def post_hashtags_are_sfw(self, browser, log_path, timeout=5):
if self.running():
all_hashtags = [
"#" + x for x in self.extract_hash_tags(
browser=browser, log_path=log_path, timeout=timeout)
]
for hashtag in all_hashtags:
if hashtag in Config.nsfw_hashtags:
Log.update(
screenshot_path=self.screenshot_path,
browser=self.browser,
log_path=self.log_path,
text="This post contains the blacklisted hashtag {}.".
format(hashtag),
)
return False
Log.update(
screenshot_path=self.screenshot_path,
browser=self.browser,
log_path=self.log_path,
text=
"This post contains none of the blacklisted hashtags. (Hashtags: {})"
.format(", ".join(all_hashtags)),
)
return True
def like(self, browser, log_path, topic, timeout=5):
if self.running():
author = self.author(browser=browser, log_path=log_path)
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.like_button_xpath)))
like_button = WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.like_button_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in like')
return
like_button.click()
src = self.extract_picture_source(browser=browser,
log_path=log_path)
Log.update(self.screenshot_path,
self.browser,
log_path,
"Liked picture/video by: " + author,
image=src)
self.update_action_list(author=author,
action_type="like",
topic=topic)
def comment(self, topic, browser, log_path, timeout=5):
if self.running():
author = self.author(browser=browser, log_path=log_path)
query = Config.comments[randint(0, len(Config.comments) - 1)]
say = query.format(
author, Config.smileys[randint(0,
len(Config.smileys) - 1)])
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.comment_xpath)))
WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.comment_xpath)))
comment_button = WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.comment_submit_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in comment')
return
comment_button.click()
comment_field = browser.find_element_by_xpath(Config.comment_xpath)
comment_field.send_keys(say)
comment_field.send_keys(Keys.RETURN)
Log.update(
self.screenshot_path, self.browser, log_path,
"Commented on " + str(author) + "s picture with: " + say)
self.update_action_list(author=author,
action_type="comment",
topic=topic)
def already_liked(self, browser, log_path, error_timeout=5):
if self.running():
try:
WebDriverWait(browser, error_timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.like_button_full_xpath)))
except TimeoutException:
return False
Log.update(self.screenshot_path, self.browser, log_path,
'Post was already liked.')
return True
def author(self, browser, log_path, timeout=5):
if self.running():
try:
author_element = WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.author_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in author')
return
return str(author_element.get_attribute("title"))
def open_unfollow_screen(self, browser, log_path, timeout=15):
if self.running():
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.following_xpath)))
heart = WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.following_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in open_unfollow_screen')
return
heart.click()
def monitor(request):
if not request.user.is_authenticated:
return
try:
n = int(request.GET['n'])
except (MultiValueDictKeyError, ValueError):
n = 20
try:
search = request.GET['search']
except (MultiValueDictKeyError, ValueError):
search = ''
if not request.user.is_authenticated:
return
path = "static/img/" + request.user.username + "/screenshot.png"
try:
time = os.path.getmtime(path)
except FileNotFoundError:
time = 0
src = path + "?mtime=" + str(time)
# pages = range(Log.number_of_pages(page_size=page_size))
username = request.user.username
log_path = Config.bot_path + "log/" + username
path = log_path + "/log.pickle"
lines = Log.get(log_path=path, page_size=n, search=search)
return render(request, 'monitor.html', {'lines': lines, 'src': src})
def unfollow(self, browser, log_path, name, timeout=5):
if self.running():
browser.get("https://www.instagram.com/" + name + "/")
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.unfollow_xpath)))
unfollow_button = WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.unfollow_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in unfollow')
return
unfollow_button.click()
Log.update(self.screenshot_path, self.browser, log_path,
"Unfollowed: " + name)
def extract_picture_source(self, browser, log_path, timeout=5):
if self.running():
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.image_div_container_xpath)))
sections = browser.find_elements_by_xpath(
Config.image_div_container_xpath)
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'Exception in extract_picture_source: ' +
str(format_exc()))
return
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in extract_picture_source')
return
for section in sections:
try:
image = section.find_element_by_tag_name("img")
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'Exception in extract_picture_source: ' +
str(format_exc()))
return
return image.get_attribute("src")
def extract_hash_tags(self, browser, log_path, timeout=5):
if self.running():
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.hashtags_xpath)))
sections = browser.find_elements_by_xpath(
Config.hashtags_xpath)
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'Exception in extract_hash_tags: ' + str(format_exc()))
return []
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in extract_hash_tags')
return []
all_hashtags = []
for section in sections:
all_hashtags.extend(
set(part[1:] for part in section.text.split()
if part.startswith('#')))
return all_hashtags
def follow(self, browser, log_path, topic, timeout=15):
if self.running():
author = self.author(browser=browser, log_path=log_path)
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.follow_xpath)))
follow_button = WebDriverWait(browser, timeout).until(
ec.element_to_be_clickable(
(By.XPATH, Config.follow_xpath)))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in follow')
return
follow_button.click()
Log.update(self.screenshot_path,
browser=self.browser,
log_path=self.log_path,
text="Followed: " + author)
self.update_action_list(author=author,
action_type="follow",
topic=topic)
self.update_accounts_to_unfollow(author=author)
self.update_followed_accounts(author=author)
def select_first(self, browser, log_path, timeout=5):
if self.running():
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.first_ele_xpath)))
pictures = browser.find_elements_by_xpath(
Config.first_ele_xpath)
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'NoSuchElementException in select_first: ' +
str(format_exc()))
return
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in select_first')
return
if len(pictures) > 9:
first_picture = pictures[9]
else:
first_picture = pictures[len(pictures) - 1]
self.focus(first_picture, browser=browser)
first_picture.click()
def post_is_sfw(self, browser, log_path, limit=0.1):
if self.running():
if not self.post_hashtags_are_sfw(browser=browser,
log_path=log_path):
return False
image_url = self.extract_picture_source(browser=browser,
log_path=log_path)
if not image_url:
Log.update(screenshot_path=self.screenshot_path,
browser=self.browser,
log_path=self.log_path,
text="Picture source could not be extracted.")
return True
sfw, nsfw = classify_nsfw(image_url)
Log.update(
screenshot_path=self.screenshot_path,
browser=self.browser,
log_path=self.log_path,
text="Analysis of this post yielded it to be {}% SFW.".format(
int(100 * sfw)),
image=image_url)
return nsfw < limit
def table_monitor_update(request):
if not request.user.is_authenticated:
return
try:
n = int(request.GET['n'])
except (MultiValueDictKeyError, ValueError):
n = 20
try:
search = request.GET['search']
except (MultiValueDictKeyError, ValueError):
search = ''
username = request.user.username
log_path = Config.bot_path + "log/" + username
path = log_path + "/log.pickle"
lines = Log.get(log_path=path, page_size=n, search=search)
return render(request, 'table_monitor_update.html', {'lines': lines})
def login(self, username, password, browser, log_path, timeout=5):
if self.running():
Log.update(self.screenshot_path, self.browser, log_path,
"Logging in")
browser.get(Config.start_url)
try:
username_field = WebDriverWait(browser, timeout).until(
ec.presence_of_element_located((By.NAME, "username")))
pass_field = WebDriverWait(browser, timeout).until(
ec.presence_of_element_located((By.NAME, "password")))
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in login')
return
username_field.send_keys(username)
pass_field.send_keys(password)
pass_field.send_keys(Keys.RETURN)
Log.update(self.screenshot_path, self.browser, log_path,
"Logged in")
def check_follows(self, browser, log_path, timeout=15):
if self.running():
try:
WebDriverWait(browser, timeout).until(
ec.presence_of_element_located(
(By.XPATH, Config.sections_xpath)))
sections = browser.find_elements_by_xpath(
Config.sections_xpath)
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'NoSuchElementException in check_follows: ' +
str(format_exc()))
return
except TimeoutException:
Log.update(self.screenshot_path, self.browser, log_path,
'Timeout in check_follows')
return
users = []
for element in sections:
try:
profile = element.find_element_by_xpath(
Config.local_name_xpath)
except NoSuchElementException:
Log.update(
self.screenshot_path, self.browser, log_path,
'NoSuchElementException in check_follows: ' +
str(format_exc()))
return
name = profile.get_attribute("title")
users.append(name)
for user in users:
if user not in self.interacting_users:
if user in self.action_list.keys():
actions = self.action_list[user]
for action in actions:
self.hashtags[action["topic"]] += 1
self.update_interacting_users(user=user)
def run(self):
self.login(browser=self.browser,
log_path=self.log_path,
password=self.password,
username=self.username)
while self.running():
try:
self.open_unfollow_screen(browser=self.browser,
log_path=self.log_path)
self.check_follows(browser=self.browser,
log_path=self.log_path)
top_hashtags = sorted(self.hashtags.keys(),
key=lambda k: self.hashtags[k],
reverse=True)[:20]
for i, topic in enumerate(top_hashtags):
self.search(query=topic,
browser=self.browser,
log_path=self.log_path)
self.select_first(browser=self.browser,
log_path=self.log_path)
delay, action = self.dispatcher.next_action()
Log.update(
self.screenshot_path, self.browser, self.log_path,
"Dispatcher selected action: {} (Sleeping {}s)".format(
action, delay))
sleep(delay)
if action == "comment":
if self.post_is_sfw(browser=self.browser,
log_path=self.log_path):
self.comment(topic=topic,
browser=self.browser,
log_path=self.log_path)
self.dispatcher.log_action("comment")
self.store_hashtags(browser=self.browser,
log_path=self.log_path)
elif action == "like":
count = 0
while self.already_liked(browser=self.browser,
log_path=self.log_path):
if not self.on_dialog_page(self.browser,
self.log_path):
break
if count > 10:
break
self.next_picture(browser=self.browser)
count += 1
if self.on_dialog_page(self.browser, self.log_path):
if self.post_is_sfw(browser=self.browser,
log_path=self.log_path):
self.like(topic=topic,
browser=self.browser,
log_path=self.log_path)
self.dispatcher.log_action("like")
self.store_hashtags(browser=self.browser,
log_path=self.log_path)
elif action == "follow":
count = 0
while self.user_followed_already(
self.author(browser=self.browser,
log_path=self.log_path)):
if not self.on_dialog_page(self.browser,
self.log_path):
break
if count > 10:
break
self.next_picture(browser=self.browser)
count += 1
if self.on_dialog_page(self.browser, self.log_path):
if self.post_is_sfw(browser=self.browser,
log_path=self.log_path):
self.follow(topic=topic,
browser=self.browser,
log_path=self.log_path)
self.dispatcher.log_action("follow")
self.store_hashtags(browser=self.browser,
log_path=self.log_path)
elif action == "unfollow":
if len(self.accounts_to_unfollow) > 50:
this_guy = self.accounts_to_unfollow[0]
self.unfollow(name=this_guy,
browser=self.browser,
log_path=self.log_path)
del self.accounts_to_unfollow[0]
self.dispatcher.log_action("unfollow")
except Exception:
Log.update(self.screenshot_path,
self.browser,
self.log_path,
text='General Exception: ' + str(format_exc()))
Log.update(self.screenshot_path,
self.browser,
self.log_path,
text='Stopped bot')
if self.vdisplay:
self.vdisplay.stop()
super(Driver, self).join()
def search(self, browser, log_path, query):
if self.running():
browser.get("https://www.instagram.com/explore/tags/" + query +
"/")
Log.update(self.screenshot_path, self.browser, log_path,
"Searching for " + query + ".")