mirror of
http://git.nowherejezfoltodf4jiyl6r56jnzintap5vyjlia7fkirfsnfizflqd.onion/nihilist/darknet-lantern.git
synced 2025-07-01 18:56:40 +00:00
[untested] RecognizeURLType and IsURLValid functions are working
This commit is contained in:
parent
08697f5c40
commit
c9a2fbcfdd
5 changed files with 179 additions and 320 deletions
284
scripts/utils.py
284
scripts/utils.py
|
|
@ -18,28 +18,108 @@ RESET = '\033[m'
|
|||
# name should contain only up to 64 alphanumeric characters
|
||||
VALID_NAME_PATTERN = re.compile(r"^[A-Za-z0-9]{1,64}$")
|
||||
|
||||
# pattern for regular urls
|
||||
# TODO: this is very simplified pattern
|
||||
URL_PATTERN = re.compile(r"^[A-Za-z0-9:\/\._%-=#?&@]+$")
|
||||
# pattern for regular urls (https://stackoverflow.com/a/3809435)
|
||||
CLEARNET_URL_PATTERN = re.compile(
|
||||
r"https?:\/\/(www\.)?[-a-zA-Z0-9@:%._\+~#=]{1,256}\.[a-zA-Z0-9()]"
|
||||
r"{1,6}\b([-a-zA-Z0-9()@:%_\+.~#?&//=]*)"
|
||||
)
|
||||
|
||||
# pattern for onion urls (56 bytes of base32 alphabet + .onion)
|
||||
ONION_URL_PATTERN = re.compile(r"^(\w+:)?(?:\/\/)?(\w+\.)?[a-z2-7]{56}\.onion")
|
||||
ONION_URL_PATTERN = re.compile(
|
||||
r"^https?:\/\/([a-zA-Z0-9-]+\.)*[a-z2-7-]{56}\.onion[^\s]*$"
|
||||
)
|
||||
|
||||
# pattern for simplex chatroom links
|
||||
SIMPLEX_CHATROOM_PATTERN = re.compile(
|
||||
r"(?:https?:\/\/(?:simplex\.chat|[^\/]+)|simplex:)\/(?:contact|invitation)#\/\?v=[\d-]+"
|
||||
r"&smp=[^&]+(?:&[^=]+=[^&]*)*(?:&data=\{[^}]*\})?"
|
||||
)
|
||||
|
||||
def print_colors(s:str=' ', bold:bool=False, is_error:bool = False, default:bool=False):
|
||||
# pattern for smp or xftp simplex server ((smp|xftp):// 44 byte key @ url [:port])
|
||||
SIMPLEX_SERVER_PATTERN = re.compile(
|
||||
r"^(smp|xftp):\/\/([a-zA-Z0-9\-_+=]{44})@([a-z2-7]{56}\.onion|"
|
||||
r"([a-zA-Z0-9\-\.]+\.[a-zA-Z0-9\-\.]+))"
|
||||
r"{1,}(?::[1-9][0-9]{0,4}|[1-5][0-9]{4}|6[0-4][0-9]{3}|"
|
||||
r"65[0-4][0-9]{2}|655[0-3][0-9]|6553[0-5])?$"
|
||||
)
|
||||
|
||||
def IsSimplexChatroomValid(url: str) -> bool:
|
||||
"""
|
||||
Helper function to print with colors
|
||||
Recognizes Simplex Chatroom link.
|
||||
Returns True if URL is a SimpleX chatroom,
|
||||
False otherwise
|
||||
"""
|
||||
if is_error:
|
||||
print(f"{RED}{s}{RESET}")
|
||||
elif bold:
|
||||
print(f"{BOLD_PURPLE}{s}{RESET}")
|
||||
elif is_error and bold:
|
||||
print(f"{BOLD_RED}{s}{RESET}")
|
||||
elif default:
|
||||
print(f'{s}')
|
||||
return SIMPLEX_CHATROOM_PATTERN.match(url)
|
||||
|
||||
def RecognizeSimplexType(url: str) -> str:
|
||||
"""
|
||||
Recognizes Simplex Server URL, returns smp, xftp or invalid
|
||||
"""
|
||||
match = SIMPLEX_SERVER_PATTERN.match(url)
|
||||
if match:
|
||||
return match.group(1)
|
||||
else:
|
||||
print(f"{PURPLE}{s}{RESET}")
|
||||
return 'invalid'
|
||||
|
||||
# stub function
|
||||
def IsXFTPServerValid(url: str) -> bool:
|
||||
"""
|
||||
Returns True if URL is a valid SimpleX XFTP Server URL
|
||||
False otherwise
|
||||
"""
|
||||
return RecognizeSimplexType(url) == 'xftp'
|
||||
|
||||
# stub function
|
||||
def IsSMPServerValid(url: str) -> bool:
|
||||
"""
|
||||
Returns True if URL is a valid SimpleX SMP Server URL
|
||||
False otherwise
|
||||
"""
|
||||
return RecognizeSimplexType(url) == 'smp'
|
||||
|
||||
def IsClearnetLinkValid(url: str) -> bool:
|
||||
"""
|
||||
Returns True if URL is a valid clearnet URL
|
||||
False otherwise
|
||||
"""
|
||||
return CLEARNET_URL_PATTERN.match(url)
|
||||
|
||||
def IsOnionLinkValid(url: str) -> bool:
|
||||
"""
|
||||
Returns True if URL is a valid onion URL
|
||||
False otherwise
|
||||
"""
|
||||
return ONION_URL_PATTERN.match(url)
|
||||
|
||||
def RecognizeURLType(url: str) -> str:
|
||||
"""
|
||||
Recognizes URL type, can return:
|
||||
- chatroom - SimpleX chatroom
|
||||
- xftp - XFTP SimpleX server
|
||||
- smp - SMP SimpleX server
|
||||
- onion - onion URL
|
||||
- clearnet - valid clearnet url
|
||||
- invalid - none of the above (probably invalid)
|
||||
"""
|
||||
# order is important here
|
||||
# (ex. simplex chatroom is also valid clearnet link)
|
||||
if IsSimplexChatroomValid(url):
|
||||
return 'chatroom'
|
||||
if IsXFTPServerValid(url):
|
||||
return 'xftp'
|
||||
if IsSMPServerValid(url):
|
||||
return 'smp'
|
||||
if IsOnionLinkValid(url):
|
||||
return 'onion'
|
||||
if IsClearnetLinkValid(url):
|
||||
return 'clearnet'
|
||||
return 'invalid'
|
||||
|
||||
def IsURLValid(url: str) -> bool:
|
||||
"""
|
||||
Checks if given URL is valid (RecognizeURLType recognizes it)
|
||||
"""
|
||||
return RecognizeURLType(url) != 'invalid'
|
||||
|
||||
|
||||
#### Checking Functions to validate that links are legit ####
|
||||
|
|
@ -54,7 +134,7 @@ def CheckUrl(url):
|
|||
}
|
||||
try:
|
||||
status = requests.get(url, proxies=proxies, timeout=5).status_code
|
||||
return bool(status == 200)
|
||||
return status == 200
|
||||
except requests.ConnectionError:
|
||||
return False
|
||||
except requests.exceptions.ReadTimeout:
|
||||
|
|
@ -82,131 +162,13 @@ def IsBannerValid(path: str) -> bool:
|
|||
return True
|
||||
|
||||
|
||||
def IsOnionValid(url: str) -> bool:
|
||||
"""
|
||||
Checks if the domain(param) is a valid onion domain and return True else False.
|
||||
"""
|
||||
try:
|
||||
# make sure the protocol is there
|
||||
if not url.startswith(('http://', 'https://')):
|
||||
url = 'http://' + url.strip().removesuffix('/')
|
||||
|
||||
domain = url.split('/')[2]
|
||||
|
||||
if ONION_URL_PATTERN.fullmatch(domain):
|
||||
parts_count = len(domain.split('.'))
|
||||
# TODO: we probably don't really need to check 62 char length
|
||||
# regex does that beforehand
|
||||
return (len(domain) == 62) and (parts_count <= 3)
|
||||
except Exception:
|
||||
return False
|
||||
|
||||
|
||||
def IsSimpleXChatroomValid(url: str) -> bool:
|
||||
"""Validate the SimpleX chatroom URL."""
|
||||
REQUIRED_SUBSTRING = "#/?v=2-7&smp=smp%3A%2F"
|
||||
|
||||
# Step 1: Check if it starts with http://, https://, or simplex:/
|
||||
if url.startswith(('http://', 'https://', 'simplex:/')):
|
||||
# Step 1.5: If http:// or https://, check for valid clearnet or onion domain
|
||||
if url.startswith(('http://', 'https://')) \
|
||||
and RecognizeUrlOnionClear(url) != 'invalid':
|
||||
return False
|
||||
elif not url.startswith('simplex:/'):
|
||||
return False # Must start with one of the valid protocols
|
||||
|
||||
# Step 2: Check for the presence of the required substring
|
||||
if REQUIRED_SUBSTRING not in url:
|
||||
return False # Required substring not found
|
||||
|
||||
# Step 3: Extract the part after "smp=smp%3A%2F"
|
||||
smp_start = url.find("smp=smp%3A%2F")
|
||||
if smp_start == -1:
|
||||
return False # Required substring not found
|
||||
|
||||
smp_start += len("smp=smp%3A%2F")
|
||||
smp_end = url.find("&", smp_start)
|
||||
if smp_end == -1:
|
||||
smp_end = len(url) # Take until the end if no "&" is found
|
||||
|
||||
smp_value = urllib.parse.unquote(url[smp_start:smp_end]) # Decode the URL-encoded string
|
||||
|
||||
# Step 3.5: Check if the smp_value contains a valid hostname
|
||||
if '@' not in smp_value:
|
||||
return False # Must contain '@' to separate fingerprint and hostname
|
||||
|
||||
fingerprint, hostname = smp_value.split('@', 1)
|
||||
if RecognizeUrlOnionClear(hostname) != 'invalid':
|
||||
return False # Invalid hostname
|
||||
|
||||
# Step 4: Check for the presence of "%2F" in the original URL
|
||||
if "%2F" not in url:
|
||||
return False # Required substring not found
|
||||
|
||||
# If all checks pass, return True
|
||||
return True
|
||||
|
||||
def RecognizeUrlOnionClear(url: str) -> str:
|
||||
"""
|
||||
Recognize if the URL is invalid, onion or clearnet.
|
||||
"""
|
||||
|
||||
# early terminate preconditions
|
||||
if len(url) < 4 or (';' in url) or ('.' not in url):
|
||||
return 'invalid'
|
||||
|
||||
# check if possibly onion url, here just perliminary check
|
||||
# IsOnionValid checks it against regex expression
|
||||
if '.onion' in url:
|
||||
if IsOnionValid(url):
|
||||
return 'onion'
|
||||
|
||||
if URL_PATTERN.fullmatch(url):
|
||||
return 'clearnet'
|
||||
|
||||
return 'invalid'
|
||||
|
||||
|
||||
def RecognizeUrlFull(url: str) -> str:
|
||||
"""
|
||||
Recognize if URL is smp, xftp, simplex groupchat, onion, clearnet or just invalid
|
||||
Depends on RecognizeUrlOnionClear
|
||||
"""
|
||||
if IsSimpleXChatroomValid(url):
|
||||
return 'chatroom'
|
||||
if url.startswith(('http://', 'https://')):
|
||||
return RecognizeUrlOnionClear(url)
|
||||
if url.startswith('xftp://'):
|
||||
if IsSimpleXServerValid(url):
|
||||
return 'xftp'
|
||||
if url.startswith('smp://'):
|
||||
if IsSimpleXServerValid(url):
|
||||
return 'smp'
|
||||
return 'invalid'
|
||||
|
||||
#def IsUrlValid(url:str)->bool:
|
||||
# """
|
||||
# Check if url is valid both dark net end clearnet.
|
||||
# """
|
||||
# pattern = re.compile("^[A-Za-z0-9:/.-]+$")
|
||||
# url = str(url)
|
||||
# if len(url) < 4:
|
||||
# return False
|
||||
# if url.endswith('.onion'):
|
||||
# return IsOnionValid(url)
|
||||
# else:
|
||||
# if not url.__contains__('.'):
|
||||
# return False
|
||||
# if pattern.fullmatch(url) is None:
|
||||
# return False
|
||||
# return True
|
||||
|
||||
|
||||
def IsStatusValid(status: str) -> bool:
|
||||
"""
|
||||
Checks if status contains only ['YES','NO']. Verbose only if False is returned
|
||||
"""
|
||||
pattern = ['YES','NO','✔️','❌','']
|
||||
pattern = ['YES','NO','']
|
||||
status = status.strip()
|
||||
if status not in pattern:
|
||||
return False
|
||||
|
|
@ -230,7 +192,7 @@ def IsScoreValid(score: str) -> bool:
|
|||
return True
|
||||
|
||||
|
||||
def IsDescriptionValid(desc:str)->bool:
|
||||
def IsDescriptionValid(desc: str) -> bool:
|
||||
"""
|
||||
Check the categories are only [a-zA-Z0-9.' ] with 256 max chars.
|
||||
"""
|
||||
|
|
@ -263,40 +225,6 @@ def IsCategoryValid(categories: list[str]) -> bool:
|
|||
return True
|
||||
|
||||
|
||||
def IsSimpleXServerValid(url: str) -> bool:
|
||||
pattern = re.compile('[0-9A-Za-z-_]*')
|
||||
url = url.strip()
|
||||
try:
|
||||
|
||||
if url.startswith(('smp://', 'xftp://')):
|
||||
# Remove the protocol part
|
||||
proless = url.split('//', 1)[-1]
|
||||
# Split the fingerprint and hostname
|
||||
parts = proless.split('@')
|
||||
if len(parts) != 2:
|
||||
return False # Must have exactly one '@' character
|
||||
|
||||
fingerprint = parts[0]
|
||||
hostname = parts[1].split(',')[0] # Get the hostname before any comma
|
||||
|
||||
# Check fingerprint length and pattern
|
||||
if len(fingerprint) == 44 and pattern.match(fingerprint):
|
||||
# Validate the hostname
|
||||
if RecognizeUrlOnionClear(hostname) != 'invalid':
|
||||
# Check for an optional comma and a valid onion domain
|
||||
if ',' in proless:
|
||||
onion_part = proless.split(',')[1].strip()
|
||||
if RecognizeUrlOnionClear(onion_part) != 'invalid':
|
||||
return False
|
||||
return True
|
||||
return False
|
||||
except Exception as e:
|
||||
print(e)
|
||||
# Any error will be a false
|
||||
return False
|
||||
|
||||
|
||||
|
||||
def IsNameValid(name: str) -> bool:
|
||||
"""
|
||||
Check the parameter name only contains [a-zA-Z0-9] and is 64 chars long.
|
||||
|
|
@ -325,3 +253,19 @@ def send_server_checks(url: str) -> tuple[str, str, str]:
|
|||
failed_response = response['resp'].get('testFailure')
|
||||
|
||||
return (response, resp_type, failed_response)
|
||||
|
||||
|
||||
def print_colors(s:str=' ', bold:bool=False, is_error:bool = False, default:bool=False):
|
||||
"""
|
||||
Helper function to print with colors
|
||||
"""
|
||||
if is_error:
|
||||
print(f"{RED}{s}{RESET}")
|
||||
elif bold:
|
||||
print(f"{BOLD_PURPLE}{s}{RESET}")
|
||||
elif is_error and bold:
|
||||
print(f"{BOLD_RED}{s}{RESET}")
|
||||
elif default:
|
||||
print(f'{s}')
|
||||
else:
|
||||
print(f"{PURPLE}{s}{RESET}")
|
||||
Loading…
Add table
Add a link
Reference in a new issue