当CASE不存在时，返回0表示SQL计数

Question

I have two tables Log and Player. Where Log stores each game play log with a playerId and date. and the Player table has the Players info as Age and Gender ..etc. I'm writing an SQL stored procedure that takes two dates and will count the LogId and group by age range and gender between the two dates. but when i run the SQL procedure, it doesn't show all the Age_Range/Gender entries when no player exists in that period.I'm trying to get all Age_Range/Gender entries with the their actual count or count= 0 if they don't exist. I've even tried changing

count(L.LogId) as Count, 

To

count(IFNULL(L.LogId, 0)) as Count,

My SQL procedure is :

CREATE PROCEDURE `SHOW_AGE_RANGE` (IN date1 CHAR(10), IN date2 CHAR(10))
 BEGIN
 SELECT
  count(L.LogId) as Count,
  CASE
     WHEN P.age BETWEEN 13 AND 18 THEN '13-18'
     WHEN P.age BETWEEN 19 AND 25 THEN '19-25'
     WHEN P.age BETWEEN 26 AND 39 THEN '26-39'
     WHEN P.age BETWEEN 40 AND 59 THEN '40-59'
     WHEN P.age > 59 THEN '60+' 
 END as Age_Range,
 CASE
    WHEN P.gender  = 0 then 'Female'
    WHEN P.gender  = 1 then 'Male'
 END  as Gender    
 FROM Log L
 LEFT JOIN Player P ON L.playerId = P.playerId 

 WHERE CAST(L.createdDate AS DATE) BETWEEN CAST(date1 AS DATE) AND CAST(date2 AS DATE)  
 GROUP BY Age_Range, Gender;    
END

Output of the SQL Procedure:

Count  Age_Range  Gender 
----------------------------------
'1'    '13-18'    'Male'
'1'    '19-25'    'Female'
'3'    '26-39'    'Female'
'2'    '40-59'    'Male'
'1'    '60+'      'Female'

The Expected Output

Count  Age_Range  Gender 
----------------------------------
'0'    '13-18'    'Female'
'1'    '13-18'    'Male'
'1'    '19-25'    'Female'
'0'    '19-25'    'Male'
'3'    '26-39'    'Female'
'0'    '26-39'    'Male'
'0'    '40-59'    'Female'
'2'    '40-59'    'Male'
'1'    '60+'      'Female'
'0'    '60+'      'Male'

Answer 1

You need to start out with your age ranges and genders. Really, you're querying off of those and the player data is just something that you're adding to them. Since you don't already have the age ranges in a table (you should probably consider adding that) then you'll need to create a virtual table for those as a subquery.

SELECT
    COUNT(L.playerId) AS cnt,
    AG.age_range,
    G.gender
FROM
(
    SELECT 13 AS min_age, 18 AS max_age, '13-18' AS age_range
    UNION ALL
    SELECT 19 AS min_age, 25 AS max_age, '19-' AS age_range
    UNION ALL
    SELECT 26 AS min_age, 39 AS max_age, '26-39' AS age_range
    UNION ALL
    SELECT 40 AS min_age, 59 AS max_age, '40-59' AS age_range
    UNION ALL
    SELECT 60 AS min_age, 999 AS max_age, '60+' AS age_range
) AS AG
CROSS JOIN
(
    SELECT 0 AS gender_value, 'Female' AS gender
    UNION ALL
    SELECT 1 AS gender_value, 'Male' AS gender
) AS G
LEFT JOIN Player P ON
    P.age BETWEEN AG.min_age AND AG.max_age AND
    P.gender = G.gender_value
LEFT OUTER JOIN Log L ON
    L.playerId = P.playerId AND
    CAST(L.createdDate AS DATE) BETWEEN CAST(date1 AS DATE) AND CAST(date2 AS DATE)
GROUP BY
    AG.age_range, G.gender

Answer 2

It looks like you want the outer join to the Log table. Swap the Player and Log tables, and relocate the predicate on the createdDate column to the ON clause.

Like this:

  FROM Player P
  LEFT 
  JOIN Log L
    ON L.playerId = P.playerId 
   AND CAST(L.createdDate AS DATE) BETWEEN CAST(date1 AS DATE) AND CAST(date2 AS DATE)
 GROUP BY Age_Range, Gender;