ryz310 / my_api_client Goto Github PK

Use Bugsnag::BreadcrumbType instead.

Refs: Module: Bugsnag::Breadcrumbs — Documentation for bugsnag (6.24.2)

Some APIs show error details in response headers, such as "www-authenticate" header:

HTTP/2 401
x-xss-protection: 1; mode=block
x-line-request-id: 3bb43d42-ca94-450f-bcc4-5f912eeb1925
x-frame-options: DENY
x-content-type-options: nosniff
www-authenticate: Bearer error=“invalid_token”, error_description=“accessToken expired(2)”
server: envoy
pragma: no-cache
expires: 0
date: Wed, 19 Oct 2022 06:39:13 GMT
content-type: application/json
content-length: 104
cache-control: no-cache, no-store, max-age=0, must-revalidate

Unfortunately, my_api_client does not support header validation.

On #10 and #13, the generator was broken but I couldn't know before release.
It needs testing on the CI in the future.

The Facebook graph API occasionally returns error which having is_transient: true like as following. It has possibility to succeed by retrying. But currentry error_handling does not support Boolean. I'd like to support it.

{
  "error": {
    "message": "An unexpected error has occurred. Please retry your request later.",
    "type": "OAuthException",
    "is_transient": true,
    "code": 2,
    "fbtrace_id": "HbP+i+BdlMX"
  }
}

Following code may not work as expected.

my_api_client_stub(ExampleApiClient, :get_user, response: { id: 12_345 })
my_api_client_stub(ExampleApiClient, :set_user)

Because #my_api_client_stub creates new double instance for each time, so the method stubbing is not take over.

Many of the following mistakes have occurred: error_handling status: 400.
error_handling status_code: 400 is correct.

I think it needs notification when specified incorrect argument.

Refs #142

Bugsnag-Ruby v6.11.0 was Added new feature whic called logging breadcrumbs.
I want to support that feature in this gem.

See: https://docs.bugsnag.com/platforms/ruby/other/#logging-breadcrumbs

refs: #22

See: https://github.com/lostisland/sawyer/blob/6c080b679a/lib/sawyer/link_parsers/simple.rb

The retry_on method has sleep time and it makes slow testing speed.
I want not to wait it in the testing.

The MyApiClient::Logger does not work. It is not output to Rails.logger.

When I write following code, testing by rspec will freeze 4 minutes:

retry_on SomeError, wait: 1.minute, attempts: 3

I want to avoid sleep on testing.

my_api_client gem を使って以下でリクエストしようとしたが、データにハッシュ形式で指定すると意図したリクエストにならない。

HTTP メソッド : POST
HTTP リクエストヘッダー : application/x-www-form-urlencoded

サンプルコード

body = {
  hoge: 'hoge',
  fuga: 'fuga'
}
headers = { 'Content-Type': 'application/x-www-form-urlencoded' }
post 'hoge/fuga', body: body, headers: headers

原因は sawyer gem の以下の箇所でデータをエンコードする際に、デフォルトの sawyer だと JSON 文字列形式に変換されてしまうため。

https://github.com/lostisland/sawyer/blob/a541a68be349a7982e24b80d94fe380a37b4805c/lib/sawyer/agent.rb#L96

とりあえずワークアラウンドとして以下のようにキーバリューのクエリ形式で渡すようにしたところ意図したとおりのリクエストができた。

body = {
  hoge: 'hoge',
  fuga: 'fuga'
}.to_a.map { |a| a.join('=') }.join('&')

Currently, MyApiClient logs a request duration as seconds.

e.g. API request POST https://example.com: "Duration 0.029339311 sec"

Almost API response times are within 1 second, so the duration format should be milliseconds.

When I try to request to http://localhost:3000/v1/line/channels by below

class ChannelApiClient < ::ApplicationApiClient
  endpoint 'http://localhost:3000/v1/line'

  def get_channels
    get 'channels', headers: headers
  end

actual request is going to be

API request `GET http://localhost/v1/line/channels`: "Start"

As you can see, port numbet (3000) is removed

I think this problem comes from

because schema_and_hostname dosen't include uri.port

I'd really appreciate it if you could fix it in your free time.

In the following code, initialization parameters may change the API response. Unfortunately, current stubbing helper can not change response parameters by the initialize parameters. I would like to support the feature in the future.

class UserApiClient < ::ApplicationApiClient
   endpoint 'https://example.com'

  attr_reader :version

  def initialize(version:)
    @version = version
  end
end

I think MyApiClient needs integration testing with REAL API on the CircleCI.
It may be implemented by serverless like a Jets framework, considering maintainability and operating costs.

概要

See. https://github.com/ryz310/my_api_client/blob/v0.20.0/README.jp.md#raise-option

README の上記箇所を参照したところ例外のレスポンスをスタブ化するには #stub_api_client_all や #stub_api_client の raise オプションとともに response オプションを利用すれば良さそうでした。

ただ、例外のレスポンスとして任意の HTTP ステータスコードを返そうと思ったのですが、以下の箇所で HTTP のステータスコードが 400 で決め打ちになっていたためできませんでした。

期待する動作

以下のように、#stub_api_client_all や #stub_api_client で例外のテストをする際にレスポンスの HTTP ステータスコードも外部から注入できるようにしたいです。

stub_api_client_all(
  ExampleApiClient,
  request: {
    raise: MyApiClient::Error,
    response: { status: 404, message: 'error' }
  }
)
# or
stub_api_client_all(
  ExampleApiClient,
  request: {
    raise: MyApiClient::Error,
    status: 404,
    response: { message: 'error' },
  }
)

MyApiClient クラスを継承した以下のような SampleApiClient クラスを作成し、RSpec を実行したところエラーに遭遇しテストが落ちました。

• sample_api_client.rb

class SampleApiClient < MyApiClient
  retry_on MyApiClient::ServerError, wait: 0.1.seconds, attempts: 3
  error_handling status: 200, json: :forbid_nil, raise: SampleApiClientError
end

• sample_api_client_spec.rb

RSpec.describe SampleApiClient, type: :api_client do
  it do
    expect { subject }
      .to be_handled_as_an_error(MyApiClient::ServerError)
      .after_retry(3).times
      .when_receive(status_code: 500)
  end
end

• error

expected to be handled as MyApiClient::ServerError, but it was handled as SampleApiClientError

ソースコードに以下の定義が status code : 200 かつ body : nil という条件定義のように見えますが、実際は status code : 500 かつ body : nil という条件にもひっかかってしまっているようです。

  error_handling status: 200, json: :forbid_nil, raise: SampleApiClientError

status code : 500 かつ body : nil の場合は上記エラーハンドルにひっかかからないように( 上記エラーハンドル定義の場合は status code : 200 かつ body :nil の場合のみひっかかる )修正をお願いできますでしょうか。

An API is expected to return following result.

{
  "age_range": {
    "max": 30,
    "min": 20
  }
}

An instance which implemented with MyApiClient returns that result as Sawyer::Resource.

result = ApiClient.request
result[:max] # => 30

However, if the API returns following result, above execution will unexpected result.

{
  "age_range": {
    "min": 20
  }
}

result = ApiClient.request
result[:max] # => [:min, 20]

The cause is into the Sawyer::Resouce implementation. Sawyer::Resouce#[] calls #method_missing and searches the key from inner fields.
Ordinarily it will return nil if does not find the key.

However, :max is Enumerable module instance so it will happen that calls Enumerable#max method.

For performance.

Refs

• https://github.com/authlete/authlete-ruby-gem/blob/master/lib/authlete.rb

See also

• "I strongly discourage the use of autoload in any standard libraries" (Re: autoload will be dead)

Some API returns 404 Not Found as normal operation.
In the case, the error handler raises error is not kindness.

I hope to return nil or false in that case.

# Example.
class SomeApiClient < MyApiClient::Base
  endpoint 'https://example.com'

  error_handling status_code: 404, return: nil

  def get_resource
    get 'path/to/resource'
  end
end

api_client = SomeApiClient.new
response = api_client.get_resource # 404 Not Found
# => nil

The function of retry_on is to retry the process and return the correct value when an error is detected. But currentry, it returns always nil.

The following code may have a problem. The # rescue_with_handler provided by ActiveSupport performs retry processing, but nobody seems to return the result value.

I'd like to write spec as following.

RSpec.describe Google::People::ApiClient, type: :api_client do  
  let(:api_client) { described_class.new(access_token: 'access_token') }
  
  describe '#get_people' do
    subject(:request!) { api_client.get_people }
    
    it do
      request!
      expect(api_client)
        .to have_requested_to(:get, 'https://people.googleapis.com/v1/people/me')
        .with(body: nil, headers: headers, query: query)
    end

    describe 'error handling' do
      context 'when the API returns 200 OK' do
        let(:body) do
          { resourceName: 'people/12345678901234567890' }.to_json
        end
        
        it { expect { request! }.not_to handle_error.when_receive(status: 200, body: body) }
      end
      
      context 'when the API returns 200 OK, but response body is empty' do
        it do 
          expect { request! }
            .to handle_error(raise: Google::People::Errors::EmptyResponse)
            .when_receive(status: 200, body: nil) 
        end
      end
      
      context 'when the API returns 400 Bad Request' do
        let(:body) do
          {
            error: {
              code: 400,
              message: 'Invalid personFields mask path: "hoge".',
              status: 'INVALID_ARGUMENT',
            },
          }.to_json
        end
        
        it do 
          expect { request! }
            .to handle_error(raise: MyApiClient::ClientError)
            .when_receive(status: 400, body: body) 
        end
      end
      
      context 'when the API returns error code: 403' do
        let(:body) do
          {
            error: {
              code: 403,
              message: 'People API has not been used in project xxx before or it is disabled.',
              status: 'PERMISSION_DENIED',
            },
          }.to_json
        end
        
        it do 
          expect { request! }
            .to handle_error(raise: Google::People::Errors::ApiIsDisabled)
            .when_receive(status: 403, body: body) 
        end
      end
    end
  end
end

Dependabot encountered the following error when parsing your .dependabot/config.yml:

Automerging is not enabled for this account. You can enable it from the [account settings](https://app.dependabot.com/accounts/ryz310/settings) screen in your Dependabot dashboard.

Please update the config file to conform with Dependabot's specification using our docs and online validator.

Refs: https://jsonapi.org/

Here’s an example response from a blog that implements JSON:API:

{
  "links": {
    "self": "http://example.com/articles",
    "next": "http://example.com/articles?page[offset]=2",
    "last": "http://example.com/articles?page[offset]=10"
  },
  "data": [{
    "type": "articles",
    "id": "1",
    "attributes": {
      "title": "JSON:API paints my bikeshed!"
    },
    "relationships": {
      "author": {
        "links": {
          "self": "http://example.com/articles/1/relationships/author",
          "related": "http://example.com/articles/1/author"
        },
        "data": { "type": "people", "id": "9" }
      },
      "comments": {
        "links": {
          "self": "http://example.com/articles/1/relationships/comments",
          "related": "http://example.com/articles/1/comments"
        },
        "data": [
          { "type": "comments", "id": "5" },
          { "type": "comments", "id": "12" }
        ]
      }
    },
    "links": {
      "self": "http://example.com/articles/1"
    }
  }],
  "included": [{
    "type": "people",
    "id": "9",
    "attributes": {
      "firstName": "Dan",
      "lastName": "Gebhardt",
      "twitter": "dgeb"
    },
    "links": {
      "self": "http://example.com/people/9"
    }
  }, {
    "type": "comments",
    "id": "5",
    "attributes": {
      "body": "First!"
    },
    "relationships": {
      "author": {
        "data": { "type": "people", "id": "2" }
      }
    },
    "links": {
      "self": "http://example.com/comments/5"
    }
  }, {
    "type": "comments",
    "id": "12",
    "attributes": {
      "body": "I like XML better"
    },
    "relationships": {
      "author": {
        "data": { "type": "people", "id": "9" }
      }
    },
    "links": {
      "self": "http://example.com/comments/12"
    }
  }]
}

For example, if an error occurs such as an access token expiring, it is useful to be able to regenerate the access token before retry processing. The current .retry_on feature is only useful in limited cases, but adding a before callback would be an enhancement.

I want to use request_to matcher without with option. But I've occurred following error:

Failure/Error:
       expect { api_request }.not_to request_to(:post, 'https://example.com/api')

     NoMethodError:
       undefined method `[]' for nil:NilClass
     # /usr/local/bundle/ruby/2.7.0/gems/my_api_client-0.17.0/lib/my_api_client/rspec/matchers/request_to.rb:11:in `block (2 levels) in <top (required)>'

The stack trace shows code that:

The log GET http:/v3.3/me/friends is wrong URL. It is not considering schema and domain name.

I, [2019-06-19T15:12:33.753081 #58]  INFO -- : API request `GET http:/v3.3/me/friends`: "Start"
I, [2019-06-19T15:12:34.225469 #58]  INFO -- : API request `GET http:/v3.3/me/friends`: "Duration 0.4365701 sec"
I, [2019-06-19T15:12:34.260660 #58]  INFO -- : API request `GET http:/v3.3/me/friends`: "Success (200)"

The reason is that the Faraday instance is initialized by Sawyer::Agent#initialize but it has not yet been called at timing of Logger instance initialization.

I feel noisy 🙄