I'm struggling with an issue where something is reading the body of an http request before Plug.Parsers.JSON gets it in the pipeline. Because of this, read_body in the plug for json times out--you can't read the body twice.
We have an HMAC implementation in an earlier plug in our pipeline and it reads the body in some cases. Is there a pattern for how use of the body is to behave in Plug? I mean, if we can only read it once, and it has to be decoded in Plug.Parsers.JSON, well...it's not going to work.
Follow on question. Do we need to include the request body when we generate the HMAC hash? I mean, it feels to me like we have to do that, but I've thought myself in a circle at this point.
thanks!
You can pass a custom :body_reader
option to Plug.Parsers
in order to cache the body for later use.
You'll want to not read the body before the Parser and instead cache the body to read later from your plug that wants to hash it.
Option:
:body_reader
- an optional replacement (or wrapper) for
Plug.Conn.read_body/2
to provide a function that gives access to the
raw body before it is parsed and discarded. It is in the standard
format of {Module, :function, [args]}
(MFA) and defaults to
{Plug.Conn, :read_body, []}
.
Example:
Sometimes you may want to customize how a parser reads the body from
the connection. For example, you may want to cache the body to perform
verification later, such as HTTP Signature Verification. This can be
achieved with a custom body reader that would read the body and store
it in the connection, such as:
defmodule CacheBodyReader do
def read_body(conn, opts) do
{:ok, body, conn} = Plug.Conn.read_body(conn, opts)
conn = update_in(conn.assigns[:raw_body], &[body | (&1 || [])])
{:ok, body, conn}
end
end
which could then be set as:
plug Plug.Parsers,
parsers: [:urlencoded, :json],
pass: ["text/*"],
body_reader: {CacheBodyReader, :read_body, []},
json_decoder: Jason
It was added in Plug v1.5.1.