Here's some more for ya.
I got SeleniumRC working in a reasonable fashion. Here's how you do it:
1) Build your test with Selenium IDE, Save it as a Python Unit test (the
IDE will write the source code for you)
2) Download selenium rc server. Put the 'selenium-server.jar' and the
attached seleniumtest.py in site-packages (for now).
2.5) Don't forget to install a recent version of Java and the selenium
Python module ('easy-install selenium').
3) In your test from step one, import the seleniumtest module and change:
unittest.main()
to:
seleniumtest.runInSeleniumRC(unittest.main)()
Caveats for now:
- I can't get Firefox or Chrome to work, the tests will only work in IE
and Opera for me. I haven't tried Safari.
- After running a test, my little module doesn't appropriately kill (at
least on windows) the seleniumRC server, so you'll have to kill that
manually. (I would appreciate any tips on this.)
I put my test in /web2py/applications/init/tests/Test_Login_Process.py
and I can run it just fine from there. This test only accesses the
webpage side of things, so it does not need to be run from a web2py
shell. However, you could very easily add controller-level testing to
it and run it from the web2py shell.
-tim
yarko wrote:
> Thanks for all this, Timbo!
>
> Looking forward to hearing more on your experience, thoughts with
> this.
>
> Yarko
>
> On Oct 19, 9:54 am, Timbo <[EMAIL PROTECTED]> wrote:
>
>> Since the start of this thread, I've only just looked into using
>> Selenium RC and how to integrate that with web2py. I haven't come up
>> with a "good" solution yet, but I'll keep you posted.
>>
>> How does selenium compare with other products? I haven't used
>> anything else, but a cursory look at the two mentioned, shows Selenium
>> as being more flexible/powerful than twill and more cross-browser than
>> TestGen4Web.
>>
>> How I use selenium now is I've downloaded selenium core and put it in
>> my static files (it's just a few webpages and lots of javascript). I
>> have a test script that runs through all possible combinations of how
>> logging into my website should behave.
>>
>> How has selenium helped my web dev? Easy question, it speeds up
>> testing by magnitudes and makes testing less boring and more
>> thorough. Also it gives you a CYA factor for when someone says
>> something doesn't work right. All you have to do is show them the
>> test. If you haven't tested a particular aspect of something, it will
>> become apparent rather quickly. The solution is to add that
>> particular regression to your test script and in no time, you'll have
>> an extensive battery of tests for your webapp.
>>
>> Web app testers such as Selenium do not only test the View...however,
>> it is through the view that they do their testing. In a true MVC
>> model, the Controller is the only thing that can be tested since it is
>> the only place where logic should exist in the process. However, most
>> developers have views that show things differently based on different
>> input from the controller. So we have logic in our views as well. It
>> goes without saying that certain behaviors in the controller produce
>> certain manifestations in the corresponding view. For example, in my
>> login application, if the user enters an incorrect password, it
>> returns with a flash-message based on how it is incorrect (illegal
>> characters, no password supplied, etc.). This flash is testable (via
>> Selenium or others) and it is testing both the view logic and the
>> controller logic. Hence, we can use Selenium to test both the
>> controller and the view. Models don't possess logic in themselves
>> (unless you want to test constraints but that's silly).
>>
>> Selenium core can only run javascript and manipulate a browser. For
>> that reason, it cannot set application state without you providing a
>> web-facing front-end to facilitate that (if you do, be wary of
>> security concerns). Selenium RC, can run as a normal Python unittest
>> and could therefore manipulate application state when run from the
>> web2py shell as Massimo pointed out. I'm going to look more into this
>> on Monday.
>>
>> @yarko:
>> The Selenium RC server is the only portion written in Java and that is
>> only necessary for tests written in languages other than HTML
>> Selenese. You could make tests via the IDE and run them purely with
>> Selenium core without having to mess with Java at all. Actually, I'd
>> like to find a way to cleanly start/stop Selenium RC Server from
>> web2py for good integrated testing. I'll let you know. But if you
>> want to get started with it, download Selenium Core and the IDE and
>> get going.
>>
>> =)
>>
>> On Oct 18, 11:15 am, morningovermidnight <[EMAIL PROTECTED]>
>> wrote:
>>
>>
>>> Thanks for pointing me to selenium (http://selenium.openqa.org/). It
>>> looks great. I've also seen some other automated web testing soultions
>>> such as twill (http://twill.idyll.org/) and TestGen4Web (http://
>>> developer.spikesource.com/wiki/index.php?title=Projects:TestGen4Web).
>>> If you don't mind, could you give me a scenario of how you test your
>>> web application with selenium ( selenium rc in particular) and how
>>> testing with selenium has helped your web dev?
>>>
>>> I ask because honestly i am still on the fence about whether or not to
>>> implement something such as selenium, or twill, or TestGen4Web. They
>>> all seem to run tests well, but -- and correct me if i'm wrong -- when
>>> thinking about the MVC architecture, it seems all of these only test
>>> the V, the view. You know, testing to see that a page displayed, that
>>> a link is present, or that a page contains some text. And while
>>> testing the view is of course important, testing the MC part of
>>> things, the application logic is just as important -- or at least a
>>> testing solution is not complete if not all components, the M, the V,
>>> and the C can be tested. Such a solution may be a combination of
>>> several different products...but it is something definitely worth
>>> looking into. Again, maybe I am only seeing Selenium on the
>>> surface...is there a way for Selenium to test application logic (e.g.
>>> setting the status of the website to "down for maintenance" then
>>> verifying that no users can access the site via login or otherwise) ?
>>>
>>> On Oct 17, 10:25 am, Timothy Farrell <[EMAIL PROTECTED]> wrote:
>>>
>>>> Massimo,
>>>>
>>>> I implied in my previous post, that I don't use web2py's built-in
>>>> testing. This brings me to a question...what is the "test" directory
>>>> for under the application root? Since the admin interface only runs
>>>> doctests, what would be a good way of running a unittest that I place in
>>>> /applications/init/tests/ ?
>>>>
>>>> This sort of setup, combined with selenium-rc, would be the ultimate in
>>>> web app testing (well almost, but it's really close).
>>>>
>>>> -tim
>>>>
>>>> Timothy Farrell wrote:
>>>>
>>>>> An excellent question. Testing in web2py is somewhat of a misnomer
>>>>> since you use web2py to create web applications but you can only truly
>>>>> test the models or controllers, not the product of all three MVC
>>>>> components. However, there isn't really a better way unless Massimo
>>>>> wants to include another external libary (and therefore dependency).
>>>>>
>>>>> What I use for webpage testing is Selenium. Selenium is a
>>>>> javascript-based scripting/testing engine. However, it has Python
>>>>> modules that you can use to write and invoke tests (which it translates
>>>>> to Javascript and runs.) It even has a cool IDE plugin for Firefox
>>>>> which makes test-making pretty easy. Take a look.
>>>>> http://www.openqa.org/selenium/
>>>>>
>>>>> -tim
>>>>>
>>>>> morningovermidnight wrote:
>>>>>
>>>>>> I've been reading earlier posts on unittests and using doctests in
>>>>>> web2py. I have tried and run some successful doctests in web2py.
>>>>>> However, it seems that the doctest is testing the value that my
>>>>>> function returns, as it should, but is there a way to write a test to
>>>>>> check an intermediate value in my function, say the value of a session
>>>>>> variable that is stored but not returned to the view?
>>>>>>
>>>>>> Also, just curious, what are some of the solutions people here in the
>>>>>> group are using for testing and for debugging web2py applications?
>>>>>>
>>>> tfarrell.vcf
>>>> < 1KViewDownload
>>>>
> >
>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups
"web2py Web Framework" group.
To post to this group, send email to web2py@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at
http://groups.google.com/group/web2py?hl=en
-~----------~----~----~----~------~----~------~--~---
''' A unittest wrapper to run Python Selenium tests in Selenium RC. '''
import os
from subprocess import Popen
import win32api
import tempfile
SELSERVER = 'selenium-server.jar'
JAR = os.path.join(os.path.dirname(__file__), SELSERVER)
PORT = 4444
def runInSeleniumRC( func, port = PORT, jar = JAR):
''' A function wrapper for unittest.main() that runs the unittest in
a Selenium RC context. Example::
if __name__ == '__main__':
seleniumtest.runInSeleniumRC(unittest.main, port=SELPORT)()
@param func: The function to be wrapped.
@param port: The port on which to run the Selenium RC server
@param jar: Where the Selenium RC jar file resides
'''
def runner(*args, **kwargs):
_p = Popen(('java.exe'
, '-jar'
, jar
, '-port'
, str(port))
,env = os.environ
,stdout = tempfile.TemporaryFile()
,stderr = tempfile.TemporaryFile()
)
res = func(*args, **kwargs)
if sys.platform == 'win32':
win32api.TerminateProcess(int(_p._handle), -1)
else:
os.kill(_p.pid)
return ret
return runner
begin:vcard
fn:Timothy Farrell
n:Farrell;Timothy
org:Statewide General Insurance Agency;IT
adr:;;4501 East 31st Street;Tulsa;OK;74135;US
email;internet:[EMAIL PROTECTED]
title:Computer Guy
tel;work:(918)492-1446
url:www.swgen.com
version:2.1
end:vcard
