An idea for simulate human behavior.

[Wathfea]

Hi, lets develop together a cool human like behavior method, what we can all use later and maybe we can add it to instagrapi as a helper method at the end.

My idea is something like this:

    #Simulate human behavior to avoid being banned
    #This function can be called with a force boolean paramter to force a human simulation
    #If the force parameter is not set, the function will randomly decide if a human simulation is needed
    
    def simulate_human(self, force=False):
        if (random.randint(1, 100) >= 85) or force:
            hashtags = [
                'nature', 
                'foodporn', 
                'ocean', 
                'forest', 
                'fishing', 
                'travel', 
                'instagood',
                'photooftheday',
                'tbt',
                'love',
                'beautiful',
                'fashion',
                'happy',
                'cute',
                'followme'
            ]

           #First simulate to get the main user feed
            self.session.get_reels_tray_feed(reason="pull_to_refresh")
            self.session.get_timeline_feed(options=["pull_to_refresh"])
            
            #Go to profile or inbox
            if random.randint(1, 100) <= 50:
                self.session.account_info()
                
                #50 % go to random hashtag
                self.random_load_hashtag_medias(hashtags)

            else:
                self.session.news_inbox_v1(mark_as_seen=True)
                
                #50 % go to random hashtag
                self.random_load_hashtag_medias(hashtags)


            #Go back to main feed
            self.session.get_reels_tray_feed(reason="pull_to_refresh")
            self.session.get_timeline_feed(options=["pull_to_refresh"])
            
    def random_load_hashtag_medias(self, hashtags):
        if random.randint(1, 100) <= 50:
            hashtag = random.sample(hashtags, 1)
            self.affy.log_msgs(f'[INFO] Search for hashtag #{hashtag[0]}.')
            self.affy.inc_request_counter()
            self.session.hashtag_medias_top(name=hashtag[0])
            self.affy.inc_request_counter()
            self.session.hashtag_medias_v1_chunk(name=hashtag[0], max_amount=6, tab_key="recent")            

Feel free to add your own ideas. :)

[muhammeddilaver]

The Instagram application makes a number of requests that run in the background before and after viewing a media or viewing a user. According to the main request to be made, I think that the request pattern created before and after that request should be imitated.

[Wathfea]

@adw0rd Do you know maybe what are those requests? Then I can add it to the function also, and we can later when it is full we can add it to instagrapi as a callable helper functio maybe. If you are up to it.

[adw0rd]

Yes, I would look at the PR, it's interesting

[muhammeddilaver]

The requests that occur in order in a login event are listed above. (except profile photo requests)
I remember that we paid attention to the patterns of requests from users when I was managing the firewall. I think there is a similar situation on Instagram. If we don't send other seemingly insignificant requests, the system marks our activities as suspicious.

We can think of this situation in follow or like requests too. Sending a follow request to a user without being displayed can be seen as an odd request.

I think that this situation is not ignored in Instagram's security policies.

[Wathfea]

Cool, we should filter out the necessary calls. I created a spreadsheet for this.
If we have the calls we can cross check there is a related method for it in instagrapi.
If no add it to instagrapi.
When we have all the calls we can create wrapper methods like the categories in the spreadsheet and finally we can use these wrappers in the simulation with random choices.

https://docs.google.com/spreadsheets/d/1k9jYtDgAee6yikEQQVSatSaqmv-wM2EeWVeDD65eBsU/edit?usp=sharing

[muhammeddilaver]

I can list requests for all sections, the content of requests, and responses to requests. But the data sent in some of the requests created here are created by the instagram app (tokens, etc.) and unfortunately I am not good at analyzing the instagram application. Frankly, it would be nice to gain experience on this subject.