Instapy: Note to developers

[redsector72]

Here is my finding, to present moment:
They are not checking if is a bot or not, they are simply enforcing multiple simultaneous login.
If you are running the official app WHILE anything else is browsing instagram, human or bot, they ban all and future actions, except the ones coming from their app.

That's probably why some people have problems and some not. The one who are real people, have trouble, the one that are only bot, works fine.

[CharlesCCC]

Uhh, interesting, how many instances you have noticed this pattern ?

[redsector72]

It's easy to reproduce. The instant you open the app, the browser version get ignored in the actions. If the browser continue the action, they ban it. Even if you are human. I've noticed it because I have opened a direct message while i was looking at the bot doing. The last picture liked was the one previous to the app launch.

[CharlesCCC]

@redsector72 Do you know how long the block will lift when you weren't using the mobile App ?

[Frituurpanda]

I can confirm the same behavior on my end. Seems that the block is active for the duration of the session as I’m able to run my script as soon as I remove the cookie file.

[CharlesCCC]

@Frituurpanda Are you saying, after you used mobile app, you will have to delete existing cookie from the log folder and re-generate it in order to run it without block ?

[jctommasi]

It's easy to reproduce. The instant you open the app, the browser version get ignored in the actions. If the browser continue the action, they ban it. Even if you are human. I've noticed it because I have opened a direct message while i was looking at the bot doing. The last picture liked was the one previous to the app launch.

I have 5 accounts running InstaPy, 3 of them are of people that use the app all the time, in this cases I get banned and can't do any automatizations at all. The other 2 accounts, are just bots and works perfectly fine.

First I was thinking it was a detection of geolocation (all accounts are from argentina) and InstaPy was runnining from DigitalOcean Servers (Germany, EEUU, etc.) but I discard this when try InstaPy with a vps from Argentina and get banned too.

[CharlesCCC]

@jctommasi thank you for sharing this insight.

Now the 5 million dollar questions is how can we get over this detection block...

[Tachenz]

Going to check this with my account overnight. If you are right, shutting down the app, then clearing cookie and having a good long Instapy session should not trigger a block...

[jctommasi]

@jctommasi thank you for sharing this insight.

Now the 5 million dollar questions is how can we get over this detection block...

Some users said in others topic: simulating mobile.. A solution just for now is being concious when open instagram app, but that is unsustainable, an interesting feature may be an api through telegram bot where you can close sessions by an "/stop" command before open IG app or best: developing an simple mobile app who detects when you open ig app and it check through an api if is a session running, if it is, alert the user:

Session is running:
opt1:stop session and open instagram app.
opt2:get out of here and let the bot make his job

i'm just brainstorming XD

[papillon-unix]

Guys, I don't want to disturb your observations but I am experiencing quite the opposite since few days: I have two accounts running Instapy, I use the official mobile app simultaneously with one of them and it works perfectly fine for this one (Instapy and app) but the other one that only runs Instapy got blocked on Instapy for any action. Can't follow, like or unfollow. But not blocked on the app...

[Tachenz]

Yeah, it was a nice theory but it does not work :
Logged off my Instagram app, cleared cookies Instapty banned within 5 follows at a speed of 1 follow per 10 minutes.

[jctommasi]

This is my script running with a test account (I developed some functions to randomize everything), I never add it to the IG app, last post was 23/12/2018 and I'm running InstaPy perfectly with chrome

#credentials
insta_username = 'username'
insta_password = 'password'

#telegram notif
telegram_chat = '-1111111'
bot_token = 'tokenbot'

# set security random countdown
countdown_to_start = random.randint(0,3600)
countdown = str(int(countdown_to_start/60))

session_id =  str(countdown_to_start)
path = sys.path[0] + '/'

#CHECK IF ACCOUNT HAS BANS
#A PHP SCRIPT RUNS EVERY 10 MINUTES, AND CHECK GENERAL.LOG IF ACC HAS WARNINGS, IF IT IS WRITE "1" in banned.txt
with open(path + 'logs/' + insta_username + '/banned.txt', 'r') as content_file:
    banned_int = int(content_file.read())
    if banned_int:
        requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": account has bans, aborting..\n===============\n".format(datetime.now().strftime("%H:%M:%S")))
        sys.exit(1)

#CHECK IF OTHER SESSION IS RUNNING, until start a session it writes 1 in onsession.txt
#sessions never gets overlayed
with open(path + 'logs/' + insta_username + '/onsession.txt', 'r') as content_file:
    onsession_int = int(content_file.read())
    if onsession_int:
        requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": onsession true, aborting..\n===============\n".format(datetime.now().strftime("%H:%M:%S")))
        sys.exit(1)

#set workspace
set_workspace("/InstaPy")

#random sleep to randomize start times (this script runs every 45 minutes from cron)
time.sleep(countdown_to_start)

#CHECK AGAIN IF OTHER SESSION IS RUNNING and write 1 onsession.txt
with open(path + 'logs/' + insta_username + '/onsession.txt', 'r') as content_file:
    onsession_int = int(content_file.read())
    if onsession_int:
        requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": onsession true, aborting..\n===============\n".format(datetime.now().strftime("%H:%M:%S")))
        sys.exit(1)
    else:
        with open(path + 'logs/' + insta_username + '/onsession.txt', 'w') as content_file:
            content_file.write(str(1))
            requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": onsession false, starting in " + countdown +"m\n===============\n".format(datetime.now().strftime("%H:%M:%S")))

def get_session():
    session = InstaPy(username=insta_username,
                      password=insta_password,
                      headless_browser=False,
                      bypass_with_mobile=True,
              nogui=True,
                      disable_image_load=False,
                      multi_logs=True)

    return session


def follow():

    sleep_time = random.randint(120,240)
    num_retries = 2

    for x in range(0, num_retries):  # try 2 times

        try:
            # get a session!
            session = get_session()
            # let's go!
            with smart_run(session):

                session.set_relationship_bounds(enabled=True,
                            potency_ratio=-0.70,
                            # potency_ratio=None,
                            delimit_by_numbers=True,
                            max_followers=6000,
                            max_following=3000,
                            min_followers=30,
                            min_following=30)

                session.set_skip_users(skip_private=True,
                                   skip_no_profile_pic=True,
                                   skip_business=False,
                                   business_percentage=100)

                session.set_quota_supervisor(enabled=True,
                sleep_after=['likes_h', 'comments_d', 'follows_h', 'unfollows_h', 'server_calls_d'], sleepyhead=True, stochastic_flow=True, notify_me=True,
                                          peak_likes=(10, None),
                                           peak_comments=(1, 1),
                                            peak_follows=(20, None),
                                             peak_unfollows=(15, None),
                                              peak_server_calls=(None, 4700))

                #DATABASE
                hitting_accounts = ['discovery', 'natgeowild', 'bbcearth', 'animalplanet','natgeoadventure', 'natgeoadventure','discoverychannelin', 'natgeochannel', 'discovery.hd', 'nasa', 'history', 'animalplanetindia', 'natgeoexpeditions', 'natgeo.la', 'natgeoindia', 'oceana', 'sciencechannel', 'time', 'insidenatgeo']

                hitting_tags = ['landscapephotography', 'landscapelovers', 'landscape_captures', 'landscapes', 'hiking', 'trees', 'mountain', 'sunrise', 'scenery', 'skyporn', 'outdoors', 'cloudporn', 'instanature', 'tree', 'water', 'forest', 'naturelover', 'panorama', 'autumn', 'skylovers', 'love', 'instagood', 'photooftheday', 'fashion', 'beautiful', 'happy', 'cute', 'tbt', 'like4like', 'followme', 'picoftheday', 'follow', 'me', 'selfie', 'summer', 'art', 'instadaily', 'friends', 'repost', 'nature', 'girls', 'fun', 'style', 'smile', 'food', 'instalike', 'family']


                # # # # # # # # # # # # # # # # # # # # # # # # # # # # #

                #RANDOMIZING ACC, TAG, AND FOLLOW NUMBERS
                rand_acc_numbers = random.randint(1,3)
                rand_tag_numbers = random.randint(1,2)
                rand_photos = random.randint(1,2)

                remove_follow_requests = random.randint(1,5)
                rand_unfollow = random.randint(1,2)
                rand_follow = random.randint(5,8)
                rand_like = random.randint(1,3)
                rand_storyview = random.randint(1,100)

                rand_like_str = str(rand_like)
                rand_follow_str = str(rand_follow)
                rand_unfollow_str = str(rand_unfollow)

                #DISORDERING
                random.shuffle(hitting_tags)
                random.shuffle(hitting_accounts)

                #CUTTING
                hitting_accounts = hitting_accounts[:rand_acc_numbers]
                hitting_tags = hitting_tags[:rand_tag_numbers]

                #just in case of debug i'm converting the selected tags in string to be able send this through telegram, also list_amout is an estimated number of procesed accounts
                hitting_tags_str = ' '.join(str(e) for e in hitting_tags)
                hitting_accounts_str = ' '.join(str(e) for e in hitting_accounts)
                list_amount = str(rand_follow*len(hitting_accounts))

                session.set_user_interact(amount=1, randomize=True, percentage=20, media='Photo')
                session.set_do_like(enabled=True, percentage=100)
                session.set_do_comment(enabled=False, percentage=0)
                session.set_do_follow(enabled=False, percentage=0)
                session.set_do_story(enabled = True, percentage = rand_storyview, simulate = False)
                session.set_action_delays(enabled=True,like=1,follow=1,unfollow=1,story=1,randomize=True, random_range=(120, 340))

                def like_feed():
                    session.like_by_feed(amount=rand_like, randomize=True, unfollow=True, interact=True)

                def like_tags():
                    session.like_by_tags(hitting_tags, amount=rand_like)

                def follow_tags():
                    session.follow_by_tags(hitting_tags, amount=rand_follow)

                def follow_followers():
                    session.follow_user_followers(hitting_accounts, amount=rand_follow, randomize=True, interact=True)

                def follow_likers():
                    session.follow_likers(hitting_accounts, photos_grab_amount = rand_photos, follow_likers_per_photo = rand_follow, randomize=True, interact=True)

                def unfollow_remove():
                    session.unfollow_users(amount=rand_unfollow, InstapyFollowed=(True, "all"), style="FIFO", unfollow_after=90*60*60, sleep_delay=600)
                    session.remove_follow_requests(amount=remove_follow_requests, sleep_delay=600)

                func_list = [follow_likers, unfollow_remove, like_tags, follow_followers, like_feed, follow_tags]

                requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": session launched succesfully\n===============\n".format(datetime.now().strftime("%H:%M:%S")))

                while func_list:
                    choice = rchoice(func_list)
                    choice();
                    func_list.remove(choice)

                str_error = None

        except Exception as str_error:
            pass

            if str_error:
                requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": ERROR, traceback detected\n===============\n".format(datetime.now().strftime("%H:%M:%S")))
                time.sleep(sleep_time)  # wait before trying to fetch the data again

        else:
            break

follow()

#onsession=0
with open(path + 'logs/' + insta_username + '/onsession.txt', 'w') as content_file:
    content_file.write(str(0))

requests.get("https://api.telegram.org/bot" + bot_token + "/sendMessage?chat_id=" + telegram_chat + "&text===============\n(id: "+ session_id +") @"+ insta_username + ": session has came to end\n===============\n".format(datetime.now().strftime("%H:%M:%S")))

Some notes:
session tasks are stored in functions to be able randomize the order of execution in every session
everything is inside a try in case of an error in connection is raised. try is inside a for, in case of error, it loop again

[Tr1pke]

@Frituurpanda Are you saying, after you used mobile app, you will have to delete existing cookie from the log folder and re-generate it in order to run it without block ?

Yes. I have the same. It goes well. Until I like 1 picture in the mobile app
Basically you have to log out on the phone until the session is done. That helps for my account.

[breuerfelix]

guys I got one really important question.

if you are on instagram with you normal Webbrowser and like a picture or do a comment and then you like a picture on your app and go back to the normal Webbrowser and do another like. is this blocked or not ? because if this is NOT blocked then we just have to focus on hiding selenium even more :) and if so ... instagram is fucking crazy :D

[papillon-unix]

@breuerfelix no, when I do what you said I don't get blocked.
It's crazy, I can't really explain the problem because I have the exact same use of two accounts (Normal browser, app, and Instapy) and one got blocked on Instapy but works perfectly on App and browser, and the second one not blocked at all...

[Tr1pke]

The moment I like one picture on the app while my bot is on my bot is blocked and I have to remove cookie and start it again every time I like a photo. Didn’t try it with following and unfollowing

[breuerfelix]

@Tr1pke so I am not talking about the bot I am talking about a regular browser :) please try this.

@papillon-unix thanks for testing that ! this means that instagram is detecting it in another way ... do you have the bot running on the current dev branch ( with the hide headless firefox extension) ?

if so, we need to improve it way more !

[lezlea]

Yeah it seems I have this problem too (although I hadn't noticed before!) so thanks for bringing it to my attention :)
Look forward to seeing if there is another solution except for simply not running at the same time!

[lezlea]

@Tr1pke so I am not talking about the bot I am talking about a regular browser :) please try this.

@papillon-unix thanks for testing that ! this means that instagram is detecting it in another way ... do you have the bot running on the current dev branch ( with the hide headless firefox extension) ?

if so, we need to improve it way more !

For me as you said I can interact in my home web browser fine and not get a block. But I get a block on my linux/proxy bot Chrome browser.

I'm going to try using my local residential IP as a proxy and see if that makes a difference.

[converge]

Yesterday I realized on my side that before doing any action, they're detecting Selenium/Webdriver somehow. I did compare the instances of a running browser without Selenium, and a running browser with Selenium.

The browser with Selenium is never able to load more than 12 profiles from my "Following" list, it doesn't load more profiles when scrolling, and I'm able to do that without Selenium.

I did compare the browser, and everything looks similar (the extension we're using is doing that), but there is something missing that we still don't know what it's.

[converge]

We did it! help testing the dev branch (0.6.0)

[lezlea]

@converge can I help - is there anything in particular you would like me to check? Or literally just run it and see if I get any blocks?

[lezlea]

@converge not sure if this is helpful or not but I tested your dev branch on an account that has had continuous action blocks recently (even after days of rest, new IP, no cookie, very low settings of 5 follows an hour).

• Normally I run in Chrome (this test was Selenium/Firefox).
• With a fresh login (new IP, no cookie) I didn't receive a validation check which I normally always do
• When testing over the last week on this account I manually click through in Chrome to follow one user to check if it is blocked or not, and I see I can follow with no block.
• I use follow_by_list with a preselected list of users, when this starts 80% of the time I get an action block with the first follow, otherwise it's within the first 3 follows!

Today with your dev version - No Action Block / the follows are valid and stick
However each follow received the notification "Phew! Last follow is not verified. [] might be temporarily blocked from following"
There was no action block popup and I've since checked the usernames and they were definitely followed.

I'll keep testing over next few hours and report back if it's useful

Really appreciate all your hard work!

[converge]

@lezlea thank you for the kind feedback! Actually, there is a lot of people behind it, I'm just part of it.

good to know everything is working again, we're fixing all these minor warnings at the moment.

[redsector72]

Hi there,
As I said, they not checking the presence of a bot, instead they are checking the absence of the human.

It Checks your behavior against its enormous set of other user behavior, and determines whether you act more like a human or a machine, the more people that use it, the larger the data set becomes, and the more the system can learn. That's why at first any "correction" seems to work, then to work less, then to stop working at all.

It's a mechanism very similar to the “Invisible reCAPTCHA" if you want to dig out more.

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

If this problem still occurs, please open a new issue