def __next__(self):
self.condition.acquire()
try:
path = self.entries[self.index]
except IndexError:
self.condition.release()
raise StopIteration
self.index += 1
self.condition.release()
return safequote(path)
def __next__(self, base_path=None):
_, path = self.next_with_index(base_path)
return safequote(path)
def __init__(
self,
url,
max_pool=1,
max_retries=5,
timeout=20,
ip=None,
proxy=None,
proxylist=None,
redirect=False,
request_by_hostname=False,
httpmethod="get",
data=None,
scheme=None,
random_agents=None,
):
self.httpmethod = httpmethod
self.data = data
self.headers = {}
parsed = urlparse(url)
# If no scheme specified, unset it first
if "://" not in url:
parsed = urlparse("{0}://{1}".format(scheme or "unknown", url))
self.base_path = parsed.path
if parsed.path.startswith("/"):
self.base_path = parsed.path[1:]
# Safe quote all special characters in base_path to prevent from being encoded
self.base_path = safequote(self.base_path)
self.host = parsed.netloc.split(":")[0]
port_for_scheme = {"http": 80, "https": 443, "unknown": 0}
if parsed.scheme not in ("unknown", "https", "http"):
raise RequestException("Unsupported URI scheme: {0}".format(
self.scheme))
# If no port specified, set default (80, 443)
try:
self.port = int(parsed.netloc.split(":")[1])
except IndexError:
self.port = port_for_scheme[parsed.scheme]
except ValueError:
raise RequestException("Invalid port number: {0}".format(
parsed.netloc.split(":")[1]))
# If no scheme is found, detect it by port number
self.scheme = parsed.scheme if parsed.scheme != "unknown" else self.get_scheme(
self.port)
# If the user neither provide the port nor scheme, guess them based
# on standard website characteristics
if not self.scheme:
self.scheme = "https" if self.get_scheme(
443) == "https" else "http"
self.port = port_for_scheme[self.scheme]
# Set the Host header, read the line 126 to know why
self.headers["Host"] = self.host
# Include port in Host header if it's non-standard
if (self.scheme == "https"
and self.port != 443) or (self.scheme == "http"
and self.port != 80):
self.headers["Host"] += ":{0}".format(self.port)
self.max_retries = max_retries
self.max_pool = max_pool
self.timeout = timeout
self.pool = None
self.proxy = proxy
self.proxylist = proxylist
self.redirect = redirect
self.random_agents = random_agents
self.auth = None
self.request_by_hostname = request_by_hostname
self.ip = ip
self.base_url = self.url = "{0}://{1}/".format(
self.scheme,
self.headers["Host"],
)
def __init__(
self,
url,
max_pool=1,
max_retries=5,
timeout=20,
ip=None,
proxy=None,
proxylist=None,
redirect=False,
request_by_hostname=False,
httpmethod="get",
data=None,
scheme=None,
):
self.httpmethod = httpmethod
self.data = data
self.headers = {}
parsed = urlparse(url)
# If no protocol specified, set http by default
if "://" not in url:
parsed = urlparse("{0}://{1}".format(scheme, url))
# If protocol is not supported
elif parsed.scheme not in ["https", "http"]:
raise RequestException({
"message":
"Unsupported URL scheme: {0}".format(parsed.scheme)
})
self.base_path = parsed.path
if parsed.path.startswith("/"):
self.base_path = parsed.path[1:]
# Safe quote all special characters in base_path to prevent from being encoded
self.base_path = safequote(self.base_path)
self.protocol = parsed.scheme
self.host = parsed.netloc.split(":")[0]
# Resolve DNS to decrease overhead
if ip:
self.ip = ip
# A proxy could have a different DNS that would resolve the name. ThereFore.
# resolving the name when using proxy to raise an error is pointless
elif not proxy and not proxylist:
try:
self.ip = socket.gethostbyname(self.host)
except socket.gaierror:
# Check if hostname resolves to IPv6 address only
try:
self.ip = socket.getaddrinfo(self.host, None,
socket.AF_INET6)[0][4][0]
except socket.gaierror:
raise RequestException({"message": "Couldn't resolve DNS"})
# If no port specified, set default (80, 443)
try:
self.port = int(parsed.netloc.split(":")[1])
except IndexError:
self.port = 443 if self.protocol == "https" else 80
except ValueError:
raise RequestException({
"message":
"Invalid port number: {0}".format(parsed.netloc.split(":")[1])
})
# Set the Host header, this will be overwritten if the user has already set the header
self.headers["Host"] = self.host
# Include port in Host header if it's non-standard
if (self.protocol == "https"
and self.port != 443) or (self.protocol == "http"
and self.port != 80):
self.headers["Host"] += ":{0}".format(self.port)
self.max_retries = max_retries
self.max_pool = max_pool
self.timeout = timeout
self.pool = None
self.proxy = proxy
self.proxylist = proxylist
self.redirect = redirect
self.random_agents = None
self.auth = None
self.request_by_hostname = request_by_hostname
self.session = requests.Session()
self.url = "{0}://{1}:{2}/".format(
self.protocol,
self.host if self.request_by_hostname else self.ip,
self.port,
)
self.base_url = "{0}://{1}:{2}/".format(
self.protocol,
self.host,
self.port,
)
self.set_adapter()