Question

I couldn’t find answer for my question, and I don’t know if my query is correct and this could be a SQLite issue, please help me solve the problem.

I have two tables in my database:

processTable {id}
taskTable {id, processId, amount, done}

There is a many-to-one relation (one process can have multiple tasks assigned). The “amount” and “done” are integer values that provides task progress information. If “done” >= “amount”, the task is done. I need to query database to get something like that:

+---------+-----------+------------+
| process | tasksDone | tasksCount |
+---------+-----------+------------+
| 1       | 1         | 3          |
+---------+-----------+------------+
| 2       | 2         | 5          |
+---------+-----------+------------+

Basing on data that I have in my tables:

processTable
+----+
| id |
+----+
| 1  |
+----+
| 2  |
+----+

tasksTable
+----+-----------+--------+------+
| id | processId | amount | done |
+----+-----------+--------+------+
| 1  | 1         | 10     | 10   | <- this task is done
+----+-----------+--------+------+
| 2  | 1         | 15     | 5    |
+----+-----------+--------+------+
| 3  | 1         | 80     | 5    |
+----+-----------+--------+------+
| 4  | 2         | 25     | 0    |
+----+-----------+--------+------+
| 5  | 2         | 60     | 60   | <- this task is done
+----+-----------+--------+------+
| 6  | 2         | 30     | 15   |
+----+-----------+--------+------+
| 7  | 2         | 40     | 40   | <- this task is done
+----+-----------+--------+------+
| 8  | 2         | 100    | 50   |
+----+-----------+--------+------+

So, I wrote this query:

SELECT processTable.id AS process, 
       COUNT(tasksTableDone.id) AS tasksDone, 
       COUNT(tasksTableAll.id) AS tasksCount 

FROM processTable

LEFT JOIN tasksTable AS tasksTableAll 
     ON   tasksTableAll.processId = processTable.id 

LEFT JOIN tasksTable AS tasksTableDone 
     ON   tasksTableDone.processId = processTable.id 
          AND
          tasksTableDone.done >= tasksTableDone.amount 

But what I’ve got is:

+---------+-----------+------------+
| process | tasksDone | tasksCount |
+---------+-----------+------------+
| 1       | 3         | 3          |
+---------+-----------+------------+
| 2       | 5         | 5          |
+---------+-----------+------------+

I was trying run the query with only one join at a time, and everything was working well.

Query with first join only:

SELECT processTable.id AS process,  
       COUNT(tasksTableAll.id) AS tasksCount 

FROM processTable

LEFT JOIN tasksTable AS tasksTableAll 
     ON   tasksTableAll.processId = processTable.id 

Result:
+---------+------------+
| process | tasksCount |
+---------+------------+
| 1       | 3          |
+---------+------------+
| 2       | 5          |
+---------+------------+

Query with second join only:

SELECT processTable.id AS process,  
       COUNT(tasksTableDone.id) AS tasksDone 

FROM processTable

LEFT JOIN tasksTable AS tasksTableDone 
     ON   tasksTableDone.processId = processTable.id 
          AND
          tasksTableDone.done >= tasksTableDone.amount 

Result:
+---------+-----------+
| process | tasksDone |
+---------+-----------+
| 1       | 1         |
+---------+-----------+
| 2       | 2         |
+---------+-----------+

How to use this two joins within one query to get proper results? I know that instead of JOIN I could use another SELECT, but I think it would be more expensive in the performance meaning.

Answer 1

You can implement a CASE statement with an aggregate:

Version using SUM()

SELECT p.id AS process,  
  sum(case when t.amount = t.done then 1 else 0 end) AS tasksDone,
  count(p.id) AS tasksCount
FROM processTable p
LEFT JOIN tasksTable t
  ON t.processId = p.id 
group by p.id

See SQL Fiddle with Demo

Version using COUNT():

SELECT p.id AS process,  
  count(case when t.amount = t.done then 1 else null end) AS tasksDone,
  count(p.id) AS tasksCount
FROM processTable p
LEFT JOIN tasksTable t
  ON t.processId = p.id 
group by p.id

See SQL Fiddle with Demo

Edit, after your comment you can wrap this in a select to get the progress:

select process,
  tasksDone,
  tasksCount,
  (tasksDone / tasksCount) progress
from
(
  SELECT p.id AS process,  
    count(case when t.amount = t.done then 1 else null end) AS tasksDone,
    count(p.id) AS tasksCount
  FROM processTable p
  LEFT JOIN tasksTable t
    ON t.processId = p.id 
  group by p.id
) src