diff options
Diffstat (limited to 'netlib')
-rw-r--r-- | netlib/http_cookies.py | 197 | ||||
-rw-r--r-- | netlib/odict.py | 35 | ||||
-rw-r--r-- | netlib/wsgi.py | 29 |
3 files changed, 242 insertions, 19 deletions
diff --git a/netlib/http_cookies.py b/netlib/http_cookies.py new file mode 100644 index 00000000..dab95ed0 --- /dev/null +++ b/netlib/http_cookies.py @@ -0,0 +1,197 @@ +""" +A flexible module for cookie parsing and manipulation. + +This module differs from usual standards-compliant cookie modules in a number of +ways. We try to be as permissive as possible, and to retain even mal-formed +information. Duplicate cookies are preserved in parsing, and can be set in +formatting. We do attempt to escape and quote values where needed, but will not +reject data that violate the specs. + +Parsing accepts the formats in RFC6265 and partially RFC2109 and RFC2965. We do +not parse the comma-separated variant of Set-Cookie that allows multiple cookies +to be set in a single header. Technically this should be feasible, but it turns +out that violations of RFC6265 that makes the parsing problem indeterminate are +much more common than genuine occurences of the multi-cookie variants. +Serialization follows RFC6265. + + http://tools.ietf.org/html/rfc6265 + http://tools.ietf.org/html/rfc2109 + http://tools.ietf.org/html/rfc2965 +""" + +# TODO +# - Disallow LHS-only Cookie values + +import re + +import odict + + +def _read_until(s, start, term): + """ + Read until one of the characters in term is reached. + """ + if start == len(s): + return "", start+1 + for i in range(start, len(s)): + if s[i] in term: + return s[start:i], i + return s[start:i+1], i+1 + + +def _read_token(s, start): + """ + Read a token - the LHS of a token/value pair in a cookie. + """ + return _read_until(s, start, ";=") + + +def _read_quoted_string(s, start): + """ + start: offset to the first quote of the string to be read + + A sort of loose super-set of the various quoted string specifications. + + RFC6265 disallows backslashes or double quotes within quoted strings. + Prior RFCs use backslashes to escape. This leaves us free to apply + backslash escaping by default and be compatible with everything. + """ + escaping = False + ret = [] + # Skip the first quote + for i in range(start+1, len(s)): + if escaping: + ret.append(s[i]) + escaping = False + elif s[i] == '"': + break + elif s[i] == "\\": + escaping = True + pass + else: + ret.append(s[i]) + return "".join(ret), i+1 + + +def _read_value(s, start, delims): + """ + Reads a value - the RHS of a token/value pair in a cookie. + + special: If the value is special, commas are premitted. Else comma + terminates. This helps us support old and new style values. + """ + if start >= len(s): + return "", start + elif s[start] == '"': + return _read_quoted_string(s, start) + else: + return _read_until(s, start, delims) + + +def _read_pairs(s, off=0, specials=()): + """ + Read pairs of lhs=rhs values. + + off: start offset + specials: a lower-cased list of keys that may contain commas + """ + vals = [] + while 1: + lhs, off = _read_token(s, off) + lhs = lhs.lstrip() + if lhs: + rhs = None + if off < len(s): + if s[off] == "=": + rhs, off = _read_value(s, off+1, ";") + vals.append([lhs, rhs]) + off += 1 + if not off < len(s): + break + return vals, off + + +def _has_special(s): + for i in s: + if i in '",;\\': + return True + o = ord(i) + if o < 0x21 or o > 0x7e: + return True + return False + + +ESCAPE = re.compile(r"([\"\\])") + + +def _format_pairs(lst, specials=(), sep="; "): + """ + specials: A lower-cased list of keys that will not be quoted. + """ + vals = [] + for k, v in lst: + if v is None: + vals.append(k) + else: + if k.lower() not in specials and _has_special(v): + v = ESCAPE.sub(r"\\\1", v) + v = '"%s"'%v + vals.append("%s=%s"%(k, v)) + return sep.join(vals) + + +def _format_set_cookie_pairs(lst): + return _format_pairs( + lst, + specials = ("expires", "path") + ) + + +def _parse_set_cookie_pairs(s): + """ + For Set-Cookie, we support multiple cookies as described in RFC2109. + This function therefore returns a list of lists. + """ + pairs, off = _read_pairs( + s, + specials = ("expires", "path") + ) + return pairs + + +def parse_set_cookie_header(str): + """ + Parse a Set-Cookie header value + + Returns a (name, value, attrs) tuple, or None, where attrs is an + ODictCaseless set of attributes. No attempt is made to parse attribute + values - they are treated purely as strings. + """ + pairs = _parse_set_cookie_pairs(str) + if pairs: + return pairs[0][0], pairs[0][1], odict.ODictCaseless(pairs[1:]) + + +def format_set_cookie_header(name, value, attrs): + """ + Formats a Set-Cookie header value. + """ + pairs = [[name, value]] + pairs.extend(attrs.lst) + return _format_set_cookie_pairs(pairs) + + +def parse_cookie_header(str): + """ + Parse a Cookie header value. + Returns a (possibly empty) ODict object. + """ + pairs, off = _read_pairs(str) + return odict.ODict(pairs) + + +def format_cookie_header(od): + """ + Formats a Cookie header value. + """ + return _format_pairs(od.lst) diff --git a/netlib/odict.py b/netlib/odict.py index 7a2f611b..dd738c55 100644 --- a/netlib/odict.py +++ b/netlib/odict.py @@ -13,7 +13,8 @@ def safe_subn(pattern, repl, target, *args, **kwargs): class ODict(object): """ - A dictionary-like object for managing ordered (key, value) data. + A dictionary-like object for managing ordered (key, value) data. Think + about it as a convenient interface to a list of (key, value) tuples. """ def __init__(self, lst=None): self.lst = lst or [] @@ -64,11 +65,20 @@ class ODict(object): key, they are cleared. """ if isinstance(valuelist, basestring): - raise ValueError("Expected list of values instead of string. Example: odict['Host'] = ['www.example.com']") - - new = self._filter_lst(k, self.lst) - for i in valuelist: - new.append([k, i]) + raise ValueError( + "Expected list of values instead of string. " + "Example: odict['Host'] = ['www.example.com']" + ) + kc = self._kconv(k) + new = [] + for i in self.lst: + if self._kconv(i[0]) == kc: + if valuelist: + new.append([k, valuelist.pop(0)]) + else: + new.append(i) + while valuelist: + new.append([k, valuelist.pop(0)]) self.lst = new def __delitem__(self, k): @@ -84,7 +94,7 @@ class ODict(object): return False def add(self, key, value): - self.lst.append([key, str(value)]) + self.lst.append([key, value]) def get(self, k, d=None): if k in self: @@ -108,10 +118,19 @@ class ODict(object): lst = copy.deepcopy(self.lst) return self.__class__(lst) + def extend(self, other): + """ + Add the contents of other, preserving any duplicates. + """ + self.lst.extend(other.lst) + def __repr__(self): + return repr(self.lst) + + def format(self): elements = [] for itm in self.lst: - elements.append(itm[0] + ": " + itm[1]) + elements.append(itm[0] + ": " + str(itm[1])) elements.append("") return "\r\n".join(elements) diff --git a/netlib/wsgi.py b/netlib/wsgi.py index bac27d5a..1b979608 100644 --- a/netlib/wsgi.py +++ b/netlib/wsgi.py @@ -1,5 +1,8 @@ from __future__ import (absolute_import, print_function, division) -import cStringIO, urllib, time, traceback +import cStringIO +import urllib +import time +import traceback from . import odict, tcp @@ -23,15 +26,18 @@ class Request(object): def date_time_string(): """Return the current date and time formatted for a message header.""" WEEKS = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun'] - MONTHS = [None, - 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', - 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec'] + MONTHS = [ + None, + 'Jan', 'Feb', 'Mar', 'Apr', 'May', 'Jun', + 'Jul', 'Aug', 'Sep', 'Oct', 'Nov', 'Dec' + ] now = time.time() year, month, day, hh, mm, ss, wd, y, z = time.gmtime(now) s = "%s, %02d %3s %4d %02d:%02d:%02d GMT" % ( - WEEKS[wd], - day, MONTHS[month], year, - hh, mm, ss) + WEEKS[wd], + day, MONTHS[month], year, + hh, mm, ss + ) return s @@ -100,6 +106,7 @@ class WSGIAdaptor(object): status = None, headers = None ) + def write(data): if not state["headers_sent"]: soc.write("HTTP/1.1 %s\r\n"%state["status"]) @@ -108,7 +115,7 @@ class WSGIAdaptor(object): h["Server"] = [self.sversion] if 'date' not in h: h["Date"] = [date_time_string()] - soc.write(str(h)) + soc.write(h.format()) soc.write("\r\n") state["headers_sent"] = True if data: @@ -130,7 +137,9 @@ class WSGIAdaptor(object): errs = cStringIO.StringIO() try: - dataiter = self.app(self.make_environ(request, errs, **env), start_response) + dataiter = self.app( + self.make_environ(request, errs, **env), start_response + ) for i in dataiter: write(i) if not state["headers_sent"]: @@ -143,5 +152,3 @@ class WSGIAdaptor(object): except Exception: # pragma: no cover pass return errs.getvalue() - - |