join a table into a field of a table containing multiple values - sql

Question

I have a POST table and a TAG table, I explain the POST table contains following fields : id , title , content , tags_ids which mean that the field tags_ids can contani multiple tags , for example the POST whose ID = 1, has following tags : tag_1, tag_2 tag_5 separated with ;
POST TABLE
id title content tag_id
---------- ---------- ---------- ----------
1 title1 Text... 1; 2; 5
2 title2 Text... 3
3 title3 Text... 1; 2
4 title4 Text... 2; 3; 4
5 title4 Text... 2; 3; 4
6 title2 Text... 3
the TAG table
id name
---------- ----------
1 tag_1
2 tag_2
3 tag_3
4 tag_4
5 tag_5
so i would like to know how many posts are registered for each case.
Here is my query
select tag, COUNT(*) AS cnt
from(
select CATEGORY.name,
case
when POST.tag_id is not null then tag.name
end as tag
from POST
left join TAG ON POST.tag_id = TAG.id
)
GROUP BY tag
;
here is the result i want to display with my query
tag cnt
-------------------- --------------
tag_1, tag_2, tag_5 1
tag_3 2
tag_1, tag_2 1
tag_2, tag_3, tag_4 2
Best regards

Answer 1

Here's one option. Read comments within code.
Sample data:
SQL> with
2 post(id, tag_id) as
3 (select 1, '1; 2; 5' from dual union all
4 select 2, '3' from dual union all
5 select 3, '1; 2' from dual union all
6 select 4, '2; 3; 4' from dual union all
7 select 5, '2; 3; 4' from dual union all
8 select 6, '3' from dual
9 ),
10 tag (id, name) as
11 (select 1, 'tag_1' from dual union all
12 select 2, 'tag_2' from dual union all
13 select 3, 'tag_3' from dual union all
14 select 4, 'tag_4' from dual union all
15 select 5, 'tag_5' from dual
16 ),
Query begins here:
17 post_distinct as
18 -- number of rows per each distinct TAG_ID
19 (select tag_id,
20 count(*) cnt
21 from post
22 group by tag_id
23 ),
24 temp as
25 -- split TAG_ID into rows
26 (select tag_id,
27 cnt,
28 trim(regexp_substr(tag_id, '[^;]+', 1, column_value)) tag_id_split
29 from post_distinct p cross join
30 table(cast(multiset(select level from dual
31 connect by level <= regexp_count(p.tag_id, ';') + 1
32 ) as sys.odcinumberlist))
33 )
34 -- finally, join tables to get the result
35 select listagg(t.name, ', ') within group (order by t.id) tag,
36 te.cnt
37 from tag t join temp te on te.tag_id_split = t.id
38 join post_distinct p on p.tag_id = te.tag_id
39 group by p.tag_id, te.cnt
40 order by p.tag_id;
TAG CNT
-------------------- ----------
tag_1, tag_2 1
tag_1, tag_2, tag_5 1
tag_2, tag_3, tag_4 2
tag_3 2
SQL>

Answer 2

You do not need to split the posts, to get your expected output you can just aggregate by the tag_ids:
SELECT tag_id,
COUNT(DISTINCT id) AS num_posts
FROM post
GROUP BY tag_id;
Which, for the sample data:
CREATE TABLE POST (id, title, content, tag_id) AS
SELECT 1, 'title1', 'Text...', '1; 2; 5' FROM DUAL UNION ALL
SELECT 2, 'title2', 'Text...', '3' FROM DUAL UNION ALL
SELECT 3, 'title3', 'Text...', '1; 2' FROM DUAL UNION ALL
SELECT 4, 'title4', 'Text...', '2; 3; 4' FROM DUAL UNION ALL
SELECT 5, 'title4', 'Text...', '2; 3; 4' FROM DUAL UNION ALL
SELECT 6, 'title2', 'Text...', '3' FROM DUAL;
CREATE TABLE TAG (id, name) AS
SELECT 1, 'tag_1' FROM DUAL UNION ALL
SELECT 2, 'tag_2' FROM DUAL UNION ALL
SELECT 3, 'tag_3' FROM DUAL UNION ALL
SELECT 4, 'tag_4' FROM DUAL UNION ALL
SELECT 5, 'tag_5' FROM DUAL;
Outputs:
TAG_ID
NUM_POSTS
1; 2
1
3
2
1; 2; 5
1
2; 3; 4
2
If you want to convert the ids to names then:
SELECT ( SELECT LISTAGG(name, '; ')
WITHIN GROUP (ORDER BY INSTR('; ' || p.tag_id || '; ', '; ' || t.id || '; '))
FROM tag t
WHERE INSTR('; ' || p.tag_id || '; ', '; ' || t.id || '; ') > 0
) AS tags,
COUNT(DISTINCT id) AS num_posts
FROM post p
GROUP BY tag_id;
Which outputs:
TAGS
NUM_POSTS
tag_1; tag_2
1
tag_1; tag_2; tag_5
1
tag_2; tag_3; tag_4
2
tag_3
2
If you want to count the posts for each individual tag then:
SELECT t.id,
COUNT(DISTINCT p.id) AS num_posts
FROM post p
INNER JOIN tag t
ON ('; ' || p.tag_id || '; ' LIKE '%; ' || t.id || '; %')
GROUP BY t.id
Which outputs:
ID
NUM_POSTS
2
4
4
2
3
4
5
1
1
2
fiddle