lefthandedgoat / canopy

f# web automation and testing library, built on top of Selenium (friendly to c# also)

canopy

Documentation available at http://lefthandedgoat.github.io/canopy/.

Getting started kit available at https://github.com/lefthandedgoat/canopyStarterKit/.

Real world examples available at https://github.com/lefthandedgoat/turtletestAutomation

Some C# examples available at https://github.com/lefthandedgoat/canopy/blob/master/tests/csharptests/Program.cs

canopy's Issues

double click

Is there a way to double click an element?

help user find misspelled elements

If a css selector fails, it would be nice of the framework helped the developer find what the problem is. for example:

if someone defines an element with the following misspelling "grwlInfo", and in the test they say

".growlInfo" == "some message"

it would be good if canopy could say:

unable to find element growlinfo, did you mean grwlinfo?

better discoverability of timeouts

Got a failure stating that an element couldn't be found in the allotted time. Would be nice if the failure would tell me that I have the ability to increase the time outs (compareTimeout, elementTimeout, pageTimeout).

Example:
"Element not found in the allotted time. If you want to increase the time, put elementTimeout <- {your new time} at the beginning of your test file, k thnks bye"

Make it work in VS 2010

Add method to highlight an element

failure if not visible, but in dom

If a failure in click, type or etc occurs, canopy should tell me if it failed because it couldn't find the element or it failed because the element was in the dom but not visible (these kind of hints would help in diagnosing problems)

elementWithText and elementsWithText

Need a way to find an element with a given innerHtml/text:

elementWithText "cssSelector" "regex"

I would say the the second parameter should be a regex string

Run only wips

new context code will run only the wips in a given context, but if you have more than one context it will run the manys/tests for other contexts too.

Ability to send keydown

Need some way to send key down

quit needs to close all browser

the quit() method needs to close all browser if there are multiple browsers open

Inconsistency in finding elements on page when elements fade in slowly with jquery

element is visible on page but is not clickable immediately it looks like.

Adding random thread.sleeps solved the issue, but maybe reliability should be built into the click method to try multiple time if an element isn't immediately found.

Here is a link to a video that I recorded showing the problem.

If the video resolution is too high using ctrl+- to make the video fit.

http://dl.dropbox.com/u/10614633/canopybug.mp4

add ability to test for exception

Unexpected suggestion for url validation

When running UI tests, I got a warning stating that I didn't do url validation when all of my helper methods validate the url, Video attached:

http://dl.dropbox.com/u/10614633/canopybug3.mp4

In a list of DOM elements, I'd like to select just one

jQuery provides the pseudo-selectors ":first" & ":last".

Is there a way to grab just one? I'd like to arbitrarily grab the nth child element.

need a sleep method

Would be nice to have a sleep method that understood seconds. So I don't have to enter System.Threading.Thread.Sleep(milliseconds).

The UI automation goes really fast, so when I'm authoring a test, there are times where I want things to happen more slowly, thoughts?

failed tests run first

UI automation test take a long time to run, would be good if canopy kept track of recent failures so it can run those tests first (this is going under the assumption that all tests are independent...so something they opt into)

Spectator Mode

It would be helpful to have a runtime mode for a script that ran at WIP speeds but used some form of on screen cursor to highlight the element that the script is currently inspecting or changing. For instance, a colored semi-transparent overlay shown over the element currently being read, with one color reserved to display a read and a different color to show that it's modifying the element.

No explicity context probably throws exception

speculation for research tomorrow. context starts null so when you go to print it you probably get null ref.

add pinRight to move the browser out of the way!

do it

Element is no longer attached to the DOM error

got latest, but now the test fails for a different reason, link to video: http://dl.dropbox.com/u/10614633/canopybug2.mp4

Ctrl, + if the video is too small

need a Count assertion on a collection of elements

I need the ability to say that a css selector "#requestedGames a" has a certain number of elements (needs to poll simliar to how *= and *!= polls)

generic assertion: let is expected actual =

let is expected actual =
if expected <> actual then System.Console.WriteLine("FAIL") else System.Console.WriteLine("PASS")
()

5 |> is 3

is true false

is true true

misleading assertion message when verifying inner html for an element that fades in

The polling mechanism fails to see the right message for an element that fades in (it looks like it's capturing the last innerhtml of the element as opposed to the last one that wasn't blank) '

Video of bug: http://dl.dropbox.com/u/10614633/canopybug4.mp4

Add feature: Warnings when you do two actions in a row, but didnt do a url check

example:

click "#page1Element"
click "#page2Element"

between is a page transition and the implicit wait of the 2nd click will wait 5 seconds, but maybe the user should do an explicit wait via on "newurl".

log all actions in a collection then do some simple pattern matching to detect potential issues and print as warnings at the end of the test run.

canopy before hook inside of runner module as opposed to canopy module?

Just thought it would be in the canopy module. I tried using intellesense and typed

canopy.

and didn't see the before hook (turns out it was in the runner module):

https://github.com/amirrajan/Oak/blob/master/Sample%20Apps/BorrowedGames/BorrowedGames.UI.Tests/Program.fs#L11

need a way to reliably determine if an element is displayed and notDisplayed

displayed "somecss" would poll until the element shows up

notDisplayed "somecss" would poll until the element goes away

elementsWithText polls?

does't look like elements with text polls, see video:
http://dl.dropbox.com/u/10614633/CanopyBug5.mp4

I've checked in the commit to borrowed games if you want to run it:
amirrajan/Oak@f8a886e

ability to tile windows

there are times where (with browser switching), I want to be able to see window side by side, this can be done through javascript and would be nice if there was a method to tile a set of browser windows. Here are links to manipulate windows via javascript:

get current resolution of monitor:
http://www.w3schools.com/jsref/prop_screen_width.asp

move window:
http://www.w3schools.com/jsref/met_win_moveto.asp

resize window:
http://www.w3schools.com/jsref/met_win_resizeto.asp

need to determine which browser instance failed

with multiple browsers (and when a failure occurs), it's hard to tell which browser had the failure. some visual indicator or something in the console stating which browser is currently selected would be good.

Can't change value in a <select> element

Need to be able to specify a value in a dropdown like
"#ID" << "Value"

If "Value" isn't a valid choice, it should fail.

app data canopy directory must exist or error

Create it for them if it doesn't exist!

Get a value from the DOM

Can we get a value from the DOM? In jQuery, I'd go
var x = $('#hiddenField').val();

Can we do this with canopy? If so, how?

Impliment dragging

This code worked on

http://scrumy.com/silenter39delayed

let (>>) a b =
let a1 = element a
let b1 = element b
(new Actions(browser)).DragAndDrop(a1, b1).Perform();

start firefox

url "http://scrumy.com/silenter39delayed"

click ".plus-button a img"

"#task_title" << "Demo"
click "#task_editor_buttons .save_button"
".handle" >> ".inprogress"

xtest to comment out tests

there are times where I want to comment out a test. would be nice if there was a method called xtest so that I can turn this:

test(fun _ -> 
    describe "registering a user"
    reset()
    registerUser "[email protected]")

into

xtest(fun _ -> 
    describe "registering a user"
    reset()
    registerUser "[email protected]")

as opposed to having to do

//test(fun _ -> 
//    describe "registering a user"
//    reset()
//    registerUser "[email protected]")

add failfast

if one test fails dont run anymore tests in that script

Context/Spec with F#

Found a need for running a shared setup...trying to think of ways to run a shared setup between tests...here is one idea:

context(fun _ ->
    describe "returning a borrowed game"

    before(fun _ ->
        logOff()
        requestGame()
        logOff()
    )

    test(fun _ ->
        describe "borrower returns game"
        loginAs "user2"
        is (elementsWithText "#wantedGames .brand" "Borrowed").Length 1
        click deleteGameLink
        is (elementsWithText "#wantedGames .brand" "Borrowed").Length 0
    )

    test(fun _ ->
        describe "lender marks game as returned"
        loginAs "user1"
        is (numberOfRequestedGames()) 1
        click gameReturnedLink
        is (numberOfRequestedGames()) 0
    )
)

global wait before performing actions such as click or write

Would be nice if there was a function that could be set that would be initially consulted before allow a click or write to continue.

For example:

waitFor <- loadingDivToDisappear()

click "#save" //this would wait until loadingDivToDisappear() returns

click take in more than just a css selector?

I started using elementsWithText and found myself wanting to pass the output into a click method.

It would be nice if I could do:

click (elementsWithText "#wantedGames .brand" "Borrowed").Head

or

click (elementWithText "#wantedGames .brand" "Borrowed")

(I may lean towards wanting this feature for write, read, etc)....thoughts?

Spectator Mode Seek Indicator

The spectator mode is great when the script actually finds what it's looking for but it's common for a script to be stuck trying to find something that is actually not on the page. It would be helpful to show a blinking green light (analogous to a hd seek indicator light) with the locator (xpath/css/etc) of the thing it's looking for.

change element function to test for visiblity, add domElement

change element function to test for visiblity, add domElement that returns anything w/o doing a vislibity test.

Tiny typo in BP message

In the message that reads "Element not found in the allotted time ..." where it says "elementTimeout <- 10" should read "elementTimeout <- 10.0"

... I know ... picking nits here.

ability to assert that an element is visible/not visible on the page

Need some way to assert that an element exists on that page, or doesn't exist. It should poll for the existence of the element (like write does).

Maybe something along the lines of:

isThere "#requestedGames div"

and

isNotThere "#requestGames div"

Add a 'contains' or 'matches' assert operation

We can currently assert that an element is exactly equal to a value with "selector == value". Can we have one that says that the element has the value somewhere inside it?

If it could honor Regex, that would be fantastic! Like ...
selector .= /regex/;
or
selector.match(/regex/)

failure of *= should print out what was found

got tripped up on using *=. I forgot to include something in the CSS selector that I used which caused the failure. It would be good if the failure message contained all the innerhtml that WAS found (it would have shown me that I had the CSS selector partially wrong)

Slower execution speed when running wip tests

UI automation goes really fast, maybe they go a tiny bit slower if they are a wip test?

Highlight element that is being read/written to during wip mode

Aaron had a great suggestion.

When running a test in wip mode, highlight the element that is being interacted with.

I am thinking set the current element to an orange color, then when done set it to yellow to show that it was interacted with at some point in the test.

suggestion to assert that I was on about:blank misleading?

Got this output:
Passed
Suggestion: as a best practice you should check that you are on a page before
accessing an element on it
you were on about:blank
then on http://localhost:3000/Account/LogOn?ReturnUrl=%2f
Suggestion: as a best practice you should check that you are on a page before
accessing an element on it
you were on http://localhost:3000/Account/LogOn?ReturnUrl=%2f
then on http://localhost:3000/Account/Register

for the following test:
url "http://localhost:3000"
click "a[href='/Account/Register']"
write "#Email" email
write "#Password" "Password"
write "#PasswordConfirmation" "Password"
click "input[value='register']"
url "http://localhost:3000/account/logoff"

Ability to verify a message in an modal alert box (and dismiss it).

Ability to verify a message in an modal alert box (and dismiss it).

maybe something like the following commands:
alert() == "Your item has been saved"

clickOk()

clickCancel()

ability to clear a textbox

Need a way to clear what's in a text box. The write method appends to what's already in the text box. Do you think it'd be better for write to just clear what's there and write the new value?

compiler warning on quit() method

I get the following warning on the quit() method:
C:\Development\Oak\Sample Apps\BorrowedGames\BorrowedGames.UI.Tests\Program.fs(48,1): warning FS0020: This expression sh
ould have type 'unit', but has type 'bool'. Use 'ignore' to discard the result of the expression, or 'let' to bind the r
esult to a name. [C:\Development\Oak\Sample Apps\BorrowedGames\BorrowedGames.UI.Tests\BorrowedGames.UI.Tests.fsproj]