-- 0.8.0: the coarse index (find_ivf). Probing every list it answers exactly what find answers; it
-- never serves a reading whose text changed; and a reading recorded after the build is found.
\set VERBOSITY terse
CREATE EXTENSION pg_phasevec;
SELECT extversion FROM pg_extension WHERE extname = 'pg_phasevec';
 extversion 
------------
 0.16.0
(1 row)

SELECT phasevec.reader_register('r256', repeat('a', 64), 256) > 0 AS r256;
 r256 
------
 t
(1 row)

CREATE TABLE docs (id int PRIMARY KEY, body text);
INSERT INTO docs SELECT g, 'doc ' || g FROM generate_series(1, 2000) g;
SELECT phasevec.watch('docs', 'docs'::regclass, 'id', 'r.body', 'r256') > 0 AS watch;
 watch 
-------
 t
(1 row)

-- deterministic codes of 256 dimensions; code256 = their signs
CREATE FUNCTION code_of(g int) RETURNS smallint[] LANGUAGE sql IMMUTABLE AS $$
    SELECT array(SELECT (((g * 31 + j * 17 * (g % 7 + 1) + (g / 13) * j) % 255) - 127)::smallint
                   FROM generate_series(1, 256) j) $$;
CREATE FUNCTION signs_of(c smallint[]) RETURNS bit(256) LANGUAGE sql IMMUTABLE AS $$
    SELECT string_agg(CASE WHEN x > 0 THEN '1' ELSE '0' END, '' ORDER BY o)::bit(256)
      FROM unnest(c) WITH ORDINALITY u(x, o) $$;
SELECT count(phasevec.record('docs', s.key, 'r256', s.text_hash, 0::bigint, code_of(s.key::int), NULL,
                             signs_of(code_of(s.key::int))))
  FROM phasevec.sources s;
 count 
-------
  2000
(1 row)

-- no index yet: find_ivf says so, it does not fall back to anything
SELECT * FROM phasevec.find_ivf('docs', 0, code_of(100), 10, false, 200, signs_of(code_of(100)), 4);
ERROR:  phasevec: watch docs has no coarse index for its reader
SELECT phasevec.ivf_build('docs') AS lists;
 lists 
-------
    45
(1 row)

-- every list probed and m = every row: exactly find's answer
SELECT (SELECT array_agg(key ORDER BY score DESC, key) FROM phasevec.find('docs', 0, code_of(100), 10, false, 2000, signs_of(code_of(100))))
     = (SELECT array_agg(key ORDER BY score DESC, key) FROM phasevec.find_ivf('docs', 0, code_of(100), 10, false, 2000, signs_of(code_of(100)), 1000))
       AS same_as_find;
 same_as_find 
--------------
 t
(1 row)

-- the question's own row comes first, with a few lists
SELECT key, round(score::numeric, 4) AS score, state FROM phasevec.find_ivf('docs', 0, code_of(100), 1, false, 200, signs_of(code_of(100)), 8);
 key | score  | state 
-----+--------+-------
 100 | 1.0000 | fresh
(1 row)

-- its text changes: in the same transaction it is no longer served (its reading is stale)
BEGIN;
UPDATE docs SET body = 'changed' WHERE id = 100;
SELECT count(*) AS served_stale FROM phasevec.find_ivf('docs', 0, code_of(100), 10, false, 2000, signs_of(code_of(100)), 1000) WHERE key = '100';
 served_stale 
--------------
            0
(1 row)

ROLLBACK;
-- a row written and read after the build: found without rebuilding (the delta)
INSERT INTO docs VALUES (5000, 'doc 5000');
SELECT count(phasevec.record('docs', s.key, 'r256', s.text_hash, 0::bigint, code_of(100), NULL, signs_of(code_of(100))))
  FROM phasevec.sources s WHERE s.key = '5000';
 count 
-------
     1
(1 row)

SELECT key FROM phasevec.find_ivf('docs', 0, code_of(100), 2, false, 200, signs_of(code_of(100)), 8) ORDER BY key;
 key  
------
 100
 5000
(2 rows)

-- coverage still refused below the declared minimum, as find does
INSERT INTO docs SELECT g, 'new ' || g FROM generate_series(6000, 6500) g;
SELECT count(*) FROM phasevec.find_ivf('docs', 0, code_of(100), 10, false, 200, signs_of(code_of(100)), 8);
ERROR:  phasevec: watch docs covers 80.0% of its rows and declared 95.0%: pending 20.0%
DROP TABLE docs;
DROP FUNCTION code_of(int);
DROP FUNCTION signs_of(smallint[]);
DROP EXTENSION pg_phasevec;