def c3():
x = '1b37373331363f78151b7f2b783431333d78397828372d363c78373e783a393b3736'
tmp = binascii.unhexlify(x)
for i in string.printable:
y = ''.join([chr(ord(a)^ord(b)) for (a,b) in zip(tmp, i*len(tmp))])
if(all(c in string.printable for c in y) and not nonsense(y)):
print(i,y)
def lang_features(story_sentences: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
lang_list = []
wtl = WhatTheLang()
for sent_dict in story_sentences:
text = sent_dict["text"]
try:
lang = wtl.predict_lang(text)
if not isinstance(lang, str):
lang = "UKN"
except:
lang = "UKN"
try:
if len(text) <= 10:
is_nonsense = False
else:
is_nonsense = nonsense(text)
except:
is_nonsense = True
is_eng = isAscii(text)
lang_dict = dict(sentence_id=sent_dict["id"], lang=lang, nonsense=is_nonsense, ascii_chars=is_eng)
lang_list.append(lang_dict)
return lang_list
def c3(x):
tmp = binascii.unhexlify(x)
res = []
for i in string.printable:
y = ''.join([chr(a^ord(b)) for (a,b) in zip(tmp, i*len(tmp))])
if(all(c in string.printable for c in y) and not nonsense(y)):
res.append((i,y.strip()))
return res
def is_meaningful(
sentence
): # exploits external library to determine if a sentence is nonsense/jibberish
if len(sentence.replace(" ", "")) < 6:
return False
try:
if nonsense(sentence):
return False
except:
return False
return True
def _test():
print(offensiveness("F**k you"))
print(offensiveness("Little bitch"))
print(offensiveness("You can find a coffee machine on the second floor"))
print(
offensiveness(
"You're so dumb you can't even find a stupid coffee machine"))
print(offensiveness("Can I fire that bitch, Charles?"))
print(offensiveness("Where can I find a coffee machine?"))
print(offensiveness("How can I file a complaint?"))
# Models aren't perfect, this model does not cover all dialects
print(offensiveness("Ray is a c**t"))
# function nonsense returns true if a string has no meaning
print(nonsense("This should return false."))
print(nonsense("ZkIOMJSIOJEKLMZKJMELLKS"))
# Even when concatenating words or using more complicated codes, the model can usually recognize meaningful strings.
print(nonsense("ioFlXFndrInfo"))
# according to the documentation the accuracy is 99%
# text has to be long enough otherwise a ValueError is raised
print(nonsense("t2shrt"))
def ratings(text):
table = []
for word in text.split():
if not word:
continue
if ignore(word):
row = [word, '--', '--']
else:
row = [
word, 'y' if in_dictionary(word) else 'n',
'n' if nonsense(word) else 'y'
]
table.append(row)
return table
def is_nonsense(sentence):
"""
Checks if a sentence is nonsense or not.
Args:
sentence: The string that is to be checked.
Returns: True if the string is not nonsense and it contains more than 6 characters
"""
try:
return nonsense(sentence)
except ValueError:
return False
def main():
res = {}
cdump_path = "/media/tong/Elements/obfs/class-dump"
file_gen = "res.json"
cnt = 0
for dirpath, dirnames, ifilenames in os.walk(cdump_path):
for ifilename in ifilenames:
file_path = os.path.abspath(os.path.join(dirpath, ifilename))
# test from file
try:
with open(file_path) as frh:
c_nonsense = 0
c_real = 0
c_total = 0
for s in frh.readlines():
s = s.strip()
if s.startswith("-") == False and s.startswith("+") == False :
continue
try:
if nonsense(s):
c_nonsense += 1
#print(s)
# print(s)
else:
#print(s)
c_real += 1
c_total += 1
except:
pass
# print(c_total)
if c_nonsense + c_real == 0:
continue
else:
res[ifilename] = float(c_real) / (c_nonsense + c_real)
cnt += 1
print("{}: {} {}".format(cnt, ifilename, res[ifilename]))
except:
continue
js_fwh = open(file_gen, 'w', encoding='utf-8')
json.dump(res, js_fwh)
def lookup_tokens(token_text):
add_token = False
if self._py_dictionary.meaning(token_text) != None:
print("Py Dictionary", token_text)
add_token = True
else:
try:
if not nonsense(token_text):
print("Not Nonsense", token_text)
add_token = True
except:
pass
if add_token:
self._allowed_tokens.add(token_text)
self._allowed_to_insert.append({"token": token_text})
self._tried_tokens.add(token_text)
self._tried_to_insert.append({"token": token_text})
async def on_message(self, message):
"""
Event message
Args:
message (discord.Message): Message content
"""
# * Vars
now = datetime.now()
if message.author.id == self.bot.user.id:
return
if message.channel.type is discord.ChannelType.private:
try:
if nonsense(message.content):
await message.channel.send(
'<:Denied:807575178561191976> - Please provide a good text!'
)
else:
jmail_channel = self.bot.get_channel(
Guild.mailling_channel)
mail_embed = discord.Embed(
title=f'New mail from {message.author.name}',
color=0x2c2f33)
mail_embed.add_field(name='Content:',
value=f'```{message.content}```',
inline=False)
mail_embed.set_footer(
text=now.strftime("%m/%d/%Y, %H:%M:%S"),
icon_url=message.author.avatar_url)
await jmail_channel.send(embed=mail_embed)
except ValueError:
await message.channel.send(
'<:Denied:807575178561191976> - Error, your text is to smart.'
)
else:
pass
def check_if_obfuscated(tmp_prog_arg):
print("start to check obfuscated")
extract_header_file(tmp_prog_arg)
tmp_header_file = open(tmp_prog_arg.extract_header_file_path)
tmp_nonsense = 0
tmp_real = 0
for tmp_s in tmp_header_file.readlines():
#if(tmp_s.find("*/ ")==-1):
# continue
#tmp_s = tmp_s[tmp_s.find("*/ ")+3:]
#if(tmp_s.find("// ") == -1):
# print("header file extract not correct")
# sys.exit()
#tmp_s = tmp_s[:tmp_s.find("// ")]
tmp_s = tmp_s.replace("\n", "")
for tmp_i in range(10):
tmp_s = tmp_s.replace(str(tmp_i), '')
if (len(tmp_s) <= 6):
continue
#print(tmp_s)
#print(len(tmp_s))
if nonsense(tmp_s):
tmp_nonsense = tmp_nonsense + 1
else:
tmp_real = tmp_real + 1
if tmp_nonsense + tmp_real == 0:
print("didn't find useful name")
sys.exit()
tmp_result = float(tmp_real) / (tmp_nonsense + tmp_real)
if tmp_result < 0.9:
print("this application is obfuscated")
tmp_prog_arg.my_app_info.is_obfuscated = True
else:
tmp_prog_arg.my_app_info.is_obfuscated = False
ct_numbers = []
for letter in ct:
number = ord(letter) - 65
ct_numbers.append(number)
a_inv = []
b = []
for i in range(26):
b.append(i)
for i in range(1, 26, 2):
inv = modinv(i, 26)
if inv is not None:
a_inv.append(inv)
pt = ""
for i in b:
for j in a_inv:
for number in ct_numbers:
if number >= 0:
pt_letter = (j * (number - i)) % 26
pt += str(chr(pt_letter + 65))
else:
pt += " "
if not nonsense(pt):
print("PT: {} -> A_inv: {}, B: {}".format(pt, j, i))
pt = ""
async def check_manually(message):
# This command used to test new antispammers AI functions
user, txt = await aio_get_user(message, allow_self=True)
if not user:
return
user_id = user['user_id']
name = user['first_name']
user_pics = await tbot(GetUserPhotosRequest(
int(user['user_id']),
offset=0,
max_id=0,
limit=100))
if user['last_name']:
name += user['last_name']
num = 0
text = "User " + await user_link_html(user['user_id'])
text += "\nName: " + name
text += "\nID: <code>" + str(user['user_id']) + '</code>'
text += '\n'
gbanned = mongodb.blacklisted_users.find_one({'user': user_id})
if gbanned:
text += "\n<b>Warn! User gbanned in SophieBot!</b>"
text += f"\nDate: <code>{gbanned['date']}</code>"
text += f"\nReason: <code>{gbanned['reason']}</code>"
text += '\n'
num += 999
else:
text += "\nUser not gbanned in SophieBot"
api_url = "https://api.unifiedban.solutions/blacklist/check/" + str(user_id)
ubanned = requests.get(api_url, headers={'Authorization': CONFIG['advanced']['utoken']})
if ubanned.text == '{"Error": "No data"}':
text += "\nUser not ubanned."
if user['first_name'].replace(' ', '').isdigit():
text += "\n<b>Warn! User have name with only numbers!</b>"
num += 80
if user['first_name'].lower() in NAMES:
text += "\n<b>Warn! User have real name (Mostly spammers try to be like real human)!</b>"
num += 75
if user_pics and len(user_pics.photos) == 1:
text += "\n<b>Warn! User have only 1 display picture!</b>"
num += 40
if user_pics and len(user_pics.photos) == 0:
text += "\n<b>Warn! User don't have any DP!</b>"
num += 25
try:
check = nonsense(name)
if check is True:
text += "\n<b>Warn! User have noncence name!</b>"
num += 85
else:
text += "\nUser have normal name"
except ValueError:
text += "\nName too short to analyse it"
# Counterweight
if '#' in name:
text += "\nUser have hashtag in name, mostly only real users have it"
num -= 20
if "☭" in name:
text += "\nGood soveit boi."
num -= 20
if "🌈" in name:
text += "\nGei detected."
num -= 20
if "🦊" in name:
text += "\nHa, this guy is a fox lover."
num -= 20
for owo in COUNTRY_EMOJIS:
if owo in name:
text += "\nHa, This guy love own country"
num -= 20
break
#
text += "\n\nDebug: Real suspicion numer: " + str(num)
if num > 100:
num = 100
if num < 0:
num = 0
text += '\n\n<b>Suspicion: </b><code>' + str(num) + "%</code>"
await message.reply(str(text))
for url in urls_list:
try:
response = requests.get(url).text
for match in get_varnames(response):
params.add(match)
except:
pass
else:
input_data = ' '.join(read_in())
for match in get_varnames(input_data):
params.add(match)
if args.smart:
print('using smart detection')
for param in params:
regex = re.compile(
'[^a-zA-Z]'
) # pattern matches only alpha because nostril only works on alpha
nostril_safe_param = regex.sub('', param)
if len(nostril_safe_param) > 6:
if not nonsense(nostril_safe_param):
print(param)
else:
print(
param
) # print everything under 7 chars, nostril won't work on anything shorter.
else:
for param in params:
print(param)
def classifier(stream_strings):
"""Rule based classification for the input string
Args:
str: A string to be processed
Returns:
str: category/ class of the string
"""
# Augment the string with splits and elimination of spaces and changing to lower case
stream_strings_org = stream_strings
stream_strings = stream_strings.lower()
stream_strings = ' '.join(stream_strings.split()).replace(',', '')
stream_strings_space = stream_strings.split(' ')
stream_strings_mod = list(
set(stream_strings_space + [stream_strings] + [stream_strings_org]))
# Variables for rule based matching
ents = []
in_vocab = []
poss = {}
alphas = {}
like_nums = {}
for token in nlp(' '.join(stream_strings_org.split()).lower(
)): #Iterate through string after converting into a spacy doc
poss[token.text] = str(token.pos_) #Part of Speech for the token
alphas[token.text] = str(token.is_alpha) #Check if token is alphabetic
like_nums[token.text] = str(token.like_num) #Check if token is numeric
if nlp.vocab.strings[token.text] in nlp.vocab:
in_vocab.append(
token.text) #Check if token is present in vocabulary
for stream_string in stream_strings_mod: # Check the entity labels
doc = nlp(stream_string)
for ent in doc.ents:
if ent.label_ not in ents:
ents.append(ent.label_)
ents = list(set(ents))
#Rule for serial number
if (len(ents) == 0 and len(in_vocab) == 0) or nonsense(
''.join(stream_strings_org.lower().split()) *
3) == True or ''.join(
stream_strings_org.split()).isnumeric() == True:
return 'serial_number'
#Rule for physical good/ Product
if (len(nlp(' '.join(stream_strings_org.split())))
== len(in_vocab)) and (len(ents) == 0 or 'PRODUCT' in ents):
return 'product'
#Rule for Address
if ('CARDINAL' in ents) or ('FAC' in ents):
return 'address'
#Rule for Geographic location
if ('GPE' in ents or 'LOC' in ents) and ('ORG' not in ents):
return 'location'
#Rule for Company Name
if 'ORG' in ents:
return 'company_name'
return 'product'
from nostril import nonsense
INPUT_FILENAME = "brute_force_results.txt"
OUTPUT_FILENAME = "meaningful_results.txt"
with open(INPUT_FILENAME) as input_file, open(OUTPUT_FILENAME, "w") as output_file:
for line in input_file:
if not nonsense(line):
output_file.write(line)
def run(self, record):
'''
The list of features will be:
- The number of domain parts, for example, www.google.com is 3.
- The overall length in characters.
- The length of the longest domain part.
- The length of the TLD, e.g. .online or .download is longer than .com.
- The randomness level of the domain.
'''
if 'analysers' not in record:
record['analysers'] = []
x_samples = []
Y_samples = []
for analyser in record['analysers']:
if analyser['analyser'] != 'WordSegmentation':
continue
for domain, segments in analyser['output'].items():
# Remove wildcard domain
domain = re.sub(r'^\*\.', '', domain)
parts = domain.split('.')
x = []
# Compute the number of domain parts
x.append(len(parts))
# Compute the length of the whole domain
x.append(len(domain))
longest = ''
# Compute the length of the longest domain parts
for part in parts:
if len(part) > len(longest):
longest = part
x.append(len(longest))
# Compute the length of the TLD
x.append(len(parts[-1]))
randomness_count = 0
# The nostril package which we are using to detect non-sense words
# in the domain only returns a boolean verdict so may be we need to
# think of how we want to quantify this
for w in segments:
try:
if len(w) >= FeaturesGenerator.NOSTRIL_LENGTH_LIMIT and nonsense(w):
randomness_count += 1
except ValueError:
continue
x.append(randomness_count / len(segments))
x_samples.append(x)
Y_samples.append(True if 'usual_suspect' in record else False)
break
record['analysers'].append({
'analyser': type(self).__name__,
'output': x_samples,
})
return record
def chat():
content = request.json
history = content['history']
query = content['query']
task_name = content['task_name']
lang_id = content['lang_id']
top_p = content['top_p']
temp = content['temp']
rept = content['rept']
lon = content['lon']
lat = content['lat']
if (type(history) == str or len(history) == 0 or history[0] == ""):
history = []
query = translate(query,
lang_id,
"en",
auto=True if task_name == "AUTOMODE" else False)
emoji_user = get_emoji_score([query])
history.append(query)
if task_name == "CovidQA": ## COVID QA
if len(query) <= 20 or nonsense(query):
resp = "Your input doesn't make sense, input something more meaninful"
else:
r2 = requests.get('https://covid19api.emos.ai/api/v1/summary/',
params={'q': query})
if (r2 != "<Response [200]>"):
resp = eval(r2.json()[0]['data'])["extractive"]
else:
resp = "This service is currenlty not available."
meta = []
viz_meta = []
task_id = 20
elif task_name == "debunker": ## COVID DEBUNKER
if len(query) <= 20 or nonsense(query):
resp = "Your input doesn't make sense, input something more meaninful"
else:
ppl = get_perplexity(query)
print(ppl)
if (float(ppl) > 170.): #### to be tuned
resp = "This claim is likely to be fake"
else:
resp = "This claim is likely to be real"
meta = []
viz_meta = []
task_id = 16
else:
resp, meta, task_name, task_id, viz_meta = generate(
history=history,
task_name=task_name,
meta_seed=1,
top_p=float(top_p),
temperature=float(temp),
repetion=float(rept),
lon=float(lon),
lat=float(lat))
toxic_ = False
if task_name == "Neural":
toxic_resp = get_toxic_score(query, resp)
if (toxic_resp != ""):
resp = toxic_resp
toxic_ = True
history.append(resp)
resp = translate(resp,
"en",
lang_id,
auto=True if task_name == "AUTOMODE" else False)
emoji_resp = get_emoji_score([resp])
jsonFormat = {
'history': history,
'response': resp,
'user_emoji': emoji_user,
'resp_emoji': emoji_resp,
'toxic': toxic_,
'task_name': task_name,
'task_id': task_id,
'meta': meta,
'viz_meta': viz_meta,
"lon": lon,
"lat": lat
}
with open("data/conversation_history/" +
"{}.json".format(timeStamped("conv")),
"w",
encoding="utf-8") as f:
json.dump(jsonFormat, f, indent=4)
return jsonify(jsonFormat) # last one is to to allow CORS [important]