Edit

BITMAP

Here is a simple example to illustrate the usage of several aggregate functions in Bitmap. For detailed function definitions or more Bitmap functions, see bitmap-functions.

Create table

The aggregation model is needed when creating table. The data type is bitmap and the aggregation function is bitmap_union.

CREATE TABLE `pv_bitmap` (
  `dt` int(11) NULL COMMENT "",
  `page` varchar(10) NULL COMMENT "",
  `user_id` bitmap BITMAP_UNION NULL COMMENT ""
) ENGINE=OLAP
AGGREGATE KEY(`dt`, `page`)
COMMENT "OLAP"
DISTRIBUTED BY HASH(`dt`) BUCKETS 2;

Note: With an amount of data, you'd better create a rollup table corresponding to high-frequent bitmap_union.

ALTER TABLE pv_bitmap ADD ROLLUP pv (page, user_id);

Data Load

TO_BITMAP (expr): Convert 0 ~ 18446744073709551615 unsigned bigint to bitmap

BITMAP_EMPTY (): Generate empty bitmap columns, used for the default value to be filled in when inserting or inputting

BITMAP_HASH (expr): Convert columns of any type to a bitmap by hashing

Stream Load

When inputting data using Stream Load, you can convert the data to a BItmap field as follows:

cat data | curl --location-trusted -u user:passwd -T - \
    -H "columns: dt,page,user_id, user_id=to_bitmap(user_id)" \
    http://host:8410/api/test/testDb/_stream_load
cat data | curl --location-trusted -u user:passwd -T - \
    -H "columns: dt,page,user_id, user_id=bitmap_hash(user_id)" \
    http://host:8410/api/test/testDb/_stream_load
cat data | curl --location-trusted -u user:passwd -T - \
    -H "columns: dt,page,user_id, user_id=bitmap_empty()" \
    http://host:8410/api/test/testDb/_stream_load

Insert Into

When inputting data using Insert Into, you need to select the corresponding mode based on the type of columns in the source table.

  • id2's column type in source table is bitmap
insert into bitmap_table1
select id, id2 from bitmap_table2;
  • id2's column type in target table is bitmap
insert into bitmap_table1 (id, id2)
values (1001, to_bitmap(1000))
, (1001, to_bitmap(2000));
  • id2's column type in source table is bitmap, and is the result of aggregation using bit_map_union().
insert into bitmap_table1
select id, bitmap_union(id2) from bitmap_table2 group by id;
  • id2's column type in source table is int, and the bitmap type is generated by to_bitmap().
insert into bitmap_table1
select id, to_bitmap(id2) from table;
  • id2's column type in source table is String, and the bitmap type is generated by bitmap_hash().
insert into bitmap_table1
select id, bitmap_hash(id2) from table;

Data Query

Syntax

`BITMAP_UNION (expr): Calculate the union of the input Bitmaps, and returns the new Bitmap.

BITMAP_UNION_COUNT (expr): Calculate the union of the input Bitmaps, and returns its cardinality, equivalent to BITMAP_COUNT (BITMAP_UNION (expr)). It is recommended to use the BITMAP_UNION_COUNT function first, for its performance is better than BITMAP_COUNT (BITMAP_UNION (expr)).

BITMAP_UNION_INT (expr): Calculate the number of different values in columns of type TINYINT, SMALLINT and INT, return the value same as COUNT (DISTINCT expr).

INTERSECT_COUNT (bitmap_column_to_count, filter_column, filter_values ...): Calculate the cardinality of the intersection of multiple bitmaps that satisfy filter_column condition. bitmap_column_to_count is a column of type bitmap, filter_column is a column of varying dimensions, and filter_values is a list of dimension values.

BITMAP_INTERSECT(expr): Calculate the intersection of this group of bitmap values and returns a new bitmap.

Example

The following SQL uses the pv_bitmap table above as an example:

Calculate the deduplicated value for user_id:

select bitmap_union_count(user_id)
from pv_bitmap;

select bitmap_count(bitmap_union(user_id))
from pv_bitmap;

Calculate the deduplicated value of id:

select bitmap_union_int(id)
from pv_bitmap;

Calculate the retention of user_id:

select intersect_count(user_id, page, 'meituan') as meituan_uv,
    intersect_count(user_id, page, 'waimai') as waimai_uv,
    intersect_count(user_id, page, 'meituan', 'waimai') as retention -- 在 'meituan' 和 'waimai' 两个页面都出现的用户数
from pv_bitmap
where page in ('meituan', 'waimai');