ElasticSearch Views

This section describes a new feature: ElasticSearch Views.

Problem

The problem this solves is caching complex data structures, often involving multiple levels of nesting of various models, in order to be able to quickly filter, sort, and then display the output on a certain page (mainly search pages). For example, you might want to have a search page, which displays users' names, their services, the name of the company they belong to with the logo of the company, the average review score, the number of components they have created, and so on. It should be possible to filter by company name, by their location, whether or not they provide certain services, and also to provide various sorting options.

Solution

The proposed solution is that instead of having generic customizations and people endpoints, you should be able to define your own data structures. Let's assume that you need a data structure like the following, to display a search page that displays programmers:

[
    {
        "name": "John Doe",
        "reviews_count": 3.5,
        "company": {
          "name": "platformOS",
          "logo":  "https://example.com/pos.jpg"
         },
        "reviews": [
          {
            "stars": 3.5,
            "author": {
              "name": "Jane"
            }
          }
        ],
        "services": [
          {
            "name": "Coding"
          }
        ]
    }
]

We could have a directory in app called, for example, cache_mappings, and in there a file like programmers_search.yml with the following content:

programmers_search:
    programmers:
        type: user
        fields:
            name:
                type: raw
                value: "{% user.first_name | append: ' ' | user.last_name %}"
                depends_on:
                  - field: first_name
                  - field: last_name

            reviews_count:
                type: float
                value: "{% graph_query get_reviews …, id: user.id %}"
                depends_on:
                  - count: review
        belongs_to:
            company:
                foreign_key: "{% user.properties.company_id %}"
                type: company
                fields:
                    name:
                        type: raw
                        value: "{% company.properties.name %}
                        depends_on:
                          - property: name
                    logo:
                        type: raw
                        value: "{% company.custom_images.logo.url %}
                        depends_on:
                          - property: logo
        collections:
            reviews:
                foreign_key: user_id
                type: review
                fields:
                    stars:
                        type: float
                belongs_to:
                    author:
                        foreign_key: 'author_id'
                        type: user
                        fields:
                            name:
                                type: raw
                                value: "{% author.fisrt_name | append: ' ' | append: author.last_name %}"
            services:
                type: service
                fields:
                    name:
                        type: raw
                        value: "{% service.properties.name %}"
                        depends_on:
                            - property: name

The example is very verbose, but it does not need to be - it is just to show the flexibility and power of such a solution.