Selenium - get all iframes in a page (even nested ones)?

Finding iframes solely by HTML element tag or attributes (including ID) appears to be unreliable.

On the other hand, recursively searching by iframe indexes works relatively fine.

def find_all_iframes(driver):
    iframes = driver.find_elements_by_xpath("//iframe")
    for index, iframe in enumerate(iframes):
        # Your sweet business logic applied to iframe goes here.
        driver.switch_to.frame(index)
        find_all_iframes(driver)
        driver.switch_to.parent_frame()

I was not able to find a website with several layers of nested frames to fully test this concept, but I was able to test it on a site with just one layer of nested frames. So, this might require a bit of debugging to deal with deeper nesting. Also, this code assumes that each of the iframes has a name attribute.

I believe that using a recursive function along these lines will solve the issue for you, and here's an example data structure to go along with it:

def frame_search(path):
    framedict = {}
    for child_frame in browser.find_elements_by_tag_name('frame'):
        child_frame_name = child_frame.get_attribute('name')
        framedict[child_frame_name] = {'framepath' : path, 'children' : {}}
        xpath = '//frame[@name="{}"]'.format(child_frame_name)
        browser.switch_to.frame(browser.find_element_by_xpath(xpath))
        framedict[child_frame_name]['children'] = frame_search(framedict[child_frame_name]['framepath']+[child_frame_name])
        ...
        do something involving this child_frame
        ...
        browser.switch_to.default_content()
        if len(framedict[child_frame_name]['framepath'])>0:
            for parent in framedict[child_frame_name]['framepath']:
                parent_xpath = '//frame[@name="{}"]'.format(parent)
                browser.switch_to.frame(browser.find_element_by_xpath(parent_xpath))
    return framedict

You'd kick it off by calling: frametree = iframe_search([]), and the framedict would end up looking something like this:

frametree = 
{'child1' : 'framepath' : [], 'children' : {'child1.1' : 'framepath' : ['child1'], 'children' : {...etc}}, 
 'child2' : 'framepath' : [], 'children' : {'child2.1' : 'framepath' : ['child2'], 'children' : {...etc}}}

A note: The reason that I wrote this to use attributes of the frames to identify them instead of just using the result of the find_elements method is that I've found in certain scenarios Selenium will throw a stale data exception after a page has been open for too long, and those responses are no longer useful. Obviously, the frame's attributes are not going to change, so it's a bit more stable to use the xpath. Hope this helps.