async def spamscan_classify(e):
""" Feed the algorithm by classifying a user either
as spam or ham """
args, user = parse_arguments(e.pattern_match.group(1), ['forward', 'reason'])
reason = args.get('reason', 'spam[gban]')
args['forward'] = args.get('forward', True)
args['user'] = user
await e.edit(CLASSIFYING_MESSAGE.format("**Fetching user information.**"))
replied_user = await get_user_from_event(e, **args)
if not replied_user:
await e.edit("**Failed to get information for user**", delete_in=3)
return
me = await bot.get_me()
if replied_user.user == me:
await e.edit("**Can't flag yourself as spam**", delete_in=3)
return
hashes = await gather_profile_pic_hashes(e, replied_user.user)
if hashes:
await e.edit(CLASSIFYING_MESSAGE.format("**Adding profile pic hashes to DB**"))
for hsh in hashes:
await add_file_hash(hsh, 'profile pic')
if spamwatch:
await e.edit(CLASSIFYING_MESSAGE.format("**Checking spamwatch.**"))
gbanned = spamwatch.get_ban(replied_user.user.id)
if not gbanned:
await e.edit(CLASSIFYING_MESSAGE.format("**Adding to SpamWatch.**"))
try:
spamwatch.add_ban(replied_user.user.id, reason)
except UnauthorizedError:
pass
await e.edit(f"**Flagged** {make_mention(replied_user.user)} **as spam**\n"
f"**Reason:** {reason}")
async def fetch_info(replied_user, **kwargs):
""" Get details from the User object. """
user = replied_user.user
id_only = kwargs.get('id', False)
show_general = kwargs.get('general', True)
show_bot = kwargs.get('bot', False)
show_misc = kwargs.get('misc', False)
show_all = kwargs.get('all', False)
mention_name = kwargs.get('mention', False)
if show_all:
show_general = True
show_bot = True
show_misc = True
full_name = str(user.first_name + ' ' + (user.last_name or ''))
if mention_name:
title = Link(full_name, f'tg://user?id={user.id}')
else:
title = Bold(full_name)
if id_only:
return KeyValueItem(title, Code(user.id))
general = SubSection(
Bold('general'), KeyValueItem('id', Code(user.id)),
KeyValueItem('first_name', Code(user.first_name)),
KeyValueItem('last_name', Code(user.last_name)),
KeyValueItem('username', Code(user.username)),
KeyValueItem('mutual_contact', Code(user.mutual_contact)),
KeyValueItem('common groups', Code(replied_user.common_chats_count)))
if spamwatch:
banobj = spamwatch.get_ban(user.id)
if banobj:
general.items.append(
KeyValueItem('gbanned', f'True / {banobj.reason}'))
else:
general.items.append(KeyValueItem('gbanned', 'False'))
bot = SubSection(
Bold('bot'), KeyValueItem('bot', Code(user.bot)),
KeyValueItem('bot_chat_history', Code(user.bot_chat_history)),
KeyValueItem('bot_info_version', Code(user.bot_info_version)),
KeyValueItem('bot_inline_geo', Code(user.bot_inline_geo)),
KeyValueItem('bot_inline_placeholder',
Code(user.bot_inline_placeholder)),
KeyValueItem('bot_nochats', Code(user.bot_nochats)))
misc = SubSection(
Bold('misc'), KeyValueItem('restricted', Code(user.restricted)),
KeyValueItem('restriction_reason', Code(user.restriction_reason)),
KeyValueItem('deleted', Code(user.deleted)),
KeyValueItem('verified', Code(user.verified)),
KeyValueItem('min', Code(user.min)),
KeyValueItem('lang_code', Code(user.lang_code)))
return Section(title, general if show_general else None,
misc if show_misc else None, bot if show_bot else None)
async def _collect_user_info(self, client, user,
**kwargs) -> Union[Section, KeyValueItem]:
id_only = kwargs.get('id', False)
show_general = kwargs.get('general', True)
show_bot = kwargs.get('bot', False)
show_misc = kwargs.get('misc', False)
show_all = kwargs.get('all', False)
if show_all:
show_general = True
show_bot = True
show_misc = True
mention_name = kwargs.get('mention', False)
full_name = await helpers.get_full_name(user)
if mention_name:
title = Link(full_name, f'tg://user?id={user.id}')
else:
title = Bold(full_name)
if spamwatch:
ban = spamwatch.get_ban(user.id)
ban_reason = ban.reason if ban else None
else:
ban_reason = None
if id_only:
return KeyValueItem(title, Code(user.id))
else:
general = SubSection(
Bold('general'), KeyValueItem('id', Code(user.id)),
KeyValueItem('first_name', Code(user.first_name)),
KeyValueItem('last_name', Code(user.last_name)),
KeyValueItem('username', Code(user.username)),
KeyValueItem('mutual_contact', Code(user.mutual_contact)),
KeyValueItem('ban_reason', Code(ban_reason))
if ban_reason else KeyValueItem('gbanned', Code('False')))
ibot = SubSection(
Bold('bot'), KeyValueItem('bot', Code(user.bot)),
KeyValueItem('bot_chat_history', Code(user.bot_chat_history)),
KeyValueItem('bot_info_version', Code(user.bot_info_version)),
KeyValueItem('bot_inline_geo', Code(user.bot_inline_geo)),
KeyValueItem('bot_inline_placeholder',
Code(user.bot_inline_placeholder)),
KeyValueItem('bot_nochats', Code(user.bot_nochats)))
misc = SubSection(
Bold('misc'), KeyValueItem('restricted',
Code(user.restricted)),
KeyValueItem('restriction_reason',
Code(user.restriction_reason)),
KeyValueItem('deleted', Code(user.deleted)),
KeyValueItem('verified', Code(user.verified)),
KeyValueItem('min', Code(user.min)),
KeyValueItem('lang_code', Code(user.lang_code)))
return Section(title, general if show_general else None,
misc if show_misc else None,
ibot if show_bot else None)
async def score_user(event, userfull):
""" Give a user a spam score based on several factors """
user = userfull.user
# Everyone starts with a score of 0. A lower score indicates
# a lower chance of being a spammer. A higher score
# indicates the opposite.
score = {}
hashes = await gather_profile_pic_hashes(event, user)
total_hashes = len(hashes)
matching_hashes = 0
# User was flagged as a scammer
if user.scam:
score.update({'flagged as scammer': 5})
# User is restricted
if user.restricted:
score.update({'restricted': 3})
# No profile pic is a +2
if total_hashes == 0:
score.update({'no profile pic': 2})
# A single profile pic can also be a red flag
elif total_hashes == 1:
score.update({'single profile pic': 2})
# If all the profile pics are the same that's another red flag
elif total_hashes >= 2 and len(set(hashes)) == 1:
score.update({'profile pics same': 2})
if matching_hashes > 0:
# If there are matching hashes that's an automatic +5
score.update(
{f'blacklisted photos ({total_hashes}/{matching_hashes})': 5})
# Lots of spammers try and look normal by having a normal(ish)
# first and last name. A first AND last name with no special
# characters is a good indicator. This is a +1.
if ((user.first_name and re.match(r"^[a-zA-Z0-9\s_]+$", user.first_name))
and
(user.last_name and re.match(r"^[a-zA-Z0-9\s_]+$", user.last_name))):
score.update({'alphanum first and last name': 1})
if user.first_name and user.last_name:
# Another thing spammers seem to have is very predictable names.
# These come in many forms like one uppercase name and one
# lowercase, all upper or lower, or having one name be
# numeric. Either way they generally have a first
# name and a last name.
if user.first_name.isupper() and user.first_name.islower():
score.update({'first upper last lower': 3})
elif user.last_name.isupper() and user.last_name.islower():
score.update({'first lower last upper': 3})
elif user.first_name.islower() and user.last_name.islower():
# This appears less bot like than all upper
score.update({'lowercase name': 2})
elif user.first_name.isupper() and user.last_name.isupper():
score.update({'uppercase name': 3})
elif user.first_name.isnumeric() or user.last_name.isnumeric():
score.update({'numeric name': 2})
# Another popular thing is bots with japanese, chinese, cyrillic,
# and arabic names. A full match here is worth +3.
if (user.first_name and is_cjk(user.first_name)
or (user.last_name and is_cjk(user.last_name))):
score.update({'ch/jp name': 3})
elif (user.first_name and is_arabic(user.first_name)
or (user.last_name and is_arabic(user.last_name))):
score.update({'arabic name': 3})
elif (user.first_name and is_cyrillic(user.first_name)
or (user.last_name and is_cyrillic(user.last_name))):
# Cyrillic names are more common, so we'll drop the score here.
score.update({'cyrillic name': 2})
if userfull.about and is_cjk(userfull.about):
score.update({'ch/jp bio': 2})
elif userfull.about and is_arabic(userfull.about):
score.update({'arabic bio': 2})
elif userfull.about and is_cyrillic(userfull.about):
score.update({'cyrillic bio': 2})
# A username ending in numbers is a +1
if user.username:
if re.match(r".*[0-9]+$", user.username):
score.update(({'sequential username': 2}))
else:
score.update({'no username': 2})
if userfull.about:
# Check the bio for red flag words. Each one of these is a +3.
total_red_flags = 0
for word in REDFLAG_WORDS:
if word in userfull.about.lower():
total_red_flags += 1
if total_red_flags > 0:
score.update(
{f'red flag words x{total_red_flags}': total_red_flags * 3})
# No bio is also an indicator worth an extra 2 points
else:
score.update({'no bio': 2})
# Check if this person is banned in spamwatch. This is
# basically a guarantee, and therefore nets a +5.
if spamwatch:
spamwatch_ban = spamwatch.get_ban(user.id)
if spamwatch_ban:
score.update({f'spamwatch ({spamwatch_ban.reason.lower()})': 5})
return score
