Using the ActiveRecord ¶

For general information on how to use yii's ActiveRecord please refer to the guide.

For defining an elasticsearch ActiveRecord class your record class needs to extend from yii\elasticsearch\ActiveRecord and implement at least the attributes() method to define the attributes of the record. The handling of primary keys is different in elasticsearch as the primary key (the _id field in elasticsearch terms) is not part of the attributes by default. However it is possible to define a path mapping for the _id field to be part of the attributes. See elasticsearch docs on how to define it. The _id field of a document/record can be accessed using getPrimaryKey() and setPrimaryKey(). When path mapping is defined, the attribute name can be defined using the primaryKey() method.

The following is an example model called Customer:

class Customer extends \yii\elasticsearch\ActiveRecord
{
    /**
     * @return array the list of attributes for this record
     */
    public function attributes()
    {
        // path mapping for '_id' is setup to field 'id'
        return ['id', 'name', 'address', 'registration_date'];
    }

    /**
     * @return ActiveQuery defines a relation to the Order record (can be in other database, e.g. redis or sql)
     */
    public function getOrders()
    {
        return $this->hasMany(Order::className(), ['customer_id' => 'id'])->orderBy('id');
    }

    /**
     * Defines a scope that modifies the `$query` to return only active(status = 1) customers
     */
    public static function active($query)
    {
        $query->andWhere(['status' => 1]);
    }
}

You may override index() and type() to define the index and type this record represents.

The general usage of elasticsearch ActiveRecord is very similar to the database ActiveRecord as described in the guide. It supports the same interface and features except the following limitations and additions(!):

• As elasticsearch does not support SQL, the query API does not support join(), groupBy(), having() and union(). Sorting, limit, offset and conditional where are all supported.
• from() does not select the tables, but the index and type to query against.
• select() has been replaced with fields() which basically does the same but fields is more elasticsearch terminology. It defines the fields to retrieve from a document.
• yii\elasticsearch\ActiveQuery::via()-relations can not be defined via a table as there are no tables in elasticsearch. You can only define relations via other records.
• As elasticsearch is not only a data storage but also a search engine there is of course support added for searching your records. There are query(), filter() and yii\elasticsearch\ActiveQuery::addFacet() methods that allows to compose an elasticsearch query. See the usage example below on how they work and check out the Query DSL on how to compose query and filter parts.
• It is also possible to define relations from elasticsearch ActiveRecords to normal ActiveRecord classes and vice versa.

NOTE: elasticsearch limits the number of records returned by any query to 10 records by default. If you expect to get more records you should specify limit explicitly in query and also relation definition. This is also important for relations that use via() so that if via records are limited to 10 the relations records can also not be more than 10.

Usage example:

$customer = new Customer();
$customer->primaryKey = 1; // in this case equivalent to $customer->id = 1;
$customer->attributes = ['name' => 'test'];
$customer->save();

$customer = Customer::get(1); // get a record by pk
$customers = Customer::mget([1,2,3]); // get multiple records by pk
$customer = Customer::find()->where(['name' => 'test'])->one(); // find by query, note that you need to configure mapping for this field in order to find records properly
$customers = Customer::find()->active()->all(); // find all by query (using the `active` scope)

// http://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-match-query.html
$result = Article::find()->query(["match" => ["title" => "yii"]])->all(); // articles whose title contains "yii"

// http://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-flt-query.html
$query = Article::find()->query([
    "fuzzy_like_this" => [
        "fields" => ["title", "description"],
        "like_text" => "This query will return articles that are similar to this text :-)",
        "max_query_terms" => 12
    ]
]);

$query->all(); // gives you all the documents
// you can add facets to your search:
$query->addStatisticalFacet('click_stats', ['field' => 'visit_count']);
$query->search(); // gives you all the records + stats about the visit_count field. e.g. mean, sum, min, max etc...

Complex queries ¶

Any query can be composed using ElasticSearch's query DSL and passed to the ActiveRecord::query() method. However, ES query DSL is notorious for its verbosity, and these oversized queries soon become unmanageable. Here is a method to make queries more maintainable. Start by defining a query class just as it is done for SQL-based ActiveRecord.

class CustomerQuery extends ActiveQuery
{
    public static function name($name)
    {
        return ['match' => ['name' => $name]];
    }

    public static function address($address)
    {
        return ['match' => ['address' => $address]];
    }

    public static function registrationDateRange($dateFrom, $dateTo)
    {
        return ['range' => ['registration_date' => [
            'gte' => $dateFrom,
            'lte' => $dateTo,
        ]]];
    }
}

Now you can use these query components to assemble the resulting query and/or filter.

$customers = Customer::find()->filter([
    CustomerQuery::registrationDateRange('2016-01-01', '2016-01-20'),
])->query([
    'bool' => [
        'should' => [
            CustomerQuery::name('John'),
            CustomerQuery::address('London'),
        ],
        'must_not' => [
            CustomerQuery::name('Jack'),
        ],
    ],
])->all();

Aggregation ¶

The aggregations framework helps provide aggregated data based on a search query. It is based on simple building blocks called aggregations, that can be composed in order to build complex summaries of the data.

Using the previously defined Customer class, let's find out how many customers have registered each day. To do that we use the terms aggregation.

$aggData = Customer::find()->addAggregation('customers_by_date', 'terms', [
    'field' => 'registration_date',
    'order' => ['_count' => 'desc'],
    'size' => 10, //top 10 registration dates
])->search(null, ['search_type' => 'count']);

In this example we are specifically requesting aggregation results only. The following code further process the data.

$customersByDate = ArrayHelper::map($aggData['aggregations']['customers_by_date']['buckets'], 'key', 'doc_count');

Now $customersByDate contains 10 dates that correspond to the the highest number of users registered.

Unusual behavior of attributes with object mapping ¶

The extension updates records using the _update endpoint. Since this endpoint is designed to perform partial updates to documents, all attributes that have an "object" mapping type in ElasticSearch will be merged with existing data. To demonstrate:

$customer = new Customer();
$customer->my_attribute = ['foo' => 'v1', 'bar' => 'v2'];
$customer->save();
// at this point the value of my_attribute in ElasticSearch is {"foo": "v1", "bar": "v2"}

$customer->my_attribute = ['foo' => 'v3', 'bar' => 'v4'];
$customer->save();
// now the value of my_attribute in ElasticSearch is {"foo": "v3", "bar": "v4"}

$customer->my_attribute = ['baz' => 'v5'];
$customer->save();
// now the value of my_attribute in ElasticSearch is {"foo": "v3", "bar": "v4", "baz": "v5"}
// but $customer->my_attribute is still equal to ['baz' => 'v5']

Since this logic only applies to objects, the solution is to wrap the object into a single-element array. Since to ElasticSearch a single-element array is the same thing as the element itself, there is no need to modify any other code.

$customer->my_attribute = [['new' => 'value']]; // note the double brackets
$customer->save();
// now the value of my_attribute in ElasticSearch is {"new": "value"}
$customer->my_attribute = $customer->my_attribute[0]; // could be done for consistensy

For more information see this discussion: https://discuss.elastic.co/t/updating-an-object-field/110735