Collect users and followers on Instagram with fake InstagramAPI

instagram.py

import sys, json
from InstagramAPI import InstagramAPI
from time import sleep
from pprint import pprint
#import pandas as pd
#import requests, urllib, bs4, ssl, json, sys, re

def connect_api(user, password):
    api = InstagramAPI(user, password)
    retries = 12
    while not api.login() and retries > 0:
        sleep(5)
        retries -= 1
    if not api and not retries:
        print("Impossible to connect to Instagram with account %s/%s" % (user, password), file=sys.stderr)
        sys.exit()
    return api

def collect_users(api, list_users):
    try:
        with open("users_info.json") as f:
            fulldata = json.load(f)
    except:
        fulldata = {}

    accounts = []

    for user_id in list_users:
        print("WORKING on", user_id)
        if user_id in fulldata:
            content = fulldata[user_id]
        else:
            retries = 12
            keys = {}
            while 'user' not in keys and retries > 0:
                api.getUsernameInfo(user_id)
                content = api.LastJson
                keys = content.keys()
                if 'user' not in keys:
                    sleep(5)
                    retries -= 1
            if 'user' not in keys and not retries:
                print("Impossible to collect user %s" % (user_id), file=sys.stderr)
                pprint(content)
                break
            fulldata[user_id] = content
            with open("users_info.json", "w") as f:
                json.dump(fulldata, f)

        dict_account = {
            'user_id': user_id,
            'username' : content['user']['username'],
            'nb_posts': content['user']['media_count'],
            'nb_followers': content['user']['follower_count'],
            'nb_following': content['user']['following_count'],
            'biography': content['user']['biography'],
            'business_account': content['user']['is_business'],
            'external_url': content['user']['external_url'],
            'email': content['user'].get('email')
        }

        accounts.append(dict_account)

        sleep(5)

    return accounts

def collect_followers(api, list_users):
    try:
        with open("followers_info.json") as f:
            fulldata = json.load(f)
    except:
        fulldata = {}

    # fulldata = {
    #   583: {
    #     "next_max_id": False,
    #     "followers": []
    #   }
    # }

    user_followers = {}
    for user_id in list_users:
        print("WORKING on", user_id)
        if user_id in fulldata:
            followers = fulldata[user_id]["followers"]
            next_max_id = fulldata[user_id]["next_max_id"]
            if not next_max_id:
                continue
        else:
            followers = []
            next_max_id = True
            fulldata[user_id] = {
                "next_max_id": True,
                "followers": []
            }
        while next_max_id:
            if next_max_id is True:
                next_max_id = ''

            retries = 12
            fail = True
            while fail and retries:
                api.getUserFollowers(user_id, maxid=next_max_id)
                content = api.LastJson
                fail = "status" in content and content["status"] == "fail"
                if fail:
                    sleep(5)
                    retries -= 1
            if fail:
                print("Impossible to collect followers for user %s after %s" % (user_id, next_max_id), file=sys.stderr)
                pprint(content)
                sys.exit()

            followers.extend(content.get("users", []))
            next_max_id = content.get("next_max_id", False)

            fulldata[user_id]["followers"] = followers
            fulldata[user_id]["next_max_id"] = next_max_id
            with open("users_info.json", "w") as f:
                json.dump(fulldata, f)

            sleep(5)
        user_followers[user_id] = followers
    return user_followers

def store_accounts_as_csv(accounts, filename):
    pass
    #pd.DataFrame(account).to_csv('info_account1.csv', index=False, sep = ";")
    #return pd.DataFrame(account)

list_ids = [337971045,10173172837,5708920156,12537108438,18341208002,18590139197,19297143602,515814807,1395905971,7072598564,857128325,49299102,2945198438,4420087292,21438298166,19984486527,1906240752,7070095870,2579722126,295635844,22904886662,3644338541,3602461392,23292267332,3508969231,19609124215,3446732741,2977775554,1620762013,7332266442,8142428011,21956610305,21531584620,7019467857,9035517470,5534057213,2304132968,8782062667,2694726042,10096501877,258193824,18753220956,1523148837,263866558,2081033702,38502642,13220407486,4562817158,4155110032,15762558953,12041648666,337202992,19555638099,5915281563,1277552728,2025431878,22604627105,12605844,1499137365,7030432419,21606357780,9284662114,3118527987,6632870837,3593864527,5454766143,16169897513,3471096946,733557832,595693700,185782160,9958639815,6619855531,6203997062,3622616553,4590712874,16661568596,6950490907,19939252778,20292931079,8618584510,180609477,2464340484,190621442,3631493356,11165539607,285795392,8419252313,1949248362,8926687023,4575581534,3892134892,16129318075,1825132273,177293615,14646254128,5464770900,8118628539,381081801,7459551266,7529634962,298848606,17641402987,4015260548,1187715018,6635654389,1550869072,19294421450,3941464878,10145861788]
#list2 = [11825812741,3170312929,4057419877,200133817,3435329739,8187844022,18884944773,2023208616,21997825236,1675249774,14982482,16617613746,6320112056,1787954632,7797163837,1420206756,18886889042,21509476686,8935229467,2008612110,10949652797,1953815149,7599797254,1679941698,5595149649,4233705265,2257577385,14297220433,2289239013,1782277430,11472918196,7383070628,4345576080,11877883391,11324818653,7925783270,21810322655,1426197079,5346012429,12784632805,341383180,7284285302,1986471972,9236083364,5621927646,1136819870,4160132196,8443960171,7878952048,1465781351,6237203402,12275018823,2125958516,3670563541,298756519,1408181630,3769751466,6229597236,1637920381,21237654260,15475791164,19444379223,2011666202,1928368037,3881198522,4004019915,11804397522,590441685,470739052,15410611763,2148687548,2159312097,36957963,13233898933,5495650,19581047063,6725488025,1744975479,6005416110,208861550,4060822521,1835779221,6871689508,7156904617,4324684081,1421059539,25715676,3261356864,7463773711,356433855,7958875238,2893288415,8934390122,199730470,10699390142,5449373771,12597970216,8274726467,1536323358,1429766734,18770400397,2085315863,4008779555,21446031699,10992854930,18645601714,12256128791,3505430247,9130591321,3268276,21262772807,3938612394,7175872002,2997977399,3537804985,16919267,3595965947,1829188208,11578348657,22423436365,54429165,4779068551,22718812402,3582172948,183286468,7305708339,6459632134,18871270208,17473435497,422794975,10895097220,5590709634,3315515018,1652500545,1997477498,2165952902,4165752894,14052972865,4148109313,13682136332,5894257320,7324486058,7404535024,214660632,19257734797,1537912168,249011472,6921660787,1244671760,1201345996,173644312,2090425980,3268618703,7318241862,21786344886,10866067911,585918169,1061865823,14410394212,19568767716,8503143014,7629393071,21674702302,2298114350,18800844274,10004813655,18018069342,13065613071,227832857,2070906273,3556242227,46715662,18009410251,5351318501,2876979029,18651327130,4776521177,21056955404,20258665363,399190308,5478787190,1994625751,11381482217,1436897537,7533513091,22181469195,20482275640,4213318973,7138251166,10902388306,19391434217,17225835759,20680281814,190568565,8519633566,4739412020,1824602793,1662357475,187842531,6214962940,13345126003,250585860,5580046326,7038737840,1482956843,1900772090,20008267221,2343688727,19211902931,484501085,12990159032,17645570698,8679648861,8177875018,370580726,11930791238,9138285120,20217218708,4145740402,6116180985,21918275413,11141005169,17288380807,14583201468,6020537746,11875170336,22705015388,4360363982,1732676087,18119295638,5809775680,23921138937,6640747203,19544852488,9376559078,9301216217,3271894668,5386987039,16469109866,1667922900,2009732757,4050478037,3057947313,7483342664,17087945209,2365713725,8944487716,1649863114,8657591059,3919764334,3888326697,8081141162,8006372831,4091842663,6021346931,8216508673,9023085638,1545982006,13765331082,21679112557,2702460094,2221070295,7057003987,8661052357,19092214883,4516208636,3485546446,326527422,3172958006,21734572830,13816058779,18530528749,7151631020,264948968,2922546436,9207738136,6086815792,238047472,3448242177,18745178554,1543667327,1070417878,11459457105,8234749648,7230644682,6105478177,7651257434,1008276905,14037201094,6717190954,8581412984,4276757483,21004850582,541808129,15920849975,4555914280,1790743870,12370732144,6866081026,2944773028,3141702123,11774810260,20344154957,2241126438,10167077288,1540479555,1700279093,1283109311,3013626593,1112915592,2104411965,16965142395,2011249859,37633326,248182511,3015785380,21976848639,8004094586,11651845564,20738590780,1563580677,581389911,18785495597,8649939925,4338377164,3134119300,7335868351,574188184,3926342822,15437636037,2950470899,24267802041,2288974491,10380175853,10718818038,4357321010,4480378892,6894234851,13823260935,4628890860,11141842932,183251821,18900242840,19706256478,32239496,497595819,9614218593,18841534112,1098188697,4060626572,15750438462,9781767448,2233178158,3179610755,3052163922,5055280170,20736104329,5402222552,208749462,1471183535,4899517131,15413328726,7521303771,1069468357,7633889076,1952490812,15895941584,2526345498,1989794302,4632688741,15234766743,6943817814,5563763370,5049388221,9063149225,3094772600,1782357729,4240549752,195198040,6157373906,5415353496,4428150952,20258241893,14487528393,1450823313,1829418687,12133754817,3311579535,8103886141,21211360293,282181916,3067203,18648081131,14296831,5589783307,2012677718,192210656,12518943408,8596544396,4333773169,1521446434,19374131817,10032387741,4377380247,4283699836,3112525277,1427518040,2132004777,12138487,3002985,4712389943,2327419438,774664520,2110142521,4203217562,206742,18032811240,3648168046]

username = sys.argv[1]
password = sys.argv[2]
api = connect_api(username, password)
#accounts = collect_users(api, list_ids)
#pprint(accounts)
#store_accounts_as_csv(accounts, "info_accounts.csv")

followers = collect_followers(api, [1523148838])
pprint(followers)