I'm trying to calculate a value using DQL on one single table. Say this table is named "TABLE". It has 5 colums :
id
people_count
animal_count
region_id
type_id
The result I'm looking for is the sum of people (when grouped by region), divided by the sum of animals (when grouped by type);
The SQL would be something like that :
SELECT SUM(people_count) /
(
SELECT SUM(animal_count)
FROM TABLE t2
GROUPED BY type_id
)
FROM TABLE t1
GROUPED BY region_id
How would you do that in Doctrine using DQL?
I resolved my problem by creating another query, executing it and including the result in the first one.
This is probably not the best way to do if you are dealing with a simple example, but it was the only solution for me regarding to my code architecture.
I have a solution but I think there is probably a best solution to resolve your problem.
In any case, make two queries and import results of first query in the second can be a solution. Unfortunately, it's a low-usage for our database. More, sometimes we must execute only one SQL to sort on column result per example.
<?php
namespace AppBundle\Repository;
use AppBundle\Entity\MyClass;
use Doctrine\ORM\EntityRepository;
class MyRepository extends EntityRepository
{
public function findWithSubQuery()
{
return $this->createQueryBuilder('a')
->addSelect(sprintf(
"(SELECT COUNT(b.id) FROM %s AS b WHERE b.a = a.id GROUP BY a.id) AS otherColumn",
MyClass::class
))
;
}
}
I use this solution. Maybe the subquery could be write with DQL ojbect rather that DQL string.
Related
I have two tables
table1
id|name|next_field
table2
id|id_of_table1|whatelse
I do a msql query to get all entries of table1 and the number of entries in table2 who has table2.id_of_table1 = table1.id
This is my query - it works fine.
$select =array('table1.*', 'COUNT(table2.id) AS `my_count_result`',);
$this->db->select($select);
if($id!=false){ $this->db->where('id',$id); }
$this->db->from('table1 as t1');
$this->db->join('table2 as t2', 't1.id = t2.id_of_table1');
return $this->db->get()->result_array();
Now I have another field which has coma-separated data
table1.next_field = info1, info2, next,...
Now I want to check in the same way like the first query how often for example "info2" is as a part inside the table1.next_field
Is it possible, how?
After all i decide to change my database structure to make the work and handle much easier.
I have a model User that has many Orders. Orders have many products, with the pivot table order-product. I don't want to preload and iterate through the orders if at all possible.
I need to return users where
signed_date === true on User
order_date on Order is after signed_date on User
order-product shows product hasn't been paid
I am failing on number 2.
In the following code, the first query within whereHas is wrong. I don't know how to reference the signed date of the user from within the where has. If I was iterating through users in a collection I could do something like ($query) use $user, but how do I do this without preloading all the users?
return User::whereNotNull('signed_date')
->whereHas('orders', function ($query) {
$query->where('order_date', '<=', 'user.signed_date');
$query->whereHas('products', function ($q) {
$q->where('paid', false);
});
})
->get(['id','fname','lname', 'title', 'signed_date']);
I would like to use eloquent if possible. If that is not possible, I would be happy for tips in solving this problem using the query builder/sql.
The Eloquent query builder has a special function called whereColumn('a', '<=', 'b') to compare columns instead of a column against a value. Using this function instead of a normal where() is necessary because of the way the query builder builds the actual query. You need to let the query builder know that you are going to pass a column name instead of a value for proper escaping and formatting of the query string.
Anyway, it seems you can also pass column names prefixed with a table name to the function, allowing you to compare columns across tables:
$query->whereColumn('orders.order_date', '<=', 'users.signed_date')
This works because you are using whereHas() in your query. Your query basically gets translated to:
SELECT id, fname, lname, title, signed_date
FROM users
WHERE signed_date NOT NULL
AND EXISTS (
SELECT 1
FROM orders
WHERE orders.order_date <= users.signed_date
AND EXISTS (
SELECT 1
FROM products
WHERE paid = 0
)
)
It might actually be not necessary at all to use the table name together with the column name in whereColumn(). But in case you'll ever add a column named the same on another table, the query might break - so IMHO it is good practice to use the table name in custom queries.
By the way, the reason this will not work together with with('relationship') is that this function results in an additional query and you obviously cannot compare columns across queries. Imagine the following:
Order::with('user')->take(5)->get();
It will be translated into the following:
SELECT *
FROM orders
LIMIT 5
SELECT *
FROM users
WHERE id IN (?, ?, ?, ?, ?)
where the five ? will be the user_ids of the orders. If the first query returns multiple rows with the same user_id, the amount of rows fetched from the users table gets reduced of course.
Note: All the queries are only examples. Might be that the query builder builds different queries based on the database type and/or escapes them differently (i.e. column names in backticks).
Is there ability to make a multiple update with subquery on Symfony3 with Doctrine query builder or DQL?
For example, I want to run this query:
UPDATE tableA
SET fieldA2 = max_field2
FROM (SELECT
field1,
max(field2) AS max_field2
FROM table
GROUP BY field1) AS subquery
WHERE subquery.field1 = tableA.field1;
I can't understand how to use $entityManager->createQuery()->update with FROM subquery.
As far as I know, it's not possible through DQL.
You need to go through a loop.
foreach($entities as entity)
{
$em->flush();
}
Else, you will consider Batch processing, or just use plain SQL. So it might be usefull to check DBAL.
TablePatient.Patient_ID(PK)
TableProviders.Encounter (joins to PK)
TableProviders.Provider_Type
TableProviders.Provider_ID
TableNames.Full_Name
TableNames.Provider_ID (joins to Table Names)
I want a query that will give, for all the Patient_IDs, the Full_Name of the provider for every Provider ID.
There are about 30 provider_types.
I have made this already using a left join a ton of left joins. It takes a long time to run and I am thinking there is a trick I am missing.
Any help?
Ok, my previous answer didn't match at all what you meant. You want to pivot the table to have on each line one Patient_ID with every Full_name for every provider_type. I assume that each patient has only one provider for one type and not more ; if more, you will have more than one row for each patient, and anyway I don't think it's really possible.
Here is my solution with pivot. The first part is to make it more understandable, so I create a table named TABLE_PATIENT in a subquery.
WITH TABLE_PATIENT AS
(
SELECT TablePatient.Patient_ID,
TableProviders.Provider_Type,
TableNames.Full_Name
FROM TablePatient LEFT JOIN
TableProviders on TablePatient.Patient_ID = TableProviders.Encounter
LEFT JOIN
TableNames on TableNames.Provider_ID = TableProviders.Provider_ID
group by TablePatient.Patient_ID,
TableProviders.Provider_Type,
TableNames.Full_Name
)
SELECT *
FROM TABLE_PATIENT
PIVOT
(
min(Full_name)
for Provider_type in ([type1], [type2],[type3])
) AS PVT
So TABLE_PATIENT just has many rows for each patient, with one provider each row, and the pivot puts everything on a single row. Tell me if something doesn't work.
You need to write every type you want in the [type1],[type2] etc. Just put them inside [], no other character needed as ' or anything else.
If you put only some types, then the query will not show providers of other types.
Tell me if something doesn't work.
If I understand what you mean, you just want to group the answer by Patient Id and then Provider ID. A full name is unique on a provider id right ?
This should be something like
SELECT TablePatient.Patient_ID,
TableProviders.Provider_ID,
TableNames.Full_Name
FROM TablePatient LEFT JOIN
TableProviders on TablePatient.Patient_ID = TableProviders.Encounter
LEFT JOIN
TableNames on TableNames.Provider_ID = TablerProviders.Provider_ID
group by TablePatient.Patient_ID,
TableProviders.Provider_ID,
TableNames.Full_Name
You can either group by TableNames.Full_Name or select First(TableNames.Full_Name) for example if indeed a full name is unique to a provider ID.
Note : I used the SQL server Syntax, there can be différences with Oracle ..
How can I write the following in Laravel's Eloquent?
SELECT *
FROM
( SELECT real_estate.property_id,
real_estate.amount_offered,
payee.summa
FROM real_estate
LEFT JOIN
(SELECT property_id,
SUM(amount) AS summa
FROM payments
GROUP BY property_id) payee ON payee.property_id = real_estate.property_id ) yoot
WHERE summa = 0.05 * amount_offered
Been on this for a while now and really can't get around it. Lemme explain the whole cause for the panic.
I have two tables, one for property and another for payments made on those properties. Now at any given time I will like to query for what properties have been paid for to a certain percentage hence the 0.05 which reps 5%. As it is the query works but I need an Eloquent alternative for it. Thanks
Anywhere you have subqueries in your SQL you'll need to use DB::raw with Eloquent. In this case you have a big subquery for the FROM statement, so the easiest way would be to do this:
DB::table(
DB::raw('SELECT real_estate.property_id, real_estate.amount_offered, payee.summa FROM real_estate LEFT JOIN (SELECT property_id, SUM(amount) AS summa FROM payments GROUP BY property_id) payee ON payee.property_id = real_estate.property_id)')
)
->where('summa', DB::raw('0.05 * amount_offered'))->get();
Notice I used DB::raw for the WHERE statment value as well. That's because you are doing a multiplication using a column name, and the value would otherwise be quoted as a string.
If you want to go a step further and build each subquery using Eloquent, then convert it to an SQL string and injecting it using DB::raw, you can do this:
$joinQuery = DB::table('payments')
->select('property_id', 'SUM(amount) AS summa')
->groupBy('property_id')
->toSql();
$tableQuery = DB::table('real_estate')
->select('real_estate.property_id', 'real_estate.amount_offered', 'payee.summa')
->leftJoin(DB::raw('(' . $joinQuery . ')'), function ($join)
{
$join->on('payee.property_id', '=', 'real_estate.property_id');
})
->toSql();
DB::table(DB::raw('(' . $tableQuery . ')'))->where('summa', DB::raw('0.05 * amount_offered'))->get();
In this case, the second approach doesn't have any benefits over the first, except perhaps that it's more readable. However, building subqueries using Eloquent, does have it's benefitfs when you'd need to bind any variable values to the query (such as conditions), because the query will be correctly built and escaped by Eloquent and you would not be prone to SQL injection.