niceSOFT/sqlite: Official Git mirror of the SQLite source tree

test/imposter1.test
142 строки5 KB
drh
Improvements to imposter tables: (1) They are read-only unless writable-schema
19 сен 2025, 20:24
19 сен 2025, 20:2449a37f3
# 2015-01-30
#
# 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 tests for SQLite library.
#
# The focus of this file is adding extra entries in the symbol table
# using sqlite3_test_control(SQLITE_TESTCTRL_IMPOSTER) and verifying that
# SQLite handles those as expected.
#

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

# Create a bunch of data to sort against
#
do_test imposter-1.0 {
  execsql {
    CREATE TABLE t1(a INTEGER PRIMARY KEY, b, c, d NOT NULL);
    CREATE INDEX t1b ON t1(b);
    CREATE UNIQUE INDEX t1c ON t1(c);
    WITH RECURSIVE c(i) AS (VALUES(1) UNION ALL SELECT i+1 FROM c WHERE i<30)
      INSERT INTO t1(a,b,c,d) SELECT i,1000+i,2000+i,3000+i FROM c;
  }
  set t1_root [db one {SELECT rootpage FROM sqlite_master WHERE name='t1'}]
  set t1b_root [db one {SELECT rootpage FROM sqlite_master WHERE name='t1b'}]
  set t1c_root [db one {SELECT rootpage FROM sqlite_master WHERE name='t1c'}]

  # Create an imposter table that uses the same b-tree as t1 but which does
  # not have the indexes
  #
  sqlite3_test_control SQLITE_TESTCTRL_IMPOSTER db main 1 $t1_root
  db eval {CREATE TABLE xt1(a,b,c,d)}

  # And create an imposter table for the t1c index.
  sqlite3_test_control SQLITE_TESTCTRL_IMPOSTER db main 1 $t1c_root
  db eval {CREATE TABLE xt1c(c,rowid,PRIMARY KEY(c,rowid))WITHOUT ROWID;}

  # Go out of imposter mode for now.
  sqlite3_test_control SQLITE_TESTCTRL_IMPOSTER db main 0 0

  # Create triggers to record changes to xt1.
  #
  db eval {
    CREATE TEMP TABLE chnglog(desc TEXT);
    CREATE TEMP TRIGGER xt1_del AFTER DELETE ON xt1 BEGIN
      INSERT INTO chnglog VALUES(
           printf('DELETE t1: rowid=%d, a=%s, b=%s, c=%s, d=%s',
                  old.rowid, quote(old.a), quote(old.b), quote(old.c),
                  quote(old.d)));
    END;
    CREATE TEMP TRIGGER xt1_ins AFTER INSERT ON xt1 BEGIN
      INSERT INTO chnglog VALUES(
           printf('INSERT t1:  rowid=%d, a=%s, b=%s, c=%s, d=%s',
                  new.rowid, quote(new.a), quote(new.b), quote(new.c),
                  quote(new.d)));
    END;
  }
} {}

# The xt1 table has separate xt1.rowid and xt1.a columns.  The xt1.rowid
# column corresponds to t1.rowid and t1.a, but the xt1.a column is always
# NULL
#
do_execsql_test imposter-1.1 {
  SELECT rowid FROM xt1 WHERE a IS NOT NULL;
} {}
do_execsql_test imposter-1.2 {
  SELECT a,b,c,d FROM t1 EXCEPT SELECT rowid,b,c,d FROM xt1;
  SELECT rowid,b,c,d FROM xt1 EXCEPT SELECT a,b,c,d FROM t1;
} {}


# Make changes via the xt1 shadow table.  This will not update the
# indexes on t1 nor check the uniqueness constraint on t1.c nor check
# the NOT NULL constraint on t1.d, resulting in a logically inconsistent
# database.
#
do_execsql_test imposter-1.3 {
  PRAGMA writable_schema=on;
  DELETE FROM xt1 WHERE rowid=5;
  INSERT INTO xt1(rowid,a,b,c,d) VALUES(99,'hello',1099,2022,NULL);
  SELECT * FROM chnglog ORDER BY rowid;
} [list \
  {DELETE t1: rowid=5, a=NULL, b=1005, c=2005, d=3005} \
  {INSERT t1:  rowid=99, a='hello', b=1099, c=2022, d=NULL} \
]

do_execsql_test imposter-1.4a {
  PRAGMA integrity_check;
} {/NULL value in t1.d/}
do_execsql_test imposter-1.4b {
  PRAGMA integrity_check;
} {/row # missing from index t1b/}
do_execsql_test imposter-1.4c {
  PRAGMA integrity_check;
} {/row # missing from index t1c/}

# Cleanup the corruption.
# Then demonstrate that the xt1c imposter table can insert non-unique
# and NULL values into the UNIQUE index.
#
do_execsql_test imposter-2.0 {
  DELETE FROM t1;
  WITH RECURSIVE c(i) AS (VALUES(1) UNION ALL SELECT i+1 FROM c WHERE i<10)
   INSERT INTO t1(a,b,c,d) SELECT i,i,i,i FROM c;
  UPDATE xt1c SET c=NULL WHERE rowid=5;
  PRAGMA integrity_check;
} {/row # missing from index t1c/}

do_execsql_test imposter-2.1 {
  DELETE FROM t1;
  WITH RECURSIVE c(i) AS (VALUES(1) UNION ALL SELECT i+1 FROM c WHERE i<10)
   INSERT INTO t1(a,b,c,d) SELECT i,i,i,i FROM c;
  UPDATE xt1c SET c=99 WHERE rowid IN (5,7,9);
  SELECT c FROM t1 ORDER BY c;
} {1 2 3 4 6 8 10 99 99 99}
do_execsql_test imposter-2.2 {
  UPDATE xt1 SET c=99 WHERE rowid IN (5,7,9);
  PRAGMA integrity_check;
} {/non-unique entry in index t1c/}

# Erase the imposter tables
#
do_test imposter-3.1 {
  sqlite3_test_control SQLITE_TESTCTRL_IMPOSTER db main 0 1
  db eval {
    DELETE FROM t1 WHERE rowid IN (5,7,9);
    PRAGMA integrity_check;
  }
} {ok}


finish_test