A small library to build Ecto query from map

I have just released a small library that help to build query from a map with structure similar to Mongo DB filter like this:

Filtery.apply(Post, %{name: {:like, "elixir"}, is_draft: false, tags: {:has, "ecto"} })

Here is the repo:

It’s quite simple and supports filtering on a single table only.
It supports basic operators eq, :gt, :gte, :in, :lt, :lte, :ne, :nin, :and, :or and some other helpful operators. And you can define your own operators too.

Please try and leave your feedback. Thank you.

Update v0.2.0

I have added a special operator ref to join multiple table

query = Filtery.apply(Post, %{comments: {:ref, %{
                                    approved: true,
                                    content: {:like, "filtery"}

comments field must be the association name in your Post schema


This is neat. It will work well for simple use cases. Its also good in way that you can build filters progressively as its a map. But its use will be limited as most production use-cases will involve multi-table filtering.

1 Like

Filtery.apply return a query so you can chain as many time as you want like this

|> Filtery.apply(%{title: {:not, {:like, "elixir"}}})
|> Filtery.apply(%{is_draft: false, comment_count: {:gt, 10}})
|> Repo.all()

I have been thinking of a simple specification to define a join filter but there is not good enough solution.
Do you have any suggestion?

1 Like

I have added new operator for querying join table. Please check update above.


That’s very nice!

1 Like

This looks similar to my library QueryBuilder :grin:
It’s possible to have joins.

Possible to make a query out of a keyword list with QueryBuilder.from_list/2.


Thanks for your library, I didn’t know it before. It has lots of cool features.

This module is actually a part of my projects. I have to write lots of code to build query from request parameter so its goal is to help building query conditions. And I want to keep it simple so my colleges can learn to use it quickly.

1 Like

It’s really nice to have ways of building queries using a more data-driven approach. It’s really awesome