计算 varchar 数组 Postgres 中不同元素的数量

Question

I have a table.

title   | tags
--------------------------
post1   | {tag1,tag2,tag3}
post2   | {tag1,tag2}
post3   | {tag1}

It was made with this:

CREATE TABLE post (
    title varchar(10),
    tags varchar(10) array[5]
);

INSERT INTO post (title, tags) VALUES ('post1', '{"tag1", "tag2", "tag3"}');
INSERT INTO post (title, tags) VALUES ('post2', '{"tag1", "tag2"}');
INSERT INTO post (title, tags) VALUES ('post3', '{"tag1"}');

How could I count the number of distinct elements in the array for the whole table? For example, the tag tag1 appears in all three rows, so it would return 3. I want to do this for every element, creating an output that looks like this:

tag     | COUNT
--------------------------
tag1    | 3
tag2    | 2
tag3    | 1

The tags array will not have any duplicates.

Answer 1

You can unnest() the array, then aggregate:

select t.tag, count(*) no_posts
from post p
cross join lateral unnest(p.tags) t(tag)
group by t.tag
order by no_posts desc

Demo on DB Fiddle :

tag  | no_posts
:--- | -------:
tag1 |        3
tag2 |        2
tag3 |        1