I am using the google-maps gem.
I am trying to mock/stub api requests unsuccessfully.
module GoogleMap
class Route
attr_accessor :start_location, :end_location
def initialize(start_location, end_location)
#start_location = start_location
#end_location = end_location
end
def driving_duration_in_seconds
route.duration.value
end
def driving_distance_in_meters
route.distance.value
end
def driving_distance_hash
return unless start_location && end_location
{ distance_in_meters: driving_distance_in_meters, duration_in_seconds: driving_duration_in_seconds }
end
private
def coordinates_as_strings(location)
"#{location.latitude},#{location.longitude}"
end
def route
#route ||= Google::Maps.route(coordinates_as_strings(start_location), coordinates_as_strings(end_location))
end
end
end
I need to stub:
WebMock::NetConnectNotAllowedError:
Real HTTP connections are disabled. Unregistered request: GET https://maps.googleapis.com/maps/api/directions/json?destination=37.19687112,116.43791248&key=<apikey>&language=en&origin=2.15362819,-81.63712649 with headers {'Accept'=>'*/*', 'Date'=>'Sat, 12 Feb 2022 21:35:55 GMT', 'User-Agent'=>'HTTPClient/1.0 (2.8.3, ruby 2.7.2 (2020-10-01))'}
You can stub this request with the following snippet:
stub_request(:get, "https://maps.googleapis.com/maps/api/directions/json?destination=37.19687112,116.43791248&key=<apikey>&language=en&origin=2.15362819,-81.63712649").
with(
headers: {
'Accept'=>'*/*',
'Date'=>'Sat, 12 Feb 2022 21:35:55 GMT',
'User-Agent'=>'HTTPClient/1.0 (2.8.3, ruby 2.7.2 (2020-10-01))'
}).
to_return(status: 200, body: "", headers: {})
If I try a most basic stub I get an error:
stub_request(:any, /maps.googleapis.com/).
to_return(status: 200, body: '', headers: {})
Google::Maps::InvalidResponseException:
unknown error: 783: unexpected token at ''
# .../gems/ruby-2.7.2/gems/google-maps-3.0.7/lib/google_maps/api.rb:64:in `rescue in response'
# .../gems/ruby-2.7.2/gems/google-maps-3.0.7/lib/google_maps/api.rb:60:in `response'
# .../.rvm/gems/ruby-2.7.2/gems/google-maps-3.0.7/lib/google_maps/api.rb:27:in `query'
I think it is erroring out because I am not passing a key in. But I don't see why I should have to pass in a valid api key into a webmock.
I also would not have my route defined by anything. And in order to test that route can return route.distance.value etc, I would need to mock with something.
For other tests I was successful in mocking instances, but to test this lib that it actually works, I feel like mocking instance methods and not that an api was actually called is a waste of a test. Maybe this is just a waste of time, and I should assume it works because I am using a gem.
But I was expecting something like this:
RSpec.describe GoogleMap do
let(:start_location) { create(:location) }
let(:end_location) { create(:location) }
context 'GoogleMaps::Route.new(start_location, end_location)' do
let(:subject) { GoogleMap::Route.new(start_location, end_location) }
# I have not been successful in stubbing this with the correct regex
# stub_request(:get, "https://maps.googleapis.com/maps/api/directions/json?destination=<lat>,<long>key=<key>&language=en&origin=<lat>,<long>").
# with(
# headers: {
# 'Accept'=>'*/*',
# 'Date'=>'Thu, 10 Feb 2022 21:09:02 GMT',
# 'User-Agent'=>'HTTPClient/1.0 (2.8.3, ruby 2.7.2 (2020-10-01))'
# }).
# to_return(status: 200, body: "", headers: {})
# stub_request(:get, %r{https:\/\/maps\.googleapis\.com\/maps\/api\/directions\/json\?destination=.+,.+&key=.+&language=en&origin=.+,.+}).
# stub_request(:any, /maps.googleapis.com/).
# to_return(status: 200, body: '', headers: {})
xit 'gives driving distance in seconds'
xit 'gives driving duration in meters'
end
end
Your WebMock is working fine. Google::Maps::InvalidResponseException is raised after WebMock has replaced the network call. At the point that exception is raised, the Google Maps API client is trying to parse what the network call returned, which is ''.
It's expecting some valid JSON to be returned. If you have your mock return {} is should get past that line. It may well stumble on some other exception later though, as the gem expects a certain schema.
You can dig that out and add in a valid response if you wanted to continue down this path. However, I'd recommend not mocking the network request as that's an implementation detail of a third party piece of code which could change at any time - making your test fail. Instead, I would mock out Google::Maps.route to return what you need it to.
Related
I created a simple ruby file (not Rails) and I am trying to test (using Rspec) a method where I am calling an API. In the test I am trying to mock the call via WebMock but it keeps giving me this error:
Requests::FilesManager#display fetches the files from the API
Failure/Error: Requests::FilesManager.new.display
ArgumentError:
wrong number of arguments (given 2, expected 1)
The files are:
#run.rb
module Requests
require "httparty"
require 'json'
class FilesManager
include HTTParty
def initialize
end
def display
response = HTTParty.get('https://api.publicapis.org/entries', format: :json)
parsed_response = JSON.parse(response.body)
puts "The secret message was: #{parsed_response["message"]}"
end
end
end
and the spec file:
require 'spec_helper'
require_relative '../run'
RSpec.describe Requests::FilesManager do
describe "#display" do
it 'fetches the files from the API' do
stub_request(:get, "https://api.publicapis.org/entries").
to_return(status: 200, body: "", headers: {})
Requests::FilesManager.new.display
end
end
end
EDIT:
So the error seems to come from the line:
JSON.parse(response.body)
If I comment it out it disappears. The problem then is that the output of the call is not a json (even with the format: :json when calling the HTTParty). I tried other solutions but nothing seems to work in making the response json. It is just a string.
Change
response = HTTParty.get('https://api.publicapis.org/entries', format: :json)
to
response = HTTParty.get('https://api.publicapis.org/entries').
I think you don't need the format: :json more so when you explicitly format the response to JSON anyway.
You need to return a json object in the body parameter of the stubbed response:
E.g: For an empty response:
stub_request(:get, "https://api.publicapis.org/entries").
to_return(status: 200, body: "".to_json, headers: {})
OR For a valid response: (Note: You may have to require json to convert a hash to json)
require 'json'
...
stub_request(:get, "https://api.publicapis.org/entries").
to_return(status: 200, body: { entries: { '0': { message: "Hello World" } } }.to_json, headers: {})
Solved!
It seems there was an error because the json gem version that HTTParty uses is too old.
Moved on to RestClient gem for the RESTful API calls. It had another conflict in the mime gem versioning.
Finally moved to Faraday and that solved my problems:
JSON.parse(response.body, :quirks_mode => true)
tl;dr Had the same issue and ended up having to upgrade webmock.
Long form:
Webmock inserts middleware into your calls, so when HTTParty makes the calls they end up going through the Webmock interfaces first.
You can verify this by trying the call standalone (withouth all the rspec config):
bundle console
irb> require "httparty"
=> true
irb> httparty.get("https://google.com")
If that standalone call succeeds, the issue is somewhere within Webmock itself.
For me, somewhere along the line of calls through Webmock was an outdated interface that was incompatible and throwing the Wrong Number of Arguments error. And this was also crashing my debugger (RubyMine).
Upgrading Webmock solved this issue (because they had fixed it in newer versions).
I have some codebase like this, and I wanna use rspec test favicon_href, but as you like, the favicon_href will call the page function, I know I can mock page function, but for this stage I wanna mock the HTTP request from the given url, so I use WebMock gem's syntax to stub HTTP request, but it seems WebMock is not compatibility with Mechanize, it always show the error in the below despite I relleay have done the stub, anyone know how can solve it or any gem can stub HTTP request on Mechanize?
Code
def favicon_href
#favicon_href ||=
begin
page.at(FAVICON_DOM).attributes['href'].value # finding <link> elements
rescue Exception
'/favicon.ico' # there are some situation the favicon's not <link>'
end
end
def page
#page ||= mechanize.get(url)
end
def mechanize
#mechanize ||= Mechanize.new
end
Error
Failure/Error: #page ||= mechanize.get(valid_url(url))
WebMock::NetConnectNotAllowedError:
Real HTTP connections are disabled. Unregistered request: GET https://tsaohucn.wordpress.com/ with headers {'Accept'=>'*/*', 'Accept-Charset'=>'ISO-8859-1,utf-8;q=0.7,*;q=0.7', 'Accept-Encoding'=>'gzip,deflate,identity', 'Accept-Language'=>'en-us,en;q=0.5', 'Connection'=>'keep-alive', 'Host'=>'tsaohucn.wordpress.com', 'Keep-Alive'=>'300', 'User-Agent'=>'Mechanize/2.7.5 Ruby/2.3.1p112 (http://github.com/sparklemotion/mechanize/)'}
You can stub this request with the following snippet:
stub_request(:get, "https://tsaohucn.wordpress.com/").
with(headers: {'Accept'=>'*/*', 'Accept-Charset'=>'ISO-8859-1,utf-8;q=0.7,*;q=0.7', 'Accept-Encoding'=>'gzip,deflate,identity', 'Accept-Language'=>'en-us,en;q=0.5', 'Connection'=>'keep-alive', 'Host'=>'tsaohucn.wordpress.com', 'Keep-Alive'=>'300', 'User-Agent'=>'Mechanize/2.7.5 Ruby/2.3.1p112 (http://github.com/sparklemotion/mechanize/)'}).
to_return(status: 200, body: "", headers: {})
registered request stubs:
stub_request(:get, "https://tsaohucn.wordpress.com/").
with(headers: {'Accept'=>'*/*', 'User-Agent'=>'Ruby'})
stub_request(:any, "http://api.stripe.com/")
stub_request(:any, "/api.stripe.com/")
============================================================
There exists an incompatibility between WebMock and net-http-persistent.
See
https://github.com/bblimke/webmock#connecting-on-nethttpstart
Add
WebMock.allow_net_connect!(:net_http_connect_on_start => true)
to your test set up.
I'm starting a new project, my first with Rails 5.1.0. I have a pb with my first request spec.
describe 'Users', type: :request do
it 'are created from external data' do
json_string = File.read('path/to/test_data/user_data.json')
params = { user: JSON.parse(json_string) }
headers = { "CONTENT_TYPE" => "application/json" }
expect do
post '/api/v1/users', params.to_s, headers
end.to change {
User.count
}.by(1)
expect(response.status).to eq 200
end
end
this spec return the error ArgumentError: wrong number of arguments (given 3, expected 1). The official documentation don't say much.
If I take out the .to_s, and send a hash, like this:
post '/api/v1/users', params, headers
I got another error:
ArgumentError: unknown keyword: user
Any thought?
I think they changed the syntax recently. Now it should use keyword args. So, something like this:
post '/api/v1/users', params: params, headers: headers
Here's a little addendum to Sergio's answer. If you are upgrading from Rails 4 to Rails 5, have lots of tests, and aren't too keen on changing them all – at least not until you've finished upgrading – I've found a way to make them work with the old method signature.
In my spec_helper I added
module FixLegacyTestRequests
def get(path, par = {}, hdr = {})
process(:get, path, params: par, headers: hdr)
end
def post(path, par = {}, hdr = {})
process(:post, path, params: par, headers: hdr)
end
def put(path, par = {}, hdr = {})
process(:put, path, params: par, headers: hdr)
end
def delete(path, par = {}, hdr = {})
process(:delete, path, params: par, headers: hdr)
end
end
and then I added this configuration for each test:
RSpec.configure do |config|
config.before :each do |example|
extend(FixLegacyTestRequests) # to be removed at some point!
end
end
My tests went back to working, and I think it should be safe because it's only applied to the currently running test and shouldn't pollute any gem's code such as with a monkey patch.
config.before(:each) do
stub_request(:post, "https://api.3rdpartysmsprovider.com/send.php?body=This%20is%20a%20test%20message&destination=60123456789&dlr='1'&output=json&password=0000000&reference=#{#text.sms_uid}&sender=silver&username=0000000").
to_return(:status => 200, :body => "01", :headers => {})
end
I am currently writing specs for a service class that sends an SMS and creates a log of it in our database. I'm trying to stub this request, however #text.sms_uid is a SecureRandom.urlsafe_base64 random code. Also I'm stubbing in config.before(:each).
Because of that, I can't specify the sms_uid in stub_request as the random sms_uid is generated after the stub is called. This causes the test to fail every time. Is there a way I can stub the request after it generates the code (in other words, after it goes through the specific method) or is there a way to stub all requests going through the domain "https://api.silverstreet.com"?
I see two options:
Stub SecureRandom.urlsafe_base64 to return a known string and use that known string when you stub_request:
config.before(:each) do
known_string = "known-string"
allow(SecureRandom).to receive(:known_string) { known_string }
stub_request(:post, "https://api.3rdpartysmsprovider.com/send.php?body=This%20is%20a%20test%20message&destination=60123456789&dlr='1'&output=json&password=0000000&reference=#{known_string}&sender=silver&username=0000000").
to_return(status: 200, body: "01", headers: {})
end
If SecureRandom.urlsafe_base64 is used in other places in your application, you'll need to stub it only in the specs where this request is generated.
Yes, you can stub any POST to that hostname
stub_request(:post, "api.3rdpartysmsprovider.com").
to_return(status: 200, body: "01", headers: {})
or even any request of any kind to that hostname
stub_request(:any, "api.3rdpartysmsprovider.com").
to_return(status: 200, body: "01", headers: {})
and webmock has a very large number of other ways to match requests.
I have a spec file with an expectation that a controller action will return success.
The POST api/v1/users/:id/features/block action in the controller calls two HTTP calls on an external API, the only difference being in the body.
I've put the two requests and responses in the same VCR cassette, but when the cassette is being used, only the first request ever gets compared against and fails when it should be matching the second, causing the tests to fail.
What I'm looking for is a way of having the multiple requests match so the controller action completes and returns successfully.
The error I'm getting is at the end.
describe "POST /api/v1/users/:id/features/block" do
before(:each) do
#user = FactoryGirl.create(:user)
post :block, user_id: #user.id, block: "0"
end
it "should return 200 OK" do
expect(response).to be_success
end
end
Simplified versions of my VCR configuration and RSpec configuration follow:
VCR.configure do |c|
c.hook_into :webmock
c.default_cassette_options = {
match_requests_on: [:method, :uri, :body_regex]
}
c.register_request_matcher :body_regex do |request_1, request_2|
# Match body against regex if cassette body is "--ruby_regex /regexhere/"
if request_2.body[/^--ruby_regex\s*\//]
regex = request_2.body.gsub(/^--ruby_regex\s*\//, '').gsub(/\/$/, '')
request_1.body[/#{regex}/] ? true : false
else
true # No regex defined, continue processing
end
end
end
RSpec.configure do |c|
c.around(:each) do |example|
options = example.metadata[:vcr] || {}
name = example.metadata[:full_description].split(/\s+/, 2).join("/").underscore.gsub(/[^\w\/]+/, "_")
VCR.use_cassette(name, options, &example)
end
end
end
A summarized version of the cassette being used in this comparison that I'm having trouble with is:
---
http_interactions:
- request:
method: post
uri: https://upstream/api
body:
string: --ruby_regex /query1.+block/
response:
status:
code: 200
body:
string: { "response": "SUCCESS" }
- request:
method: post
uri: https://upstream/api
body:
string: --ruby_regex /query2.+block/
response:
status:
code: 200
body:
string: { "response": "SUCCESS" }
recorded_at: Fri, 05 Sep 2014 08:26:12 GMT
recorded_with: VCR 2.8.0
Error during tests:
An HTTP request has been made that VCR does not know how to handle
...
VCR is using the current cassette: (Correct cassette file path)
...
Under the current configuration VCR can not find a suitable HTTP interaction to replay and is prevented from recording new requests.
I don't want to record new requests because then the second one overwrites the first instead of adding the second request to the end of the cassette.