Hello.
TSearch2 allows to search a table of tsvectors by a single tsquery.
I need to solve the reverse problem.
I have a large table of tsquery. I need to find all tsqueries in that table that match a single document tsvector:
CREATE TABLE "test"."test_tsq" (
"id" SERIAL,
"q" TSQUERY NOT NULL,
CONSTRAINT "test_tsq_pkey" PRIMARY KEY("id")
);
insert into test.test_tsq(q)
select to_tsquery(g || 'x' || g) from generate_series(100000, 900000) as g;
explain analyze
select * from test.test_tsq
where to_tsvector('400000x400000') @@ q
This gets a strange explain analyze:
QUERY PLAN
Seq Scan on test_tsq (cost=0.00..17477.01 rows=800 width=36) (actual time=68.698..181.458 rows=1 loops=1)
Filter: ('''400000x400000'':1'::tsvector @@ q)
Total runtime: 181.484 ms
No matter if I use GIST index on test_tsq.q or not: the explain analyze result is the same.
So, why "rows=800"? The table contains much more rows...
TSearch2 allows to search a table of tsvectors by a single tsquery.
I need to solve the reverse problem.
I have a large table of tsquery. I need to find all tsqueries in that table that match a single document tsvector:
CREATE TABLE "test"."test_tsq" (
"id" SERIAL,
"q" TSQUERY NOT NULL,
CONSTRAINT "test_tsq_pkey" PRIMARY KEY("id")
);
insert into test.test_tsq(q)
select to_tsquery(g || 'x' || g) from generate_series(100000, 900000) as g;
explain analyze
select * from test.test_tsq
where to_tsvector('400000x400000') @@ q
This gets a strange explain analyze:
QUERY PLAN
Seq Scan on test_tsq (cost=0.00..17477.01 rows=800 width=36) (actual time=68.698..181.458 rows=1 loops=1)
Filter: ('''400000x400000'':1'::tsvector @@ q)
Total runtime: 181.484 ms
No matter if I use GIST index on test_tsq.q or not: the explain analyze result is the same.
So, why "rows=800"? The table contains much more rows...
No comments:
Post a Comment