The purpose of hanami in this gem?

[pat]

Hi there 👋🏻

We're considering using openapi_first in a Rails app - particularly given the support for OpenAPI v3.1 - but I'm wondering why hanami-router is a dependency for the validation process? To be clear: I'm a fan of hanami, and use it in other projects - it's just that I don't want to bring in a tonne of new indirect dependencies for such a focused need, and I don't immediately grok the purpose of it here. If you have the time to enlighten me, that'd be much appreciated :)

[ahx]

Hi @pat. Sorry for getting back to you so late. I did not see the message :/.

This is a very good question. openapi_first uses hanami-router for find the OpenApi operation for the current request. hanami-router implements a trie algorithm which makes this super efficient.
Please don't hesitate to raise any concerns or general feedback here if you have any thoughts about this.

About OpenAPI 3.1 support: json_schemer supports OpenAPI 3.0 and 3.1 since version 2.0 and I am about to update openapi_first to use that version. So we should be able to support all nuances of 3.1

[SalvatoreT]

I'm also struggling with all the dependencies pulled in just for the one usage of hanami-router. I think it should be possible to remove the dependency.

I started on a spike but ran out of time to have it handle the 404/405 and other non-happy cases.

diff --git a/lib/openapi_first/router.rb b/lib/openapi_first/router.rb
index 6da7c44..26f6dcc 100644
--- a/lib/openapi_first/router.rb
+++ b/lib/openapi_first/router.rb
@@ -2,7 +2,6 @@
 
 require 'rack'
 require 'multi_json'
-require 'hanami/router'
 require_relative 'body_parser_middleware'
 
 module OpenapiFirst
@@ -77,23 +76,31 @@ module OpenapiFirst
     end
 
     def build_router(operations)
-      router = Hanami::Router.new.tap do |r|
-        operations.each do |operation|
-          normalized_path = operation.path.gsub('{', ':').gsub('}', '')
-          r.public_send(
-            operation.method,
-            normalized_path,
-            to: build_route(operation)
-          )
-        end
+      @routes = {}
+      operations.each do |operation|
+        normalized_path = operation.path.gsub('{', ':').gsub('}', '')
+        @routes[[operation.method.downcase.to_sym, normalized_path]] = build_route(operation)
       end
+
+      route_request_lambda = lambda { |env| route_request(env) }
+
       raise_error = @raise
       Rack::Builder.app do
-        use(BodyParserMiddleware, raise_error:)
-        run router
+        use(BodyParserMiddleware, raise_error: raise_error)
+        run route_request_lambda
       end
     end
 
+    def route_request(env)
+      req = Rack::Request.new(env)
+      key = [req.request_method.downcase.to_sym, req.path]
+      route = @routes[key]
+      return route.call(env) if route
+
+      raise_error(env) if @raise
+      @app.call(env) if @not_found == :continue
+    end
+
     def build_route(operation)
       lambda do |env|
         env[OPERATION] = operation

[ahx]

@SalvatoreT I have not tried the patch yet, but that should not work for paths with path parameters, or does it?
Having a request path like /pets/42/food should match a /pets/{id}/food.

[SalvatoreT]

Probably not! It's half baked, for sure. I'm more trying to point towards a direction than land a specific implementation.

[ahx]

Thanks for the reply @SalvatoreT. I will try to find a solution that does not need Hanami-router @pat, but using a trie algorithm like hanami-router does makes sense to me.
I will trie try to come up with a simpler (naive) version and try to measure how that impacts performance.

[ahx]

I will remove hanami-router in #197

[pat]

Thanks @ahx, greatly appreciate you going to the effort of reworking things!

[SalvatoreT]

Yes, thank you!

[ahx]

I have released 1.0 the other day and things got much better. hanami-router was removed and also Router is gone. I would love to get your feedback on this if you find the time.

[SalvatoreT]

I just started leave, so I won't be able to take a swing for a bit. 😓