MySQL joining on specific result - mysql

Question

I am trying to build a query to extract data from a form builder where I can join on a field value only where another field value is equal to a specific. My query returns no rows.
users
------------------
|id | name |
------------------
|40 | John |
|45 | Michael |
|47 | Bob |
------------------
data_table
----------------------------------------------------
|id | submission | field_type | field_value |
----------------------------------------------------
|1 | 12345 | user | 40 |
|2 | 12345 | score | 5 |
|3 | 12345 | completed | 1 |
|4 | 23456 | user | 45 |
|5 | 23456 | score | 3 |
|6 | 23456 | completed | 0 |
|7 | 45678 | user | 47 |
|8 | 45678 | score | 2 |
|9 | 45678 | completed | 1 |
----------------------------------------------------
Desired result
---------------
|Name | Score |
---------------
|John | 5 |
|Bob | 2 |
---------------
Select
u.name,
dt2.field_value as score
from
users u
left join
data_table dt on u.id=dt.field_value and dt.field_type='user'
left join
data_table dt2 on dt.submission=dt2.submission and dt2.field_type='score'
where
(dt.field_type='completed' and dt.field_value=1)
http://sqlfiddle.com/#!2/d6e72f/3

Answer 1

You could do something like that (if you really can't change your data structure, which looks... strange).
You'll have a subquery on data_table, with two self join on that table (as you need 3 different rows with conditions)
select u.name, s.score
from users u
join (
select dt.field_value as user_id, dt1.field_value as score
from data_table dt
join data_table dt1 on dt1.submission = dt.submission
join data_table dt2 on dt2.submission = dt1.submission
where dt.field_type='user' and
dt1.field_type = 'score' and
dt2.field_type='completed' and
dt2.field_value = 1
) s
on s.user_id = u.id
see SqlFiddle

Answer 2

Working fiddle http://sqlfiddle.com/#!2/d6e72f/19/0
I've seen this a lot when users get to dynamically add additional attributes to an existing structure. you must first unpivot the additional data and then you can treat it as a normal table. since mysql doesn't support an unpivot, i used the normal work around.
Select u.name, score
from users u
INNER JOIN (
Select submission,
max(case when field_Type='user' then field_value end) as user,
max(case when field_Type='score' then field_value end) as score,
max(case when field_Type='completed' then field_value end) as completed
FROM data_table
group by submission) dt
on dt.user = u.id
and dt.completed = 1
This assumes that for a given submission there can't be more than one field_Type value combination. if there is, this will only return the max value.
Basically what this does is unpivot the data into a table structure that we can then join back to.
The reason we do max or min is so that we get one row back instead of multiple rows for a given submission. Again, this simply unpivots the data and combines the rows back into one. But is based on an assumption that no field_type and field_value will be duplicated within a submission.

Answer 3

Let's take a look at what you are actually doing so we can see what is going wrong:
Select
u.name,
dt2.field_value as score
from
users u
Get a list of users
left join
data_table dt on u.id=dt.field_value and dt.field_type='user'
Join only rows from data_table of type 'user'
left join
data_table dt2 on dt.submission=dt2.submission and dt2.field_type='score'
Join only rows from data_table of type 'score'.
Now your result set looks something like:
User, DT (data table rows for type 'USER'), DT2 (data table rows for type 'SCORE')
where
(dt.field_type='completed' and dt.field_value=1)
Filter the results to include only users where dt.field_type (previously filtered to only include type 'user') have type 'complete'.
Basically your joins filter out all 'complete' rows in 'data_table', so your where statement finds no matches. That is just an explanation of what is happening. On to your problem.
Looking at your schema, you have a few options. As much as I am not a fan of the design, here is how I would write your query:
SELECT U.name, SCORE_DT.field_value AS score
FROM user U
JOIN data_table DT ON DT.field_value=U.id AND DT.field_type="USER"
JOIN data_table SCORE_DT ON SCORE_DT.submission=DT.submission AND SCORE_DT.field_type="SCORE"
JOIN data_table COMPLETED_DT.submission=DT.submission AND COMPLETED_DT.field_type="COMPLETED" AND COMPLETED_DT.field_value=1
Realistically, it would make your life much easier to change the table design, as this data structure requires you to build queries that perform pivot operations for every column you are interested in. For a small data set like this one it is doable, but as the number of columns in your form increases it will become incredibly tedious to work with.

Answer 4

Another variation that works is ...
select x.name
, d.field_value
from data_table d
join (select u.name
, d2.submission
from users u
join data_table d2
on d2.field_value = u.id
and d2.field_type = 'user'
join data_table d3
on d2.submission = d3.submission
and d3.field_type = 'completed'
and d3.field_value = '1'
) x
on x.submission = d.submission
and d.field_type = 'score'
see SqlFiddle
For your set of data, you might find this or xQbert's answer to perform differently.
I would give them both a try. Based on your data, try to get the inner most query to return the smallest data set possible. For example, if you know that only a small subset of the data_table records will have 'completed' = '1', then a 3rd nested select might not be unreasonable if it results in a smaller result for MySql to work with.