Lib/_LWPCookieJar.py

   1 """Load / save to libwww-perl (LWP) format files.
   2
   3 Actually, the format is slightly extended from that used by LWP's
   4 (libwww-perl's) HTTP::Cookies, to avoid losing some RFC 2965 information
   5 not recorded by LWP.
   6
   7 It uses the version string "2.0", though really there isn't an LWP Cookies
   8 2.0 format.  This indicates that there is extra information in here
   9 (domain_dot and # port_spec) while still being compatible with
  10 libwww-perl, I hope.
  11
  12 """
  13
  14 import time, re, logging
  15 from cookielib import (reraise_unmasked_exceptions, FileCookieJar, LoadError,
  16      Cookie, MISSING_FILENAME_TEXT, join_header_words, split_header_words,
  17      iso2time, time2isoz)
  18
  19 def lwp_cookie_str(cookie):
  20     """Return string representation of Cookie in an the LWP cookie file format.
  21
  22     Actually, the format is extended a bit -- see module docstring.
  23
  24     """
  25     h = [(cookie.name, cookie.value),
  26          ("path", cookie.path),
  27          ("domain", cookie.domain)]
  28     if cookie.port is not None: h.append(("port", cookie.port))
  29     if cookie.path_specified: h.append(("path_spec", None))
  30     if cookie.port_specified: h.append(("port_spec", None))
  31     if cookie.domain_initial_dot: h.append(("domain_dot", None))
  32     if cookie.secure: h.append(("secure", None))
  33     if cookie.expires: h.append(("expires",
  34                                time2isoz(float(cookie.expires))))
  35     if cookie.discard: h.append(("discard", None))
  36     if cookie.comment: h.append(("comment", cookie.comment))
  37     if cookie.comment_url: h.append(("commenturl", cookie.comment_url))
  38
  39     keys = cookie._rest.keys()
  40     keys.sort()
  41     for k in keys:
  42         h.append((k, str(cookie._rest[k])))
  43
  44     h.append(("version", str(cookie.version)))
  45
  46     return join_header_words([h])
  47
  48 class LWPCookieJar(FileCookieJar):
  49     """
  50     The LWPCookieJar saves a sequence of"Set-Cookie3" lines.
  51     "Set-Cookie3" is the format used by the libwww-perl libary, not known
  52     to be compatible with any browser, but which is easy to read and
  53     doesn't lose information about RFC 2965 cookies.
  54
  55     Additional methods
  56
  57     as_lwp_str(ignore_discard=True, ignore_expired=True)
  58
  59     """
  60
  61     def as_lwp_str(self, ignore_discard=True, ignore_expires=True):
  62         """Return cookies as a string of "\n"-separated "Set-Cookie3" headers.
  63
  64         ignore_discard and ignore_expires: see docstring for FileCookieJar.save
  65
  66         """
  67         now = time.time()
  68         r = []
  69         for cookie in self:
  70             if not ignore_discard and cookie.discard:
  71                 continue
  72             if not ignore_expires and cookie.is_expired(now):
  73                 continue
  74             r.append("Set-Cookie3: %s" % lwp_cookie_str(cookie))
  75         return "\n".join(r+[""])
  76
  77     def save(self, filename=None, ignore_discard=False, ignore_expires=False):
  78         if filename is None:
  79             if self.filename is not None: filename = self.filename
  80             else: raise ValueError(MISSING_FILENAME_TEXT)
  81
  82         f = open(filename, "w")
  83         try:
  84             # There really isn't an LWP Cookies 2.0 format, but this indicates
  85             # that there is extra information in here (domain_dot and
  86             # port_spec) while still being compatible with libwww-perl, I hope.
  87             f.write("#LWP-Cookies-2.0\n")
  88             f.write(self.as_lwp_str(ignore_discard, ignore_expires))
  89         finally:
  90             f.close()
  91
  92     def _really_load(self, f, filename, ignore_discard, ignore_expires):
  93         magic = f.readline()
  94         if not re.search(self.magic_re, magic):
  95             msg = "%s does not seem to contain cookies" % filename
  96             raise LoadError(msg)
  97
  98         now = time.time()
  99
 100         header = "Set-Cookie3:"
 101         boolean_attrs = ("port_spec", "path_spec", "domain_dot",
 102                          "secure", "discard")
 103         value_attrs = ("version",
 104                        "port", "path", "domain",
 105                        "expires",
 106                        "comment", "commenturl")
 107
 108         try:
 109             while 1:
 110                 line = f.readline()
 111                 if line == "": break
 112                 if not line.startswith(header):
 113                     continue
 114                 line = line[len(header):].strip()
 115
 116                 for data in split_header_words([line]):
 117                     name, value = data[0]
 118                     standard = {}
 119                     rest = {}
 120                     for k in boolean_attrs:
 121                         standard[k] = False
 122                     for k, v in data[1:]:
 123                         if k is not None:
 124                             lc = k.lower()
 125                         else:
 126                             lc = None
 127                         # don't lose case distinction for unknown fields
 128                         if (lc in value_attrs) or (lc in boolean_attrs):
 129                             k = lc
 130                         if k in boolean_attrs:
 131                             if v is None: v = True
 132                             standard[k] = v
 133                         elif k in value_attrs:
 134                             standard[k] = v
 135                         else:
 136                             rest[k] = v
 137
 138                     h = standard.get
 139                     expires = h("expires")
 140                     discard = h("discard")
 141                     if expires is not None:
 142                         expires = iso2time(expires)
 143                     if expires is None:
 144                         discard = True
 145                     domain = h("domain")
 146                     domain_specified = domain.startswith(".")
 147                     c = Cookie(h("version"), name, value,
 148                                h("port"), h("port_spec"),
 149                                domain, domain_specified, h("domain_dot"),
 150                                h("path"), h("path_spec"),
 151                                h("secure"),
 152                                expires,
 153                                discard,
 154                                h("comment"),
 155                                h("commenturl"),
 156                                rest)
 157                     if not ignore_discard and c.discard:
 158                         continue
 159                     if not ignore_expires and c.is_expired(now):
 160                         continue
 161                     self.set_cookie(c)
 162         except:
 163             reraise_unmasked_exceptions((IOError,))
 164             raise LoadError("invalid Set-Cookie3 format file %s" % filename)