I need to do an automated script that fills two text fields and clicks a button on a web page, and stores all resulting text to a string variable.
I know how to do this with Watir, but the problem is that this script will be running on a Windows server (with no physical monitor attached).
So this needs some kind of "emulated browser" without actual browser window... I have never before done anything like this, but after google search I think that Ruby gems "mechanize" or "capybara" might be able to do the trick.
But because I don't have any experience with either capybara or mehcanize, I'm asking a little help here...
Here is what I'm trying to do, written in Watir code. I would really appreciate it if someone could tell me how to do the same thing with either "mechanize" or "capybara". Or, if there is some other way to do this, all suggestion are welcomed:
require "watir"
ie = Watir::Browser.new
ie.goto "http://www.vero.fi/vere/Tarkistus/VeronumeronTarkistus.aspx/"
ie.text_field(:id, "tbNimi").set "John Smith"
ie.text_field(:id, "tbVerotunnus").set "123456789012"
ie.button(:id, "btnHae").click
info = ie.text
You could also use Celerity. It drives headless browser using Watir API.
Can Selenium Help With with Selenium Server Running and Firefox Running in headless state
I have basically wrote an article in this over here
Hope this help
I use PhantomJS for this (with the Capybara driver poltergeist). It runs a headless WebKit (Safari and Chrome's browser engine) and Capybara tells it what to do. It's the simplest-to-setup implementation of this concept that I've found.
The code would look something like:
visit "http://www.vero.fi/vere/Tarkistus/VeronumeronTarkistus.aspx/"
fill_in "tbNimi", :with => "John Smith"
fill_in "tbVerotunnus", :with => "123456789012"
click_on "btnHae"
info = page.html
Related
I'm trying to (ab)use the capybara web testing framework to automate some tasks on github that are not accessible via the github API and which require me to be logged in and click on buttons to send AJAX requests.
Since capybara/selenium is a testing framework it helpfully creates a temporary session which has no cookies in it. I'd like to either stop it from doing that, or else I'd like to know how to load my cookie store into the browser session that it creates.
All I'm trying to do is this:
#!/usr/bin/env ruby
require 'selenium-webdriver'
driver = Selenium::WebDriver.for :chrome
driver.navigate.to "https://github.com"
Or this:
#!/usr/bin/env ruby
require 'capybara'
Capybara.register_driver :selenium do |app|
Capybara::Selenium::Driver.new(app, :browser => :chrome)
end
session = Capybara::Session.new(:selenium)
session.visit "https://www.github.com"
In both cases I get the github.com landing page you'd see as a logged-out user or incognito mode in the browser. I'd like to get my logged-in landing page like I just fired up a web browser myself and navigated to that URL.
Since I have 2FA setup on github that makes automating the login process from the github landing page somewhat annoying, so I'd like to avoid automating logging into github. The tasks that I want to automate do not require re-authenticating via 2FA.
ANSWER:
For MacOSX+Ruby+Selenium this works:
#!/usr/bin/env ruby
require 'selenium-webdriver'
caps = Selenium::WebDriver::Remote::Capabilities.chrome("chromeOptions" => {"debuggerAddress" => "127.0.0.1:20480"}, detach: false)
driver = Selenium::WebDriver.for :chrome, :desired_capabilities => caps
driver.navigate.to "https://github.com"
Then fire up chrome with this:
% /Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome --user-data-dir=/Users/lamont/Library/Application\ Support/Google/Chrome --profile-directory=Default --remote-debugging-port=20480
Obviously the paths will need to be adjusted because they're OSX-centric and have my homedir in them.
There is also a bug in the selenium-webdriver gem for ruby where it inserts a 'detach' option which gets into a fight with 'debuggerAddress':
/Users/lamont/.rvm/gems/ruby-2.2.4/gems/selenium-webdriver-2.53.0/lib/selenium/webdriver/remote/response.rb:70:in `assert_ok': unknown error: cannot parse capability: chromeOptions (Selenium::WebDriver::Error::UnknownError)
from unknown error: unrecognized chrome option: detach
The lib/selenium/webdriver/chrome/bridge.rb file can be edited to take that out as a quick hack:
chrome_options['binary'] = Chrome.path if Chrome.path
chrome_options['nativeEvents'] = true if native_events
chrome_options['verbose'] = true if verbose
#chrome_options['detach'] = detach.nil? || !!detach
chrome_options['noWebsiteTestingDefaults'] = true if no_website_testing_defaults
chrome_options['prefs'] = prefs if prefs
To implement something similar in Ruby, check out this page that goes over that. Thanks to lamont for letting me know in the comments.
You can start chrome using a specific Chrome profile. I am not sure what the ruby implementation would look like, but in python it looks something like:
from selenium import webdriver
from selenium.webdriver.chrome.options import Options as ChromeOptions
options = ChromeOptions()
# more on this line here later.
options.add_experimental_option('debuggerAddress', '127.0.0.1:7878')
driver = webdriver.Chrome(chrome_options=otpions)
In order for this to work you need to do a few things.
manually start chrome from terminal/command prompt with these command line arguments
--user-data-dir=/path/to/any/custom/directory/home/user/Desktop/Chromedir --profile-directory="Profile 1" --remote-debugging-port=7878
make sure "Profile 1" is already existing in the same --user-data-dir (make sure user Profile 1 has necessary chrome://components/
to run any apps that require those components)
you can use any free port in place of 7878
verify that http://localhost:7878 is running and returns value.
This should manually launch chrome with the "Profile 1" profile, and so long as it has logged into the site in question, it will stay logged in like a normal user so long as you follow these instructions to run the tests.
I used this to write a quick netflix bot that clicks the "continue playing" button when it pops up, and it's the only way to get DRM content to play as far as I have found. But it retains the cookies for the login, and also launches chrome with whatever components the profile is set up to have.
I have tried launching chrome with specific profiles before using different methodologies, but this was the only way to really force it to work how I wanted it to.
Edit: There are methods for saving cookie info as well although I don't know how well they work. Check out this link for more info, as my solution is probably not the best solution even if it works.
The show_me_the_cookies gem provides cross-driver cookie manipulation and can let you add new cookies. The one thing to be aware of when using selenium is that you need to visit the domain before you can create cookie for it, so you'll need to do something like
visit "https://www.github.com"
create_cookie(...)
visit "https://www.github.com"
for it to work - first visit just puts the browser/driver in a state where you can create the cookie, second visit actually goes to the page with the cookies set.
I had to tweak the OP's answer (from within her question) to get this going with Ruby in 2022.
Prerequisites
Chromedriver installed and allowed to run even though it's not signed:
> brew install chromedriver
> xattr -d com.apple.quarantine /usr/local/bin/chromedriver
Chrome launched and accepting commands on a specific port:
> /Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome --user-data-dir=~/Library/Application\ Support/Google/Chrome --profile-directory=Default --remote-debugging-port=20480
This created a new profile in Chrome so I signed in to my account and got the browser set up, ready to start interacting with the (legacy EdTech) site I'm trying to automate.
Actual use
require 'selenium-webdriver'
caps = Selenium::WebDriver::Remote::Capabilities.chrome("goog:chromeOptions" => {"debuggerAddress" => "127.0.0.1:20480"})
driver = Selenium::WebDriver.for :chrome, capabilities: caps
driver.navigate.to "https://www.google.com"
I am new to Capybara. I have a question about why my Capybara doesn't work when I use it together with Selenium Webdriver.
This is my sample code:
Given(/^I am on the Youtube homepage$/) do
# visit 'http://www.youtube.com'
driver = Selenium::WebDriver.for :firefox
driver.navigate.to('http://www.youtube.com')
end
When(/^I search for "([^"]*)"$/) do |search_term|
fill_in 'search_query', :with => search_term
click_on 'search-btn'
end
Then(/^videos of large rodents are returned$/) do
expect(page).to have_content 'Making Friends with a Capybara'
end
When I run it, it just open Firefox and go to Youtube homepage. But it gets error:
Capybara::ElementNotFound: Unable to find field "search_query".
Everything works with visit 'http://www.youtube.com' command.
You're creating a driver, telling it to navigate to a page and then it's going out of scope so it's getting deleted. the visit line works because it's using the current capybara driver which stays around between test steps. Rather than creating a driver manually you should be registering a driver with Capybara and then specifying which driver to use for the specific test. See drivers
Since capybara sets up a selenium driver by default for use with firefox you can just do
Capybara.default_driver = :selenium
somewhere before running your tests to make all your tests run using selenium with firefox, or since capybara registers selenium as the default javascript driver you can tag any scenarios you want to run in firefox with #javascript as shown here
#javascript
Scenario: do something something
Given ...
I'm having a problem getting Cucumber to find fields with the HTML5 type="number". I'm not a big fan of the way they look in the browser, but I have a few fields that need the number keyboard on mobile, and this seems to be the easiest way to get it. I'm using SimpleForm to build forms and when I set :as => :text everything works, but if I set :as => :number, the fields don't get filled out. I don't get any error, the field just doesn't get filled.
To be specific, when I have a step like this:
And I fill in "application_form_age" with "52"
then this tag won't get filled in:
<input class="numeric integer required" id="application_form_age" min="0" name="application_form[age]" size="50" step="1" type="number">
but this one works:
<input class="string required" id="application_form_age" name="application_form[age]" size="50" type="text">
Also, it only happens in #javascript scenarios. In situations where #javascript isn't necessary and the scenario doesn't launch a browser, that works fine too.
Versions of things:
capybara (2.2.1)
cucumber (1.3.14)
selenium-webdriver (2.41.0)
simple_form (2.1.1)
webrat (0.7.3)
Firefox 29.0
I'm stumped. I tried yanking out a bunch of my application JS and CSS to see if something I'm doing is breaking it, but no luck with that. I'm just patching it out by forcing those fields not to have HTML5 type number in my test environment, but I don't want to live like that. Is anyone else seeing this? I haven't been able to find any references to it, which makes it seem like it's something I'm doing. But I haven't been able to figure it out.
Ok, I have found firefox has an option to disable number input field support: 'dom.forms.number'.
So if you add the following lines in your env.rb, number input gets disabled and tests work again.
Capybara.register_driver :selenium do |app|
profile = Selenium::WebDriver::Firefox::Profile.new
profile["dom.forms.number"] = false
Capybara::Selenium::Driver.new(app, :browser => :firefox, :profile => profile)
end
First off, I think you have some confusion regarding those frameworks. Cucumber is a BDD framework, which doesn't automate browsers in any way, so this question has nothing to do with it (This is why I removed it from your question title).
Looks like you are using Capybara, which is an ATDD framework. You might probably consider showing us the Capybara code you use in order diagnose your problem.
Under the hood, I assume you use Selenium WebDriver, I can confirm that Selenium works fine with <input type="number"> (Tested with Firefox 28, which is the one selenium-webdriver (2.41.0) supports to).
require 'selenium-webdriver'
driver = Selenium::WebDriver.for :firefox
DEMO_PAGE = <<-eos
data:text/html,
<input class="numeric integer required" id="application_form_age" min="0" name="application_form[age]" size="50" step="1" type="number">
eos
driver.get(DEMO_PAGE)
driver.find_element(:tag_name, 'input').send_keys('25')
So you might want to create a similar demo using Capybara to test this functionality.
If the demo works, then we need take a closer look at your application.
Othwewise, please raise a ticket for Capybara developers.
I have JavaScript in my application that submits a form when the user hits Ctrl-S or Cmd-S. I want to write an automated test for this using RSpec, Capybara, and Capybara Webkit. I don't think I can just have Capybara execute JavaScript to trigger Ctrl-S/Cmd-S because that's not normally allowed with JavaScript in Chrome as a security concern. I see with Selenium there are page.driver.browser.action.key_down/key_up methods available. Is there anything similar with Capybara Webkit? If not, how can I send Ctrl-S and Cmd-S to the browser in my test?
Edit: I also can't get this to work using the regular Selenium driver with Firefox:
describe 'edit a template and hit Ctrl-S', js: true do
render_views
it 'saves the template' do
visit my_path
page.execute_script("$('#hidden_textarea').val('Fabulous new content')")
builder = page.driver.browser.action
builder.key_down(:control).send_keys('s').key_up(:control).perform
expect(page).to have_text('Record was saved.')
expect(page).to have_text('Fabulous new content')
end
end
It looks like the builder.key_down(:control).send_keys('s').key_up(:control).perform isn't doing anything--the page loads in Firefox but just sits there. This is with Firefox 19 on OS X with selenium-webdriver 2.35.1.
Any suggestions on how to get this to work in either Firefox or Chrome, with either Selenium or Capybara Webkit?
I'm trying to do a similar thing in Chrome using Capybara and SitePrism. This actually works for me in Firefox though:
page.element.native.send_keys :command, 'a'
so I suggest trying this
builder.native.send_keys :control, 's'
I am new in using watir but I can only imagine this to be a bug:
require "watir"
Watir::Browser.default = 'safari'
b = Watir::Browser.new
b.goto()
=> nil
b.link(:title, "Start").click
Leads me to the next page as expected but on the following page no link works even they are there
b.link(:title, "Do something").exist?
=> true
When I then enter
b.link(:title, "Do something").click
nothing happens, even the href attribute of course links to the next page (to be more precise, it is the same page but different request parameters)
Identifying the link with :xpath looks similar.
The same thing works fine with "firefox" as browser.
I use Safari 5.0.1, ruby 1.8.7 and installed beside others safariwatir-0.3.8
In fact controlling safari partially works but I have no glue why it hangs on the second page. Here is an example using a scenario in the internet:
this works fine:
browser.goto("http://reiseauskunft.bahn.de/bin/query.exe/dn?S=Frankfurt(Main)Hbf&Z=Nrnberg%20Hbf&date=%2B30&time=1500&start=1&")
=> nil
this also works fine - the link does exist:
browser.link(:title, /Mit Tickets zum Normalpreis/).exist?
=> true
this does not work - the browser window stays unchanged:
browser.link(:title, /Mit Tickets zum Normalpreis/).click
=> nil
I would like to try fire_event or fireEvent but irb says invalid method. I guess the hints I found about firing an event are outdated?
As far as I know, the only way to drive Safari is to use safariwatir gem (works only on Mac OS X). Try this (from SafariWatir wiki page at Watir wiki):
require 'rubygems'
require 'safariwatir'
browser = Watir::Safari.new
browser.goto("http://google.com")