roaringbitmap

pig build pkg pg_roaringbitmap         # 构建 RPM 包

pig repo add pgdg -u          # 添加 PGDG 仓库并更新缓存

CREATE EXTENSION roaringbitmap;

CREATE EXTENSION IF NOT EXISTS roaringbitmap;

SET roaringbitmap.output_format = 'array';
SELECT rb_build(ARRAY[1, 2, 3, 4, 5]);

CREATE TABLE cohorts (
  segment text PRIMARY KEY,
  users32 roaringbitmap,
  users64 roaringbitmap64
);

INSERT INTO cohorts(segment, users32)
VALUES ('trial', rb_build(ARRAY[1, 2, 3, 100, 200]));

INSERT INTO cohorts(segment, users32)
SELECT 'active', rb_build_agg(id)
FROM generate_series(1, 100000) AS id;

SELECT rb_cardinality(users32) FROM cohorts WHERE segment = 'active';
SELECT rb_to_array(users32) FROM cohorts WHERE segment = 'trial';
SELECT rb_iterate(users32) FROM cohorts WHERE segment = 'trial';

SELECT rb_build(ARRAY[1,2,3]) | rb_build(ARRAY[3,4,5]);  -- 并集
SELECT rb_build(ARRAY[1,2,3]) & rb_build(ARRAY[3,4,5]);  -- 交集
SELECT rb_build(ARRAY[1,2,3]) # rb_build(ARRAY[3,4,5]);  -- 异或
SELECT rb_build(ARRAY[1,2,3]) - rb_build(ARRAY[3,4,5]);  -- 差集

SELECT rb_build(ARRAY[1,2,3]) | 9;                       -- 添加元素
SELECT rb_build(ARRAY[1,2,3]) - 2;                       -- 移除元素

SELECT rb_build(ARRAY[1,2,3]) @> rb_build(ARRAY[2,3]);
SELECT rb_build(ARRAY[2,3]) <@ rb_build(ARRAY[1,2,3]);
SELECT rb_build(ARRAY[1,2,3]) && rb_build(ARRAY[3,4,5]);

SELECT rb_and_cardinality(a.users32, b.users32);
SELECT rb_or_cardinality(a.users32, b.users32);
SELECT rb_xor_cardinality(a.users32, b.users32);
SELECT rb_andnot_cardinality(a.users32, b.users32);

SELECT rb_range(users32, 100, 1000);
SELECT rb_range_cardinality(users32, 100, 1000);
SELECT rb_fill(users32, 100, 200);
SELECT rb_clear(users32, 100, 200);
SELECT rb_flip(users32, 100, 200);

SELECT rb_min(users32), rb_max(users32), rb_rank(users32, 500), rb_index(users32, 500);
SELECT rb_jaccard_dist(a.users32, b.users32);

SELECT rb_build_agg(user_id) FROM events;
SELECT rb_or_agg(users32) FROM cohorts;
SELECT rb_and_agg(users32) FROM cohorts;
SELECT rb_xor_agg(users32) FROM cohorts;

SELECT rb64_build_agg(user_id::bigint) FROM events;
SELECT rb64_or_agg(users64) FROM cohorts;

UPDATE cohorts
SET users32 = rb_runoptimize(users32);

UPDATE cohorts
SET users64 = rb64_runoptimize(users64);