def test_update_location(id):
anon = Agent()
location = Location(id)
data = anon.update(location)
Location.clear_cache()
def test_update_tag(name):
anon = Agent()
tag = Tag(name)
data = anon.update(tag)
Tag.clear_cache()
def test_update():
anon = Agent()
anon.update()
assert (not getattr(anon, "_rhx_gis", None) is None)
assert (not getattr(anon, "_csrf_token", None) is None)
def profile_exists(username):
try:
agent = Agent()
account = Account(username)
agent.update(account)
return True
except:
return False
def test_get_likes(shortcode):
anon = Agent()
media = Media(shortcode)
data, pointer = anon.get_likes(media)
assert (media.likes_count >= len(data))
Media.clear_cache()
def test_update_photo_set(shortcode):
anon = Agent()
photo_set = Media(shortcode)
data = anon.update(photo_set)
assert (not photo_set.is_video)
Media.clear_cache()
def test_update_video(shortcode):
anon = Agent()
video = Media(shortcode)
data = anon.update(video)
assert (video.is_video)
Media.clear_cache()
def test_get_comments_long(count, shortcode):
anon = Agent()
media = Media(shortcode)
data, pointer = anon.get_comments(media, count=count)
assert (min(media.comments_count, count) == len(data))
assert ((pointer is None) == (media.likes_count <= count))
Media.clear_cache()
def test_get_media_account(count, username):
anon = Agent()
account = Account(username)
data, pointer = anon.get_media(account, count=count)
assert (min(account.media_count, count) == len(data))
assert ((pointer is None) == (account.media_count <= count))
Account.clear_cache()
Media.clear_cache()
def test_get_media_location_long(count, id):
anon = Agent()
location = Location(id)
data, pointer = anon.get_media(location, count=count)
assert (min(location.media_count, count) == len(data))
assert ((pointer is None) == (location.media_count <= count))
Location.clear_cache()
Media.clear_cache()
def test_get_media_tag_long(count, name):
anon = Agent()
tag = Tag(name)
data, pointer = anon.get_media(tag, count=count)
assert (min(tag.media_count, count) == len(data))
assert ((pointer is None) == (tag.media_count <= count))
Tag.clear_cache()
Media.clear_cache()
def test_get_comments_pointer(count, shortcode):
anon = Agent()
media = Media(shortcode)
pointer = None
data = []
for i in range(count):
tmp, pointer = anon.get_comments(media, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (media.likes_count <= count))
Media.clear_cache()
def test_get_media_location_pointer(count, id):
anon = Agent()
location = Location(id)
pointer = None
data = []
for i in range(count):
tmp, pointer = anon.get_media(location, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (location.media_count <= count))
Location.clear_cache()
Media.clear_cache()
def test_get_media_tag_pointer(count, name):
anon = Agent()
tag = Tag(name)
pointer = None
data = []
for i in range(count):
tmp, pointer = anon.get_media(tag, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (tag.media_count <= count))
Tag.clear_cache()
Media.clear_cache()
def test_get_media_account_pointer(count, username):
anon = Agent()
account = Account(username)
pointer = None
data = []
for i in range(count):
tmp, pointer = anon.get_media(account, pointer=pointer)
data.extend(tmp)
assert ((pointer is None) == (account.media_count <= count))
Account.clear_cache()
Media.clear_cache()
def get_post_comments(post, num=None, path=None):
agent = Agent()
comments = set()
pointer = None
if num == None:
comments_count = int(post.comments_count)
else:
comments_count = num
limit = 50
batch_num = math.ceil(comments_count / limit)
for i in range(batch_num):
if i == batch_num - 1:
count = comments_count - limit * (batch_num - 1)
batch_comments, pointer = agent.get_comments(post,
pointer=pointer,
count=count)
else:
batch_comments, pointer = agent.get_comments(post,
pointer=pointer,
count=limit)
for j, item in enumerate(batch_comments):
comments.add(Comment(item.id))
comments_info = {}
for i, item in enumerate(comments):
comment_info = copy.copy(item)
comment_info.media = str(comment_info.media)
comment_info.owner = str(comment_info.owner)
comments_info[i] = comment_info.__dict__
comments_dict = {"comments": comments_info}
comments_json = json.dumps(comments_dict, indent=2)
if path == None:
path = './data'
pathlib.Path(path + '/comments').mkdir(parents=True, exist_ok=True)
postcode = post.code
filename = path + '/comments/' + postcode + '__last_comments.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(comments_json)
return comments
def get_wave_warned(posts, users_profiles):
agent = Agent()
warned = {}
for post in posts:
warned.update(
InstaHelper.get_warned_by_post(agent, post, users_profiles))
return warned
def get_posts_by_tag_name(tagname, num=None, path=None):
agent = Agent()
agent.update(Tag(tagname))
tag = Tag(tagname)
media = set()
pointer = None
if num == None:
media_count = tag.media_count
else:
media_count = num
limit = 50
batch_num = math.ceil(media_count / limit)
for i in range(batch_num):
if i == batch_num - 1:
count = media_count - limit * (batch_num - 1)
batch_media, pointer = agent.get_media(tag,
pointer=pointer,
count=count)
else:
batch_media, pointer = agent.get_media(tag,
pointer=pointer,
count=limit)
for j, item in enumerate(batch_media):
print("Getting media: " + str(i * 50 + j + 1) + " / " +
str(media_count))
agent.update(Media(item.code))
media.add(Media(item.code))
media_posts = {}
for i, item in enumerate(media):
post_info = copy.copy(item)
post_info.likes = dict(post_info.likes)
post_info.comments = dict(post_info.comments)
post_info.location = str(post_info.location)
media_posts[i] = post_info.__dict__
media_dict = {"posts": media_posts}
media_json = json.dumps(media_dict, indent=2)
print(media_json)
if path == None:
path = './data/tag__' + tagname
pathlib.Path(path).mkdir(parents=True, exist_ok=True)
filename = path + '/tag__' + tagname + '__last_posts.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(media_json)
return media
def get_posts_for_wave(usernames):
links = []
posts = []
agent = Agent()
for user in usernames:
post, link = InstaHelper._get_last_post_link(agent, user)
posts.append(post)
links.append(link)
return [str(x) for x in posts], links
def get_picture(post, path):
agent = Agent()
name = post.owner.login
picture = post.display_url
pathlib.Path(path + '/pictures').mkdir(parents=True, exist_ok=True)
picname = path + '/pictures/' + name + '__' + post.code + '.jpg'
urllib.request.urlretrieve(picture, picname)
return picture
def get_tag_info(tagname, path=None):
agent = Agent()
agent.update(Tag(tagname))
tag = Tag(tagname)
tag_info = copy.copy(tag)
tag_info.media = dict(tag_info.media)
tag_dict = {"tag": tag_info.__dict__}
tag_json = json.dumps(tag_dict, indent=2)
if path == None:
path = './data/tag__' + tagname
pathlib.Path(path).mkdir(parents=True, exist_ok=True)
filename = path + '/tag__' + tagname + '__tag_info.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(tag_json)
return tag
def test_update_account(username):
anon = Agent()
account = Account(username)
data = anon.update(account)
assert (not data is None)
assert (not account.id is None)
assert (not account.full_name is None)
assert (not account.profile_pic_url is None)
assert (not account.profile_pic_url_hd is None)
assert (not account.biography is None)
assert (not account.follows_count is None)
assert (not account.followers_count is None)
assert (not account.media_count is None)
assert (not account.is_private is None)
assert (not account.is_verified is None)
assert (not account.country_block is None)
Account.clear_cache()
def get_account_info(username, path=None):
agent = Agent()
agent.update(Account(username))
account = Account(username)
account_info = copy.copy(account)
account_info.media = dict(account_info.media)
account_info.follows = dict(account_info.follows)
account_info.followers = dict(account_info.followers)
account_dict = {"account": account_info.__dict__}
account_json = json.dumps(account_dict, indent=2)
if path == None:
path = './data/' + username
pathlib.Path(path).mkdir(parents=True, exist_ok=True)
filename = path + '/' + username + '__account_info.json'
with open(filename, 'w', newline='', encoding='utf8') as f:
f.write(account_json)
return account
def agent(settings):
return Agent(settings=settings)
import os
import instaparser
from instaparser.agents import Agent
from instaparser.entities import Tag, Media
def preprocess_text(text):
text = text.lower().replace("ё", "е")
text = re.sub('((www\.[^\s]+)|(https?://[^\s]+))', 'URL', text)
text = re.sub('@[^\s]+', 'USER', text)
text = re.sub('#[^\s]+', 'TAG', text)
text = re.sub('[^a-zA-Zа-яА-Я1-9]+', ' ', text)
text = re.sub('[a-z]+', ' ', text)
text = re.sub('\n', ' ', text)
return text.strip()
agent = Agent()
def get_comments_by_tag(word, numberOfPost):
try:
os.mkdir("{}".format(word))
except FileExistsError:
pass
tag = Tag(word)
medias = agent.get_media(obj=tag, count = numberOfPost ) #get tuple with post identifiers and tag identifier
counter = 0
for media in medias[0]:
comments = ""
post = agent.update(obj=Media(media))#get post with comments
for t in post["edge_media_to_comment"]["edges"]:
comment = str(t["node"]["text"]) #create str from comments
comment = preprocess_text(comment)
def has_post(username):
agent = Agent()
account = Account(username)
media1, _ = agent.get_media(account)
return len(media1) != 0
def is_profile_private(username):
agent = Agent()
account = Account(username)
agent.update(account)
return account.is_private
from random import random
from time import sleep
from openpyxl import Workbook
from collections import namedtuple
from datetime import datetime
#VIPS = ['pryanya93','ksenia_lykina_tennis','golubovskaya_polina','pribylova96','lera.solovyeva','pivovarovaofficial',
# 'anastasia__frolova','kasatkina','polinaleykina']
VIPS = [
'pivovarovaofficial', 'anastasia__frolova', 'kasatkina', 'polinaleykina'
]
#VIP = 'mariasharapova'
#VIP = 'katemakarova1'
FRIENDS = 20 # Количество друзей с наибольшей перепиской
agent = Agent()
#exception_manager.repeats = 10 # Количество повторов при ошибке - поставил в библиотеке
for vip in VIPS:
print(datetime.now().strftime("%H:%M:%S"), vip,
'------------------------------')
account = Account(vip)
all_medias_codes = []
all_medias = []
try:
medias, pointer = agent.get_media(account)
except Exception as e:
print(datetime.now().strftime("%H:%M:%S"), 'Сбой')
medias, pointer = agent.get_media(account)
for media in medias:
