3

The title says it all - I am looking for a way (or ways) to load, enable, add (and likewise: disable / remove) extensions after the webdriver has been created.

Context: Python, selenium (per tags). Pyautogui for certain parts would be unavoidable, but would need to be robust. I have a draft working soln but its rough and a little GPU dependent (point: can/will not accept "this cannot be done with Chrome" as I know it can be done, but would like to see some soln/code to this effect).

I am perfectly aware of the various methods to load an extension prior to the driver being instantiated / created. i.e.

  1. o.add_extension(crx file location)
  2. o.add_argument(unpacked folder loc)
  3. o.add_argument('--user-data-dir=' + profileX)

where o = Options, profileX has extensions pre-installed and so forth. So will be clear - I am completely uninterested in proposals to this effect.

Any ideas/proposals in regards to the question in the title/above?

Ideally (but not absolutely critical) would be a programmatic solution that is not entirely GUI based (robotic/PyAutogui dependent).

You can assume the extension is available on the chrome store for this exercise.

Use-case: The extension I'm testing involves registering points and to protect fair / intended usage terms, its designed to cease functioning whenever it recognises automated manipulation (which is perfectly possible with 'headed' chrome, e.g. you just assign it a hotkey in chrome settings, and use Pyautogui to activate, or locate it's html page and interact directly etc.). GIVEN this, and the very many, many tests I have, it seems a waste of resource me having to reload the tests, or set up profiles, or stop and start the tests every so often so that I can include 'fresh' copies of the crx files or deploy one of the other methods I mention above. To date / my best knowledge, there is no solution for this. Happy to be pleasantly surprised if someone can point me to a 'duplicate' Q in this regard.

JB-007
  • 2,156
  • 1
  • 6
  • 22
  • I have a draft working solution but am hoping to see alternate methods; otherwise, if there is any evidence/expression of popular 'needs'/interest by others for something like above, will post online in this forum. fyi - this would then showcase/feature virtually all 'basic' known methods (including the boring pre-loaded methods I mention above too)... post comments below accordingly... Ta – JB-007 Jun 13 '21 at 08:56

1 Answers1

1

NOTES

  • Problem addressed with three possible approaches (thus soln. appears lengthy, but individual methods are not excessive/particular lengthy; what's more, code can be easily adapted as req.)
  • Further, this starts in the context of a single-browser environment / setup;
  • Ensues with variable and method definitions that extend the premise / thinking to encompass implementation within a multi-threaded / multi-browser environment

PRELIMINARY

import glob2,itertools,os, pyautogui, shutil, time, pyperclip, subprocess, keyboard as kb
from datetime import datetime
from zipfile import ZipFile

now = datetime.now()
dt_string: str = now.strftime("%d-%b-%Y, %H:%M:%S")
print(dt_string)
from selenium import webdriver
from selenium.webdriver import DesiredCapabilities
from selenium.webdriver.chrome.options import Options
from selenium.webdriver.support.wait import WebDriverWait
from concurrent.futures import ThreadPoolExecutor
from os import path
#references: date-time format: https://www.programiz.com/python-programming/datetime/current-datetime

datetime reference


METHODS

Method 1: add after chrome launches via Chrome-store

def ext_update_method1(ps, section=''):

global w, p2s
text_temp, p2s =  [''] * len(ps), []

def button_click(p, parms = ['',True]):
    #parms[0] = ID, parms[1] = click?
    ID,click = parms[0], parms[1]
    global text_temp
    text_temp[p] = ''
    ID = ['omghfjlpggmjjaagoclmmobgdodcjboh'] if ID == '' else ID #default i= Browsec if no ID entered
    d[p].get(f'https://chrome.google.com/webstore/detail/{ID[0]}')
    w[p] = WebDriverWait(d[p], 10)
    w[p].until(lambda x: x.execute_script("return document.readyState") == "complete")
    button = w[p].until(lambda x: x.find_element_by_class_name('g-c-R'))
    text_temp[p] = button.text
    if click == True:  button.click()

def add_remove():
    global p2s
    for i, y in enumerate(ys):  activ(i, ys), (ss, 0.5), keys([(pr, 'tab'), (pr, 'space')])
    after2, ys2 = gwwt(''), []
    for x in after2:
        if (x not in before) and (x not in ys_remove): ys2 = ys2 + [x, ]
    #    ys2 = ys2 + [x,] if (all(x not in before) and (x not in ys_remove)) else ys2
    for i, y in enumerate(ys2): activ(i, ys), (ss, 0.5), keys([(pr, 'tab'), (pr, 'space')])
    ss(1)
    after2, ys2 = gwwt(''), []
    for x in after2:
        if (x not in before) and (x not in ys_remove) and (x not in a[0] for a in wd): ys2 = ys2 + [x, ]
    for i, y in enumerate(ys2):
        if all(y != x[0] for x in wd): y.close()
    if len(ys_remove) > 0:
        for i, y in enumerate(ys_remove):
            activ(i, ys_remove), pr('space'), ss(0.5)
            p2s = p2s + [i, ]
        if section == 1:
            return 'complete 2nd iteration' #prevent endless loop (see next comment)
        else:
            print('removed add-in, now adding back again....')
            cc(2)
            thread_all(ps, close_extra)
            ext_update_method1(p2s, 1) #loop back on self to add
    else:
        return 'complete 1st iteration'

def close_extra(p):
    global w
    w_old = w[p]
    w[p] = WebDriverWait(d[p], 10)
    tabs_close_extra(p)
    tabs_close_extra(p)
    activ(p), pr('a'), activ(p), pr('enter')
    w[p].until(lambda x: x.execute_script("return document.readyState") == "complete")
    w[p] = w_old

thread_all(ps, tabs_close_extra)
before = gwwt('')
thread_all(ps, button_click)
cc(2) #cc(5)
after = gwwt('')
ys, ys_remove, p2s = [], [], []
ys_remove = gwwt('remove')
print(f'ys_remove = {ys_remove}')
for x in after:
    if (x not in ys_remove) and (x not in before): ys = ys + [x, ]
print(add_remove())
cc(2)
thread_all(ps, close_extra)
thread_all(ps,button_click,['',False])
qs = []
for x in text_temp: qs = qs + [1,] if 'add' in x.lower() else qs
print(f'qs: {qs}')
if section == 2: return #prevent never-ending loop (see next comment)
if len(qs) > 0: ext_update_method1(qs,2) #loop back on self in case 'add extension' still present in web-store

Notes: Can download multiple versions of same extension on same browser

  1. Caution: 3rd parameter in n_i_c is set to 1 by default: this will clear your downloads folder. Set to 0 if don't want this to happen
  2. Arguments:
    • args: n_i_c[0] = number of times you wish to download a given extension onto the same browser
    • n_i_c[1] = item of global list: ext_details, see vars()n_i_c2 = 1 to clear folder
    • n_i_c[3]= 0 => only download unpacked folder (not loading it too)
    • Requires methods ('Ancillary Code', Vars) below - code can be modified to exclude these.

Method 2: Download & load .crx file after Chrome launch

def ext_update_method2(ps, n_i_c=[2, 0, 1,1], section = ''):  # same order as it would 

appear for single set of extensions (iro one tab)...
        # parms = [2,0,0,0] =>
        # 0. repetitions - e.g. 2 downloads : uBlock
        # 1. 0 [index value - e.g.  uBlock  = 4, below] downloaded and unzipped/installed on profile p
        # 2: clear folder
        # 3: is_load  = 0 => only download unpacked folder (not loading it too)

    global start, w, files_old, ev, d, parent, txt
    global tabs_before
    global files, files_new, index_latest_files, index_redo, index_hks
    hks2, index_ext, clear_folder, is_load = n_i_c[0], n_i_c[1], n_i_c[2], n_i_c[3]  # ID, alias =  cred[0], cred[1], cred[2]
    hks2 = hks if hks == '' else hks2
    ID, alias = ext_details[index_ext][1], ext_details[index_ext][0]
    url_robwu = '''https://robwu.nl/crxviewer/?crx=https://chrome.google.com/webstore/detail/''' + str(ID)
    if clear_folder == 1:
        folder_clear(del_dir=os.path.join(os.path.expanduser("~"), 'Downloads', '*'))
    def thread_all_0(p): #set up stuff
        global d, w, index_hks2
        if 'error' in str(tabs_close_extra(p)):
            tab_close(p)
        d[p].implicitly_wait(5)
        w[p] = WebDriverWait(d[p], 5)
        index_hks2[p] = ''
        d[p].get('chrome://extensions')
        code_0 = '''var tags = document.querySelector('body').querySelector('extensions-manager')['shadowRoot'].querySelector('cr-view-manager').querySelector('extensions-item-list')["shadowRoot"].querySelector('#container').querySelectorAll('extensions-item'); return tags.length;'''
        index_hks[p]=ev[p](code_0)
        print(index_hks[p])
    #
    thread_all(ps, thread_all_0)
    #
    def thread_all_1(p): #open website to download crx (multi-thread to expedite)
        d[p].get(url_robwu)

    #
    if len(ps) >= 2: #prevent overloading website - only do 3 at a time if more than 2 browsers being used
        groups = list(data_mygrouper(2, ps))
        for x in groups: thread_all(x, thread_all_1,workers = 3, chunk = 3)
    else:
        thread_all(ps, thread_all_1)
    files_old = glob2.glob(os.path.join(os.path.expanduser("~"), 'Downloads', '*.zip'))
    def thread_all_2(p):
        try:
            w[p].until(lambda x: x.find_element_by_id("download-link")).click()
            print('lambda wait passed !!')
        except:
            print('lambda wait failed :(')
            return
    start_temp = time.time()
    while (time.time() - start_temp <= 90) and (len(glob2.glob(os.path.join(os.path.expanduser("~"), 'Downloads', '*.zip'))) -  len(files_old) < hks2 * (len(ps))) or (kb.is_pressed('escape') == True):
        if kb.is_pressed('escape') == True: break
        thread_all(ps, thread_all_2), ss(1)
    files = glob2.glob(os.path.join(os.path.expanduser("~"), 'Downloads', '*.zip'))
    files_new = []
    for keep in files:
        if str(keep) not in files_old: files_new = files_new + [str(keep), ]
    # thread_all(files_new, z_extension_unzip, alias)
    path_dir_core = os.path.join(os.path.expanduser("~"), 'Downloads')
    index_latest_files = [''] * len(files_new)
    #
    def thread_all_3(j):  # unzip
        global index_latest_files
        index_latest_files[j] = os.path.join(path_dir_core, alias + str(j))
        with ZipFile(files_new[j], 'r') as zipObj:
            zipObj.extractall(index_latest_files[j])
        os.remove(files_new[j])
    #
    thread_all(list(range(len(files_new))), thread_all_3)
    if is_load != 1: return
    code1 = '''if(document.querySelector('body').querySelector('extensions-manager')["shadowRoot"].querySelector('extensions-toolbar').attributes.length !=2){document.querySelector("body").querySelector("extensions-manager")["shadowRoot"].querySelector("extensions-toolbar")["shadowRoot"].querySelector("cr-toolbar").querySelector(".more-actions").querySelector("#devMode")["shadowRoot"].querySelector("#knob").click() } else {'dev mode already activated!'}'''
    code2 = '''document.querySelector('body').querySelector('extensions-manager')["shadowRoot"].querySelector('extensions-toolbar')["shadowRoot"].querySelector("#loadUnpacked").click()'''
    #
    def thread_all_4(p):
        d[p].get('chrome://extensions/'), ss(1)
        ev[p](code1)
    for y in range(hks2):
        thread_all(ps, thread_all_4)
        windows_before = gwwt('Select the extension directory')
        for p in ps:
            try:
                if path.exists(os.path.join(path_dir_core, index_latest_files[p + len(ps) * (y)])):
                    try:
                        ev[p](code2)
                    except:
                        continue
                    start_temp = time.time()
                    while (time.time() - start_temp <= 30) and (
                            len(gwwt('Select the extension directory')) - len(windows_before)) < 1 and (
                            'select the extension directory' not in gawt().lower()): pass  # len(ps) - len(errors_chrome()): pass
                    tabs_before = len(d[p].window_handles)
                    windows_A = gwwt('')
                    try:
                        keys([(tw, index_latest_files[p + len(ps) * (y)]), (ss, 0.5), (pr, 'enter'), (pr, 'tab'),(pr, 'space')], 0.75)
                        if len(windows_A) < len(gwwt('')):
                            d[p].refresh()
                            print('would break1')
                        else:
                            window_target = gwwt('Select the extension directory')[0]
                            try:
                                window_target.activate()
                            except:
                                window_target.minimize()
                                window_target.restore()
                            finally:
                                pyperclip.copy(index_latest_files[p  + len(ps) * (y)]), ss(0.5)
                                keys([(hk, 'ctrl', 'l'), (kd, 'shift'), (pr, 'tab', 4), (ku, 'shift'), (hk, 'ctrl', 'v'), (pr, 'enter'), (pr, 'tab'), (pr, 'space')])
                    except:
                        d[p].refresh()
                        pr('escape', 2)
                        print('would break2')
                    tabs_close_extra(p)
                else:  continue
            except:  continue
    def thread_all_5(p):
        global qs
        d[p].get('chrome://extensions/shortcuts')
        code_0 = '''return document.querySelector('body').firstElementChild["shadowRoot"].querySelector("cr-view-manager").querySelector('extensions-keyboard-shortcuts')['shadowRoot'].querySelector('#container').querySelectorAll('.shortcut-card').length'''
        index_hks[p] = int(ev[p](code_0)) - index_hks[p]
        qs = qs + [p,] if index_hks[p] < hks2 else qs
        print(f'index_hks[{p}] = {index_hks[p]}')
        index4[p] = 0
    qs = []
    if clear_folder == 1:
        thread_all(ps, thread_all_5)
    if section == 1: return
    if len(qs) > 0:  ext_update_method2(qs, n_i_c = [hks2, index_ext, 0, 1], section = 1)
    qs = []
    for p in range(len(ps)): qs = qs + [p,]

Method 3: remove programmatically

def ext_remove(ps,ext):
    for p in range(len(ps)):
        try:
            activ(p)
            d[p].get('chrome://extensions/')
            #pr('f12'),ss(1)
            hk('ctrl', 'shift', 'j'), ss(1)
            code = '''var tags = document.querySelector('body').querySelector('extensions-manager')['shadowRoot'].querySelector('cr-view-manager').querySelector('extensions-item-list')["shadowRoot"].querySelector('#container').querySelectorAll('extensions-item'); for (i = 0;i <= tags.length; i ++) {try {if(tags[i]['shadowRoot'].querySelector('#a11yAssociation').innerText.toLowerCase().indexOf("''' + str(ext) +'''") > 0) {tags[i]['shadowRoot'].querySelector('#remove-button').click();break} } catch {}}'''
            pyperclip.copy(code), ss(0.5)
            # pr('tab'), ss(0.5)
            temp_rep = 1 if type(ext) == str else len(ext)
            for _ in range(temp_rep):
                hk('ctrl', 'v'), ss(0.5)
                pr('enter'), ss(0.5)
                pr('space'), ss(0.5)
                tabs_close_extra(p)

            pr('f12'), ss(0.5) #hk('ctrl', 'shift', 'j'), ss(0.5)
        except: pass

ANCILLARY CODE

  1. Activate window
  2. Shorthand: simulate keyboard strokes
  3. Methods: sub-divide lists, close tabs, thread-pool (multi-threading), and tile windows*

*(java scripts incl. - to be saved in same location as this script

Activate window

pyautogui.FAILSAFE = False

def activ(p, index_wd=''):
    global   index_activ
    try:
        if index_wd == '':
            index_wd = wd[p][0]
        else:
            index_wd = index_wd[p]
    except:
        index_wd = wd[p]

    print(f'active {w}')
    index_activ[p] = False
    try:
        index_wd.activate()
        index_activ[p] = True
        return index_activ[p]
    except:
        try:
            index_wd.minimize()
            index_wd.restore()
            index_activ[p] = True
            return index_activ[p]
        except:
            index_activ[p] = False
    index_activ[p] = False
    return index_activ[p]

Quick keys

def keys(actions, delay=0.5):

# global BL3
print(f'def keys()')  # {actions}') #, profile {p}')
outcome = []
for action in actions:
    print(f'action {action}')
    if len(action) == 4:
        outcome = outcome + [action[0](action[1], action[2], action[3]), ]
    elif len(action) == 3:
        outcome = outcome + [action[0](action[1], action[2]), ]
    elif (len(action) == 2) and (action[1] != ''):
        outcome = outcome + [action[0](action[1]), ]
    else:
        outcome = outcome + [action[0](), ]
    ss(delay)  # await ass(delay)
return outcome

Sub-divide lists

def data_mygrouper(n,
                   iterable):
    args = [iter(iterable)] * n
    return ([e for e in t if e != None] for t in itertools.zip_longest(*args))

stack overflow reference (data_mygrouper)

ThreadPool

def thread_all(ps, fn, parm='', actions=[], workers=6, chunk=1):
    # https://stackoverflow.com/questions/42056738/how-to-pass-a-function-with-more-than-one-argument-to-python-concurrent-futures/42056975
    print(f'thread_all({ps}, {fn}, {parm}, {actions}')
    if parm == '':
        with ThreadPoolExecutor(max_workers=max(1, workers)) as executor:
            return executor.map(fn, ps, timeout=90, chunksize=max(1, chunk))
    else:
        with ThreadPoolExecutor(max_workers=max(1, workers)) as executor:
            return executor.map(fn, ps, itertools.repeat(parm, L), timeout=90, chunksize=max(1, chunk))

thread-pool_exector ref: here

close tabs

def tab_close(p):
    print(f'def tab_close {p}')
    # d[p].switch_to.window(parent[p])
    for h in d[p].window_handles:
        d[p].switch_to.window(h)
        parent[p] = h
        [ev[p]] = [d[p].execute_script]
        break
    # parent[p] = d[p].current_window_handle
    for h in d[p].window_handles:
        if h != parent[p]:
            d[p].switch_to.window(h)
            if len(d[p].window_handles) > 1: d[p].close()
            else: break
    d[p].switch_to.window(parent[p])



def tabs_close_extra(p):
    # alternative method (much slower): https://stackoverflow.com/questions/12729265/switch-tabs-using-selenium-webdriver-with-java
    try:
        for h in d[p].window_handles:
            d[p].switch_to.window(h)
            parent[p] = h
            break
        # parent[p] = d[p].current_window_handle
        tabs_original = len(d[p].window_handles)
        if tabs_original > 1:
            for h in d[p].window_handles:
                if h != parent[p]:
                    d[p].switch_to.window(h)
                    activ(p), hk('ctrl', 'w')
                    if len(d[p].window_handles) == tabs_original:
                        d[p].close()
                    # d[p].switch_to.window(parent[p])
            d[p].switch_to.window(parent[p])
    except:
        print(f'error in tabs_close_extra {p}')

Variables

def vars():
    print('def vars()')
    global index, index2, index3, index4, index5, index6
    global index_activ, index_activ2, hks, pycharm_win
    global index_latest_files, index_hks, index_hks2
    global parent, d,  w, w2
    global path_chrome, path_exec, path_core, path_temp, profile, done, caps, pycharm_win
    global ps, p2s, qs
    global L, tile_type, compact, outcome
    global kd, ku, hk, pr, tw, wr, cl
    global gats, gaws, gwwt, gawt, gaw
    global cc, ss, start, start2, start_overall
    global wd, wd2, ev, errors, errors2, errors3
    global ext_0, ext_1, ext_2, ext_3,ext_details
    caps = DesiredCapabilities.CHROME.copy()
    [cc, ss, kd, ku, hk, pr, tw, wr, cl] = [pyautogui.countdown, time.sleep, 
    pyautogui.keyDown, pyautogui.keyUp, pyautogui.hotkey, pyautogui.press,,pyautogui.typewrite, pyautogui.write, pyautogui.click]
    [gats, gaws, gwwt] = [pyautogui.getAllTitles, pyautogui.getAllWindows, pyautogui.getWindowsWithTitle]
    [gawt, gaw] = [pyautogui.getActiveWindowTitle, pyautogui.getActiveWindow]
    L = 4
    qs, p2s, ps, = list(range(L)), list(range(L)), list(range(L))
    text_temp, start = [''] * L, [0] * L
    index_hks, index_hks2 = [hks] * L, [0] * L
    index_latest_files = ['']
    parent, w, ev, d, wd = [''] * L, [''] * L, [''] * L, [''] * (L + 1), [['']]*L
    start_overall, outcome = '', ''
    try: pycharm_win = [gaw()]
    except: pass
    path_chrome = "C:\Program Files\Google\Chrome\Application\chrome.exe"
    path_core = os.path.join(os.path.expanduser("~"), 'PyCharmProjects', 'Parallel')
    path_exec = os.path.join(path_core, "chromedriver.exe")
    ext_details = [ ['ublock', 'cjpalhdlnbpafiamejdnhcphjbkeiagm'], ['keyboard_shortcuts', 'dkoadhojigekhckndaehenfbhcgfeepl'], ['browsec', 'omghfjlpggmjjaagoclmmobgdodcjboh'], ['itrace', 'njkmjblmcfiobddjgebnoeldkjcplfjb']]
    ext_0= os.path.join(path_core, "Browsec")
    ext_1 = os.path.join(path_core, "Keyboard")
    ext_2 = os.path.join(path_core, "itrace")
    ext_3 = os.path.join(path_core, "ublock")

MAIN CODE

if __name__ == '__main__':
    ps =''
    vars()
    if len(ps) > 3:
        ps_new = list(data_mygrouper(int(round(len(ps) / 2, ndigits=0)), ps))
        for x in range(2): thread_all(ps_new[x], new_profile, workers=len(ps_new[x]) + 1, chunk=len(ps_new[x]) + 1), cc(5)
    # new_profile
    else:   thread_all(ps, new_profile, workers=L + 1, chunk=L+1)
    #thread_all(ps,new_profile)
    tile_windows('v')
    ext_update_method1(ps)
    ext_remove(ps,'browsec')
    ext_update_method2(ps)
JB-007
  • 2,156
  • 1
  • 6
  • 22