sqlite/ext/vec1/test/vec1cat.test

# 2026-01-17
#
# 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.
#
#***********************************************************************
#
#

source [file join [file dirname [info script]] vec1_common.tcl]
set testprefix vec1cat
vec1_load db

do_execsql_test 1.0 {
  PRAGMA table_info = vec1cat
} {
  0 database {} 0 {} 0 
  1 name {} 0 {} 0 
  2 model {} 0 {} 0 
  3 qprofile {} 0 {} 0 
}

do_execsql_test 1.1 {
  CREATE VIRTUAL TABLE x1 USING vec1(x);
  SELECT database, name FROM vec1cat;
} {
  main x1
}

db close
sqlite3 db test.db
vec1_load db

do_execsql_test 1.2 {
  SELECT database, name, model FROM vec1cat;
} {
  main x1 {{"index": "none", "distance": "l2"}}
}

do_execsql_test 1.3 {
  INSERT INTO x1(cmd, arg) VALUES( 'rebuild', '{index:"flat", distance:"cos"}');
}

do_execsql_test 1.4 {
  SELECT database, name, model FROM vec1cat;
} {
  main x1 {{"index": "flat", "distance": "cos"}}
}

db close
sqlite3 db test.db
vec1_load db

do_execsql_test 1.5 {
  SELECT database, name, model FROM vec1cat;
} {
  main x1 {{"index": "flat", "distance": "cos"}}
}

do_execsql_test 1.6 {
  INSERT INTO x1(cmd, arg) VALUES('rebuild', (
    WITH s(i) AS (SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<1000)
    SELECT vec1_train(random_vector(16), '{nbucket:8, codesize:8}') FROM s
  ));
  SELECT database, name, model FROM vec1cat;
} {
  main x1
  {{"index": "trained", "distance": "l2", "nbucket": 8, "quantizer": "pq", "codesize": 8, "residual": true, "wht": false}}
}

do_execsql_test 1.7 {
  CREATE VIRTUAL TABLE temp.x2 USING vec1(x);
  SELECT rowid, database, name, model FROM vec1cat;
} {
  1 temp x2
  {{"index": "none", "distance": "l2"}}
  2 main x1
  {{"index": "trained", "distance": "l2", "nbucket": 8, "quantizer": "pq", "codesize": 8, "residual": true, "wht": false}}
}

foreach {tn spec res} {

  1 {nbucket:0, codesize:8} {
    "index": "trained", "distance": "l2", "nbucket": 0, 
    "quantizer": "pq", "codesize": 8, 
    "residual": false, "wht": false
  }

  2 {nbucket:8, codesize:0} {
    "index": "trained", "distance": "l2", "nbucket": 8, 
    "quantizer": "none", "codesize": 0, 
    "residual": false, "wht": false
  }

  3 {nbucket:0, codesize:8, opq:1} {
      "index": "trained", "distance": "l2", "nbucket": 0, 
      "quantizer": "opq", "codesize": 8, 
      "residual": false, "wht": false
  }

  4 {codesize:2, quantizer: "bq", distance: "cos"} {
      "index": "trained", "distance": "cos", "nbucket": 0, 
      "quantizer": "bq", "codesize": 2,
      "residual": false, "wht": true
  }

  5 {codesize:1, nbucket: 16, quantizer: "bq", distance: "l2", wht: 0} {
      "index": "trained", "distance": "l2", "nbucket": 16, 
      "quantizer": "bq", "codesize": 1,
      "residual": true, "wht": false
  }

  6 {codesize:0, nbucket: 0} {
    "index":"flat", "distance":"l2"
  }

  7 {} {
    "index":"flat", "distance":"l2"
  }

} {

  set model [db one {
    WITH s(i) AS (SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<1000)
    SELECT vec1_train(random_vector(16), '{'||$spec||'}') FROM s
  }]

  set output [db one { SELECT json('{' || $res || '}') }]

  do_execsql_test 1.8.$tn.1 {
    INSERT INTO x1(cmd, arg) VALUES( 'rebuild', $model );
    SELECT json(model) FROM vec1cat WHERE name='x1';
  } [list $output]

  if {[db one {SELECT json_extract($output, '$.index')=='trained'}]} {
    set model [db one {
      WITH s(i) AS (SELECT 1 UNION ALL SELECT i+1 FROM s WHERE i<1000)
        SELECT vec1_train(random_vector(16), $output) FROM s
    }]
  } else {
    set model $output
  }
  do_execsql_test 1.8.$tn.2 {
    INSERT INTO x1(cmd, arg) VALUES( 'rebuild', $model );
    SELECT json(model) FROM vec1cat WHERE name='x1';
  } [list $output]
}

#-------------------------------------------------------------------------
reset_db
vec1_load db

do_execsql_test 2.0 {
  CREATE VIRTUAL TABLE x1 USING vec1(x);
  CREATE VIRTUAL TABLE x2 USING vec1(y);
  CREATE VIRTUAL TABLE x3 USING vec1(z);
}

do_execsql_test 2.1 {
  SELECT rowid, name FROM vec1cat
} {
  1 x3 2 x2 3 x1
}

do_test 2.2 {
  list [catch {
    db eval {
      SELECT rowid, name FROM vec1cat
    } {
      if {$name=="x2"} {
        db eval {
          DROP TABLE x2;
          DROP TABLE x1;
        }
      }
    }
  } msg] $msg
} {1 {database table is locked}}

do_execsql_test 2.3 {
  SELECT qprofile FROM vec1cat WHERE name='x1';
} {{{
  "total": {"cnt": 0, "cycle": 0},
  "init": {"cnt": 0, "cycle": 0},
  "coarse": {"cnt": 0, "cycle": 0},
  "idxread": {"cnt": 0, "cycle": 0},
  "metaread": {"cnt": 0, "cycle": 0},
  "metascan": {"cnt": 0, "cycle": 0},
  "lut": {"cnt": 0, "cycle": 0},
  "pqscan": {"cnt": 0, "cycle": 0},
  "finalsort": {"cnt": 0, "cycle": 0}
}}}

#-------------------------------------------------------------------------
reset_db
vec1_load db
do_execsql_test 3.0 {
  CREATE VIRTUAL TABLE x2 USING vec1(y);
  INSERT INTO x2 VALUES( random_vector(13) );
  INSERT INTO x2 VALUES( random_vector(13) );
  INSERT INTO x2 VALUES( random_vector(13) );
  INSERT INTO x2 VALUES( random_vector(13) );
  INSERT INTO x2 VALUES( random_vector(13) );

  INSERT INTO x2(cmd, arg) VALUES('rebuild', (
    WITH s(i) AS ( VALUES (1) UNION ALL SELECT i+1 FROM s WHERE i<512 )
    SELECT vec1_train( random_vector(13), '{nbucket:8}' ) FROM s
  ))
}

do_execsql_test 3.1 {
  SELECT count(distance) FROM x2;
} 5

#-------------------------------------------------------------------------
reset_db
vec1_load db
do_execsql_test 4.0 {
  CREATE VIRTUAL TABLE x1 USING vec1(y, a);

  INSERT INTO x1(cmd, arg) VALUES('rebuild', (
    WITH s(i) AS ( VALUES (1) UNION ALL SELECT i+1 FROM s WHERE i<512 )
    SELECT vec1_train( 
      random_vector(32), '{nbucket:8, codesize:8, distance:"cos"}' 
    ) FROM s
  ))
}

do_execsql_test 4.1 {
  INSERT INTO x1 (rowid, y, a) VALUES(10, random_vector(32), 4556);
} 

do_execsql_test 4.2 {
  SELECT key FROM json_each( (SELECT distance FROM x1) )
} {bucket coarse_error reconstruction_error code}

do_execsql_test 4.3 {
  SELECT typeof(distance) FROM x1(random_vector(32), '{nprobe:8, K:5}')
} {real}


finish_test