Dojo4 Blog | All Posts

Unpublished APIs or, How I Learned to Stop Worrying and Love the Bomb.

Sometimes you have some boring repetitive online task or you need to automate some or all of a workflow on a website. Scripting these interactions can prevent us from wasting tons of man hours on mundanely clicking through the same pages on a site.

A lot of us will reach for Mechanize right off the bat. Then you find that the UI for the site you're trying to automate a task for is JavaScript heavy or AJAXy. Now what do you use?

Phantomjs to the rescue. Phantomjs is a fast, easily scriptable, headless WebKit browser. When the job fails for whatever reason, you can snap a screenshot to review. For us this meant emailing it to an admin since the jobs are running on servers and not locally. We are also using Phantomjs.rb to handle the path and output from Phantomjs.

Here's a bit of code from what we are doing.

class InvitationAcceptor
  attr_accessor :invitation, :tmpdir

  def initialize(invitation)
    self.invitation = invitation
  end

  def accept!
    fail_image_path = false

    Dir.mktmpdir(invitation.id.to_s) do |tmpdir|
      self.tmpdir = tmpdir

      Phantomjs.inline(js) do |line|
        puts line
        fail_image_path = line.gsub('FAIL:', '').chomp if line.starts_with?('FAIL')
      end

      if fail_image_path
        Mailer.invitaion_failed(invitation, fail_image_path).deliver
      else
        invitation.invitation_accepted!
      end
    end
  end

  private

    def js
      fail_image = "#{tmpdir}/failed.png"

      <<-JS
        // https://github.com/ariya/phantomjs/blob/master/examples/waitfor.js
        function waitFor(testFx, onReady, timeOutMillis) {
          var maxtimeOutMillis = timeOutMillis ? timeOutMillis : 10000,
            start = new Date().getTime(),
            condition = false
          var interval = setInterval(function() {
            if ((new Date().getTime() - start < maxtimeOutMillis) && !condition) {
              condition = (typeof(testFx) === "string" ? eval(testFx) : testFx());
            } else {
              if (! condition) {
                fail()
              } else {
                console.log("'waitFor()' waited " + (new Date().getTime() - start) + "ms.");
                typeof(onReady) === "string" ? eval(onReady) : onReady();
                clearInterval(interval);
              }
            }
          }, 250);
        };

        var page = require('webpage').create(),
          testindex = 0,
          loadInProgress = false,
          tries = 0;

        // this is important for our app because the site we're automating tries to block automation.
        page.settings.userAgent = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_9_2) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/34.0.1847.116 Safari/537.36'

        var fail = function() {
          // send the fail message and image path to the Ruby script
          console.log("FAIL:#{fail_image}");
          // save the screenshot to the file system so it can be emailed
          page.render("#{fail_image}");
          phantom.exit()
        }

        page.onConsoleMessage = function(msg) {
          console.log(msg);
        };

        page.onLoadStarted = function() {
          loadInProgress = true;
          console.log("load started");
        };

        page.onLoadFinished = function(status) {
          loadInProgress = false;
          console.log("load finished");
        };

        var steps = [
          function() {
            console.log("Load Login Page");

            page.open("#{invitation.invitation_received.subject}")
          },
          function() {
            console.log("Verify the invitation is still valid");

            success = page.evaluate(function() {
              if (null == document.getElementById('hasSomeElement')) {
                return false;
              }

              return true;
            })

            if (! success) {
              fail();
            }

          },
          function() {
            console.log("Enter Credentials");

            page.evaluate(function() {
              document.getElementById('hasSomeElement').checked;
              document.getElementById('hasSomeElement').checked;
              document.getElementById('email').value = '#{invitation.email}';
              document.getElementById('password').value = '#{invitation.password}';
              document.getElementById('login').click();
            });
          },
          function() {
            console.log("Accepting the terms");

            page.evaluate(function() {
              document.getElementById('accept').click()
            });
          },
          function() {
            console.log("Waiting for the JS craziness to complete");

            waitFor(function() {
              return page.evaluate(function() {
                return document.getElementsByClassName('someExpectedContent').length != 0
              });
            }, function() {
              phantom.exit();
            });
          }
        ]

        interval = setInterval(function() {
          if (!loadInProgress && typeof steps[testindex] == "function") {
            steps[testindex]();
            // useful for debugging
            page.render("#{tmpdir}/step" + (testindex + 1) + ".png");
            testindex++;
          }
        }, 250);
      JS
    end

end

You can also inject scripts into the page, with injectJs or includeJs but I prefer to stick to what is already there or just use the basics to avoid any clashes with what may be in the page already.