Query Builder¶

Selecting Rows From A Table¶

use Cake\ORM\TableRegistry;

// Prior to 3.6 use TableRegistry::get('Articles')
$query = TableRegistry::getTableLocator()->get('Articles')->find();

foreach ($query as $article) {
    debug($article->title);
}

For the remaining examples, assume that $articles is a Table. When inside controllers, you can use $this->Articles instead of $articles.

Almost every method in a Query object will return the same query, this means that Query objects are lazy, and will not be executed unless you tell them to:

$query->where(['id' => 1]); // Return the same query object
$query->order(['title' => 'DESC']); // Still same object, no SQL executed

You can of course chain the methods you call on Query objects:

$query = $articles
    ->find()
    ->select(['id', 'name'])
    ->where(['id !=' => 1])
    ->order(['created' => 'DESC']);

foreach ($query as $article) {
    debug($article->created);
}

If you try to call debug() on a Query object, you will see its internal state and the SQL that will be executed in the database:

debug($articles->find()->where(['id' => 1]));

// Outputs
// ...
// 'sql' => 'SELECT * FROM articles where id = ?'
// ...

You can execute a query directly without having to use foreach on it. The easiest way is to either call the all() or toList() methods:

$resultsIteratorObject = $articles
    ->find()
    ->where(['id >' => 1])
    ->all();

foreach ($resultsIteratorObject as $article) {
    debug($article->id);
}

$resultsArray = $articles
    ->find()
    ->where(['id >' => 1])
    ->toList();

foreach ($resultsArray as $article) {
    debug($article->id);
}

debug($resultsArray[0]->title);

In the above example, $resultsIteratorObject will be an instance of Cake\ORM\ResultSet, an object you can iterate and apply several extracting and traversing methods on.

Often, there is no need to call all(), you can simply iterate the Query object to get its results. Query objects can also be used directly as the result object; trying to iterate the query, calling toList() or some of the methods inherited from Collection, will result in the query being executed and results returned to you.

Selecting A Single Row From A Table¶

You can use the first() method to get the first result in the query:

$article = $articles
    ->find()
    ->where(['id' => 1])
    ->first();

debug($article->title);

Getting A List Of Values From A Column¶

// Use the extract() method from the collections library
// This executes the query as well
$allTitles = $articles->find()->extract('title');

foreach ($allTitles as $title) {
    echo $title;
}

You can also get a key-value list out of a query result:

$list = $articles->find('list');

foreach ($list as $id => $title) {
    echo "$id : $title"
}

For more information on how to customize the fields used for populating the list refer to Finding Key/Value Pairs section.

Queries Are Collection Objects¶

Once you get familiar with the Query object methods, it is strongly encouraged that you visit the Collection section to improve your skills in efficiently traversing the data. In short, it is important to remember that anything you can call on a Collection object, you can also do in a Query object:

// Use the combine() method from the collections library
// This is equivalent to find('list')
$keyValueList = $articles->find()->combine('id', 'title');

// An advanced example
$results = $articles->find()
    ->where(['id >' => 1])
    ->order(['title' => 'DESC'])
    ->map(function ($row) { // map() is a collection method, it executes the query
        $row->trimmedTitle = trim($row->title);
        return $row;
    })
    ->combine('id', 'trimmedTitle') // combine() is another collection method
    ->toArray(); // Also a collections library method

foreach ($results as $id => $trimmedTitle) {
    echo "$id : $trimmedTitle";
}

Queries Are Lazily Evaluated¶

Query objects are lazily evaluated. This means a query is not executed until one of the following things occur:

• The query is iterated with foreach().

• The query’s execute() method is called. This will return the underlying statement object, and is to be used with insert/update/delete queries.

• The query’s first() method is called. This will return the first result in the set built by SELECT (it adds LIMIT 1 to the query).

• The query’s all() method is called. This will return the result set and can only be used with SELECT statements.

• The query’s toList() or toArray() method is called.

Until one of these conditions are met, the query can be modified without additional SQL being sent to the database. It also means that if a Query hasn’t been evaluated, no SQL is ever sent to the database. Once executed, modifying and re-evaluating a query will result in additional SQL being run.

If you want to take a look at what SQL CakePHP is generating, you can turn database query logging on.

Selecting Specific Fields¶

By default a query will select all fields from a table, the exception is when you call the select() function yourself and pass certain fields:

// Only select id and title from the articles table
$articles->find()->select(['id', 'title']);

If you wish to still select all fields from a table after having called select($fields), you can pass the table instance to select() for this purpose:

// Only all fields from the articles table including
// a calculated slug field.
$query = $articlesTable->find();
$query
    ->select(['slug' => $query->func()->concat(['title' => 'identifier', '-', 'id' => 'identifier'])])
    ->select($articlesTable); // Select all fields from articles

If you want to select all but a few fields on a table, you can use selectAllExcept():

$query = $articlesTable->find();

// Get all fields except the published field.
$query->selectAllExcept($articlesTable, ['published']);

You can also pass an Association object when working with contained associations.

Using SQL Functions¶

CakePHP’s ORM offers abstraction for some commonly used SQL functions. Using the abstraction allows the ORM to select the platform specific implementation of the function you want. For example, concat is implemented differently in MySQL, PostgreSQL and SQL Server. Using the abstraction allows your code to be portable:

// Results in SELECT COUNT(*) count FROM ...
$query = $articles->find();
$query->select(['count' => $query->func()->count('*')]);

A number of commonly used functions can be created with the func() method:

rand()

Generate a random value between 0 and 1 via SQL.

sum()

Calculate a sum. Assumes arguments are literal values.

avg()

Calculate an average. Assumes arguments are literal values.

min()

Calculate the min of a column. Assumes arguments are literal values.

max()

Calculate the max of a column. Assumes arguments are literal values.

count()

Calculate the count. Assumes arguments are literal values.

concat()

Concatenate two values together. Assumes arguments are bound parameters.

coalesce()

Coalesce values. Assumes arguments are bound parameters.

dateDiff()

Get the difference between two dates/times. Assumes arguments are bound parameters.

now()

Defaults to returning date and time, but accepts ‘time’ or ‘date’ to return only those values.

extract()

Returns the specified date part from the SQL expression.

dateAdd()

Add the time unit to the date expression.

dayOfWeek()

Returns a FunctionExpression representing a call to SQL WEEKDAY function.

$query = $articles->find()->innerJoinWith('Categories');
$concat = $query->func()->concat([
    'Articles.title' => 'identifier',
    ' - CAT: ',
    'Categories.name' => 'identifier',
    ' - Age: ',
    $query->func()->dateDiff([
        'NOW()' => 'literal',
        'Articles.created' => 'identifier'
    ])
]);
$query->select(['link_title' => $concat]);

SELECT CONCAT(
    Articles.title,
    :c0,
    Categories.name,
    :c1,
    (DATEDIFF(NOW(), Articles.created))
) FROM articles;

$query = $articles->find();
$year = $query->func()->year([
    'created' => 'identifier'
]);
$time = $query->func()->date_format([
    'created' => 'identifier',
    "'%H:%i'" => 'literal'
]);
$query->select([
    'yearCreated' => $year,
    'timeCreated' => $time
]);

SELECT YEAR(created) as yearCreated,
       DATE_FORMAT(created, '%H:%i') as timeCreated
FROM articles;

Ordering Results¶

To apply ordering, you can use the order method:

$query = $articles->find()
    ->order(['title' => 'ASC', 'id' => 'ASC']);

When calling order() multiple times on a query, multiple clauses will be appended. However, when using finders you may sometimes need to overwrite the ORDER BY. Set the second parameter of order() (as well as orderAsc() or orderDesc()) to Query::OVERWRITE or to true:

$query = $articles->find()
    ->order(['title' => 'ASC']);
// Later, overwrite the ORDER BY clause instead of appending to it.
$query = $articles->find()
    ->order(['created' => 'DESC'], Query::OVERWRITE);

The orderAsc and orderDesc methods can be used when you need to sort on complex expressions:

$query = $articles->find();
$concat = $query->func()->concat([
    'title' => 'identifier',
    'synopsis' => 'identifier'
]);
$query->orderAsc($concat);

To build complex order clauses, use a Closure to build order expressions:

$query->orderAsc(function (QueryExpression $exp, Query $query) {
    return $exp->addCase(...);
});

Limiting Results¶

To limit the number of rows or set the row offset you can use the limit() and page() methods:

// Fetch rows 50 to 100
$query = $articles->find()
    ->limit(50)
    ->page(2);

As you can see from the examples above, all the methods that modify the query provide a fluent interface, allowing you to build a query through chained method calls.

Aggregates - Group and Having¶

When using aggregate functions like count and sum you may want to use group by and having clauses:

$query = $articles->find();
$query->select([
    'count' => $query->func()->count('view_count'),
    'published_date' => 'DATE(created)'
])
->group('published_date')
->having(['count >' => 3]);

Case Statements¶

The ORM also offers the SQL case expression. The case expression allows for implementing if ... then ... else logic inside your SQL. This can be useful for reporting on data where you need to conditionally sum or count data, or where you need to specific data based on a condition.

If we wished to know how many published articles are in our database, we could use the following SQL:

SELECT
COUNT(CASE WHEN published = 'Y' THEN 1 END) AS number_published,
COUNT(CASE WHEN published = 'N' THEN 1 END) AS number_unpublished
FROM articles

To do this with the query builder, we’d use the following code:

$query = $articles->find();
$publishedCase = $query->newExpr()
    ->addCase(
        $query->newExpr()->add(['published' => 'Y']),
        1,
        'integer'
    );
$unpublishedCase = $query->newExpr()
    ->addCase(
        $query->newExpr()->add(['published' => 'N']),
        1,
        'integer'
    );

$query->select([
    'number_published' => $query->func()->count($publishedCase),
    'number_unpublished' => $query->func()->count($unpublishedCase)
]);

The addCase function can also chain together multiple statements to create if .. then .. [elseif .. then .. ] [ .. else ] logic inside your SQL.

If we wanted to classify cities into SMALL, MEDIUM, or LARGE based on population size, we could do the following:

$query = $cities->find()
    ->where(function (QueryExpression $exp, Query $q) {
        return $exp->addCase(
            [
                $q->newExpr()->lt('population', 100000),
                $q->newExpr()->between('population', 100000, 999000),
                $q->newExpr()->gte('population', 999001),
            ],
            ['SMALL',  'MEDIUM', 'LARGE'], # values matching conditions
            ['string', 'string', 'string'] # type of each value
        );
    });
# WHERE CASE
#   WHEN population < 100000 THEN 'SMALL'
#   WHEN population BETWEEN 100000 AND 999000 THEN 'MEDIUM'
#   WHEN population >= 999001 THEN 'LARGE'
#   END

Any time there are fewer case conditions than values, addCase will automatically produce an if .. then .. else statement:

$query = $cities->find()
    ->where(function (QueryExpression $exp, Query $q) {
        return $exp->addCase(
            [
                $q->newExpr()->eq('population', 0),
            ],
            ['DESERTED', 'INHABITED'], # values matching conditions
            ['string', 'string'] # type of each value
        );
    });
# WHERE CASE
#   WHEN population = 0 THEN 'DESERTED' ELSE 'INHABITED' END

Fetching Arrays Instead of Entities¶

While ORMs and object result sets are powerful, creating entities is sometimes unnecessary. For example, when accessing aggregated data, building an Entity may not make sense. The process of converting the database results to entities is called hydration. If you wish to disable this process you can do this:

$query = $articles->find();
$query->enableHydration(false); // Results as arrays instead of entities
$result = $query->toList(); // Execute the query and return the array

After executing those lines, your result should look similar to this:

[
    ['id' => 1, 'title' => 'First Article', 'body' => 'Article 1 body' ...],
    ['id' => 2, 'title' => 'Second Article', 'body' => 'Article 2 body' ...],
    ...
]

Adding Calculated Fields¶

After your queries, you may need to do some post-processing. If you need to add a few calculated fields or derived data, you can use the formatResults() method. This is a lightweight way to map over the result sets. If you need more control over the process, or want to reduce results you should use the Map/Reduce feature instead. If you were querying a list of people, you could calculate their age with a result formatter:

// Assuming we have built the fields, conditions and containments.
$query->formatResults(function (\Cake\Collection\CollectionInterface $results) {
    return $results->map(function ($row) {
        $row['age'] = $row['birth_date']->diff(new \DateTime)->y;
        return $row;
    });
});

As you can see in the example above, formatting callbacks will get a ResultSetDecorator as their first argument. The second argument will be the Query instance the formatter was attached to. The $results argument can be traversed and modified as necessary.

Result formatters are required to return an iterator object, which will be used as the return value for the query. Formatter functions are applied after all the Map/Reduce routines have been executed. Result formatters can be applied from within contained associations as well. CakePHP will ensure that your formatters are properly scoped. For example, doing the following would work as you may expect:

// In a method in the Articles table
$query->contain(['Authors' => function ($q) {
    return $q->formatResults(function (\Cake\Collection\CollectionInterface $authors) {
        return $authors->map(function ($author) {
            $author['age'] = $author['birth_date']->diff(new \DateTime)->y;
            return $author;
        });
    });
}]);

// Get results
$results = $query->all();

// Outputs 29
echo $results->first()->author->age;

As seen above, the formatters attached to associated query builders are scoped to operate only on the data in the association. CakePHP will ensure that computed values are inserted into the correct entity.

Using Identifiers in Expressions¶

When you need to reference a column or SQL identifier in your queries you can use the identifier() method:

$query = $countries->find();
$query->select([
        'year' => $query->func()->year([$query->identifier('created')])
    ])
    ->where(function ($exp, $query) {
        return $exp->gt('population', 100000);
    });

You can use identifier() in comparisons to aggregations too:

$query = $this->Orders->find();
$query->select(['Customers.customer_name', 'total_orders' => $query->func()->count('Orders.order_id')])
    ->contain('Customers')
    ->group(['Customers.customer_name'])
    ->having(['total_orders >=' => $query->identifier('Customers.minimum_order_count')]);

Automatically Creating IN Clauses¶

When building queries using the ORM, you will generally not have to indicate the data types of the columns you are interacting with, as CakePHP can infer the types based on the schema data. If in your queries you’d like CakePHP to automatically convert equality to IN comparisons, you’ll need to indicate the column data type:

$query = $articles->find()
    ->where(['id' => $ids], ['id' => 'integer[]']);

// Or include IN to automatically cast to an array.
$query = $articles->find()
    ->where(['id IN' => $ids]);

The above will automatically create id IN (...) instead of id = ?. This can be useful when you do not know whether you will get a scalar or array of parameters. The [] suffix on any data type name indicates to the query builder that you want the data handled as an array. If the data is not an array, it will first be cast to an array. After that, each value in the array will be cast using the type system. This works with complex types as well. For example, you could take a list of DateTime objects using:

$query = $articles->find()
    ->where(['post_date' => $dates], ['post_date' => 'date[]']);

Automatic IS NULL Creation¶

When a condition value is expected to be null or any other value, you can use the IS operator to automatically create the correct expression:

$query = $categories->find()
    ->where(['parent_id IS' => $parentId]);

The above will create parent_id` = :c1 or parent_id IS NULL depending on the type of $parentId

Automatic IS NOT NULL Creation¶

When a condition value is expected not to be null or any other value, you can use the IS NOT operator to automatically create the correct expression:

$query = $categories->find()
    ->where(['parent_id IS NOT' => $parentId]);

The above will create parent_id` != :c1 or parent_id IS NOT NULL depending on the type of $parentId

Raw Expressions¶

When you cannot construct the SQL you need using the query builder, you can use expression objects to add snippets of SQL to your queries:

$query = $articles->find();
$expr = $query->newExpr()->add('1 + 1');
$query->select(['two' => $expr]);

Expression objects can be used with any query builder methods like where(), limit(), group(), select() and many other methods.

Returning the Total Count of Records¶

Using a single query object, it is possible to obtain the total number of rows found for a set of conditions:

$total = $articles->find()->where(['is_active' => true])->count();

The count() method will ignore the limit, offset and page clauses, thus the following will return the same result:

$total = $articles->find()->where(['is_active' => true])->limit(10)->count();

This is useful when you need to know the total result set size in advance, without having to construct another Query object. Likewise, all result formatting and map-reduce routines are ignored when using the count() method.

Moreover, it is possible to return the total count for a query containing group by clauses without having to rewrite the query in any way. For example, consider this query for retrieving article ids and their comments count:

$query = $articles->find();
$query->select(['Articles.id', $query->func()->count('Comments.id')])
    ->matching('Comments')
    ->group(['Articles.id']);
$total = $query->count();

After counting, the query can still be used for fetching the associated records:

$list = $query->all();

Sometimes, you may want to provide an alternate method for counting the total records of a query. One common use case for this is providing a cached value or an estimate of the total rows, or to alter the query to remove expensive unneeded parts such as left joins. This becomes particularly handy when using the CakePHP built-in pagination system which calls the count() method:

$query = $query->where(['is_active' => true])->counter(function ($query) {
    return 100000;
});
$query->count(); // Returns 100000

In the example above, when the pagination component calls the count method, it will receive the estimated hard-coded number of rows.

Caching Loaded Results¶

When fetching entities that don’t change often you may want to cache the results. The Query class makes this simple:

$query->cache('recent_articles');

Will enable caching on the query’s result set. If only one argument is provided to cache() then the ‘default’ cache configuration will be used. You can control which caching configuration is used with the second parameter:

// String config name.
$query->cache('recent_articles', 'dbResults');

// Instance of CacheEngine
$query->cache('recent_articles', $memcache);

In addition to supporting static keys, the cache() method accepts a function to generate the key. The function you give it will receive the query as an argument. You can then read aspects of the query to dynamically generate the cache key:

// Generate a key based on a simple checksum
// of the query's where clause
$query->cache(function ($q) {
    return 'articles-' . md5(serialize($q->clause('where')));
});

The cache method makes it simple to add cached results to your custom finders or through event listeners.

When the results for a cached query are fetched the following happens:

1. The Model.beforeFind event is triggered.

2. If the query has results set, those will be returned.

3. The cache key will be resolved and cache data will be read. If the cache data is not empty, those results will be returned.

4. If the cache misses, the query will be executed and a new ResultSet will be created. This ResultSet will be written to the cache and returned.

Passing Conditions to Contain¶

When using contain() you are able to restrict the data returned by the associations and filter them by conditions. To specify conditions, pass an anonymous function that receives as the first argument a query object, \Cake\ORM\Query:

// In a controller or table method.
// Prior to 3.5.0 you would use contain(['Comments' => function () { ... }])

$query = $articles->find()->contain('Comments', function (Query $q) {
    return $q
        ->select(['body', 'author_id'])
        ->where(['Comments.approved' => true]);
});

This also works for pagination at the Controller level:

$this->paginate['contain'] = [
    'Comments' => function (Query $query) {
        return $query->select(['body', 'author_id'])
        ->where(['Comments.approved' => true]);
    }
];

It is also possible to restrict deeply-nested associations using the dot notation:

$query = $articles->find()->contain([
    'Comments',
    'Authors.Profiles' => function (Query $q) {
        return $q->where(['Profiles.is_published' => true]);
    }
]);

In the above example, you’ll still get authors even if they don’t have a published profile. To only get authors with a published profile use matching(). If you have defined custom finders in your associations, you can use them inside contain():

// Bring all articles, but only bring the comments that are approved and
// popular.
$query = $articles->find()->contain('Comments', function (Query $q) {
    return $q->find('approved')->find('popular');
});

If you need full control over the query that is generated, you can tell contain() to not append the foreignKey constraints to the generated query. In that case you should use an array passing foreignKey and queryBuilder:

$query = $articles->find()->contain([
    'Authors' => [
        'foreignKey' => false,
        'queryBuilder' => function (Query $q) {
            return $q->where(...); // Full conditions for filtering
        }
    ]
]);

If you have limited the fields you are loading with select() but also want to load fields off of contained associations, you can pass the association object to select():

// Select id & title from articles, but all fields off of Users.
$query = $articles->find()
    ->select(['id', 'title'])
    ->select($articles->Users)
    ->contain(['Users']);

Alternatively, if you have multiple associations, you can use enableAutoFields():

// Select id & title from articles, but all fields off of Users, Comments
// and Tags.
$query->select(['id', 'title'])
    ->contain(['Comments', 'Tags'])
    ->enableAutoFields(true) // Prior to 3.4.0 use autoFields(true)
    ->contain(['Users' => function(Query $q) {
        return $q->autoFields(true);
    }]);

Sorting Contained Associations¶

When loading HasMany and BelongsToMany associations, you can use the sort option to sort the data in those associations:

$query->contain([
    'Comments' => [
        'sort' => ['Comments.created' => 'DESC']
    ]
]);

Filtering by Associated Data¶

A fairly common query case with associations is finding records ‘matching’ specific associated data. For example if you have ‘Articles belongsToMany Tags’ you will probably want to find Articles that have the CakePHP tag. This is extremely simple to do with the ORM in CakePHP:

// In a controller or table method.

$query = $articles->find();
$query->matching('Tags', function ($q) {
    return $q->where(['Tags.name' => 'CakePHP']);
});

You can apply this strategy to HasMany associations as well. For example if ‘Authors HasMany Articles’, you could find all the authors with recently published articles using the following:

$query = $authors->find();
$query->matching('Articles', function ($q) {
    return $q->where(['Articles.created >=' => new DateTime('-10 days')]);
});

Filtering by deep associations is surprisingly easy, and the syntax should be already familiar to you:

// In a controller or table method.
$query = $products->find()->matching(
    'Shops.Cities.Countries', function ($q) {
        return $q->where(['Countries.name' => 'Japan']);
    }
);

// Bring unique articles that were commented by 'markstory' using passed variable
// Dotted matching paths should be used over nested matching() calls
$username = 'markstory';
$query = $articles->find()->matching('Comments.Users', function ($q) use ($username) {
    return $q->where(['username' => $username]);
});

The data from the association that is ‘matched’ will be available on the _matchingData property of entities. If both match and contain the same association, you can expect to get both the _matchingData and standard association properties in your results.

Using innerJoinWith¶

Sometimes you need to match specific associated data but without actually loading the matching records like matching(). You can create just the INNER JOIN that matching() uses with innerJoinWith():

$query = $articles->find();
$query->innerJoinWith('Tags', function ($q) {
    return $q->where(['Tags.name' => 'CakePHP']);
});

innerJoinWith() allows you to use the same parameters and dot notation:

$query = $products->find()->innerJoinWith(
    'Shops.Cities.Countries', function ($q) {
        return $q->where(['Countries.name' => 'Japan']);
    }
);

You can combine innerJoinWith() and contain() with the same association when you want to match specific records and load the associated data together. The example below matches Articles that have specific Tags and loads the same Tags:

$filter = ['Tags.name' => 'CakePHP'];
$query = $articles->find()
    ->distinct($articles->getPrimaryKey())
    ->contain('Tags', function (Query $q) use ($filter) {
        return $q->where($filter);
    })
    ->innerJoinWith('Tags', function (Query $q) use ($filter) {
        return $q->where($filter);
    });

$query
    ->select(['country_name' => 'Countries.name'])
    ->innerJoinWith('Countries');

Using notMatching¶

The opposite of matching() is notMatching(). This function will change the query so that it filters results that have no relation to the specified association:

// In a controller or table method.

$query = $articlesTable
    ->find()
    ->notMatching('Tags', function ($q) {
        return $q->where(['Tags.name' => 'boring']);
    });

The above example will find all articles that were not tagged with the word boring. You can apply this method to HasMany associations as well. You could, for example, find all the authors with no published articles in the last 10 days:

$query = $authorsTable
    ->find()
    ->notMatching('Articles', function ($q) {
        return $q->where(['Articles.created >=' => new \DateTime('-10 days')]);
    });

It is also possible to use this method for filtering out records not matching deep associations. For example, you could find articles that have not been commented on by a certain user:

$query = $articlesTable
    ->find()
    ->notMatching('Comments.Users', function ($q) {
        return $q->where(['username' => 'jose']);
    });

Since articles with no comments at all also satisfy the condition above, you may want to combine matching() and notMatching() in the same query. The following example will find articles having at least one comment, but not commented by a certain user:

$query = $articlesTable
    ->find()
    ->notMatching('Comments.Users', function ($q) {
        return $q->where(['username' => 'jose']);
    })
    ->matching('Comments');

Keep in mind that contrary to the matching() function, notMatching() will not add any data to the _matchingData property in the results.

Using leftJoinWith¶

On certain occasions you may want to calculate a result based on an association, without having to load all the records for it. For example, if you wanted to load the total number of comments an article has along with all the article data, you can use the leftJoinWith() function:

$query = $articlesTable->find();
$query->select(['total_comments' => $query->func()->count('Comments.id')])
    ->leftJoinWith('Comments')
    ->group(['Articles.id'])
    ->enableAutoFields(true); // Prior to 3.4.0 use autoFields(true);

The results for the above query will contain the article data and the total_comments property for each of them.

leftJoinWith() can also be used with deeply nested associations. This is useful, for example, for bringing the count of articles tagged with a certain word, per author:

$query = $authorsTable
    ->find()
    ->select(['total_articles' => $query->func()->count('Articles.id')])
    ->leftJoinWith('Articles.Tags', function ($q) {
        return $q->where(['Tags.name' => 'awesome']);
    })
    ->group(['Authors.id'])
    ->enableAutoFields(true); // Prior to 3.4.0 use autoFields(true);

This function will not load any columns from the specified associations into the result set.

Adding Joins¶

In addition to loading related data with contain(), you can also add additional joins with the query builder:

$query = $articles->find()
    ->join([
        'table' => 'comments',
        'alias' => 'c',
        'type' => 'LEFT',
        'conditions' => 'c.article_id = articles.id',
    ]);

You can append multiple joins at the same time by passing an associative array with multiple joins:

$query = $articles->find()
    ->join([
        'c' => [
            'table' => 'comments',
            'type' => 'LEFT',
            'conditions' => 'c.article_id = articles.id',
        ],
        'u' => [
            'table' => 'users',
            'type' => 'INNER',
            'conditions' => 'u.id = articles.user_id',
        ]
    ]);

As seen above, when adding joins the alias can be the outer array key. Join conditions can also be expressed as an array of conditions:

$query = $articles->find()
    ->join([
        'c' => [
            'table' => 'comments',
            'type' => 'LEFT',
            'conditions' => [
                'c.created >' => new DateTime('-5 days'),
                'c.moderated' => true,
                'c.article_id = articles.id'
            ]
        ],
    ], ['c.created' => 'datetime', 'c.moderated' => 'boolean']);

When creating joins by hand and using array based conditions, you need to provide the datatypes for each column in the join conditions. By providing datatypes for the join conditions, the ORM can correctly convert data types into SQL. In addition to join() you can use rightJoin(), leftJoin() and innerJoin() to create joins:

// Join with an alias and string conditions
$query = $articles->find();
$query->leftJoin(
    ['Authors' => 'authors'],
    ['Authors.id = Articles.author_id']);

// Join with an alias, array conditions, and types
$query = $articles->find();
$query->innerJoin(
    ['Authors' => 'authors'],
    [
    'Authors.promoted' => true,
    'Authors.created' => new DateTime('-5 days'),
    'Authors.id = Articles.author_id'
    ],
    ['Authors.promoted' => 'boolean', 'Authors.created' => 'datetime']);

It should be noted that if you set the quoteIdentifiers option to true when defining your Connection, join conditions between table fields should be set as follow:

$query = $articles->find()
    ->join([
        'c' => [
            'table' => 'comments',
            'type' => 'LEFT',
            'conditions' => [
                'c.article_id' => new \Cake\Database\Expression\IdentifierExpression('articles.id')
            ]
        ],
    ]);

This ensures that all of your identifiers will be quoted across the Query, avoiding errors with some database Drivers (PostgreSQL notably)

Binding values¶

It is possible to protect against many unsafe situations by using bindings. Similar to binding values to prepared statements, values can be bound to queries using the Cake\Database\Query::bind() method.

The following example would be a safe variant of the unsafe, SQL injection prone example given above:

$query
    ->where([
        'MATCH (comment) AGAINST (:userData)',
        'created < NOW() - :moreUserData'
    ])
    ->bind(':userData', $userData, 'string')
    ->bind(':moreUserData', $moreUserData, 'datetime');

Unions¶

Unions are created by composing one or more select queries together:

$inReview = $articles->find()
    ->where(['need_review' => true]);

$unpublished = $articles->find()
    ->where(['published' => false]);

$unpublished->union($inReview);

You can create UNION ALL queries using the unionAll() method:

$inReview = $articles->find()
    ->where(['need_review' => true]);

$unpublished = $articles->find()
    ->where(['published' => false]);

$unpublished->unionAll($inReview);

Subqueries¶

Subqueries are a powerful feature in relational databases and building them in CakePHP is fairly intuitive. By composing queries together, you can make subqueries:

// Prior to 3.6.0 use association() instead.
$matchingComment = $articles->getAssociation('Comments')->find()
    ->select(['article_id'])
    ->distinct()
    ->where(['comment LIKE' => '%CakePHP%']);

$query = $articles->find()
    ->where(['id IN' => $matchingComment]);

Subqueries are accepted anywhere a query expression can be used. For example, in the select() and join() methods.

Adding Locking Statements¶

Most relational database vendors support taking out locks when doing select operations. You can use the epilog() method for this:

// In MySQL
$query->epilog('FOR UPDATE');

The epilog() method allows you to append raw SQL to the end of queries. You should never put raw user data into epilog().

Executing Complex Queries¶

While the query builder makes it easy to build most queries, very complex queries can be tedious and complicated to build. You may want to execute the desired SQL directly.

Executing SQL directly allows you to fine tune the query that will be run. However, doing so doesn’t let you use contain or other higher level ORM features.