROM t2_data WHERE id=1;
} {6}
do_execsql_test 2.2 {
  INSERT INTO t2 VALUES(rnddoc(10));
  SELECT length(block) FROM t2_data WHERE id=1;
} {2}


#-------------------------------------------------------------------------
# Test that missing leaf pages are recognized as corruption.
#
reset_db
do_test 3.0 { create_t1 } {}

do_execsql_test 3.1 {
  SELECT count(*) FROM t1_data;
} {105}

proc do_3_test {tn} {
  set i 0
  foreach ::rowid [db eval "SELECT rowid FROM t1_data WHERE rowid>100"] {
    incr i
    do_test $tn.$i {
      db eval BEGIN
      db eval {DELETE FROM t1_data WHERE rowid = $::rowid}
      list [
        catch { db eval {SELECT rowid FROM t1 WHERE t1 MATCH 'x*'} } msg
      ] $msg
    } {1 {database disk image is malformed}}
    catch { db eval ROLLBACK }
  }
}

do_3_test 3.2

do_execsql_test 3.3 {
  INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
  INSERT INTO t1 SELECT x FROM t1;
  INSERT INTO t1(t1) VALUES('optimize');
} {}

do_3_test 3.4

do_test 3.5 {
  execsql { 
    DELETE FROM t1;
    INSERT INTO t1(t1, rank) VALUES('pgsz', 40);
  }
  for {set i 0} {$i < 1000} {incr i} {
    set rnd [expr int(rand() * 1000)]
    set doc [string repeat "x$rnd " [expr int(rand() * 3) + 1]]
    execsql { INSERT INTO t1(rowid, x) VALUES($i, $doc) }
  }
} {}

do_3_test 3.6

do_test 3.7 {
  execsql {
    INSERT INTO t1(t1, rank) VALUES('pgsz', 40);
    INSERT INTO t1 SELECT x FROM t1;
    INSERT INTO t1(t1) VALUES('optimize');
  }
} {}

do_3_test 3.8

do_test 3.9 {
  execsql { 
    DELETE FROM t1;
    INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
  }
  for {set i 0} {$i < 100} {incr i} {
    set rnd [expr int(rand() * 100)]
    set doc "x[string repeat $rnd 20]"
    execsql { INSERT INTO t1(rowid, x) VALUES($i, $doc) }
  }
} {}

do_3_test 3.10

#-------------------------------------------------------------------------
# Test that segments that end unexpectedly are identified as corruption.
#
reset_db
do_test 4.0 {
  execsql { 
    CREATE VIRTUAL TABLE t1 USING fts5(x);
    INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
  }
  for {set i 0} {$i < 100} {incr i} {
    set rnd [expr int(rand() * 100)]
    set doc "x[string repeat $rnd 20]"
    execsql { INSERT INTO t1(rowid, x) VALUES($i, $doc) }
  }
  execsql { INSERT INTO t1(t1) VALUES('optimize') }
} {}

set nErr 0
for {set i 1} {1} {incr i} {
  set struct [db one {SELECT block FROM t1_data WHERE id=10}]
  binary scan $struct c* var
  set end [lindex $var end]
  if {$end<=$i} break
  lset var end [expr $end - $i]
  set struct [binary format c* $var]

  db close
  sqlite3 db test.db

  db eval {
    BEGIN;
    UPDATE t1_data SET block = $struct WHERE id=10;
  }
  do_test 4.1.$i {
    incr nErr [catch { db eval { SELECT rowid FROM t1 WHERE t1 MATCH 'x*' } }]
    set {} {}
  } {}
  catch { db eval ROLLBACK }
}
do_test 4.1.x { expr $nErr>45 } 1

#-------------------------------------------------------------------------
#

# The first argument passed to this command must be a binary blob 
# containing an FTS5 leaf page. This command returns a copy of this
# blob, with the pgidx of the leaf page replaced by a single varint
# containing value $iVal.
#
proc rewrite_pgidx {blob iVal} {
  binary scan $blob SS off1 szLeaf
  if {$iVal<0 || $iVal>=128} {
    error "$iVal out of range!"
  } else {
    set pgidx [binary format c $iVal]
  }

  binary format a${szLeaf}a* $blob $pgidx
}

reset_db
do_execsql_test 5.1 {
  CREATE VIRTUAL TABLE x1 USING fts5(x);
  INSERT INTO x1(x1, rank) VALUES('pgsz', 40);
  BEGIN;
  INSERT INTO x1 VALUES('xaaa xabb xccc xcdd xeee xeff xggg xghh xiii xijj');
  INSERT INTO x1 SELECT x FROM x1;
  INSERT INTO x1 SELECT x FROM x1;
  INSERT INTO x1 SELECT x FROM x1;
  INSERT INTO x1 SELECT x FROM x1;
  INSERT INTO x1(x1) VALUES('optimize');
  COMMIT;
}

#db eval { SELECT fts5_decode(id, block) b from x1_data } { puts $b }
#
db func rewrite_pgidx rewrite_pgidx  
set i 0
foreach rowid [db eval {SELECT rowid FROM x1_data WHERE rowid>100}] {
  foreach val {2 100} {
    do_test 5.2.$val.[incr i] {
      catchsql {
        BEGIN;
        UPDATE x1_data SET block=rewrite_pgidx(block, $val) WHERE id=$rowid;
        SELECT rowid FROM x1 WHERE x1 MATCH 'xa*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xb*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xc*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xd*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xe*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xf*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xg*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xh*';
        SELECT rowid FROM x1 WHERE x1 MATCH 'xi*';
      }
      set {} {}
    } {}
    catch { db eval ROLLBACK }
  }
}

#------------------------------------------------------------------------
#
reset_db
do_execsql_test 6.1.0 {
  CREATE VIRTUAL TABLE t1 USING fts5(a);
  INSERT INTO t1 VALUES('bbbbb ccccc');
  SELECT quote(block) FROM t1_data WHERE rowid>100;
} {X'000000180630626262626201020201056363636363010203040A'}
do_execsql_test 6.1.1 {
  UPDATE t1_data SET block = 
  X'000000180630626262626201020201056161616161010203040A'
  WHERE rowid>100;
}
do_catchsql_test 6.1.2 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}

#-------
reset_db
do_execsql_test 6.2.0 {
  CREATE VIRTUAL TABLE t1 USING fts5(a);
  INSERT INTO t1(t1, rank) VALUES('pgsz', 32);
  INSERT INTO t1 VALUES('aa bb cc dd ee');
  SELECT pgno, quote(term) FROM t1_idx;
} {2 X'' 4 X'3064'}
do_execsql_test 6.2.1 {
  UPDATE t1_idx SET term = X'3065' WHERE pgno=4;
}
do_catchsql_test 6.2.2 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}

#-------
reset_db
do_execsql_test 6.3.0 {
  CREATE VIRTUAL TABLE t1 USING fts5(a);
  INSERT INTO t1 VALUES('abc abcdef abcdefghi');
  SELECT quote(block) FROM t1_data WHERE id>100;
}    {X'0000001C043061626301020204036465660102030703676869010204040808'}
do_execsql_test 6.3.1 {
  BEGIN;
    UPDATE t1_data SET block = 
      X'0000001C043061626301020204036465660102035003676869010204040808'
      ------------------------------------------^^---------------------
    WHERE id>100;
}
do_catchsql_test 6.3.2 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}
do_execsql_test 6.3.3 {
  ROLLBACK;
  BEGIN;
    UPDATE t1_data SET block = 
      X'0000001C043061626301020204036465660102030750676869010204040808'
      --------------------------------------------^^-------------------
    WHERE id>100;
}
do_catchsql_test 6.3.3 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}
do_execsql_test 6.3.4 {
  ROLLBACK;
  BEGIN;
    UPDATE t1_data SET block = 
      X'0000001C043061626301020204036465660102030707676869010204040850'
      --------------------------------------------------------------^^-
    WHERE id>100;
}
do_catchsql_test 6.3.5 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}
do_execsql_test 6.3.6 {
  ROLLBACK;
  BEGIN;
    UPDATE t1_data SET block = 
      X'0000001C503061626301020204036465660102030707676869010204040808'
      ----------^^-----------------------------------------------------
    WHERE id>100;
}
do_catchsql_test 6.3.5 {
  INSERT INTO t1(t1) VALUES('integrity-check');
} {1 {database disk image is malformed}}


#------------------------------------------------------------------------
#
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
}

db func rnddoc rnddoc
do_test 7.0 {
  execsql {
    CREATE VIRTUAL TABLE t5 USING fts5(x);
    INSERT INTO t5 VALUES( rnddoc(10000) );
    INSERT INTO t5 VALUES( rnddoc(10000) );
    INSERT INTO t5 VALUES( rnddoc(10000) );
    INSERT INTO t5 VALUES( rnddoc(10000) );
    INSERT INTO t5(t5) VALUES('optimize');
  }
} {}

do_test 7.1 {
  foreach i [db eval { SELECT rowid FROM t5_data WHERE rowid>100 }] {
    db eval BEGIN  
    db eval {DELETE FROM t5_data WHERE rowid = $i}
    set r [catchsql { INSERT INTO t5(t5) VALUES('integrity-check')} ]
    if {$r != "1 {database disk image is malformed}"} { error $r }
    db eval ROLLBACK  
  }
} {}

}

#------------------------------------------------------------------------
# Corruption within the structure record.
#
reset_db
do_execsql_test 8.1 {
  CREATE VIRTUAL TABLE t1 USING fts5(x, y);
  INSERT INTO t1 VALUES('one', 'two');
}

do_test 9.1.1 {
  set    blob "12345678"    ;# cookie
  append blob "0105"        ;# 1 level, total of 5 segments
  append blob "06"          ;# write counter
  append blob "0002"        ;# first level has 0 segments merging, 2 other.
  append blob "450108"      ;# first segment
  execsql "REPLACE INTO t1_data VALUES(10, X'$blob')"
} {}
do_catchsql_test 9.1.2 {
  SELECT * FROM t1('one AND two');
} {1 {database disk image is malformed}}

do_test 9.2.1 {
  set    blob "12345678"    ;# cookie
  append blob "0205"        ;# 2 levels, total of 5 segments
  append blob "06"          ;# write counter
  append blob "0001"        ;# first level has 0 segments merging, 1 other.
  append blob "450108"      ;# first segment
  execsql "REPLACE INTO t1_data VALUES(10, X'$blob')"
} {}
do_catchsql_test 9.2.2 {
  SELECT * FROM t1('one AND two');
} {1 {database disk image is malformed}}

sqlite3_fts5_may_be_corrupt 0
finish_test