SQL query for finding checkins from friends

I have the following tables in a PostgreSQL:

Categories | Locations   | Checkins     | Users  | Friendships
id         | id          | id           | id     | user_id
name       | category_id | location_id  | gender | friend_id
icon       | name        | user_id      |        |

Now, i want to retrieve the following information about a venue

• How many female and male users a location has
• Category name and icon
• Location name
• How many friends have checked in at a location (from a given user id)

Except the last point, I solved it. But I have troubles to count the friends from a given user id. I tried it with this query:

SELECT distinct locations.id,
 max(locations.name) as name,
 max(locations.location) as location,
 max(categories.name) as cat,
 max(categories.icon) as caticon,
 SUM(CASE WHEN users.gender = 'm' THEN 1 ELSE 0 END) AS male,
 SUM(CASE WHEN users.gender = 'f' THEN 1 ELSE 0 END) AS female,
 SUM(CASE WHEN friendships.user_id = 1 OR friend开发者_如何学Goships.friend_id=1 THEN 1 ELSE 0 END) AS friends
 FROM locations
 INNER JOIN checkins ON checkins.location_id = locations.id 
 INNER JOIN users ON users.id = checkins.user_id 
 INNER JOIN categories ON categories.id = locations.category_id
 LEFT JOIN friendships ON friendships.user_id = users.id OR friendships.friend_id = users.id
 WHERE locations.id=7
 GROUP BY locations.id

But I get a wrong number of the count for female users. Any idea what I'm doing wrong? I think I need a left join for the friendships table, because if a user has no friends (or no user is given) it should only return 0 for the friend count.

Hope I made myself clear, thx, tux

---

SELECT
  L.id,
  L.name,
  c.name AS cat,
  c.icon AS caticon,
  COUNT(CASE u.gender WHEN 'm' THEN 1 END) AS male,
  COUNT(CASE u.gender WHEN 'f' THEN 1 END) AS female,
  COUNT(f.user_id) AS friends
FROM Locations L
  INNER JOIN Categories c ON c.id = L.category_id
  INNER JOIN Checkins ch ON ch.location_id = L.id
  INNER JOIN Users u ON u.id = ch.user_id
  LEFT JOIN Friendships f ON f.user_id = @user_id AND f.friend_id = ch.user_id
                          OR f.user_id = ch.user_id AND f.friend_id = @user_id
WHERE L.id = @location_id
GROUP BY L.id, L.name, c.name, c.icon

---

Drop the distinct in the first line. You already have a group by clause for the same field. Let me know if that helps.