search value within all keys in a nested elasticsearch object

Question

I want to know if it is possible to search value in all array elements in a nested Elasticsearch object in one single query? See my situation next:

In ElasticSearch database I've configured a nested object on column name:

$set_index = [
    'index' => 'table',
    'body' => [
        'mappings' => [
            'table' => [
                '_source' => [
                    'enabled' => true
                ],
                'properties' => [
                    "name" => [ // name of column
                        "type" => "nested" 
                    ],
                ]
            ]
        ]
    ]
];

Then I insert following data (through PHP):

$data1 = array(
    'descripotion' => 'Row 1',
    'name' => [
        'en' => 'First name data',
        'ru' => 'Первое имя',
        'de' => 'Eins data',
        'it' => 'something else',
    ]
);

$data2 = array(
    'descripotion' => 'Row 2',
    'name' => [
        'en' => 'Second name data',
    ]
);

And then I query on this data using the POST method through Postman Chrome extension:

POST http://localhost:9200/_search

Body is:

{
    "query":{
        "nested": {
            "path": "name",
            "query": {
                "bool": {
                    "must": [
                            { "match": { "name.en": "First" }}
                    ]
                }
            }
        }
    }
}

This is working perfectly. In above query - I am querying data based on English translation in column name.

So if I wanted to search using the Russian language, the code would have been following:

{ "match": { "name.ru": "First" }}

Now, what I want to do - is to query on all translations at the same time (there can be 1 to 100 languages provided). Something like:

{ "match": { "name.*": "First" }}

question - is That possible, using my current config? If not with current config, then how? I know that if I convert name column back to normal type (instead of nested) - then it becomes possible, as all translations are concentrated into a single string type value. But in that case - i am loosing the possibility to be selective! I want both :) Preferably, without data duplication.
question - if I am searching for all translations at the same time - is it possible to prioritize one language over another? So if word data is in 5 languages - I want the document which contained English search result - to be first in the list.

Alex · Accepted Answer · 2017-03-26 12:10:15Z

Alright,

After going through countless tutorials I have come to the conclusion that ElasticSearch can not work with Associative arrays. Instead - Multidimensional arrays shall be used.

Here is an example. Before:

$arr = array(
    'name' => [
        'en' => 'First name',
        'ru' => 'Первое имя',
    ]
);

Converted to multidimensional:

$arr = array(
    'name' => [ 
        [
            'lang' => 'en',
            'value' => 'First name data'
        ],
        [
            'lang' => 'ru',
            'value' => 'Первое имя'
        ]
    ]
);

Now, back to my questions. Ending ElasticSearch query looks like this:

{
    "query":{
        "nested": {
            "path": "name",
            "query": {
                "bool": {
                    "should": [
                            { "term": { "name.lang": "it" }}
                    ],
                    "must": [
                            { "match": { "name.value": "first" }}
                    ]

                }
            }
        }
    }
}

Answers:

Answer to question 1 - remove should section from query above.
Answer to question 2 - keep the should section. Set the preferred language code there.

And if we want to search by specific language then should should be set to must.

Collectives™ on Stack Overflow

search value within all keys in a nested elasticsearch object

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related