I'm using group_by in a DBIx::Class resultset search. The result returned for each group is always the row in the group with the lowest id (i.e the oldest row in the group). I'm looking for a way to get the row with the highest id (i.e. the newest row in the group) instead.
The problem is fundamentally the same as this:
Retrieving the last record in each group
...except that I'm using DBIx::Class not raw SQL.
To put the question in context:
I have a table of music reviews
review
------
id
artist_id
album_id
pub_date
...other_columns...
There can be multiple reviews for any given artist_id/album_id.
I want the most recent reviews, in descending date order, with no more than one review per artist_id/album_id.
I tried to do this using:
$schema->resultset('Review')->search(
undef,
{
group_by => [ qw/ artist_id album_id / ],
order_by => { -desc => 'pub_date' },
}
);
This nearly works, but returns the oldest review in each group instead of the newest.
How can I get the newest?
For this to work you are relying on broken database behaviour. You should not be able to select columns from a table when you use group by unless they use an aggregate function (min, max etc.) or are specified in the group by clause.
In MySQL, even the manual admits this is wrong - though it supports it.
What I think you need to do is get the latest dates of the reviews, with max(pub_date):
my $dates = $schema->resultset('Review')->search({},
{
select => ['artist_id', 'album_id', {max => 'pub_date'}],
as => [ qw(artist_id album_id recent_pub_date) ],
group_by => [ qw(artist_id album_id) ],
}
);
Then loop through to get the review:
while (my $review_date = $dates->next) {
my $review = $schema->resultset('Review')->search({
artist_id => $review_date->artist_id,
album_id => $review_date->album_id,
pub_date => $review_date->get_column('recent_pub_date'),
})->first;
}
Yep - it's more queries but it makes sense - what if two reviews are on the same date - how should the DB know which one to return in the select statement?
Related
this is my first question on stackoverflow, never used it before but this issue is making me tear my hair out.
I'm building an infinite scroll component for a react app I'm working on a I'm trying to make a Postgres DB query work.
I have 2 tables - Challenges, and UserChallenges.
Challenges have many User Challenges.
I need to get a subsection of Challenges (from start to end) with each Challenge having a count of the number of "participants" (number of associated UserChallenges), and also a count of all challenges.
Something like this:
{
rows: [Challenge, Challenge, Challenge],
count: n
}
Where each challenge includes the total number of userChallenges as "participants" and count is a count of all challenges.
Here is the query:
let json_query = {
attributes: {
include: [[Sequelize.fn("COUNT", Sequelize.col("user_challenges.id")), "participants"]]
},
include: [{
model: UserChallenge, attributes: []
}],
order: [['timestamp', 'DESC']],
offset: start,
limit: end
}
The start and end quantities are the start and end of the pagination.
I'm running this query as follows:
var challengeInstances = await Challenge.findAndCountAll(json_query)
This results in the following error:
name: 'SequelizeDatabaseError',
parent: error: missing FROM-clause entry for table "user_challenges"
and this is the sql it's saying it's running:
`SELECT "challenge".* FROM (SELECT "challenge"."id", "challenge".*, COUNT("user_challenges"."id"), "challenge"."participants" FROM "challenges" AS "challenge" GROUP BY "challenge"."id" ORDER BY "challenge"."end_date" DESC LIMIT '4' OFFSET '0') AS "challenge" LEFT OUTER JOIN "user_challenges" AS "user_challenges" ON "challenge"."id" = "user_challenges"."challenge_id" ORDER BY "challenge"."end_date" DESC;`,
Sequelize or raw queries are both good.
Do let me know if you need any more information and thank you so so much.
you can use sequelize literal like this & remove object from attributes just paste this code for attributes .
attributes: [
[
sequelize.literal(`(
SELECT COUNT(id)
FROM user_challenges
WHERE
// your condition of foreign key like (user_challenges.participants_id = participants.id)
)`),
'numberOfParticipants'
]
]
I have a table called Transaction with relation User, Transaction has a field called balance.
Data looks like:
id user_id balance
1 22 365
2 22 15
3 22 900
4 32 100
4 32 50
I need all users associative data and last insert balance field of User. For example here id=3 is last inserted data for user_id=22.
In raw SQL I have tried this:
select * from transactions where id in (select max(id) from transactions group by user_id)
If I add here a inner join I know I can also retrieve User data. But how can I do this in CakePHP?
IMHO, subqueries are ugly in CakePHP 2.x. You may as well hard code the SQL statement and execute it through query(), as suggested by #AgRizzo in the comments.
However, when it comes to retrieving the last (largest, oldest, etc.) item in a group, there is a more elegant solution.
In this SQL Fiddle, I've applied the technique described in
Retrieving the last record in each group
The CakePHP 2.x equivalent would be:
$this->Transaction->contains('User');
$options['fields'] = array("User.id", "User.name", "Transaction.balance");
$options['joins'] = array(
array('table' => 'transactions',
'alias' => 'Transaction2',
'type' => 'LEFT',
'conditions' => array(
'Transaction2.user_id = Transaction2.user_id',
'Transaction.id < Transaction2.id'
)
),
);
$options['conditions'] = array("Transaction2.id IS NULL");
$transactions=$this->Transaction->find('all', $options);
I've been playing around with the RavenDB Northwind database, and am having trouble getting orders sorted by Freight.
My index:
Map = orders => from o in orders
select new {
o.Freight
};
IndexSortOptions.Add(x => x.Freight, SortOptions.Double);
Indexes.Add(x => x.Freight, FieldIndexing.NotAnalyzed);
My query:
return sess.Query<Order>("Orders/ByFreight")
.OrderByDescending(x => x.Freight)
.Select(x => x.Freight);
It get the following order back:
[
32.38,
11.61,
65.83,
41.34 ... ]
Which is clearly not correct. In the studio, I can define the order to be by Freight, and it orders just fine. However, if in the studio I specify the range as over Freight_Range, I get these same results back. It appears to me that Raven is selecting the Freight_Range field to sort by rather than the Freight field. Why?
Remove the line starting with Indexes.Add. You don't need to do that.
Use this syntax instead of the IndexSortOptions:
Sort(x=> x.Freight, SortOptions.Double);
Make sure your Freight field is indeed a double as defined on your Order class.
I have following SQL Query:
SELECT campaigns.* , campaign_countries.points, offers.image
FROM campaigns
JOIN campaign_countries ON campaigns.id = campaign_countries.campaign_id
JOIN countries ON campaign_countries.country_id = countries.id
JOIN offers ON campaigns.offer_id = offers.id
WHERE countries.code = 'US'
This works perfectly well. I want its rails active record version some thing like:
Campaign.includes(campaign_countries: :country).where(countries: {code: "US"})
Above code runs more or less correct query (did not try to include offers table), issue is returned result is collection of Campaign objects so obviously it does not include Points
My tables are:
campaigns --HAS_MANY--< campaign_countries --BELONGS_TO--< countries
campaigns --BELONGS_TO--> offers
Any suggestions to write AR version of this SQL? I don't want to use SQL statement in my code.
I some how got this working without SQL but surely its poor man's solution:
in my controller I have:
campaigns = Campaign.includes(campaign_countries: :country).where(countries: {code: country.to_s})
render :json => campaigns.to_json(:country => country)
in campaign model:
def points_for_country country
CampaignCountry.joins(:campaign, :country).where(countries: {code: country}, campaigns: {id: self.id}).first
end
def as_json options={}
json = {
id: id,
cid: cid,
name: name,
offer: offer,
points_details: options[:country] ? points_for_country(options[:country]) : ""
}
end
and in campaign_countries model:
def as_json options={}
json = {
face_value: face_value,
actual_value: actual_value,
points: points
}
end
Why this is not good solution? because it invokes too many queries:
1. It invokes query when first join is performed to get list of campaigns specific to country
2. For each campaign found in first query it will invoke one more query on campaign_countries table to get Points for that campaign and country.
This is bad, Bad and BAD solution. Any suggestions to improve this?
If You have campaign, You can use campaign.campaign_countries to get associated campaign_countries and just get points from them.
> campaign.campaign_countries.map(&:points)
=> [1,2,3,4,5]
Similarly You will be able to get image from offers relation.
EDIT:
Ok, I guess now I know what's going on. You can use joins with select to get object with attached fields from join tables.
cs = Campaign.joins(campaign_countries: :country).joins(:offers).select('campaigns.*, campaign_countries.points, offers.image').where(countries: {code: "US"})
You can than reference additional fields by their name on Campaign object
cs.first.points
cs.first.image
But be sure, that additional column names do not overlap with some primary table fields or object methods.
EDIT 2:
After some more research I came to conclusion that my first version was actually correct for this case. I will use my own console as example.
> u = User.includes(:orders => :cart).where(:carts => { :id => [5168, 5167] }).first
> u.orders.length # no query is performed
=> 2
> u.orders.count # count query is performed
=> 5
So when You use includes with condition on country, in campaign_countries are stored only campaign_countries that fulfill Your condition.
Try this:
Campaign.joins( [{ :campaign_countries => :countries}, :offers]).where('`countries`.`code` = ?', "US")
I have a problem concerning CakePHP SQL-queries. I need to fetch products from the database where shop_id is given and then count the products. All I need is Product.url and its count.
This will do the trick in plain SQL:
SELECT url,COUNT(*) as count FROM products GROUP BY url ORDER BY count DESC;
This one I used to get all products relating to shops:
$this->Product->find('all', array('conditions' => array('Product.shop_id'=>$id)));
That works correctly, but I need to convert that SQL-query above to CakePHP.
I tried something like this:
$this->Product->find('count', array('conditions' => array('Product.shop_id'=>$id),
'fields'=>array('Product.url','Product.id'),
'order'=>'count DESC',
'group'=>'Product.url'));
That returns only an int. But if I run that SLQ-query presented above in mysql server, I get two columns: url and count. How do I get the same results with CakePHP?
You can try this:
$data = $this->Post->query(
"SELECT COUNT(id),MONTH(created) FROM posts GROUP BY YEAR(created), MONTH(created);"
);
The most easiest way to do this:
$this->Product->query("SELECT url,COUNT(*) as count FROM products GROUP BY url ORDER BY count DESC;");
...at least for me.
Try the following code:
$this->Product->virtualFields['CNT_PRODUCTS'] = 0;
$this->Product->find('count', array('conditions' => array('Product.shop_id' => $id),
'fields' => array('Product.id', 'Product.url', 'count(*) as CNT_PRODUCTS'),
'order' => array('CNT_PRODUCTS' => 'DESC'),
'group' => 'Product.url'));