MYSQL子查询并计算包含值的行数

Question

I have two tables oc_order o and oc_order_total ot

oc_order o has fields o.customer_id, o.date_added, o.email,o.total

And oc_order_total ot has fields ot.code and ot.value

I want to show results only if customer orders more than 3 times that is if customer_id repeats thrice or more in the result and show ot.value where ot.code = 'shipping'

I am trying to do following

SELECT COUNT(o.customer_id) AS 'Orders Count', o.date_added, o.email,o.total, ot.value
FROM oc_order o
Inner join oc_order_total ot ON ot.order_id = o.order_id
WHERE count(o.customer_id) > 3 AND ot.value = (select value from oc_order_total where code = 'shipping' )
GROUP BY o.customer_id

I am getting Invalid use of group error and I think I am not using subquery correctly in where clause.

Answer 1

You can not use SUM / COUNT in WHERE statement. You need to use for this HAVING operator. Try this query:

SELECT COUNT(o.customer_id) AS 'Orders Count', o.date_added, o.email,o.total, ot.value

FROM oc_order o
INNER JOIN oc_order_total ot ON ot.order_id = o.order_id AND 

WHERE ot.code IN (select value from oc_order_total where code = 'shipping' )

GROUP BY o.customer_id
HAVING count(o.customer_id) > 3

EDIT: Adding example with ot.value not affected by GROUP BY .

SELECT o.order_id, q.orders_count AS 'Orders Count', o.date_added, o.email,o.total, ot.value

FROM oc_order o
INNER JOIN oc_order_total ot ON ot.order_id = o.order_id
INNER JOIN (SELECT o.customer_id, COUNT(o.customer_id) AS orders_count

FROM oc_order o
INNER JOIN oc_order_total ot ON ot.order_id = o.order_id

WHERE ot.code IN (select value from oc_order_total where code = 'shipping' )

GROUP BY o.customer_id
HAVING count(o.customer_id) > 3) AS q ON q.customer_id = o.customer_id

Basically what happens here is that you pre-filter customers with previous query and then use these pre-filtered list to get individual orders for these customers meeting your criteria. On this individual orders_id you can perform any operation without grouping cause you already eliminated customers not meeting your needs. Hope it helps.

Answer 2

NO, you are getting error cause your GROUP BY doesn't contains all the column listed in SELECT list and you can't use COUNT() aggregate function like that in WHERE condition; it's only allowed in HAVING clause. You can modify your query like

SELECT o.date_added
     , o.email
     , o.total
     , ot.value
  FROM oc_order o
  JOIN oc_order_total ot 
    ON ot.order_id = o.order_id
  JOIN
     ( SELECT COUNT(customer_id) AS 'Orders Count'
            , customer_id
         FROM oc_order
        GROUP 
           BY customer_id 
     ) xxx 
    ON xxx .customer_id = o.customer_id
 WHERE  xxx.`Orders Count` > 3 
   AND ot.code = 'shipping';

Answer 3

For one thing, aggregates cannot be referenced in the WHERE clause.

The predicates in the WHERE clause are evaluated when the rows are accessed. At the time the rows are being retrieved, MySQL doesn't have any information about the value returned by aggregate functions (eg COUNT() ).

The COUNT() aggregate will be evaluated after the rows are accessed, and after the GROUP BY operation.

An aggregate can be referenced in a HAVING clause, following the GROUP BY clause.

 GROUP BY ...
HAVING COUNT(...) > 3

Note that the HAVING clause is evaluated after the GROUP BY operation, and after the values of the aggregate expressions are evaluated. Much different than the WHERE clause.

---

Also, the subquery is a bit odd. Because it's being referenced in an equality comparison, the subquery can return at most one row. We don't see anything in the query that would prevent that, unless code is guaranteed to be unique in `ot_.

If code is unique, then we wouldn't need a subquery at all, we could just test for code.

 WHERE ot.code = 'shipping'

If there are multiple rows with "shipping" in the code column, we don't see any guarantee that the value column on those rows will be the same. To test for any of the possible values, we could use an IN operator instead of the equality (scalar) comparison. eg

 WHERE ot.value IN ( SELECT v.value 
                       FROM oc_order_total v
                      WHERE v.code = 'shipping'
                   )

But that still looks really odd. What's strange is that it's using the same table as the outer query. If we are using a subquery to lookup the set of value s related to a string code , that's usually a separate lookup table. And we'd normally prefer a JOIN operation rather than an IN (subquery) . Very strange.

---

Also, the non-aggregate expressions in the SELECT list

 o.date_added, o.email, o.total, ot.value

Do not appear in the GROUP BY clause. Most relational databases will throw an error "non-aggregate in select not in group by" something of that ilk.

But a MySQL extension will allow the query to run, but the values returned for the non-aggregates is indeterminate. MySQL will return values for those expressions based on some row included in the collapsed set, but there's no guarantee which row that will be.

We can also get MySQL to throw an error like other databases, by including ONLY_FULL_GROUP_BY in the sql_mode variable.