mitmproxy/netlib/http/request.py

390 lines
13 KiB
Python
Raw Normal View History

2015-09-25 22:39:04 +00:00
from __future__ import absolute_import, print_function, division
import re
2015-09-25 22:39:04 +00:00
import warnings
import six
from six.moves import urllib
from netlib import utils
from netlib.http import cookies
from netlib.odict import ODict
from .. import encoding
from .headers import Headers
from .message import Message, _native, _always_bytes, MessageData, MessageMultiDict
2015-09-25 22:39:04 +00:00
# This regex extracts & splits the host header into host and port.
# Handles the edge case of IPv6 addresses containing colons.
# https://bugzilla.mozilla.org/show_bug.cgi?id=45891
host_header_re = re.compile(r"^(?P<host>[^:]+|\[.+\])(?::(?P<port>\d+))?$")
2015-09-25 22:39:04 +00:00
2016-04-11 21:14:18 +00:00
class RequestData(MessageData):
2015-09-25 22:39:04 +00:00
def __init__(self, first_line_format, method, scheme, host, port, path, http_version, headers=None, content=None,
timestamp_start=None, timestamp_end=None):
2016-02-08 03:16:58 +00:00
if not isinstance(headers, Headers):
headers = Headers(headers)
2015-09-25 22:39:04 +00:00
self.first_line_format = first_line_format
self.method = method
self.scheme = scheme
self.host = host
self.port = port
self.path = path
self.http_version = http_version
self.headers = headers
self.content = content
self.timestamp_start = timestamp_start
self.timestamp_end = timestamp_end
class Request(Message):
"""
An HTTP request.
"""
def __init__(self, *args, **kwargs):
2016-04-02 11:50:53 +00:00
self.data = RequestData(*args, **kwargs)
2015-09-25 22:39:04 +00:00
def __repr__(self):
if self.host and self.port:
hostport = "{}:{}".format(self.host, self.port)
else:
hostport = ""
path = self.path or ""
2015-09-26 15:39:50 +00:00
return "Request({} {}{})".format(
2015-09-25 22:39:04 +00:00
self.method, hostport, path
)
def replace(self, pattern, repl, flags=0):
"""
Replaces a regular expression pattern with repl in the headers, the
request path and the body of the request. Encoded content will be
decoded before replacement, and re-encoded afterwards.
Returns:
The number of replacements made.
"""
# TODO: Proper distinction between text and bytes.
c = super(Request, self).replace(pattern, repl, flags)
self.path, pc = utils.safe_subn(
pattern, repl, self.path, flags=flags
)
c += pc
return c
2015-09-25 22:39:04 +00:00
@property
def first_line_format(self):
"""
HTTP request form as defined in `RFC7230 <https://tools.ietf.org/html/rfc7230#section-5.3>`_.
origin-form and asterisk-form are subsumed as "relative".
"""
return self.data.first_line_format
@first_line_format.setter
def first_line_format(self, first_line_format):
self.data.first_line_format = first_line_format
@property
def method(self):
"""
HTTP request method, e.g. "GET".
"""
2015-09-28 11:53:59 +00:00
return _native(self.data.method).upper()
2015-09-25 22:39:04 +00:00
@method.setter
def method(self, method):
self.data.method = _always_bytes(method)
@property
def scheme(self):
"""
HTTP request scheme, which should be "http" or "https".
"""
return _native(self.data.scheme)
@scheme.setter
def scheme(self, scheme):
self.data.scheme = _always_bytes(scheme)
@property
def host(self):
"""
2015-09-26 15:39:50 +00:00
Target host. This may be parsed from the raw request
(e.g. from a ``GET http://example.com/ HTTP/1.1`` request line)
2015-09-25 22:39:04 +00:00
or inferred from the proxy mode (e.g. an IP in transparent mode).
2015-11-01 17:20:00 +00:00
Setting the host attribute also updates the host header, if present.
2015-09-25 22:39:04 +00:00
"""
if six.PY2: # pragma: no cover
2015-09-25 22:39:04 +00:00
return self.data.host
if not self.data.host:
return self.data.host
try:
return self.data.host.decode("idna")
except UnicodeError:
return self.data.host.decode("utf8", "surrogateescape")
@host.setter
def host(self, host):
if isinstance(host, six.text_type):
try:
# There's no non-strict mode for IDNA encoding.
# We don't want this operation to fail though, so we try
# utf8 as a last resort.
host = host.encode("idna", "strict")
except UnicodeError:
host = host.encode("utf8", "surrogateescape")
self.data.host = host
# Update host header
if "host" in self.headers:
if host:
self.headers["host"] = host
else:
self.headers.pop("host")
@property
def port(self):
"""
Target port
"""
return self.data.port
@port.setter
def port(self, port):
self.data.port = port
@property
def path(self):
"""
HTTP request path, e.g. "/index.html".
2016-05-09 18:15:20 +00:00
Guaranteed to start with a slash, except for OPTIONS requests, which may just be "*".
2015-09-25 22:39:04 +00:00
"""
2016-03-20 22:22:50 +00:00
if self.data.path is None:
return None
else:
return _native(self.data.path)
2015-09-25 22:39:04 +00:00
@path.setter
def path(self, path):
self.data.path = _always_bytes(path)
2015-09-26 15:39:50 +00:00
@property
def url(self):
"""
The URL string, constructed from the request's URL components
"""
2016-04-20 22:27:39 +00:00
if self.first_line_format == "authority":
return "%s:%d" % (self.host, self.port)
2015-09-26 15:39:50 +00:00
return utils.unparse_url(self.scheme, self.host, self.port, self.path)
@url.setter
def url(self, url):
self.scheme, self.host, self.port, self.path = utils.parse_url(url)
def _parse_host_header(self):
"""Extract the host and port from Host header"""
if "host" not in self.headers:
return None, None
host, port = self.headers["host"], None
m = host_header_re.match(host)
if m:
host = m.group("host").strip("[]")
if m.group("port"):
port = int(m.group("port"))
return host, port
2015-09-26 15:39:50 +00:00
@property
def pretty_host(self):
"""
Similar to :py:attr:`host`, but using the Host headers as an additional preferred data source.
This is useful in transparent mode where :py:attr:`host` is only an IP address,
but may not reflect the actual destination as the Host header could be spoofed.
"""
host, port = self._parse_host_header()
if not host:
return self.host
if not port:
port = 443 if self.scheme == 'https' else 80
# Prefer the original address if host header has an unexpected form
return host if port == self.port else self.host
2015-09-26 15:39:50 +00:00
@property
def pretty_url(self):
"""
Like :py:attr:`url`, but using :py:attr:`pretty_host` instead of :py:attr:`host`.
"""
if self.first_line_format == "authority":
return "%s:%d" % (self.pretty_host, self.port)
return utils.unparse_url(self.scheme, self.pretty_host, self.port, self.path)
2015-09-26 15:39:50 +00:00
@property
def query(self):
# type: () -> MessageMultiDict
2015-09-26 15:39:50 +00:00
"""
The request query string as an :py:class:`MessageMultiDict` object.
2015-09-26 15:39:50 +00:00
"""
return MessageMultiDict("query", self)
@property
def _query(self):
2015-09-26 15:39:50 +00:00
_, _, _, _, query, _ = urllib.parse.urlparse(self.url)
return tuple(utils.urldecode(query))
2015-09-26 15:39:50 +00:00
@query.setter
def query(self, value):
query = utils.urlencode(value)
2015-09-26 15:39:50 +00:00
scheme, netloc, path, params, _, fragment = urllib.parse.urlparse(self.url)
_, _, _, self.path = utils.parse_url(
urllib.parse.urlunparse([scheme, netloc, path, params, query, fragment]))
2015-09-26 15:39:50 +00:00
@property
def cookies(self):
# type: () -> MessageMultiDict
2015-09-26 15:39:50 +00:00
"""
The request cookies.
An empty :py:class:`MessageMultiDict` object if the cookie monster ate them all.
2015-09-26 15:39:50 +00:00
"""
return MessageMultiDict("cookies", self)
@property
def _cookies(self):
h = self.headers.get_all("Cookie")
return tuple(cookies.parse_cookie_headers(h))
2015-09-26 15:39:50 +00:00
@cookies.setter
def cookies(self, value):
self.headers["cookie"] = cookies.format_cookie_header(value)
2015-09-26 15:39:50 +00:00
@property
def path_components(self):
"""
The URL's path components as a tuple of strings.
2015-09-26 15:39:50 +00:00
Components are unquoted.
"""
_, _, path, _, _, _ = urllib.parse.urlparse(self.url)
# This needs to be a tuple so that it's immutable.
# Otherwise, this would fail silently:
# request.path_components.append("foo")
return tuple(urllib.parse.unquote(i) for i in path.split("/") if i)
2015-09-26 15:39:50 +00:00
@path_components.setter
def path_components(self, components):
components = map(lambda x: urllib.parse.quote(x, safe=""), components)
path = "/" + "/".join(components)
scheme, netloc, _, params, query, fragment = urllib.parse.urlparse(self.url)
_, _, _, self.path = utils.parse_url(
urllib.parse.urlunparse([scheme, netloc, path, params, query, fragment]))
2015-09-26 15:39:50 +00:00
2015-09-25 22:39:04 +00:00
def anticache(self):
"""
Modifies this request to remove headers that might produce a cached
response. That is, we remove ETags and If-Modified-Since headers.
"""
delheaders = [
"if-modified-since",
"if-none-match",
]
for i in delheaders:
self.headers.pop(i, None)
def anticomp(self):
"""
Modifies this request to remove headers that will compress the
resource's data.
"""
self.headers["accept-encoding"] = "identity"
def constrain_encoding(self):
"""
Limits the permissible Accept-Encoding values, based on what we can
decode appropriately.
"""
accept_encoding = self.headers.get("accept-encoding")
if accept_encoding:
self.headers["accept-encoding"] = (
', '.join(
e
for e in encoding.ENCODINGS
if e in accept_encoding
)
)
@property
def urlencoded_form(self):
"""
The URL-encoded form data as an :py:class:`MessageMultiDict` object.
None if the content-type indicates non-form data.
2015-09-25 22:39:04 +00:00
"""
is_valid_content_type = "application/x-www-form-urlencoded" in self.headers.get("content-type", "").lower()
if is_valid_content_type:
return MessageMultiDict("urlencoded_form", self)
2015-09-25 22:39:04 +00:00
return None
@property
def _urlencoded_form(self):
return tuple(utils.urldecode(self.content))
2015-09-25 22:39:04 +00:00
@urlencoded_form.setter
def urlencoded_form(self, value):
2015-09-25 22:39:04 +00:00
"""
Sets the body to the URL-encoded form data, and adds the appropriate content-type header.
This will overwrite the existing content if there is one.
"""
self.headers["content-type"] = "application/x-www-form-urlencoded"
self.content = utils.urlencode(value)
2015-09-25 22:39:04 +00:00
@property
def multipart_form(self):
"""
The multipart form data as an :py:class:`MultipartFormDict` object.
None if the content-type indicates non-form data.
2015-09-25 22:39:04 +00:00
"""
is_valid_content_type = "multipart/form-data" in self.headers.get("content-type", "").lower()
if is_valid_content_type:
return MessageMultiDict("multipart_form", self)
2015-09-25 22:39:04 +00:00
return None
@property
def _multipart_form(self):
return utils.multipartdecode(self.headers, self.content)
2015-09-25 22:39:04 +00:00
@multipart_form.setter
2016-02-08 01:10:48 +00:00
def multipart_form(self, value):
2015-09-25 22:39:04 +00:00
raise NotImplementedError()
# Legacy
def get_query(self): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".get_query is deprecated, use .query instead.", DeprecationWarning)
return self.query or ODict([])
def set_query(self, odict): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".set_query is deprecated, use .query instead.", DeprecationWarning)
self.query = odict
def get_path_components(self): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".get_path_components is deprecated, use .path_components instead.", DeprecationWarning)
return self.path_components
def set_path_components(self, lst): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".set_path_components is deprecated, use .path_components instead.", DeprecationWarning)
self.path_components = lst
def get_form_urlencoded(self): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".get_form_urlencoded is deprecated, use .urlencoded_form instead.", DeprecationWarning)
return self.urlencoded_form or ODict([])
def set_form_urlencoded(self, odict): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".set_form_urlencoded is deprecated, use .urlencoded_form instead.", DeprecationWarning)
self.urlencoded_form = odict
def get_form_multipart(self): # pragma: no cover
2015-09-25 22:39:04 +00:00
warnings.warn(".get_form_multipart is deprecated, use .multipart_form instead.", DeprecationWarning)
return self.multipart_form or ODict([])