Req.Steps (req v0.5.15)

View Source

The collection of built-in steps.

Req is composed of:

  • Req - the high-level API

  • Req.Request - the low-level API and the request struct

  • Req.Steps - the collection of built-in steps (you're here!)

  • Req.Test - the testing conveniences

Summary

Request Steps

auth(request)

Sets request authentication.

cache(request)

Performs HTTP caching using if-modified-since header.

checksum(request)

Sets expected response body checksum.

compress_body(request)

Compresses the request body.

compressed(request)

Asks the server to return compressed response.

encode_body(request)

Encodes the request body.

put_aws_sigv4(request)

Signs request with AWS Signature Version 4.

put_base_url(request)

Sets base URL for all requests.

put_params(request)

Adds params to request query string.

put_path_params(request)

Uses a templated request path.

put_plug(request)

Sets adapter to run_plug/1.

put_range(request)

Sets the "Range" request header.

put_user_agent(request)

Sets the user-agent header.

run_finch(request)

Runs the request using Finch.

run_plug(request)

Runs the request against a plug instead of over the network.

Response Steps

decode_body(request_response)

Decodes response body based on the detected format.

decompress_body(request_response)

Decompresses the response body based on the content-encoding header.

handle_http_errors(request_response)

Handles HTTP 4xx/5xx error responses.

redirect(request_response)

Follows redirects.

verify_checksum(arg)

Verifies the response body checksum.

Error Steps

retry(request_response_or_error)

Retries a request in face of errors.

Request Steps

auth(request)

Sets request authentication.

Request Options

  • :auth - sets the authorization header:

    • string - sets to this value;

    • {:basic, userinfo} - uses Basic HTTP authentication;

    • {:bearer, token} - uses Bearer HTTP authentication;

    • :netrc - load credentials from .netrc at path specified in NETRC environment variable. If NETRC is not set, load .netrc in user's home directory;

    • {:netrc, path} - load credentials from path

    • fn -> {:bearer, "eyJ0eXAi..." } end - a 0-arity function that returns one of the aforementioned types.

Examples

iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: {:basic, "foo:foo"}).status
401
iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: {:basic, "foo:bar"}).status
200
iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: fn -> {:basic, "foo:bar"} end).status
200

iex> Req.get!("https://httpbin.org/bearer", auth: {:bearer, ""}).status
401
iex> Req.get!("https://httpbin.org/bearer", auth: {:bearer, "foo"}).status
200
iex> Req.get!("https://httpbin.org/bearer", auth: fn -> {:bearer, "foo"} end).status
200

iex> System.put_env("NETRC", "./test/my_netrc")
iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: :netrc).status
200

iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: {:netrc, "./test/my_netrc"}).status
200
iex> Req.get!("https://httpbin.org/basic-auth/foo/bar", auth: fn -> {:netrc, "./test/my_netrc"} end).status
200

cache(request)

Performs HTTP caching using if-modified-since header.

Only successful (200 OK) responses are cached.

This step also prepends a response step that loads and writes the cache. Be careful when prepending other response steps, make sure the cache is loaded/written as soon as possible.

Options

  • :cache - if true, performs simple caching using if-modified-since header. Defaults to false.

  • :cache_dir - the directory to store the cache, defaults to <user_cache_dir>/req (see: :filename.basedir/3)

Examples

iex> url = "https://elixir-lang.org"
iex> response1 = Req.get!(url, cache: true)
iex> response2 = Req.get!(url, cache: true)
iex> response1 == response2
true

checksum(request)

Sets expected response body checksum.

Request Options

  • :checksum - if set, this is the expected response body checksum.

    Can be one of:

    • "md5:(...)"
    • "sha1:(...)"
    • "sha256:(...)"

Examples

iex> resp = Req.get!("https://httpbin.org/json", checksum: "sha1:9274ffd9cf273d4a008750f44540c4c5d4c8227c")
iex> resp.status
200

iex> Req.get!("https://httpbin.org/json", checksum: "sha1:bad")
** (Req.ChecksumMismatchError) checksum mismatch
expected: sha1:bad
actual:   sha1:9274ffd9cf273d4a008750f44540c4c5d4c8227c

compress_body(request)

Compresses the request body.

Request Options

  • :compress_body - if set to true, compresses the request body using gzip. Defaults to false.

compressed(request)

Asks the server to return compressed response.

Supported formats:

  • gzip

  • br (if brotli is installed)

  • zstd (if ezstd is installed)

Request Options

  • :compressed - if set to true, sets the accept-encoding header with compression algorithms that Req supports. Defaults to true.

    When streaming response body (into: fun | collectable), compressed defaults to false.

Examples

Req automatically decompresses response body (decompress_body/1 step) so let's disable that by passing raw: true.

By default, we ask the server to send compressed response. Let's look at the headers and the raw body. Notice the body starts with <<31, 139>> (<<0x1F, 0x8B>>), the "magic bytes" for gzip:

iex> response = Req.get!("https://elixir-lang.org", raw: true)
iex> Req.Response.get_header(response, "content-encoding")
["gzip"]
iex> response.body |> binary_part(0, 2)
<<31, 139>>

Now, let's pass compressed: false and notice the raw body was not compressed:

iex> response = Req.get!("https://elixir-lang.org", raw: true, compressed: false)
iex> response.body |> binary_part(0, 15)
"<!DOCTYPE html>"

The Brotli and Zstandard compression algorithms are also supported if the optional packages are installed:

Mix.install([
  :req,
  {:brotli, "~> 0.3.0"},
  {:ezstd, "~> 1.0"}
])

response = Req.get!("https://httpbin.org/anything")
response.body["headers"]["Accept-Encoding"]
#=> "zstd, br, gzip"

encode_body(request)

Encodes the request body.

Request Options

  • :form - if set, encodes the request body as application/x-www-form-urlencoded (using URI.encode_query/1).

  • :form_multipart - if set, encodes the request body as multipart/form-data.

    It accepts name / value pairs. value can be one of:

    • integer (automatically encoded as string)

    • iodata

    • File.Stream

    • Enumerable

    • {value, options} tuple.

      value can be any of the values mentioned above.

      Supported options are: :filename, :content_type, and :size.

      When value is an Enumerable, option :size can be set with the binary size of the value. The size will be used to calculate and send the content-length header which might be required for some servers. There is no need to pass :size for integer, iodata, and File.Stream values as it's automatically calculated.

  • :json - if set, encodes the request body as JSON (using Jason.encode_to_iodata!/1), sets the accept header to application/json, and the content-type header to application/json.

Examples

Encoding form (application/x-www-form-urlencoded):

iex> Req.post!("https://httpbin.org/anything", form: [a: 1]).body["form"]
%{"a" => "1"}

Encoding form (multipart/form-data):

iex> fields = [a: 1, b: {"2", filename: "b.txt"}]
iex> resp = Req.post!("https://httpbin.org/anything", form_multipart: fields)
iex> resp.body["form"]
%{"a" => "1"}
iex> resp.body["files"]
%{"b" => "2"}

Encoding streaming form (multipart/form-data):

iex> stream = Stream.cycle(["abc"]) |> Stream.take(3)
iex> fields = [file: {stream, filename: "b.txt"}]
iex> resp = Req.post!("https://httpbin.org/anything", form_multipart: fields)
iex> resp.body["files"]
%{"file" => "abcabcabc"}

# with explicit :size
iex> stream = Stream.cycle(["abc"]) |> Stream.take(3)
iex> fields = [file: {stream, filename: "b.txt", size: 9}]
iex> resp = Req.post!("https://httpbin.org/anything", form_multipart: fields)
iex> resp.body["files"]
%{"file" => "abcabcabc"}

Encoding JSON:

iex> Req.post!("https://httpbin.org/post", json: %{a: 2}).body["json"]
%{"a" => 2}

put_aws_sigv4(request)

Signs request with AWS Signature Version 4.

Request Options

  • :aws_sigv4 - if set, the AWS options to sign request:

    • :access_key_id - the AWS access key id.

    • :secret_access_key - the AWS secret access key.

    • :token - if set, the AWS security token, for example returned from AWS STS.

    • :service - the AWS service. We try to automatically detect the service (e.g. s3.amazonaws.com host sets service to :s3)

    • :region - the AWS region. Defaults to "us-east-1".

    • :datetime - the request datetime, defaults to DateTime.utc_now(:second).

Examples

iex> req =
...>   Req.new(
...>     base_url: "https://s3.amazonaws.com",
...>     aws_sigv4: [
...>       access_key_id: System.get_env("AWS_ACCESS_KEY_ID"),
...>       secret_access_key: System.get_env("AWS_SECRET_ACCESS_KEY")
...>     ]
...>   )
iex>
iex> %{status: 200} = Req.put!(req, url: "/bucket1/key1", body: "Hello, World!")
iex> resp = Req.get!(req, url: "/bucket1/key1").body
"Hello, World!"

Request body streaming also works though content-length header must be explicitly set:

iex> path = "a.txt"
iex> File.write!(path, String.duplicate("a", 100_000))
iex> size = File.stat!(path).size
iex> chunk_size = 10 * 1024
iex> stream = File.stream!(path, chunk_size)
iex> %{status: 200} = Req.put!(req, url: "/key1", headers: [content_length: size], body: stream)
iex> byte_size(Req.get!(req, "/bucket1/key1").body)
100_000

put_base_url(request)

Sets base URL for all requests.

Request Options

  • :base_url - if set, the request URL is merged with this base URL.

    The base url can be a string, a %URI{} struct, a 0-arity function, or a {mod, fun, args} tuple describing a function to call.

Examples

iex> req = Req.new(base_url: "https://httpbin.org")
iex> Req.get!(req, url: "/status/200").status
200
iex> Req.get!(req, url: "/status/201").status
201

put_params(request)

Adds params to request query string.

Request Options

  • :params - params to add to the request query string. Defaults to [].

Examples

iex> Req.get!("https://httpbin.org/anything/query", params: [x: 1, y: 2]).body["args"]
%{"x" => "1", "y" => "2"}

put_path_params(request)

Uses a templated request path.

By default, params in the URL path are expressed as strings prefixed with :. For example, :code in https://httpbin.org/status/:code. If you want to use the {code} syntax, set path_params_style: :curly. Param names must start with a letter and can contain letters, digits, and underscores; this is true both for :colon_params as well as {curly_params}.

Path params are replaced in the request URL path. The path params are specified as a keyword list of parameter names and values, as in the examples below. The values of the parameters are converted to strings using the String.Chars protocol (to_string/1).

Request Options

  • :path_params - params to add to the templated path. Defaults to [].

  • :path_params_style (available since v0.5.1) - how path params are expressed. Can be one of:

    • :colon - (default) for Plug-style parameters, such as :code in https://httpbin.org/status/:code.

    • :curly - for OpenAPI-style parameters, such as {code} in https://httpbin.org/status/{code}.

Examples

iex> Req.get!("https://httpbin.org/status/:code", path_params: [code: 201]).status
201

iex> Req.get!("https://httpbin.org/status/{code}", path_params: [code: 201], path_params_style: :curly).status
201

put_plug(request)

Sets adapter to run_plug/1.

See run_plug/1 for more information.

Request Options

  • :plug - if set, the plug to run the request through.

put_range(request)

Sets the "Range" request header.

Request Options

  • :range - can be one of the following:

    • a string - returned as is

    • a first..last range - converted to "bytes=<first>-<last>"

Examples

iex> response = Req.get!("https://httpbin.org/range/100", range: 0..3)
iex> response.status
206
iex> response.body
"abcd"
iex> Req.Response.get_header(response, "content-range")
["bytes 0-3/100"]

put_user_agent(request)

Sets the user-agent header.

Request Options

  • :user_agent - sets the user-agent header. Defaults to "req/0.5.15".

Examples

iex> Req.get!("https://httpbin.org/user-agent").body
%{"user-agent" => "req/0.5.15"}

iex> Req.get!("https://httpbin.org/user-agent", user_agent: "foo").body
%{"user-agent" => "foo"}

run_finch(request)

Runs the request using Finch.

This is the default Req adapter. See "Adapter" section in the Req.Request module documentation for more information on adapters.

Finch returns Mint.TransportError exceptions on HTTP connection problems. These are automatically converted to Req.TransportError exceptions. Similarly, HTTP-protocol-related errors, Mint.HTTPError and Finch.Error, and converted to Req.HTTPError.

HTTP/1 Pools

On HTTP/1 connections, Finch creates a pool per {scheme, host, port} tuple. These pools are kept around to re-use connections as much as possible, however they are not automatically terminated. To do so, you can configure custom Finch pool:

{:ok, _} =
  Finch.start_link(
    name: MyFinch,
    pools: %{
      default: [
        # terminate idle {scheme, host, port} pool after 60s
        pool_max_idle_time: 60_000
      ]
    }
  )

Req.get!("https://httpbin.org/json", finch: MyFinch)

More commonly you'd add the the custom Finch pool as part of your supervision tree in your application.ex:

children = [
  {Finch,
   name: MyFinch,
   pools: %{
     default: [size: 70]
   }}
]

That way you can also configure a bigger pool size for the HTTP pool. You just mustn't forget to pass along finch: MyFinch as discussed above. You could use Req.default_options/1 to make it a global default but it's generally discouraged.

For documentation about the possible pool options and their meaning, please check out the Finch docs on pool configuration options.

Request Options

  • :finch - the name of the Finch pool. Defaults to a pool automatically started by Req.

  • :connect_options - dynamically starts (or re-uses already started) Finch pool with the given connection options:

    • :timeout - socket connect timeout in milliseconds, defaults to 30_000.

    • :protocols - the HTTP protocols to use, defaults to [:http1].

    • :hostname - Mint explicit hostname, see Mint.HTTP.connect/4 for more information.

    • :transport_opts - Mint transport options, see Mint.HTTP.connect/4 for more information.

    • :proxy_headers - Mint proxy headers, see Mint.HTTP.connect/4 for more information.

    • :proxy - Mint HTTP/1 proxy settings, a {schema, address, port, options} tuple. See Mint.HTTP.connect/4 for more information.

    • :client_settings - Mint HTTP/2 client settings, see Mint.HTTP.connect/4 for more information.

  • :inet6 - if set to true, uses IPv6.

    If the request URL looks like IPv6 address, i.e., say, [::1], it defaults to true and otherwise defaults to false. This is a shortcut for setting connect_options: [transport_opts: [inet6: true]].

  • :pool_timeout - pool checkout timeout in milliseconds, defaults to 5000.

  • :receive_timeout - socket receive timeout in milliseconds, defaults to 15_000.

  • :unix_socket - if set, connect through the given UNIX domain socket.

  • :pool_max_idle_time - the maximum number of milliseconds that a pool can be idle before being terminated, used only by HTTP1 pools. Default to :infinity.

  • :finch_private - a map or keyword list of private metadata to add to the Finch request. May be useful for adding custom data when handling telemetry with Finch.Telemetry.

  • :finch_request - a function that executes the Finch request, defaults to using Finch.request/3.

    The function should accept 4 arguments:

    • request - the %Req.Request{} struct

    • finch_request - the Finch request

    • finch_name - the Finch name

    • finch_options - the Finch options

    And it should return either {request, response} or {request, exception}.

Examples

Custom :receive_timeout:

iex> Req.get!(url, receive_timeout: 1000)

Connecting through UNIX socket:

iex> Req.get!("http:///v1.41/_ping", unix_socket: "/var/run/docker.sock").body
"OK"

Custom connection options:

iex> Req.get!(url, connect_options: [timeout: 5000])

iex> Req.get!(url, connect_options: [protocols: [:http2]])

Connecting without certificate check (useful in development, not recommended in production):

iex> Req.get!(url, connect_options: [transport_opts: [verify: :verify_none]])

Connecting with custom certificates:

iex> Req.get!(url, connect_options: [transport_opts: [cacertfile: "certs.pem"]])

Connecting through a proxy with basic authentication:

iex> Req.new(
...>  url: "https://elixir-lang.org",
...>  connect_options: [
...>    proxy: {:http, "your.proxy.com", 8888, []},
...>    proxy_headers: [{"proxy-authorization", "Basic " <> Base.encode64("user:pass")}]
...>  ]
...> )
iex> |> Req.get!()

Transport errors are represented as Req.TransportError exceptions:

iex> Req.get("https://httpbin.org/delay/1", receive_timeout: 0, retry: false)
{:error, %Req.TransportError{reason: :timeout}}

run_plug(request)

Runs the request against a plug instead of over the network.

This step is a Req adapter. It is set as the adapter by the put_plug/1 step if the :plug option is set.

It requires :plug dependency:

{:plug, "~> 1.0"}

Request Options

  • :plug - the plug to run the request through. It can be one of:

    • A function plug: a fun(conn) or fun(conn, options) function that takes a Plug.Conn and returns a Plug.Conn.

    • A module plug: a module name or a {module, options} tuple.

    Req automatically calls Plug.Conn.fetch_query_params/2 before your plug, so you can get query params using conn.query_params.

    Req also automatically parses request body using Plug.Parsers for JSON, urlencoded and multipart requests and you can access it with conn.body_params. The raw request body of the request is available by calling Req.Test.raw_body/1 with the conn in your tests.

Examples

This step is particularly useful to test plugs:

defmodule Echo do
  def call(conn, _) do
    "/" <> path = conn.request_path
    Plug.Conn.send_resp(conn, 200, path)
  end
end

test "echo" do
  assert Req.get!("http:///hello", plug: Echo).body == "hello"
end

You can define plugs as functions too:

test "echo" do
  echo = fn conn ->
    "/" <> path = conn.request_path
    Plug.Conn.send_resp(conn, 200, path)
  end

  assert Req.get!("http:///hello", plug: echo).body == "hello"
end

which is particularly useful to create HTTP service stubs, similar to tools like Bypass.

Response streaming is also supported however at the moment the entire response body is emitted as one chunk:

test "echo" do
  plug = fn conn ->
    conn = Plug.Conn.send_chunked(conn, 200)
    {:ok, conn} = Plug.Conn.chunk(conn, "echo")
    {:ok, conn} = Plug.Conn.chunk(conn, "echo")
    conn
  end

  assert Req.get!(plug: plug, into: []).body == ["echoecho"]
end

When testing JSON APIs, it's common to use the Req.Test.json/2 helper:

test "JSON" do
  plug = fn conn ->
    Req.Test.json(conn, %{message: "Hello, World!"})
  end

  resp = Req.get!(plug: plug)
  assert resp.status == 200
  assert resp.headers["content-type"] == ["application/json; charset=utf-8"]
  assert resp.body == %{"message" => "Hello, World!"}
end

You can simulate network errors by calling Req.Test.transport_error/2 in your plugs:

test "network issues" do
  plug = fn conn ->
    Req.Test.transport_error(conn, :timeout)
  end

  assert Req.get(plug: plug, retry: false) ==
           {:error, %Req.TransportError{reason: :timeout}}
end

Response Steps

decode_body(request_response)

Decodes response body based on the detected format.

Supported formats:

FormatDecoder
jsonJason.decode/2
tar, tgz:erl_tar.extract/2
zip:zip.unzip/2
gzip:zlib.gunzip/1
zst:ezstd.decompress/1 (if ezstd is installed)
csvNimbleCSV.RFC4180.parse_string/2 (if nimble_csv is installed)

The format is determined based on the content-type header of the response. For example, if the content-type is application/json, the response body is decoded as JSON. The built-in decoders also understand format extensions, such as decoding as JSON for a content-type of application/vnd.api+json. To do this, Req falls back to MIME.extensions/1; check the documentation for that function for more information.

This step is disabled on response body streaming. If response body is not a binary, in other words it has been transformed by another step, it is left as is.

Request Options

  • :decode_body - if set to false, disables automatic response body decoding. Defaults to true.

  • :decode_json - options to pass to Jason.decode/2, defaults to [].

  • :raw - if set to true, disables response body decoding. Defaults to false.

    Note: setting raw: true also disables response body decompression in the decompress_body/1 step.

Examples

Decode JSON:

iex> response = Req.get!("https://httpbin.org/json")
...> response.body["slideshow"]["title"]
"Sample Slide Show"

Decode gzip:

iex> response = Req.get!("https://httpbin.org/gzip")
...> response.body["gzipped"]
true

decompress_body(request_response)

Decompresses the response body based on the content-encoding header.

This step is disabled on response body streaming. If response body is not a binary, in other words it has been transformed by another step, it is left as is.

Supported formats:

FormatDecoder
gzip, x-gzip:zlib.gunzip/1
br:brotli.decode/1 (if brotli is installed)
zstd:ezstd.decompress/1 (if ezstd is installed)
otherReturns data as is

This step updates the following headers to reflect the changes:

  • content-encoding is removed
  • content-length is removed

Options

  • :raw - if set to true, disables response body decompression. Defaults to false.

    Note: setting raw: true also disables response body decoding in the decode_body/1 step.

Examples

iex> response = Req.get!("https://httpbin.org/gzip")
iex> response.body["gzipped"]
true

If the brotli package is installed, Brotli is also supported:

Mix.install([
  :req,
  {:brotli, "~> 0.3.0"}
])

response = Req.get!("https://httpbin.org/brotli")
Req.Response.get_header(response, "content-encoding")
#=> ["br"]
response.body["brotli"]
#=> true

handle_http_errors(request_response)

Handles HTTP 4xx/5xx error responses.

Request Options

  • :http_errors - how to handle HTTP 4xx/5xx error responses. Can be one of the following:

    • :return (default) - return the response

    • :raise - raise an error

Examples

iex> Req.get!("https://httpbin.org/status/404").status
404

iex> Req.get!("https://httpbin.org/status/404", http_errors: :raise)
** (RuntimeError) The requested URL returned error: 404
Response body: ""

redirect(request_response)

Follows redirects.

The original request method may be changed to GET depending on the status code:

CodeMethod handling
301, 302, 303Changed to GET
307, 308Method not changed

Request Options

  • :redirect - if set to false, disables automatic response redirects. Defaults to true.

  • :redirect_trusted - by default, authorization credentials are only sent on redirects with the same host, scheme and port. If :redirect_trusted is set to true, credentials will be sent to any host.

  • :redirect_log_level - the log level to emit redirect logs at. Can also be set to false to disable logging these messages. Defaults to :debug.

  • :max_redirects - the maximum number of redirects, defaults to 10. If the limit is reached, the pipeline is halted and a Req.TooManyRedirectsError exception is returned.

Examples

iex> Req.get!("http://api.github.com").status
# 23:24:11.670 [debug] redirecting to https://api.github.com/
200

iex> Req.get!("https://httpbin.org/redirect/4", max_redirects: 3)
# 23:07:59.570 [debug] redirecting to /relative-redirect/3
# 23:08:00.068 [debug] redirecting to /relative-redirect/2
# 23:08:00.206 [debug] redirecting to /relative-redirect/1
** (RuntimeError) too many redirects (3)

iex> Req.get!("http://api.github.com", redirect_log_level: false)
200

iex> Req.get!("http://api.github.com", redirect_log_level: :error)
# 23:24:11.670 [error]  redirecting to https://api.github.com/
200

verify_checksum(arg)

Verifies the response body checksum.

See checksum/1 for more information.

Error Steps

retry(request_response_or_error)

Retries a request in face of errors.

This function can be used as either or both response and error step.

Request Options

  • :retry - can be one of the following:

    • :safe_transient (default) - retry safe (GET/HEAD) requests on one of:

      • HTTP 408/429/500/502/503/504 responses

      • Req.TransportError with reason: :timeout | :econnrefused | :closed

      • Req.HTTPError with protocol: :http2, reason: :unprocessed

    • :transient - same as :safe_transient except retries all HTTP methods (POST, DELETE, etc.)

    • fun - a 2-arity function that accepts a Req.Request and either a Req.Response or an exception struct and returns one of the following:

      • true - retry with the default delay controller by default delay option described below.

      • {:delay, milliseconds} - retry with the given delay.

      • false/nil - don't retry.

    • false - don't retry.

  • :retry_delay - if not set, which is the default, the retry delay is determined by the value of the Retry-After header on HTTP 429/503 responses. If the header is not set, the default delay follows a simple exponential backoff: 1s, 2s, 4s, 8s, ...

    :retry_delay can be set to a function that receives the retry count (starting at 0) and returns the delay, the number of milliseconds to sleep before making another attempt.

  • :retry_log_level - the log level to emit retry logs at. Can also be set to false to disable logging these messages. Defaults to :warning.

  • :max_retries - maximum number of retry attempts, defaults to 3 (for a total of 4 requests to the server, including the initial one.)

Examples

With default options:

iex> Req.get!("https://httpbin.org/status/500,200").status
# 19:02:08.463 [warning] retry: got response with status 500, will retry in 2000ms, 2 attempts left
# 19:02:10.710 [warning] retry: got response with status 500, will retry in 4000ms, 1 attempt left
200

Delay with jitter:

iex> delay = fn n -> trunc(Integer.pow(2, n) * 1000 * (1 - 0.1 * :rand.uniform())) end
iex> Req.get!("https://httpbin.org/status/500,200", retry_delay: delay).status
# 08:43:19.101 [warning] retry: got response with status 500, will retry in 941ms, 2 attempts left
# 08:43:22.958 [warning] retry: got response with status 500, will retry in 1877s, 1 attempt left
200