Related
<?php
$query = $this->restaurant->leftJoin('reviews', 'reviews.restaurant_id', '=', 'restaurants.id');
$query = $user_id == null ? $query : $query->where("user_id", $user_id);
$query = $query->select('restaurants.*')
->selectRaw('ROUND(AVG(reviews.rating), 1) as average_rating')
->groupBy('restaurants.id')
->havingRaw('AVG(reviews.rating) >= ?', [$request->from_rating])
->havingRaw('AVG(reviews.rating) <= ?', [$request->to_rating])
->orderByDesc("average_rating")
->paginate(
$this->perPage,
['id', 'name', 'image', 'description', 'filesystem', 'created_at'],
'page',
$request->page
);
return $query;
My query looks like this. I have two tables: Restaurants and Reviews. The Reviews table stores restaurant_id. The problem occurs when we have, for example, 10 restaurants and we have 1 Review only for one of the Restaurants. After running this query, only 1 restaurant is returned, because I use havingRaw which filters Restaurants according to the average rating. So if a Restaurant doesn't have any rating, that Restaurant doesn't get returned.
How do I make that so all Restaurants still return, but their average rating becomes 0?
you can use CASE WHEN to populate mysql with more specific condition. Also you may change the datatype of 'rating' table to int(5), in this case all the empty value will be shown as 0. So that will go as you want according to your question.
Im quite new to laravel and so far I realy enjoy eloquent and querybuilder, but as soon as the queries become more complex, my head starts to hurt... I have just finished 2 working examples after quite some time, maybe you guys can help me optimize it. I will first give the examples, then the (dis)advantages
So first the DB::select... which honestly, I think is more readable then the 2nd example.. or maybe I am just doing the builder thing wrong xD.
$shared_slugs = Magic::in_query($this->shared_slugs);
$items = DB::select("
SELECT * FROM f_admin_items
WHERE
active = 1
AND (
slug IN $shared_slugs
OR
:treshhold <= :urank AND id IN (SELECT admin_item_id FROM f_user_access WHERE user_id = :uid)
OR
:treshhold > :urank AND `rank` >= :urank
)
ORDER BY `order` ASC
", [
'treshhold' => $this->threshold_rank,
'urank' => $user_rank,
'uid' => $user_id
]);
Overall pretty effective with the named parameter binding. Basicly the menu items always have to be active, plus (1 OR 2 OR 3 ). For example dashboard is accepted as shared slug.. Otherwise there are some ranking checks.
Now I have been trying hard to do this with eloquent en query builder =/ I didn't think setting up a relation was not necessary because I only use this for the menu and a middleware. In this case the f_user_access table containing only admin_item_id and user_id, doesn't really function as a pivot and isn't used anywhere else.
$items =
$this->where( 'active', 1 )
->where( function ($query) {
$query->whereIn( 'slug', $this->shared_slugs )
->orWhere(function ($query) {
$query->whereRaw( $this->threshold_rank.' <= '.$this->user_rank )
->whereIn('id', function ($query) {
$query->select('admin_item_id')->from('f_user_access')->where('user_id', $this->user_id)->get();
});
})
->orWhere(function ($query) {
$query->whereRaw( $this->threshold_rank.' > '.$this->user_rank )
->where( 'rank', '>=', $this->user_rank );
});
})->orderBy( 'order', 'ASC' )->get();
What I like about this second one is the fact I can pass $shared_slugs as an array, I don't have to convert it into a string first. But apart from that I really hate the looks of this, where(function($query){}) etc etc... On top of that the $ids passed to this function, are not accessible in the where.functions, so I had to define them in the class first.
The first one I like because of the named binding and it doesn't read that bad to be :S, also the vars are accessible.. Downside, I have to call this function to convert the shared slugs into an array.
Is it really that bad to not use eloquent and querybuilder? at least in some cases... and what would you guys do to make the second example better? =/
UPDATE::
Due to the answers and feedback, I ditched the raw sql. The current function has 5 scopes in it, and a (small) model is made for the user_access.
$items =
$this->active() //Only active items AND... 1, 2 or 3
->where( function ($query) {
$query->shared_slug() // 1
->orWhere(function ($query) { // OR 2
$query->access_required()
->whereIn('id', UserAccess::get_access( $this->user_id ) );
})
->orWhere(function ($query) { // OR 3
$query->no_access_required()
->whereRaw( 'rank >= '.$this->user_rank );
});
})->ASC()->get();
The major benefit of using the query builder is it abstracts you away from the language used by your storage of choice, i.e. MySQL, Oracle, SQLite, etc. If you ever switch database types, you can be stuck with a lot of refactoring raw SQL. Believe me, it's not pretty when you start on a project and learn this is the case.
There are always caveats, however, which is precisely why Eloquent has the capability to work with raw statements, as well.
This is the native sql:
$sql = "Select count(name) from users Where email = 't#t.com' and user_id = 10";
I have this laravel code:
$checker = Customer::whereEmailAndUserId("t#t.com",10)->count("name");
Is this a correct way to do it in laravel?
You have to use where helper function and pass an array of checks. For example in your code it will be:
$checker = Customer::where([
['email', '=', 't#t.com'],
['user_id' '=', '10']
])->count();
Note: Please use the appropriate column name as it in table.
Assuming Customer model represents table users, you'll get query with eloquent like this:
Customer::where('email', 't#t.com')->where('user_id', 10)->select(\DB::raw('count(name)'))->get();
The option you are trying is incorrect
here is the right option
$users = \App\Customer::where('email','t#t.com')
->where('user_id',10)
->count()
Explanation of above code
App\Customer is the Model class and I am trying to read records where email = 't#t.com you can use various comparison operators like <,> and so on and you can also use the same function to for string pattern matching also
Eg.
$users = \App\Customer::where('email','%t.com')
->where('user_id',10)
->count()
You can use the same where function for Null Value test also
Eg.
$users = \App\Customer::where('email','=', null)
->where('user_id',10)
->count()
The above where clause will be converted to is null test of the SQL
You can read more here
i have a problem trying to get records from a model based on a related table.
I have two tables one called leads and one called recycle_logs, my app will basically send the same record in the leads table like once a month, and when it does so i'll store a new record in recycle_logs.
The problem is that i need to select all leads with a specific campaign_id value and that have a specific status that is different from invalid, so far so good, now the problem is i need to get them only if they don't have any recycleLogs associated with them OR if the last recycleLog associated with that particular lead is older than 30 days ago for instance.
What i currently have looks somewhat like this.
$leads = $this->leadModel->where(Lead::CAMPAIGN_ID, $this->campaignID)
->where(Lead::DUPLICATED, Lead::DUPLICATED_NO)
->where(Lead::LEAD_STATUS, "!=" ,Lead::LEAD_STATUS_INVALID)
->orderBy(Lead::CREATED_AT, 'desc')
->with(
['leadRecyclingLog' => function($query) {
$query->where(LeadRecyclingLog::CREATED_AT, '<', (new Carbon())->subDays($this->configRecyclingDays))
->orWhere(LeadRecyclingLog::ID, null);
}]
)
->get();
What exactly am i doing wrong? It always selects the same number of records regardless of me adding or removing recycleLogs
I've managed to get it done through a raw SQL query which i'll post below in case it helps anyone, i'd still like to know how to do it in Eloquent/Query Builder.
SELECT * FROM `leads` LEFT JOIN `lead_recycling_logs` ON `leads`.`guid` = `lead_recycling_logs`.`original_lead_guid` WHERE `leads`.`campaign_id` = :campaignID AND `leads`.`duplicated` = 0 AND `leads`.`lead_status` != :invalidStatus AND (`lead_recycling_logs`.`id` IS NULL OR `lead_recycling_logs`.`created_at` < :recyclingDate) ORDER BY `leads`.`created_at` DESC
Try this:
$leads = $this->leadModel->where(Lead::CAMPAIGN_ID, $this->campaignID)
->where(Lead::DUPLICATED, Lead::DUPLICATED_NO)
->where(Lead::LEAD_STATUS, "!=" ,Lead::LEAD_STATUS_INVALID)
->orderBy(Lead::CREATED_AT, 'desc')
->where(function($q) {
$q->whereHas('leadRecyclingLog', function($q) {
$q->where(LeadRecyclingLog::CREATED_AT, '<', (new Carbon())->subDays($this->configRecyclingDays));
})
->orWhereHas('leadRecyclingLog', '<', 1); // Where count of the relationship is smaller than 1
})->get();
I assumed the first part of the query is working well (up until the relationship).
What you're looking for is ->whereHas(relationship), not ->with(relationship). ->with(relationship) will attach the associated results to the original model (the query for the original model will not be affected by ->with()). ->whereHas(relationship) filters the original model by the condition.
Got it to work through #devk 's help
$leads = $this->leadModel->where(Lead::CAMPAIGN_ID, $this->campaignID)
->where(Lead::DUPLICATED, Lead::DUPLICATED_NO)
->where(Lead::LEAD_STATUS, "!=" ,Lead::LEAD_STATUS_INVALID)
->orderBy(Lead::CREATED_AT, 'desc')
->where(function($q) {
$q->whereHas('leadRecyclingLog', function($q) {
$q->where(LeadRecyclingLog::CREATED_AT, '<', (new Carbon())->subDays($this->configRecyclingDays));
})
->doesntHave('leadRecyclingLog', 'or');
})->get();
I am creating a system of newsfeed, and as you can easily guess, it is beyond my skills.
Please be kind to put me on the right track or provide something I can go on with.
I have several hundred events (model name is Event1, table 'events')
I also have a pivot table in which users can assign any event's importance (values 0,1,2,3)
The relevant columns of the pivot table user_attitudes (Model Userattitude) are
id, item_type, item_id, importance, attitude, creator_id
An example three record are:
456 - event - 678 - 2 - 4
457 - event - 690 - 3 - 15
458 - event - 690 - 1 - 4
459 - participant - 45 - 1 - 4
Plain English: Total aggregated importance of the event #690 is '4', while the event #678 is '2'.
Therefore in my ranking the event #690 should be listed as first.
Just to see the bigger pic: the user #4 also rated participant # 45 as importance = 1.
The table services many models - the above example include two - just to give a better image of what I have.
WHAT I NEED:
I wish to print a ranking of top 5 events (and later other models). I wish to be able to use two methods of calculating the total score:
by counting the actual value (0,1,2,3)
by counting any value above 0 as 1 point.
I want to generate views which filter events by this criteria:
at least one user set the importance to '0' (I need it to flag an event as untrustworthy)
events which has not been rated yet
reverse of the above - events which are rated by at least one user
events listed by number of users who assigned any importance to it
This is easy, but still I have no idea how to make it happen. The same filters as the above #2, but related to a particular user decisions:
list 5 or 10 events (random or newest) which has not yet been rated by the user
maybe something like this would be an answer:
$q->where('creator_id', '=', Auth::user()->id);
Relevant code:
As I don't really grasp the merged relations, I might fail to show everything needed to provide help - ask for more code in comments.
Models:
Event1 (table 'events'):
public function importances()
{
return $this->morphMany('Userattitude', 'item');
}
public function user_importance($user)
{
return $this->morphMany('Userattitude', 'item')->where('creator_id', ($user ? $user->id : NULL))->first();
}
User: (table 'users' - standard user table)
public function importances()
{
return $this->hasMany('Userattitude', 'creator_id');
}
In model Userattitude (different from User, table name 'user_attitudes')
public function events()
{
return $this->morphTo('item')->where('item_type', 'event');
}
public function event()
{
return $this->belongsTo ('Event1', 'item_id');
}
PROBLEMS IN REPLY TO #lucas answer:
PROBLEM 1.
table name 'items' keeps me confused as in my project 'items' are events (model Event1), the participants (model Entity) and other objects.
Can we stick to my naming until I get hold of the knowledge you are providing?
it also contains column named attitudes, which is used for blacklisting particular items.
For instance, an item of type 'entity' (possible participant of multiple events) can be voted by user two-wise:
- by importance set by an user (we are doing this now, values available to use are 0,1,2,3)
- by attitude of an user toward (possible value (-1, 0, 1)
Such solution allows me to compute karma of each item. For instance -1 x 3 = -3 (worst possible karma value), while 1 x 2 = 2 (medium positive karma).
In consequence I am unable to use queries with the users method. It is still too confusing to me, sorry. We diverted too far from my original mental image.
Consider this query:
$events = Event1::has('users', '<', 1)->get();
If in Event1 I declare
public function users()
{
return $this->morphToMany('User', 'item', null, null, 'creator_id');
}
Note: User is the standard users table, where username, password and email are stored
I get this error:
[2014-12-28 05:02:48] production.ERROR: FATAL DATABASE ERROR: 500 = SQLSTATE[42S02]: Base table or view not found: 1146 Table 'niepoz_niepozwalam.items' doesn't exist (SQL: select * from `Events` where (select count(*) from `users` inner join `items` on `users`.`id` = `items`.`creator_id` where `items`.`item_id` = `Events`.`id` and `items`.`item_type` = Event1) >= 1) [] []
if I change the method definition to
public function users()
{
return $this->morphToMany('Userattitude', 'item', null, null, 'creator_id');
}
Note: Userattitude is model (table name is 'user_attitudes') where i store user judgments. This table contains columns 'importance' and 'attitude'.
I get the same error.
If I change the method to
public function users()
{
return $this->morphToMany('User', 'Userattitudes', null, null, 'creator_id');
}
I get this:
[2014-12-28 05:08:28] production.ERROR: FATAL DATABASE ERROR: 500 = SQLSTATE[42S22]: Column not found: 1054 Unknown column 'user_attitudes.Userattitudes_id' in 'where clause' (SQL: select * from Events where (select count(*) from users inner join user_attitudes on users.id = user_attitudes.creator_id where user_attitudes.Userattitudes_id = Events.id and user_attitudes.Userattitudes_type = Event1) >= 1) [] []
Possible solution:
the 'user_attitudes' table alias with name 'items'.
I could create a view with the required name.
I did it, but now the query produces no results.
PROBLEM 2
should I rename creator_id into user_id - or keep both columns and keep duplicated information in them? The creator_id follows conventions and I use it to create records... how to resolve this dillema?
PROBLEM 3.
As far as I understand, if I want to get a USER-RELATED list of top-5 events,
I need to ad another line to the code, which narrows search scope to records created by a particular logged in user:
Auth::user()->id)
The code would look like this:
All with importance 0
$events = Event1::whereHas('users', function($q){
$q->where('importance', 0);
$q->where('creator_id', '=', Auth::user()->id);
})->get();
right?
PROBLEM 5:
Ok, I am now able to output a query like these:
$rank_entities = Entity::leftJoin('user_attitudes', function($q){
$q->on('entity_id', '=', 'entities.id');
$q->where('item_type', '=', 'entity');
})
->selectRaw('entities.*, SUM(user_attitudes.importance) AS importance')
->groupBy('entities.id')
->orderBy('importance', 'desc')
->take(6)
->get();
and in the foreach loop I can display the total importance count with this code:
{{$e->importance or '-'}}
But How I could display count of an alternative query: SUM of values from another column, named attitude, which can be computed in this SEPARATE query:
In other words, in my #foreach loop I need to display both $e->importance and a computed SUM(user_attitudes.attitude) AS karma, which for now can be received with this query:
$rank_entities = Entity::leftJoin('userattitudes', function($q){
$q->on('entity_id', '=', 'entities.id');
$q->where('item_type', '=', 'entity');
})
->selectRaw('entities.*, SUM(userattitudes.karma) AS karma')
->groupBy('entities.id')
->orderBy('karma', 'desc')
->take(5)
->get();
My solution would be to create some extra columns in the 'entities' table:
- karma_negative
- karma_positive
to store/update total amount of votes each time someone is voting.
First, let's talk about the setup. I wasn't entirely sure how and if your's works but I created this on my testing instance and it worked, so I recommend you change yours accordingly:
Database
events
That's a simple one (and you probably already have it like this
id (primary key)
name (or something like that)
etc
users
I'm not sure if in your example that is Userattitude but I don't think so...
id (primary key)
email (?)
etc
items
This is the important one. The pivot table. The name can be different but to keep it simple and follow conventions it should be the plural of the polymorphic relation (in your case item => items)
id (actually not even necessary, but I left it in there)
item_type
item_id
importance
creator_id (consider changing that to user_id. This would simplify the relationship declaration)
Models
I think you have to read the docs again. You had several weird relations declared. Here's how I did it:
Event1
By default Laravel uses the classname (get_class($object)) as value for the ..._type column in the database. To change that you need to define $morphClass in your models.
class Event1 extends Eloquent {
protected $table = 'events';
protected $morphClass = 'event';
public function users()
{
return $this->morphToMany('User', 'item', null, null, 'creator_id');
}
}
User
class User extends Eloquent implements UserInterface, RemindableInterface {
// ... default laravel stuff ...
public function events(){
return $this->morphedByMany('Event1', 'item', null, null, 'creator_id');
}
}
Queries
Alright now we can get started. First one additional information. I used Eloquent relations whenever possible. In all the queries a join() is made it would be slower to use relations because certain things (like counting or calculating the maximum) would have to be done in PHP after the query. And MySQL does a pretty good job (also performance wise) at those things.
Top 5 by total value
$events = Event1::leftJoin('items', function($q){
$q->on('item_id', '=', 'events.id');
$q->where('item_type', '=', 'event');
})
->selectRaw('events.*, SUM(items.importance) AS importance')
->groupBy('events.id')
->orderBy('importance', 'desc')
->take(5)
->get();
Top 5 by number of votes over 0
$events = Event1::leftJoin('items', function($q){
$q->on('item_id', '=', 'events.id');
$q->where('item_type', '=', 'event');
$q->where('importance', '>', 0);
})
->selectRaw('events.*, COUNT(items.id) AS importance')
->groupBy('events.id')
->orderBy('importance', 'desc')
->take(5)
->get();
All with importance 0
$events = Event1::whereHas('users', function($q){
$q->where('importance', 0);
})->get();
All without any votes
$events = Event1::has('users', '<', 1)->get();
All with 1+ votes
$events = Event1::has('users')->get();
All ordered by number of votes
$events = Event1::leftJoin('items', function($q){
$q->on('item_id', '=', 'events.id');
$q->where('item_type', '=', 'event');
})
->selectRaw('events.*, COUNT(items.id) AS count')
->groupBy('events.id')
->orderBy('count', 'desc')
->get();
Newest 5 without votes
If you are using Eloquents timestamps created_at:
$events = Event1::has('users', '<', 1)->latest()->take(5)->get();
If you're not (order by greatest id):
$events = Event1::has('users', '<', 1)->latest('id')->take(5)->get();
Random 5 without votes
$events = Event1::has('users', '<', 1)->orderByRaw('RAND()')->take(5)->get();
I did not add any explanations to the queries on purpose. If you want to know more about something specific or need help, please write a comment
PROBLEM 4: SOLVED! (credit to #lukasgeiter)
If you wish to display a ranking of items and limit the results to a particular tag defined in a pivot table, this is the solution:
$events = Event1 (table name = 'events')
For example, the tag would be war: defined in table
eventtags
Event nature are defined as
id = '1' is name = 'wars'
id = '2' is name = 'conflicts'
pivot table, which assigns multiple tags:
event_eventtags they are defined as id = '4'
Example records for event_eventtags:
id - event_id - eventtag_id
1 - 45 - 1
2 - 45 - 2
Plain English: the Event1 #45 is tagged as war(#1) and conflict(#2)
Now in order to print a list of 10 wars you should define your query in this way:
$events= Entity::join('event_eventtags', function($q){
$q->on('entity_id', '=', 'entities.id');
$q->where('entitycapacitytypes_id', '=', 1);
})->leftJoin('user_attitudes', function($q){
$q->on('item_id', '=', 'entities.id');
$q->where('item_type', '=', 'entity');
})
->selectRaw('entities.*, SUM(user_attitudes.importance) AS importance')
->groupBy('entities.id')
->orderBy('importance', 'desc')
->take(10)
->get();
The user_attitudes is part of voting system described in the original question. You can remove it and sort events by another method.