从复杂的MySQL中删除重复项选择查询

Question

All my projects are stored in the com_projects table. Each project can have multiple locations, which are stored in the com_location table. The com_country table is just a list of country names, which is use by the com_location table to pull in the name of the countries.

In the com_project table I have a category_id field that can have a value from 1-4.

The SELECT query below is my attempt to count the total number of times a category is assigned per country.

---------------------------------------------------------------
| country | category_1 | category_2 | category_3 | category_4 |
---------------------------------------------------------------
| USA     |     20     |      5     |     3      |     0      |
---------------------------------------------------------------
| UK      |     1      |      12    |     0      |     0      |

etc....

SELECT b.country_id, c.name,
SUM(case when a.category_id = 1 then 1 else 0 end) as category_1,
SUM(case when a.category_id = 2 then 1 else 0 end) as category_2,
SUM(case when a.category_id = 3 then 1 else 0 end) as category_3,
SUM(case when a.category_id = 4 then 1 else 0 end) as category_4
FROM com_project a
Inner JOIN com_location b 
ON a.id = b.project_id
INNER JOIN com_country c
ON c.id = b.country_id
WHERE a.state = 1
AND b.state = 1
GROUP BY b.country_id

I would only like a category to be incremented if a project has one or more locations in a different countries. The problem I have is that many projects have multiple location in the same country and it's artificially increasing the results.

How can I tweak my SELECT statement to prevent duplicates from the same country?

By the way I tried adding DISTINCT immediately after the SELECT and it didn't help.

Answer 1

I think a where clause will do for filtering out projects that are always in one country:

SELECT l.country_id, c.name,
       SUM(p.category_id = 1) as category_1,
       SUM(p.category_id = 2) as category_2,
       SUM(p.category_id = 3) as category_3,
       SUM(p.category_id = 4) as category_4
FROM com_project p Inner JOIN
     com_location l
     ON p.id = l.project_id INNER JOIN
     com_country c
     ON c.id = l.country_id
WHERE p.state = 1 AND l.state = 1 AND
      EXISTS (SELECT 1
              FROM com_location l2
              WHERE l2.project_id = l.project_id AND l2.country <> l.country
             )
GROUP BY l.country_id, c.name;

This just checks that each project being aggregated has at least one other country.