def parse_init(line):
try:
method, url, protocol = string.split(line)
except ValueError:
return None
httpversion = parse_http_protocol(protocol)
if not httpversion:
return None
if not utils.isascii(method):
return None
return method, url, httpversion
def _parse_init(self, line):
try:
method, url, protocol = string.split(line)
except ValueError:
return None
httpversion = self._parse_http_protocol(protocol)
if not httpversion:
return None
if not utils.isascii(method):
return None
return method, url, httpversion
def parse_init_http(line):
"""
Returns (method, url, httpversion)
"""
v = parse_init(line)
if not v:
return None
method, url, httpversion = v
if not utils.isascii(url):
return None
if not (url.startswith("/") or url == "*"):
return None
return method, url, httpversion
def _parse_init_http(self, line):
"""
Returns (method, url, httpversion)
"""
v = self._parse_init(line)
if not v:
return None
method, url, httpversion = v
if not utils.isascii(url):
return None
if not (url.startswith("/") or url == "*"):
return None
return method, url, httpversion
def parse_url(url):
"""
Returns a (scheme, host, port, path) tuple, or None on error.
Checks that:
port is an integer 0-65535
host is a valid IDNA-encoded hostname with no null-bytes
path is valid ASCII
"""
try:
scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
except ValueError:
return None
if not scheme:
return None
if '@' in netloc:
# FIXME: Consider what to do with the discarded credentials here Most
# probably we should extend the signature to return these as a separate
# value.
_, netloc = string.rsplit(netloc, '@', maxsplit=1)
if ':' in netloc:
host, port = string.rsplit(netloc, ':', maxsplit=1)
try:
port = int(port)
except ValueError:
return None
else:
host = netloc
if scheme == "https":
port = 443
else:
port = 80
path = urlparse.urlunparse(('', '', path, params, query, fragment))
if not path.startswith("/"):
path = "/" + path
if not _is_valid_host(host):
return None
if not utils.isascii(path):
return None
if not _is_valid_port(port):
return None
return scheme, host, port, path
def parse_url(url):
"""
Returns a (scheme, host, port, path) tuple, or None on error.
Checks that:
port is an integer 0-65535
host is a valid IDNA-encoded hostname with no null-bytes
path is valid ASCII
"""
try:
scheme, netloc, path, params, query, fragment = urlparse.urlparse(url)
except ValueError:
return None
if not scheme:
return None
if '@' in netloc:
# FIXME: Consider what to do with the discarded credentials here Most
# probably we should extend the signature to return these as a separate
# value.
_, netloc = string.rsplit(netloc, '@', maxsplit=1)
if ':' in netloc:
host, port = string.rsplit(netloc, ':', maxsplit=1)
try:
port = int(port)
except ValueError:
return None
else:
host = netloc
if scheme == "https":
port = 443
else:
port = 80
path = urlparse.urlunparse(('', '', path, params, query, fragment))
if not path.startswith("/"):
path = "/" + path
if not _is_valid_host(host):
return None
if not utils.isascii(path):
return None
if not _is_valid_port(port):
return None
return scheme, host, port, path
def read_request(
self,
include_body=True,
body_size_limit=None,
allow_empty=False,
):
"""
Parse an HTTP request from a file stream
Args:
include_body (bool): Read response body as well
body_size_limit (bool): Maximum body size
wfile (file): If specified, HTTP Expect headers are handled
automatically, by writing a HTTP 100 CONTINUE response to the stream.
Returns:
Request: The HTTP request
Raises:
HttpError: If the input is invalid.
"""
timestamp_start = time.time()
if hasattr(self.tcp_handler.rfile, "reset_timestamps"):
self.tcp_handler.rfile.reset_timestamps()
httpversion, host, port, scheme, method, path, headers, body = (
None, None, None, None, None, None, None, None)
request_line = self._get_request_line()
if not request_line:
if allow_empty:
return http.EmptyRequest()
else:
raise tcp.NetLibDisconnect()
request_line_parts = self._parse_init(request_line)
if not request_line_parts:
raise HttpError(
400,
"Bad HTTP request line: %s" % repr(request_line)
)
method, path, httpversion = request_line_parts
if path == '*' or path.startswith("/"):
form_in = "relative"
if not utils.isascii(path):
raise HttpError(
400,
"Bad HTTP request line: %s" % repr(request_line)
)
elif method == 'CONNECT':
form_in = "authority"
r = self._parse_init_connect(request_line)
if not r:
raise HttpError(
400,
"Bad HTTP request line: %s" % repr(request_line)
)
host, port, httpversion = r
path = None
else:
form_in = "absolute"
r = self._parse_init_proxy(request_line)
if not r:
raise HttpError(
400,
"Bad HTTP request line: %s" % repr(request_line)
)
_, scheme, host, port, path, _ = r
headers = self.read_headers()
if headers is None:
raise HttpError(400, "Invalid headers")
expect_header = headers.get_first("expect", "").lower()
if expect_header == "100-continue" and httpversion == (1, 1):
self.tcp_handler.wfile.write(
'HTTP/1.1 100 Continue\r\n'
'\r\n'
)
self.tcp_handler.wfile.flush()
del headers['expect']
if include_body:
body = self.read_http_body(
headers,
body_size_limit,
method,
None,
True
)
if hasattr(self.tcp_handler.rfile, "first_byte_timestamp"):
# more accurate timestamp_start
timestamp_start = self.tcp_handler.rfile.first_byte_timestamp
timestamp_end = time.time()
return http.Request(
form_in,
method,
scheme,
host,
port,
path,
httpversion,
headers,
body,
timestamp_start,
timestamp_end,
)
def read_request(
self,
include_body=True,
body_size_limit=None,
allow_empty=False,
):
"""
Parse an HTTP request from a file stream
Args:
include_body (bool): Read response body as well
body_size_limit (bool): Maximum body size
wfile (file): If specified, HTTP Expect headers are handled
automatically, by writing a HTTP 100 CONTINUE response to the stream.
Returns:
Request: The HTTP request
Raises:
HttpError: If the input is invalid.
"""
timestamp_start = time.time()
if hasattr(self.tcp_handler.rfile, "reset_timestamps"):
self.tcp_handler.rfile.reset_timestamps()
httpversion, host, port, scheme, method, path, headers, body = (
None, None, None, None, None, None, None, None)
request_line = self._get_request_line()
if not request_line:
if allow_empty:
return http.EmptyRequest()
else:
raise tcp.NetLibDisconnect()
request_line_parts = self._parse_init(request_line)
if not request_line_parts:
raise HttpError(400,
"Bad HTTP request line: %s" % repr(request_line))
method, path, httpversion = request_line_parts
if path == '*' or path.startswith("/"):
form_in = "relative"
if not utils.isascii(path):
raise HttpError(
400, "Bad HTTP request line: %s" % repr(request_line))
elif method == 'CONNECT':
form_in = "authority"
r = self._parse_init_connect(request_line)
if not r:
raise HttpError(
400, "Bad HTTP request line: %s" % repr(request_line))
host, port, httpversion = r
path = None
else:
form_in = "absolute"
r = self._parse_init_proxy(request_line)
if not r:
raise HttpError(
400, "Bad HTTP request line: %s" % repr(request_line))
_, scheme, host, port, path, _ = r
headers = self.read_headers()
if headers is None:
raise HttpError(400, "Invalid headers")
expect_header = headers.get_first("expect", "").lower()
if expect_header == "100-continue" and httpversion == (1, 1):
self.tcp_handler.wfile.write('HTTP/1.1 100 Continue\r\n' '\r\n')
self.tcp_handler.wfile.flush()
del headers['expect']
if include_body:
body = self.read_http_body(headers, body_size_limit, method, None,
True)
if hasattr(self.tcp_handler.rfile, "first_byte_timestamp"):
# more accurate timestamp_start
timestamp_start = self.tcp_handler.rfile.first_byte_timestamp
timestamp_end = time.time()
return http.Request(
form_in,
method,
scheme,
host,
port,
path,
httpversion,
headers,
body,
timestamp_start,
timestamp_end,
)
