def test_tag():
tag_name = 'cat'
res = ie.tag(tag_name)
assert isinstance(res, tuple)
assert isinstance(res.data, dict)
assert isinstance(res.cursor, six.string_types)
time.sleep(1)
data, _ = ie.tag(tag_name, res.cursor)
assert isinstance(data, dict)
def run(self):
data, cursor = ie.tag(self.term)
edges = data['edge_hashtag_to_media']['edges']
posts = []
for post in edges:
body = post['node']['edge_media_to_caption']['edges'][0]['node']['text']
text = TextBlob(body)
sentiment = text.sentiment
b = {
"id":post['node']['id'],
"body": body,
"display_url": post['node']['display_url'],
"caption": post['node']['accessibility_caption'],
"sentiment": sentiment
}
posts.append(b)
return posts
def Search(Query_List, Task_ID, Type, **kwargs):
Data_to_Cache = []
Cached_Data = []
if kwargs.get('Limit'):
if int(kwargs["Limit"]) > 0:
Limit = kwargs["Limit"]
else:
Limit = 10
else:
Limit = 10
Directory = General.Make_Directory(Plugin_Name.lower())
logger = logging.getLogger()
logger.setLevel(logging.INFO)
Log_File = General.Logging(Directory, Plugin_Name.lower())
handler = logging.FileHandler(os.path.join(Directory, Log_File), "w")
handler.setLevel(logging.DEBUG)
formatter = logging.Formatter("%(levelname)s - %(message)s")
handler.setFormatter(formatter)
logger.addHandler(handler)
Cached_Data = General.Get_Cache(Directory, Plugin_Name)
if not Cached_Data:
Cached_Data = []
Query_List = General.Convert_to_List(Query_List)
for Query in Query_List:
if Type == "User":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.user(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response, indent=4, sort_keys=True)
Output_file = General.Main_File_Create(Directory, Local_Plugin_Name, CSE_JSON_Output_Response, Query, ".json")
Posts = CSE_Response[0]["edge_owner_to_timeline_media"]["edges"]
Output_Connections = General.Connections(Query, Local_Plugin_Name, "instagram.com", "Data Leakage", Task_ID, Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = "https://www.instagram.com/p/" + Shortcode + "/"
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(Limit):
if Output_file:
Output_Connections.Output(Output_file, URL, General.Get_Title(URL))
Data_to_Cache.append(URL)
Current_Step += 1
elif Type == "Tag":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.tag(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response, indent=4, sort_keys=True)
Output_file = General.Main_File_Create(Directory, Local_Plugin_Name, CSE_JSON_Output_Response, Query, ".json")
Posts = CSE_Response[0]["edge_hashtag_to_media"]["edges"]
Output_Connections = General.Connections(Query, Local_Plugin_Name, "instagram.com", "Data Leakage", Task_ID, Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = "https://www.instagram.com/p/" + Shortcode + "/"
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(Limit):
if Output_file:
Output_Connections.Output(Output_file, URL, General.Get_Title(URL))
Data_to_Cache.append(URL)
Current_Step += 1
elif Type == "Location":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = location(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response, indent=4, sort_keys=True)
Output_file = General.Main_File_Create(Directory, Local_Plugin_Name, CSE_JSON_Output_Response, Query, ".json")
Posts = CSE_Response[0]["edge_location_to_media"]["edges"]
Output_Connections = General.Connections(Query, Local_Plugin_Name, "instagram.com", "Data Leakage", Task_ID, Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = "https://www.instagram.com/p/" + Shortcode + "/"
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(Limit):
if Output_file:
Output_Connections.Output(Output_file, URL, General.Get_Title(URL))
Data_to_Cache.append(URL)
Current_Step += 1
elif Type == "Media":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.media(Query)
if CSE_Response:
CSE_JSON_Output_Response = json.dumps(CSE_Response, indent=4, sort_keys=True)
Output_file = General.Main_File_Create(Directory, Local_Plugin_Name, CSE_JSON_Output_Response, Query, ".json")
URL = "https://www.instagram.com/p/" + Query + "/"
if URL not in Cached_Data and URL not in Data_to_Cache:
if Output_file:
Output_Connections = General.Connections(Query, Local_Plugin_Name, "instagram.com", "Data Leakage", Task_ID, Local_Plugin_Name.lower())
Output_Connections.Output(Output_file, URL, General.Get_Title(URL))
Data_to_Cache.append(URL)
else:
logging.warning(General.Date() + " - " + __name__.strip('plugins.') + " - Invalid response.")
else:
logging.warning(General.Date() + " - " + __name__.strip('plugins.') + " - Invalid type provided.")
if Cached_Data:
General.Write_Cache(Directory, Data_to_Cache, Plugin_Name, "a")
else:
General.Write_Cache(Directory, Data_to_Cache, Plugin_Name, "w")
class HashTagSearchExample(HashTagSearch):
def __init__(self):
super().__init__()
self.total_posts = 0
def save_results(self, instagram_results):
super().save_results(instagram_results)
for i, post in enumerate(instagram_results):
self.total_posts += 1
print("%i - %s" % (self.total_posts, post.processed_text()))
import instagram_explore as ie
# Search tag name
res = ie.tag('cat')
print(res.data) # All
# Next page
data, cursor = ie.tag('cat', res.cursor)
# Image only
images = ie.tag_images('cat').data
import instagram_explore as ie
# Search location id
res = ie.location('213163910')
print(res.data)
print(
"run stalkSetup.py before the first time running this, only need to run it once"
)
import instagram_explore as ie
from instabot import Bot
import urllib
image = urllib.request.URLopener()
bot = Bot()
boobTag = input("From what hashtag? ")
result = ie.tag(boobTag)
bot.login(username="******", password="******")
boobCount = input("How many boobs?")
for i in range(int(boobCount)):
try:
sexy = result.data['edge_hashtag_to_top_posts']['edges'][i]['node'][
'thumbnail_src']
imgName = (str(i) + ".jpg")
image.retrieve(sexy, imgName)
except:
print("no worky worky")
def Search(Query_List, Task_ID, Type, **kwargs):
try:
Data_to_Cache = []
Directory = General.Make_Directory(Plugin_Name.lower())
logger = logging.getLogger()
logger.setLevel(logging.INFO)
Log_File = General.Logging(Directory, Plugin_Name.lower())
handler = logging.FileHandler(os.path.join(Directory, Log_File), "w")
handler.setLevel(logging.DEBUG)
formatter = logging.Formatter("%(levelname)s - %(message)s")
handler.setFormatter(formatter)
logger.addHandler(handler)
Cached_Data = General.Get_Cache(Directory, Plugin_Name)
Query_List = General.Convert_to_List(Query_List)
Limit = General.Get_Limit(kwargs)
for Query in Query_List:
if Type == "User":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.user(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response,
indent=4,
sort_keys=True)
Main_File = General.Main_File_Create(
Directory, Local_Plugin_Name, CSE_JSON_Output_Response,
Query, The_File_Extensions["Main"])
Posts = CSE_Response[0]["edge_owner_to_timeline_media"][
"edges"]
Output_Connections = General.Connections(
Query, Local_Plugin_Name,
"instagram.com", "Social Media - Person", Task_ID,
Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = f"https://www.instagram.com/p/{Shortcode}/"
Title = "IG | " + General.Get_Title(URL)
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(
Limit):
Response = requests.get(URL, headers=headers).text
Output_file = General.Create_Query_Results_Output_File(
Directory, Query, Local_Plugin_Name, Response,
Shortcode, The_File_Extensions["Query"])
if Output_file:
Output_Connections.Output([Main_File, Output_file],
URL, Title,
Plugin_Name.lower())
Data_to_Cache.append(URL)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Failed to create output file. File may already exist."
)
Current_Step += 1
elif Type == "Tag":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.tag(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response,
indent=4,
sort_keys=True)
Main_File = General.Main_File_Create(
Directory, Local_Plugin_Name, CSE_JSON_Output_Response,
Query, The_File_Extensions["Main"])
Posts = CSE_Response[0]["edge_hashtag_to_media"]["edges"]
Output_Connections = General.Connections(
Query, Local_Plugin_Name,
"instagram.com", "Social Media - Person", Task_ID,
Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = f"https://www.instagram.com/p/{Shortcode}/"
Title = "IG | " + General.Get_Title(URL)
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(
Limit):
Response = requests.get(URL, headers=headers).text
Output_file = General.Create_Query_Results_Output_File(
Directory, Query, Local_Plugin_Name, Response,
Shortcode, The_File_Extensions["Query"])
if Output_file:
Output_Connections.Output([Main_File, Output_file],
URL, Title,
Plugin_Name.lower())
Data_to_Cache.append(URL)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Failed to create output file. File may already exist."
)
Current_Step += 1
elif Type == "Location":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = location(Query)
CSE_JSON_Output_Response = json.dumps(CSE_Response,
indent=4,
sort_keys=True)
Main_File = General.Main_File_Create(
Directory, Local_Plugin_Name, CSE_JSON_Output_Response,
Query, The_File_Extensions["Main"])
Posts = CSE_Response[0]["edge_location_to_media"]["edges"]
Output_Connections = General.Connections(
Query, Local_Plugin_Name, "instagram.com",
"Social Media - Place", Task_ID, Local_Plugin_Name.lower())
Current_Step = 0
for Post in Posts:
Shortcode = Post["node"]["shortcode"]
URL = f"https://www.instagram.com/p/{Shortcode}/"
Title = "IG | " + General.Get_Title(URL)
if URL not in Cached_Data and URL not in Data_to_Cache and Current_Step < int(
Limit):
Response = requests.get(URL, headers=headers).text
Output_file = General.Create_Query_Results_Output_File(
Directory, Query, Local_Plugin_Name, Response,
Shortcode, The_File_Extensions["Query"])
if Output_file:
Output_Connections.Output([Main_File, Output_file],
URL, Title,
Plugin_Name.lower())
Data_to_Cache.append(URL)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Failed to create output file. File may already exist."
)
Current_Step += 1
elif Type == "Media":
Local_Plugin_Name = Plugin_Name + "-" + Type
CSE_Response = instagram_explore.media(Query)
if CSE_Response:
CSE_JSON_Output_Response = json.dumps(CSE_Response,
indent=4,
sort_keys=True)
Main_File = General.Main_File_Create(
Directory, Local_Plugin_Name, CSE_JSON_Output_Response,
Query, The_File_Extensions["Main"])
URL = f"https://www.instagram.com/p/{Query}/"
Title = "IG | " + General.Get_Title(URL)
if URL not in Cached_Data and URL not in Data_to_Cache:
Response = requests.get(URL, headers=headers).text
Output_file = General.Create_Query_Results_Output_File(
Directory, Query, Local_Plugin_Name, Response,
Shortcode, The_File_Extensions["Query"])
if Output_file:
Output_Connections = General.Connections(
Query, Local_Plugin_Name, "instagram.com",
"Social Media - Media", Task_ID,
Local_Plugin_Name.lower())
Output_Connections.Output([Main_File, Output_file],
URL, Title,
Plugin_Name.lower())
Data_to_Cache.append(URL)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Failed to create output file. File may already exist."
)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Invalid response."
)
else:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - Invalid type provided."
)
if Cached_Data:
General.Write_Cache(Directory, Data_to_Cache, Plugin_Name, "a")
else:
General.Write_Cache(Directory, Data_to_Cache, Plugin_Name, "w")
except Exception as e:
logging.warning(
f"{General.Date()} - {__name__.strip('plugins.')} - {str(e)}")