Selenium Pagemodel implementation for Python.
This project helps Test Engineers by abstracting out the implementation – in this case Selenium – from the actual set of Actions necessary to perform a single Test Scenario. The Selenium-PageModel library does this by defining a set of well defined PageModel constructs that Test Engineers can extend/implement to describe a Website. These PageModel constructs includes the following:
Using these constructs, you can describe a WebPage as having the following web-elements:
- A Header Widget containing:
- A Home Link
- A Login Link
- A Register Link
- A form Widget containing:
- A userName TextField
- A password TextField
- A submitButton Button
- A potential Error message TextElement (in the event of a login failure)
Once a PageModel is defined, a login test for an imaginary website may look like this:
homePage.open(); loginPage.waitForPageLoad().validate(); loginPage.userName.type("myuser"); loginPage.password.type("mypassword"); loginPage.submitButton.click(); homePage.waitForPageLoad().isLoggedIn();
Please see the project’s Homepage for more information: Selenium-PageModel
The library can be installed via:
pip install seleniumpm
Or if you want to install from src:
pip install git+https://github.com/gradeawarrior/python-seleniumpm.git
Here is the ever so popular Google example using seleniumpm:
from selenium import webdriver from seleniumpm.examples.google_page import GooglePage """ Setup for Remote execution against a local standalone-selenium-server and using the PhantomJS driver. This can be changed of course to using the driver of your choice (e.g. Chrome or Firefox) """ driver = webdriver.Remote(command_executor="http://localhost:4444/wd/hub", desired_capabilities=webdriver.DesiredCapabilities.PHANTOMJS) # Instantiate Google Page google = GooglePage(driver, url="https://www.google.com") # Open + wait for page load + validate Google google.open().wait_for_page_load().validate() # Print the page title print google.get_title() # Search for 'Cheese!' search_str = "Cheese!" google.search_field.type(search_str) google.search_field.submit() # Ensure that the page is refreshed from your search print google.wait_for_title(search_str).get_title()
Creating your PageObject’s
The GooglePage used in the above example looks like the following:
from selenium.webdriver.common.by import By from seleniumpm.webpage import Webpage from seleniumpm.webelements.textfield import TextField from seleniumpm.locator import Locator class GooglePage(Webpage): def __init__(self, driver, url=None): super(GooglePage, self).__init__(driver, url) self.search_field = TextField(driver, Locator.by_name('q')) def get_result_links(self): links =  elements = self.driver.find_elements(By.XPATH, "//h3[contains(@class, 'r')]/a") for element in elements: links.append(element.get_attribute("href")) return links
You should notice that most of the operations except for get_result_links() are not visible on the GooglePage class. That is because the basic behaviors are either part of the Webpage or the TextField (aka an Element) type.
For more information about writing your PageObject’s in SeleniumPM, please direct your attention to Creating-your-PageObject-with-SeleniumPM
The Selenium PageModel implementation is not limited to just one language. Here are other language implementations:
- Java - Java-SeleniumPM
- Ruby - In consideration depending on needs and popularity.
Contributing to SeleniumPM
- Check out the latest master to make sure the feature hasn’t been implemented or the bug hasn’t been fixed yet.
- Check out the issue tracker to make sure someone already hasn’t requested it and/or contributed it.
- Fork the project.
- Start a feature/bugfix branch.
- Commit and push until you are happy with your contribution.
- Make sure to add tests for it. This is important so I don’t break it in a future version unintentionally.
- Please try not to mess with the version or history. If you want to have your own version, or is otherwise necessary, that is fine, but please isolate to its own commit so that I can cherry-pick around it.
A huge shoutout to Peter Downs for his very easy-to-follow instructions for submitting a Python package to the community. See first time with pypi for his instructions.
Also see the following:
seleniumpm installs the following upstream packages as of the latest release:
Copyright (c) 2017 Peter Salas. See LICENSE for further details.
- Added highlight() feature to all Element types
- Support defining locators as tuples
- Adding utility for scanning a path and generating json output of all the PageObject structures in the project
- Added get_webelement_by_text() to Element types
- Moving upload of wheel to use twine instead of setup.py. See the following link for why: https://packaging.python.org/distributing/#wheels
- Fixed issue with validating a Widget with locator = None on a Webpage
- Removed snapshot feature
- Implementing a wait_for_page_load (Default: False) on Webpage open(), refresh(), and reload() operations
- Webpage.refresh() and Webpage.reload() did not return self
- Turning off snapshot feature during is_page() and is_widget_loaded() operations. These operations often time trigger TimeoutException but it should not trigger a snapshot to be taken.
- Implementing to_json() and to_json_pp() for Element, Widget, Locator, and Webpage types
- Adding additional wait_for_widget_load() timer to calculate load time of a widget
- Adding more logging across seleniumpm. Most are WARNING levels, so this will need to be set accordingly
- Implemented a is_widget_loaded() for all Widget types. This also includes both Panel and IFrame types
- Added ability to enumerate over all elements defined on a webpage and/or widget, and then return all elements that failed their checks for presence and visibility. The ability is built directly into wait_for_page_load(), validate(), and wait_for_widget_load(). This check is controlled via parameter and global selenium config failfast_check_element (Default: True). This is set to True by default for old behavior of throwing exception on the first element that is not present/visible. Setting this to False will enable this feature and will check each element before raising the expected TimeoutException with a message containing all the elements that failed. NOTE: Depending on the timeout, this could potentially be a compounded amount of time before the method returns, hence why this feature is disabled by default.
- Element.dict() had an issue if the locator was None
- Widget.validate() force_check_visibility was accidentally set to True when expected to be False. This ended up causing unexpected conditions in a test where some elements were marked as invisible, but validate() was still checking for visibility
- For validate() and wait_for_page_load() operations, adding the meta-data for the Class and attribute name of the element that is being checked. This is so that if there is a TimeoutException, we can also report where we can find this element to fix in the code
- Add Logger for better debugging capability for testing purposes.
- Moving click_index() from Clickable to Element type.
- Adding Clickable.click_index() and Element.get_index_of_text() to indirectly address StaleElementReferenceException. Unfortunately, the problem may still arise and some experts say that StaleElementReferenceException is by design and should generally be “fixed” on a case-by-case basis and should not be fixed in a wrapper framework like seleniumpm nor will it be addressed in the Selenium core library.
- Adding the following ‘wait_for’ methods to Element: wait_for_webelements(), wait_for_text(), and wait_for_texts()
- Element.wait_for_present_and_visible() did not have a return self
- Adding mark_check() and mark_visible() to complement the existing mark_do_not_check() and mark_invisible()
- Adding unselect() to Checkbox and select() to both RadioButton and Checkbox types.
- Adding a new Label type. Right now, this is effectively the same as a TextElement or an Element
- Fixing a missed-case configuring disable_check_for_selenium_webdriver (Default: False) on a Webpage
- Adding new config disable_check_for_selenium_webdriver (Default: False) which now makes is possible to disable this check on an Element. Apparently, there are situations where a developer may pass in a non-webdriver object that overrides __getattr__() to returns driver.__getattribute__(item)
- Adding a check_myself flag (Default: True) on a Widget/Panel/IFrame type. This is to support certain scenarios where a developer may want to validate the collection of elements found in an Widget/Panel/IFrame, but not actually validate itself. Again, this is because Widget/Panel/IFrame is of type Element and thus has a locator.
- Adding a get_html() capability to return back either an entire page or the inner-html of a specific element. This is implemented on a Webpage, an IFrame, and Element types.
- Adding a wait_for_iframe_load() to IFrame type
- Added stop-watch capabilities (e.g. start, stop, split) to a Webpage and Element types. In addition, a basic page load timer has been implemented. Basically, every wait_for_page_load() and wait_for_iframe_load() will calculate a duration time automatically.
- Removed requirement that a Widget/Panel/IFrame type must define a Locator. This is to support situations where a developer doesn’t want to validate the location of the Widget, but instead just wants to validate the objects that are within a Widget. This is similar to a Webpage validation.
- Fixed scenario for __getattr__ where a developer could decorate a method with @property, and thus could execute code that results in an Error/Exception.
- Added try/finally block to the validate() operation on an IFrame. This is to handle the situation when a sub-webpage (a.k.a. an iFrame) fails a validation; in this situation, we want to make sure that we allow going back to the top-level-webpage in the event of a Error.
- Added a get_attribute_contains() and a get_attribute_is() on Element type. The intention is to implement an all-in-one solution for both retrieving an attribute and validating whether something contains or is something. This operation is useful for dynamic elements that use css classes for changing the state of a page (e.g. <div class=’svgOverviewView-status-icon fa fa-exclamation-triangle svg-status-warning’>); One could define a generic reference to this element, and then use get_attribute_contains() to check if it now contains ‘fa-exclamation-triangle’ css class.
- Implementation of seleniumpm.config module. This module can be imported via import seleniumpm.config as seleniumconfig; and variables can be changed like so: seleniumconfig.page_timeout_in_sec = 60
- Added a global element_timeout_in_ms and page_timeout_in_ms to seleniumpm.config module. Defaults are still 10s for element timeout and 30s for page timeout.
- Added a global screenshot_dir, screenshot_enabled, and test_screenshot_enabled to selenium.config module. The default values are ‘./screenshots’ and True respectively
- Added a global debug_logging_function to selenium.config module
- Added a refresh() operation on a Webpage (a.k.a. a page refresh). This of course is simply a driver.refresh() operation on an opened page.
- Added a get_current_url() operation on a Webpage
- Added get_number(), get_numbers(), get_int(), and get_float() to an Element. This is so that you can quickly and easily get numbers from element(s) text that represent numbers
- Added an “access element” abstraction layer so that you can retrieve any element defined any levels deep (e.g. A button defined within a Panel that is under an IFrame). The idea is simple: If I want to click a button 2-levels-deep, this can be done either by (1) page.iframe.panel1.login_button.click(), or (2) directly from the top-level page as if it was a local attribute via page.login_button.click()
- Added take_screenshot() feature to both a Webpage and Element types. The screenshot capability is controlled by seleniumpm.config.screenshot_enabled and is set to True by default.
- Added a take_screenshot_on_test_error annotation for annotating tests. The screenshot capability for a test is controlled by seleniumpm.config.test_screenshot_enabled and is set to True by default.
- Added pretty-print way of visualizing all attributes and methods for a Webpage. This is useful for documentation as well as for debugging
- Fixing issue with get_attribute() on Element. This method did not have an expected return statement
- Fixing missing implementation to do proper validation of an IFrame that is embedded on a Webpage
- Implement a get_texts() which returns back all the text (in a List) given a locator. This is available for all Element types
- Implement hover_over() capability to Element
- Fixed several critical issues related to get_element_attr(), validate(), and wait_for_page_load() operations on both a Webpage and Widget types
- Add more unittests for IFrame
- Fix wait_for_present_and_visible() timeout for the check for visibility. There are scenarios in which an element may be present, but not immediately visible
- Adding click() operations to a Widget. This is to support the fact that sometimes an entire section of a page can be “clickable”
- Adding new IFrame type
- Fixing issue with get_element_attr() method on both a Widget and a Webpage. The problem was that if you marked Widget type as do_not_check, then it should respect both the Widget and all of its embedded elements. The issue was that it would respect that the Widget was marked as do_not_check, but would still enumerate over all of its sub-elements.
- Fixing another issue with get_element_attr() method on both a Widget and a Webpage. The problem was related to Widget’s that are marked “invisible”. Similar to do_not_check, sub-elements should not be checked if the top-level Widget is already “invisible”.
- Both Webpage and Widget validate() methods should return self
- Fixing force_check_visibility parameter in Webpage.wait_for_page_load(). The default was set to True instead of False
- Fixed HISTORY.rst and README.rst files that are malformed sometime after 2.5.1 release
- Hopefully fixing other issues with 2.7.0 not being installable
- Adding a new ability to mark an Element type as ‘invisible’ so that validations on a Widget or Webpage can be done simply on presence
- Adding a new ability to mark an Element type as ‘do not check’ so that you could potentially define an Element that you simply don’t validate using the default validation() operation defined on a Widget or a Webpage. This feature should be used sparingly as there are better design patterns to handle most cases that you may think that you need to avoid validating the presence of specific Element types.
- Adding get_tuple() to Locator object. This is to make passing between SeleniumPM and the Selenium libraries eaiser
- Adding simpler way of defining Locators. Locators can now be created via Locator.by_xpath(path) or Locator.by_css_selector(path). The following types are supported:
- Adding wait_for_selected() to Element
- Adding wait_for_clickable() and click_invisible() to Clickable
- Adding send_keys_delayed() and type_delayed() to TextField
- Adding new Panel type that simply extends Widget. Conceptually they’re exactly the same, but Panel appears to be a more generally acceptable term for a section of a page
Fixed: - Simplifying README for more of a project overview. Details should be located on the wiki
- Fixing issue with setup.py throwing error missing HISTORY.rst from package data
2.5.1 (2017-03-07) - BROKEN
- Fixing issue with set_focus() or scroll_into_view(). They were apparently not included in 2.5.0 release
2.5.0 (2017-03-07) - BROKEN
- Addition of RadioButton type
- Addition of Dropdown type
- Addition of Image type
- Add new method get_element_attr() to Webpage and Widget type. This will give developers access to all define Element attributes on a Webpage or within a Widget. This method all supports retrieving a specific Element type (e.g. Button, Link, Checkbox)
- Changing default wait_for_page_load() and validate() methods to use the above mentioned get_element_attr(). This can still be overridden, and does not affect previous implementations.
- Adding new seleniumpm.examples.widgets package
- Element class now implements a get_action_chains() method to return back an ActionChains type.
- Element class now implements a set_focus() or scroll_into_view() functionality, for those pesky webelements that are need to be visible, but are corrently scrolled off page somehow.
- Adding type-checking to constructor of the Element, Widget, and Webpage types. These classes will now throw an AttributeError if not passed in a legitimate RemoteWebdriver, URL, or Locator type as parameters.
- Fixing issue appending two .rst files together to generate the long_description
- Using setuptools for setup.py.
- Using disutils.core instead of setuptools for setup.py. Hoping this fixes pretty-print of rst files on PyPi
- Better support for Table type and interacting with them on a page. This includes support for ‘search’ operations and enumerating over rows and columns
- Additional methods to Locator object to assist in managing them
- implemented get_webelement() and get_webelements() for all Elements. This will return the Selenium WebElement object(s).
- Implementation of object equality for all Selenium Page Model classes
- UnitTests are now using PhantomJS (Headless) target
- Removal of requestest dependency to keep the project simple
- The Widget type was missing in 2.3.0 release
- Expanding of the UnitTest coverage to ensure libraries are working correctly
- Fixing issue with get_text() in Python Selenium. Apparently, this call in Python (versus Java) is simply called ‘text’
- Conversion of README and HISTORY files to rst. This is so that they are rendered correctly on PyPi server
- Provides a full implementation of the current Java v2.3 of Selenium PageModel
- First release of seleniumpm for the world
- Contains minimum proof-of-concept for testing search on Google