github.com/jdgcs/sqlite3@v1.12.1-0.20210908114423-bc5f96e4dd51/testdata/tcl/fts3atoken.test (about)

     1  # 2007 June 21
     2  #
     3  # The author disclaims copyright to this source code.  In place of
     4  # a legal notice, here is a blessing:
     5  #
     6  #    May you do good and not evil.
     7  #    May you find forgiveness for yourself and forgive others.
     8  #    May you share freely, never taking more than you give.
     9  #
    10  #*************************************************************************
    11  # This file implements regression tests for SQLite library. The focus 
    12  # of this script is testing the pluggable tokeniser feature of the 
    13  # FTS3 module.
    14  #
    15  # $Id: fts3atoken.test,v 1.1 2007/08/20 17:38:42 shess Exp $
    16  #
    17  
    18  set testdir [file dirname $argv0]
    19  source $testdir/tester.tcl
    20  
    21  # If SQLITE_ENABLE_FTS3 is defined, omit this file.
    22  ifcapable !fts3 {
    23    finish_test
    24    return
    25  }
    26  
    27  set ::testprefix fts3atoken
    28  
    29  proc escape_string {str} {
    30    set out ""
    31    foreach char [split $str ""] {
    32      scan $char %c i
    33      if {$i<=127} {
    34        append out $char
    35      } else {
    36        append out [format {\x%.4x} $i]
    37      }
    38    }
    39    set out
    40  }
    41  
    42  #--------------------------------------------------------------------------
    43  # Test cases fts3atoken-1.* are the warm-body test for the SQL scalar
    44  # function fts3_tokenizer(). The procedure is as follows:
    45  #
    46  #   1: Verify that there is no such fts3 tokenizer as 'blah'.
    47  #
    48  #   2: Query for the built-in tokenizer 'simple'. Insert a copy of the
    49  #      retrieved value as tokenizer 'blah'.
    50  #
    51  #   3: Test that the value returned for tokenizer 'blah' is now the
    52  #      same as that retrieved for 'simple'.
    53  #
    54  #   4: Test that it is now possible to create an fts3 table using 
    55  #      tokenizer 'blah' (it was not possible in step 1).
    56  #
    57  #   5: Test that the table created to use tokenizer 'blah' is usable.
    58  #
    59  sqlite3_db_config db SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER 1
    60  do_test fts3atoken-1.1 {
    61    catchsql {
    62      CREATE VIRTUAL TABLE t1 USING fts3(content, tokenize blah);
    63    }
    64  } {1 {unknown tokenizer: blah}}
    65  do_test fts3atoken-1.2 {
    66    execsql {
    67      SELECT fts3_tokenizer('blah', fts3_tokenizer('simple')) IS NULL;
    68    }
    69  } {0}
    70  do_test fts3atoken-1.3 {
    71    execsql {
    72      SELECT fts3_tokenizer('blah') == fts3_tokenizer('simple');
    73    }
    74  } {1}
    75  do_test fts3atoken-1.4 {
    76    catchsql {
    77      CREATE VIRTUAL TABLE t1 USING fts3(content, tokenize blah);
    78    }
    79  } {0 {}}
    80  do_test fts3atoken-1.5 {
    81    execsql {
    82      INSERT INTO t1(content) VALUES('There was movement at the station');
    83      INSERT INTO t1(content) VALUES('For the word has passed around');
    84      INSERT INTO t1(content) VALUES('That the colt from ol regret had got');
    85      SELECT content FROM t1 WHERE content MATCH 'movement'
    86    }
    87  } {{There was movement at the station}}
    88  
    89  unset -nocomplain simple blah2name simplename
    90  set simplename "simple"
    91  set blah2name "blah2"
    92  set simple [db one {SELECT fts3_tokenizer('simple')}]
    93  sqlite3_db_config db SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER 0
    94  do_catchsql_test 1.6 {
    95    SELECT fts3_tokenizer('blah', fts3_tokenizer('simple')) IS NULL;
    96  } {1 {fts3tokenize disabled}}
    97  do_test fts3atoken-1.7 {
    98    execsql {
    99      SELECT fts3_tokenizer('blah2', $simple) IS NULL;
   100    }
   101  } {1}
   102  
   103  # With ENABLE_FTS3_TOKENIZER off, the fts3_tokenzer(1) function
   104  # returns NULL unless the first parameter is a bound parameter.
   105  # If the first parameter is a bound parameter, then fts3_tokenizer(1)
   106  # returns the actual pointer value as a BLOB.
   107  #
   108  do_test fts3atoken-1.8 {
   109    execsql {
   110      SELECT fts3_tokenizer($blah2name) == fts3_tokenizer($simplename),
   111             typeof(fts3_tokenizer($blah2name)),
   112             typeof(fts3_tokenizer('blah2')),
   113             typeof(fts3_tokenizer($simplename)),
   114             typeof(fts3_tokenizer('simple'));
   115    }
   116  } {1 blob null blob null}
   117  
   118  # With ENABLE_FTS3_TOKENIZER on, fts3_tokenizer() always returns
   119  # the BLOB pointer, regardless the parameter
   120  #
   121  sqlite3_db_config db SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER 1
   122  do_test fts3atoken-1.9 {
   123    execsql {
   124      SELECT fts3_tokenizer('blah2') == fts3_tokenizer('simple'),
   125             typeof(fts3_tokenizer($blah2name)),
   126             typeof(fts3_tokenizer('blah2')),
   127             typeof(fts3_tokenizer($simplename)),
   128             typeof(fts3_tokenizer('simple'));
   129    }
   130  } {1 blob blob blob blob}
   131  
   132  # 2019-12-31:  The fts3_tokenizer() function can never be invoked from
   133  # within a trigger or view.
   134  #
   135  do_catchsql_test fts3atoken-1.10 {
   136    CREATE VIEW v110(x) AS
   137        SELECT fts3_tokenizer('tok110', fts3_tokenizer('simple')) IS NULL;
   138  } {0 {}}
   139  do_catchsql_test fts3atoken-1.11 {
   140    SELECT * FROM v110;
   141  } {1 {unsafe use of fts3_tokenizer()}}
   142  do_catchsql_test fts3atoken-1.12 {
   143    CREATE TABLE t110(a,b);
   144    CREATE TRIGGER r110 AFTER INSERT ON t110 BEGIN
   145        SELECT fts3_tokenizer('tok110', fts3_tokenizer('simple')) IS NULL;
   146    END;
   147  } {0 {}}
   148  do_catchsql_test fts3atoken-1.13 {
   149    INSERT INTO t110(a,b) VALUES(1,2);
   150  } {1 {unsafe use of fts3_tokenizer()}}
   151  do_catchsql_test fts3atoken-1.14 {
   152    SELECT * FROM t110;
   153  } {0 {}}
   154  
   155  #--------------------------------------------------------------------------
   156  # Test cases fts3atoken-2.* test error cases in the scalar function based
   157  # API for getting and setting tokenizers.
   158  #
   159  do_test fts3atoken-2.1 {
   160    catchsql {
   161      SELECT fts3_tokenizer('nosuchtokenizer');
   162    }
   163  } {1 {unknown tokenizer: nosuchtokenizer}}
   164  
   165  #--------------------------------------------------------------------------
   166  # Test cases fts3atoken-3.* test the three built-in tokenizers with a
   167  # simple input string via the built-in test function. This is as much
   168  # to test the test function as the tokenizer implementations.
   169  #
   170  sqlite3_db_config db SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER 1
   171  do_test fts3atoken-3.1 {
   172    execsql {
   173      SELECT fts3_tokenizer_test('simple', 'I don''t see how');
   174    }
   175  } {{0 i I 1 don don 2 t t 3 see see 4 how how}}
   176  do_test fts3atoken-3.2 {
   177    execsql {
   178      SELECT fts3_tokenizer_test('porter', 'I don''t see how');
   179    }
   180  } {{0 i I 1 don don 2 t t 3 see see 4 how how}}
   181  ifcapable icu {
   182    do_test fts3atoken-3.3 {
   183      execsql {
   184        SELECT fts3_tokenizer_test('icu', 'I don''t see how');
   185      }
   186    } {{0 i I 1 don't don't 2 see see 3 how how}}
   187  }
   188  
   189  #--------------------------------------------------------------------------
   190  # Test cases fts3atoken-4.* test the ICU tokenizer. In practice, this
   191  # tokenizer only has two modes - "thai" and "everybody else". Some other
   192  # Asian languages (Lao, Khmer etc.) require the same special treatment as 
   193  # Thai, but ICU doesn't support them yet.
   194  #
   195  ifcapable icu {
   196  
   197    proc do_icu_test {name locale input output} {
   198      set ::out [db eval { SELECT fts3_tokenizer_test('icu', $locale, $input) }]
   199      do_test $name {
   200        lindex $::out 0
   201      } $output
   202    }
   203    
   204    do_icu_test fts3atoken-4.1 en_US  {}   {}
   205    do_icu_test fts3atoken-4.2 en_US {Test cases fts3} [list \
   206      0 test Test 1 cases cases 2 fts3 fts3
   207    ]
   208  
   209    # The following test shows that ICU is smart enough to recognise
   210    # Thai chararacters, even when the locale is set to English/United 
   211    # States.
   212    #
   213    set input "\u0e2d\u0e30\u0e44\u0e23\u0e19\u0e30\u0e04\u0e23\u0e31\u0e1a"
   214    set output    "0 \u0e2d\u0e30\u0e44\u0e23 \u0e2d\u0e30\u0e44\u0e23 "
   215    append output "1 \u0e19\u0e30 \u0e19\u0e30 "
   216    append output "2 \u0e04\u0e23\u0e31\u0e1a \u0e04\u0e23\u0e31\u0e1a"
   217  
   218    do_icu_test fts3atoken-4.3 th_TH  $input $output
   219    do_icu_test fts3atoken-4.4 en_US  $input $output
   220  
   221    # ICU handles an unknown locale by falling back to the default.
   222    # So this is not an error.
   223    do_icu_test fts3atoken-4.5 MiddleOfTheOcean  $input $output
   224  
   225    set    longtoken "AReallyReallyLongTokenOneThatWillSurelyRequire"
   226    append longtoken "AReallocInTheIcuTokenizerCode"
   227  
   228    set    input "short tokens then "
   229    append input $longtoken
   230    set    output "0 short short "
   231    append output "1 tokens tokens "
   232    append output "2 then then "
   233    append output "3 [string tolower $longtoken] $longtoken"
   234  
   235    do_icu_test fts3atoken-4.6 MiddleOfTheOcean  $input $output
   236    do_icu_test fts3atoken-4.7 th_TH  $input $output
   237    do_icu_test fts3atoken-4.8 en_US  $input $output
   238  
   239    do_execsql_test 5.1 {
   240      CREATE VIRTUAL TABLE x1 USING fts3(name,TOKENIZE icu en_US);
   241      insert into x1 (name) values (NULL);
   242      insert into x1 (name) values (NULL);
   243      delete from x1;
   244    }
   245  
   246    proc cp_to_str {codepoint_list} {
   247      set fmt [string repeat %c [llength $codepoint_list]]
   248      eval [list format $fmt] $codepoint_list
   249    }
   250  
   251    do_test 5.2 {
   252      set str [cp_to_str {19968 26085 32822 32645 27874 23433 20986}]
   253      execsql { INSERT INTO x1 VALUES($str) }
   254    } {}
   255  }
   256  
   257  do_test fts3atoken-internal {
   258    execsql { SELECT fts3_tokenizer_internal_test() }
   259  } {ok}
   260  
   261  #-------------------------------------------------------------------------
   262  # Test empty tokenizer names.
   263  #
   264  do_catchsql_test 6.1.1 {
   265    CREATE VIRTUAL TABLE t3 USING fts4(tokenize="");
   266  } {1 {unknown tokenizer: }}
   267  do_catchsql_test 6.1.2 {
   268    CREATE VIRTUAL TABLE t3 USING fts4(tokenize=);
   269  } {1 {unknown tokenizer: }}
   270  do_catchsql_test 6.1.3 {
   271    CREATE VIRTUAL TABLE t3 USING fts4(tokenize="   ");
   272  } {1 {unknown tokenizer:    }}
   273  
   274  do_catchsql_test 6.2.1 {
   275    SELECT fts3_tokenizer(NULL);
   276  } {1 {unknown tokenizer: }}
   277  
   278  sqlite3_db_config db SQLITE_DBCONFIG_ENABLE_FTS3_TOKENIZER 1
   279  do_catchsql_test 6.2.2 {
   280    SELECT fts3_tokenizer(NULL, X'1234567812345678');
   281  } {1 {argument type mismatch}}
   282  do_catchsql_test 6.2.3 {
   283    SELECT fts3_tokenizer(NULL, X'12345678');
   284  } {1 {argument type mismatch}}
   285  
   286  
   287  finish_test