Default HTTP client implementation using Req.

This module is used by default when you create a new Humaans client without specifying a custom HTTP client.

Rate-limit retries

By default this client enables Req's :transient retry strategy with max_retries: 3. That means responses with status 408, 429, 500, 502, 503, and 504 are automatically retried with exponential backoff, honouring the Retry-After header when the API includes one. For the Humaans API the practical effect is that bursty bulk reads no longer fail outright on a single rate-limit hit.

To opt out, set retry: false in your :req_options:

Humaans.new(access_token: "...", req_options: [retry: false])

Customization

For most tweaks (timeouts, retries, plugins), pass :req_options directly to Humaans.new/1. They are merged into the Req client config:

Humaans.new(
  access_token: "...",
  req_options: [connect_options: [timeout: 30_000], retry: :transient]
)

If you need to wrap or replace the entire request pipeline, implement your own module satisfying Humaans.HTTPClient.Behaviour and pass it via :http_client:

defmodule MyCustomReqClient do
  @behaviour Humaans.HTTPClient.Behaviour

  @impl true
  def request(client, opts) do
    Req.new(base_url: client.base_url, auth: {:bearer, client.access_token})
    |> Req.merge(opts)
    # Add custom Req plugins or middleware here
    |> Req.request()
  end
end