add MultiDict

This commit introduces MultiDict, a multi-dictionary similar to
ODict, but with improved semantics (as in the Headers class).
MultiDict fixes a few issues that were present in the Request/Response
API. In particular, `request.cookies["foo"] = "bar"` has previously been a
no-op, as the cookies property returned a mutable _copy_ of the cookies.

10 files changed, 300 insertions, 148 deletions

diff --git a/netlib/encoding.py b/netlib/encoding.py
index 14479e00..98502451 100644
--- a/netlib/encoding.py
+++ b/netlib/encoding.py
@@ -5,7 +5,6 @@ from __future__ import absolute_import
 from io import BytesIO
 import gzip
 import zlib
-from .utils import always_byte_args
 
 
 ENCODINGS = {"identity", "gzip", "deflate"}
diff --git a/netlib/http/cookies.py b/netlib/http/cookies.py
index 4451f1da..fd531146 100644
--- a/netlib/http/cookies.py
+++ b/netlib/http/cookies.py
@@ -1,6 +1,4 @@
-from six.moves import http_cookies as Cookie
 import re
-import string
 from email.utils import parsedate_tz, formatdate, mktime_tz
 
 from .. import odict
@@ -179,20 +177,27 @@ def format_set_cookie_header(name, value, attrs):
     return _format_set_cookie_pairs(pairs)
 
 
+def parse_cookie_headers(cookie_headers):
+    cookie_list = []
+    for header in cookie_headers:
+        cookie_list.extend(parse_cookie_header(header))
+    return cookie_list
+
+
 def parse_cookie_header(line):
     """
         Parse a Cookie header value.
-        Returns a (possibly empty) ODict object.
+        Returns a list of (lhs, rhs) tuples.
     """
     pairs, off_ = _read_pairs(line)
-    return odict.ODict(pairs)
+    return pairs
 
 
-def format_cookie_header(od):
+def format_cookie_header(lst):
     """
         Formats a Cookie header value.
     """
-    return _format_pairs(od.lst)
+    return _format_pairs(lst)
 
 
 def refresh_set_cookie_header(c, delta):
diff --git a/netlib/http/headers.py b/netlib/http/headers.py
index 72739f90..7e39c371 100644
--- a/netlib/http/headers.py
+++ b/netlib/http/headers.py
@@ -1,9 +1,3 @@
-"""
-
-Unicode Handling
-----------------
-See also: http://lucumr.pocoo.org/2013/7/2/the-updated-guide-to-unicode/
-"""
 from __future__ import absolute_import, print_function, division
 
 import re
@@ -13,23 +7,22 @@ try:
 except ImportError:  # pragma: no cover
     from collections import MutableMapping  # Workaround for Python < 3.3
 
-
 import six
+from ..multidict import MultiDict
+from ..utils import always_bytes
 
-from netlib.utils import always_byte_args, always_bytes, Serializable
+# See also: http://lucumr.pocoo.org/2013/7/2/the-updated-guide-to-unicode/
 
 if six.PY2:  # pragma: no cover
     _native = lambda x: x
     _always_bytes = lambda x: x
-    _always_byte_args = lambda x: x
 else:
     # While headers _should_ be ASCII, it's not uncommon for certain headers to be utf-8 encoded.
     _native = lambda x: x.decode("utf-8", "surrogateescape")
     _always_bytes = lambda x: always_bytes(x, "utf-8", "surrogateescape")
-    _always_byte_args = always_byte_args("utf-8", "surrogateescape")
 
 
-class Headers(MutableMapping, Serializable):
+class Headers(MultiDict):
     """
     Header class which allows both convenient access to individual headers as well as
     direct access to the underlying raw data. Provides a full dictionary interface.
@@ -49,7 +42,7 @@ class Headers(MutableMapping, Serializable):
         >>> h["host"]
         "example.com"
 
-        # Headers can also be creatd from a list of raw (header_name, header_value) byte tuples
+        # Headers can also be created from a list of raw (header_name, header_value) byte tuples
         >>> h = Headers([
             [b"Host",b"example.com"],
             [b"Accept",b"text/html"],
@@ -77,7 +70,6 @@ class Headers(MutableMapping, Serializable):
         For use with the "Set-Cookie" header, see :py:meth:`get_all`.
     """
 
-    @_always_byte_args
     def __init__(self, fields=None, **headers):
         """
         Args:
@@ -89,19 +81,25 @@ class Headers(MutableMapping, Serializable):
                 If ``**headers`` contains multiple keys that have equal ``.lower()`` s,
                 the behavior is undefined.
         """
-        self.fields = fields or []
-
-        for name, value in self.fields:
-            if not isinstance(name, bytes) or not isinstance(value, bytes):
-                raise ValueError("Headers passed as fields must be bytes.")
+        super(Headers, self).__init__(fields)
 
         # content_type -> content-type
         headers = {
-            _always_bytes(name).replace(b"_", b"-"): value
+            _always_bytes(name).replace(b"_", b"-"): _always_bytes(value)
             for name, value in six.iteritems(headers)
             }
         self.update(headers)
 
+    @staticmethod
+    def _reduce_values(values):
+        # Headers can be folded
+        return ", ".join(values)
+
+    @staticmethod
+    def _kconv(key):
+        # Headers are case-insensitive
+        return key.lower()
+
     def __bytes__(self):
         if self.fields:
             return b"\r\n".join(b": ".join(field) for field in self.fields) + b"\r\n"
@@ -111,98 +109,40 @@ class Headers(MutableMapping, Serializable):
     if six.PY2:  # pragma: no cover
         __str__ = __bytes__
 
-    @_always_byte_args
-    def __getitem__(self, name):
-        values = self.get_all(name)
-        if not values:
-            raise KeyError(name)
-        return ", ".join(values)
-
-    @_always_byte_args
-    def __setitem__(self, name, value):
-        idx = self._index(name)
-
-        # To please the human eye, we insert at the same position the first existing header occured.
-        if idx is not None:
-            del self[name]
-            self.fields.insert(idx, [name, value])
-        else:
-            self.fields.append([name, value])
-
-    @_always_byte_args
-    def __delitem__(self, name):
-        if name not in self:
-            raise KeyError(name)
-        name = name.lower()
-        self.fields = [
-            field for field in self.fields
-            if name != field[0].lower()
-        ]
+    def __delitem__(self, key):
+        key = _always_bytes(key)
+        super(Headers, self).__delitem__(key)
 
     def __iter__(self):
-        seen = set()
-        for name, _ in self.fields:
-            name_lower = name.lower()
-            if name_lower not in seen:
-                seen.add(name_lower)
-                yield _native(name)
-
-    def __len__(self):
-        return len(set(name.lower() for name, _ in self.fields))
-
-    # __hash__ = object.__hash__
-
-    def _index(self, name):
-        name = name.lower()
-        for i, field in enumerate(self.fields):
-            if field[0].lower() == name:
-                return i
-        return None
-
-    def __eq__(self, other):
-        if isinstance(other, Headers):
-            return self.fields == other.fields
-        return False
-
-    def __ne__(self, other):
-        return not self.__eq__(other)
-
-    @_always_byte_args
+        for x in super(Headers, self).__iter__():
+            yield _native(x)
+
     def get_all(self, name):
         """
         Like :py:meth:`get`, but does not fold multiple headers into a single one.
         This is useful for Set-Cookie headers, which do not support folding.
-
         See also: https://tools.ietf.org/html/rfc7230#section-3.2.2
         """
-        name_lower = name.lower()
-        values = [_native(value) for n, value in self.fields if n.lower() == name_lower]
-        return values
+        name = _always_bytes(name)
+        return [
+            _native(x) for x in
+            super(Headers, self).get_all(name)
+        ]
 
-    @_always_byte_args
     def set_all(self, name, values):
         """
         Explicitly set multiple headers for the given key.
         See: :py:meth:`get_all`
         """
-        values = map(_always_bytes, values)  # _always_byte_args does not fix lists
-        if name in self:
-            del self[name]
-        self.fields.extend(
-            [name, value] for value in values
-        )
-
-    def get_state(self):
-        return tuple(tuple(field) for field in self.fields)
-
-    def set_state(self, state):
-        self.fields = [list(field) for field in state]
+        name = _always_bytes(name)
+        values = [_always_bytes(x) for x in values]
+        return super(Headers, self).set_all(name, values)
 
-    @classmethod
-    def from_state(cls, state):
-        return cls([list(field) for field in state])
+    def insert(self, index, key, value):
+        key = _always_bytes(key)
+        value = _always_bytes(value)
+        super(Headers, self).insert(index, key, value)
 
-    @_always_byte_args
     def replace(self, pattern, repl, flags=0):
         """
         Replaces a regular expression pattern with repl in each "name: value"
@@ -211,6 +151,8 @@ class Headers(MutableMapping, Serializable):
         Returns:
             The number of replacements made.
         """
+        pattern = _always_bytes(pattern)
+        repl = _always_bytes(repl)
         pattern = re.compile(pattern, flags)
         replacements = 0
 
diff --git a/netlib/http/http1/read.py b/netlib/http/http1/read.py
index 6e3a1b93..d30976bd 100644
--- a/netlib/http/http1/read.py
+++ b/netlib/http/http1/read.py
@@ -316,14 +316,14 @@ def _read_headers(rfile):
             if not ret:
                 raise HttpSyntaxException("Invalid headers")
             # continued header
-            ret[-1][1] = ret[-1][1] + b'\r\n ' + line.strip()
+            ret[-1] = (ret[-1][0], ret[-1][1] + b'\r\n ' + line.strip())
         else:
             try:
                 name, value = line.split(b":", 1)
                 value = value.strip()
                 if not name:
                     raise ValueError()
-                ret.append([name, value])
+                ret.append((name, value))
             except ValueError:
                 raise HttpSyntaxException("Invalid headers")
     return Headers(ret)
diff --git a/netlib/http/http2/connections.py b/netlib/http/http2/connections.py
index f900b67c..6643b6b9 100644
--- a/netlib/http/http2/connections.py
+++ b/netlib/http/http2/connections.py
@@ -201,13 +201,13 @@ class HTTP2Protocol(object):
         headers = request.headers.copy()
 
         if ':authority' not in headers:
-            headers.fields.insert(0, (b':authority', authority.encode('ascii')))
+            headers.insert(0, b':authority', authority.encode('ascii'))
         if ':scheme' not in headers:
-            headers.fields.insert(0, (b':scheme', request.scheme.encode('ascii')))
+            headers.insert(0, b':scheme', request.scheme.encode('ascii'))
         if ':path' not in headers:
-            headers.fields.insert(0, (b':path', request.path.encode('ascii')))
+            headers.insert(0, b':path', request.path.encode('ascii'))
         if ':method' not in headers:
-            headers.fields.insert(0, (b':method', request.method.encode('ascii')))
+            headers.insert(0, b':method', request.method.encode('ascii'))
 
         if hasattr(request, 'stream_id'):
             stream_id = request.stream_id
@@ -224,7 +224,7 @@ class HTTP2Protocol(object):
         headers = response.headers.copy()
 
         if ':status' not in headers:
-            headers.fields.insert(0, (b':status', str(response.status_code).encode('ascii')))
+            headers.insert(0, b':status', str(response.status_code).encode('ascii'))
 
         if hasattr(response, 'stream_id'):
             stream_id = response.stream_id
@@ -420,7 +420,7 @@ class HTTP2Protocol(object):
                 self._handle_unexpected_frame(frm)
 
         headers = Headers(
-            [[k.encode('ascii'), v.encode('ascii')] for k, v in self.decoder.decode(header_blocks)]
+            (k.encode('ascii'), v.encode('ascii')) for k, v in self.decoder.decode(header_blocks)
         )
 
         return stream_id, headers, body
diff --git a/netlib/http/message.py b/netlib/http/message.py
index da9681a0..262ef3e1 100644
--- a/netlib/http/message.py
+++ b/netlib/http/message.py
@@ -4,6 +4,7 @@ import warnings
 
 import six
 
+from ..multidict import MultiDict
 from .headers import Headers
 from .. import encoding, utils
 
@@ -235,3 +236,37 @@ class decoded(object):
     def __exit__(self, type, value, tb):
         if self.ce:
             self.message.encode(self.ce)
+
+
+class MessageMultiDict(MultiDict):
+    """
+    A MultiDict that provides a proxy view to the underlying message.
+    """
+
+    def __init__(self, attr, message):
+        if False:
+            # We do not want to call the parent constructor here as that
+            # would cause an unnecessary parse/unparse pass.
+            # This is here to silence linters. Message
+            super(MessageMultiDict, self).__init__(None)
+        self._attr = attr
+        self._message = message  # type: Message
+
+    @staticmethod
+    def _kconv(key):
+        # All request-attributes are case-sensitive.
+        return key
+
+    @staticmethod
+    def _reduce_values(values):
+        # We just return the first element if
+        # multiple elements exist with the same key.
+        return values[0]
+
+    @property
+    def fields(self):
+        return getattr(self._message, "_" + self._attr)
+
+    @fields.setter
+    def fields(self, value):
+        setattr(self._message, self._attr, value)
diff --git a/netlib/http/request.py b/netlib/http/request.py
index a42150ff..26ec12cf 100644
--- a/netlib/http/request.py
+++ b/netlib/http/request.py
@@ -11,7 +11,7 @@ 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
+from .message import Message, _native, _always_bytes, MessageData, MessageMultiDict
 
 # This regex extracts & splits the host header into host and port.
 # Handles the edge case of IPv6 addresses containing colons.
@@ -224,45 +224,54 @@ class Request(Message):
 
     @property
     def query(self):
+        # type: () -> MessageMultiDict
         """
-        The request query string as an :py:class:`ODict` object.
-        None, if there is no query.
+        The request query string as an :py:class:`MessageMultiDict` object.
         """
+        return MessageMultiDict("query", self)
+
+    @property
+    def _query(self):
         _, _, _, _, query, _ = urllib.parse.urlparse(self.url)
-        if query:
-            return ODict(utils.urldecode(query))
-        return None
+        return tuple(utils.urldecode(query))
 
     @query.setter
-    def query(self, odict):
-        query = utils.urlencode(odict.lst)
+    def query(self, value):
+        query = utils.urlencode(value)
         scheme, netloc, path, params, _, fragment = urllib.parse.urlparse(self.url)
         _, _, _, self.path = utils.parse_url(
                 urllib.parse.urlunparse([scheme, netloc, path, params, query, fragment]))
 
     @property
     def cookies(self):
+        # type: () -> MessageMultiDict
         """
         The request cookies.
-        An empty :py:class:`ODict` object if the cookie monster ate them all.
+
+        An empty :py:class:`MessageMultiDict` object if the cookie monster ate them all.
         """
-        ret = ODict()
-        for i in self.headers.get_all("Cookie"):
-            ret.extend(cookies.parse_cookie_header(i))
-        return ret
+        return MessageMultiDict("cookies", self)
+
+    @property
+    def _cookies(self):
+        h = self.headers.get_all("Cookie")
+        return tuple(cookies.parse_cookie_headers(h))
 
     @cookies.setter
-    def cookies(self, odict):
-        self.headers["cookie"] = cookies.format_cookie_header(odict)
+    def cookies(self, value):
+        self.headers["cookie"] = cookies.format_cookie_header(value)
 
     @property
     def path_components(self):
         """
-        The URL's path components as a list of strings.
+        The URL's path components as a tuple of strings.
         Components are unquoted.
         """
         _, _, path, _, _, _ = urllib.parse.urlparse(self.url)
-        return [urllib.parse.unquote(i) for i in path.split("/") if i]
+        # 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)
 
     @path_components.setter
     def path_components(self, components):
@@ -309,34 +318,42 @@ class Request(Message):
     @property
     def urlencoded_form(self):
         """
-        The URL-encoded form data as an :py:class:`ODict` object.
-        None if there is no data or the content-type indicates non-form data.
+        The URL-encoded form data as an :py:class:`MessageMultiDict` object.
+        None if the content-type indicates non-form data.
         """
         is_valid_content_type = "application/x-www-form-urlencoded" in self.headers.get("content-type", "").lower()
-        if self.content and is_valid_content_type:
-            return ODict(utils.urldecode(self.content))
+        if is_valid_content_type:
+            return MessageMultiDict("urlencoded_form", self)
         return None
 
+    @property
+    def _urlencoded_form(self):
+        return tuple(utils.urldecode(self.content))
+
     @urlencoded_form.setter
-    def urlencoded_form(self, odict):
+    def urlencoded_form(self, value):
         """
         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(odict.lst)
+        self.content = utils.urlencode(value)
 
     @property
     def multipart_form(self):
         """
-        The multipart form data as an :py:class:`ODict` object.
-        None if there is no data or the content-type indicates non-form data.
+        The multipart form data as an :py:class:`MultipartFormDict` object.
+        None if the content-type indicates non-form data.
         """
         is_valid_content_type = "multipart/form-data" in self.headers.get("content-type", "").lower()
-        if self.content and is_valid_content_type:
-            return ODict(utils.multipartdecode(self.headers,self.content))
+        if is_valid_content_type:
+            return MessageMultiDict("multipart_form", self)
         return None
 
+    @property
+    def _multipart_form(self):
+        return utils.multipartdecode(self.headers, self.content)
+
     @multipart_form.setter
     def multipart_form(self, value):
         raise NotImplementedError()
diff --git a/netlib/http/response.py b/netlib/http/response.py
index 2f06149e..20074dca 100644
--- a/netlib/http/response.py
+++ b/netlib/http/response.py
@@ -70,6 +70,7 @@ class Response(Message):
     def reason(self, reason):
         self.data.reason = _always_bytes(reason)
 
+    # FIXME
     @property
     def cookies(self):
         """
@@ -88,6 +89,7 @@ class Response(Message):
                 ret.append([name, [value, attrs]])
         return ODict(ret)
 
+    # FIXME
     @cookies.setter
     def cookies(self, odict):
         values = []
diff --git a/netlib/multidict.py b/netlib/multidict.py
new file mode 100644
index 00000000..a7158bc5
--- /dev/null
+++ b/netlib/multidict.py
@@ -0,0 +1,163 @@
+from __future__ import absolute_import, print_function, division

+

+from abc import ABCMeta, abstractmethod

+

+from typing import Tuple

+

+try:

+    from collections.abc import MutableMapping

+except ImportError:  # pragma: no cover

+    from collections import MutableMapping  # Workaround for Python < 3.3

+

+import six

+

+from .utils import Serializable

+

+

+@six.add_metaclass(ABCMeta)

+class MultiDict(MutableMapping, Serializable):

+    def __init__(self, fields=None):

+

+        # it is important for us that .fields is immutable, so that we can easily

+        # detect changes to it.

+        self.fields = tuple(fields) if fields else tuple()  # type: Tuple[Tuple[bytes, bytes], ...]

+

+        for key, value in self.fields:

+            if not isinstance(key, bytes) or not isinstance(value, bytes):

+                raise TypeError("MultiDict fields must be bytes.")

+

+    def __repr__(self):

+        fields = tuple(

+            repr(field)

+            for field in self.fields

+        )

+        return "{cls}[{fields}]".format(

+            cls=type(self).__name__,

+            fields=", ".join(fields)

+        )

+

+    @staticmethod

+    @abstractmethod

+    def _reduce_values(values):

+        pass

+

+    @staticmethod

+    @abstractmethod

+    def _kconv(v):

+        pass

+

+    def __getitem__(self, key):

+        values = self.get_all(key)

+        if not values:

+            raise KeyError(key)

+        return self._reduce_values(values)

+

+    def __setitem__(self, key, value):

+        self.set_all(key, [value])

+

+    def __delitem__(self, key):

+        if key not in self:

+            raise KeyError(key)

+        key = self._kconv(key)

+        self.fields = tuple(

+            field for field in self.fields

+            if key != self._kconv(field[0])

+        )

+

+    def __iter__(self):

+        seen = set()

+        for key, _ in self.fields:

+            key_kconv = self._kconv(key)

+            if key_kconv not in seen:

+                seen.add(key_kconv)

+                yield key

+

+    def __len__(self):

+        return len(set(self._kconv(key) for key, _ in self.fields))

+

+    def __eq__(self, other):

+        if isinstance(other, MultiDict):

+            return self.fields == other.fields

+        return False

+

+    def __ne__(self, other):

+        return not self.__eq__(other)

+

+    def get_all(self, key):

+        """

+        Return the list of items for a given key.

+        If that key is not in the MultiDict,

+        the return value will be an empty list.

+        """

+        key = self._kconv(key)

+        return [

+            value

+            for k, value in self.fields

+            if self._kconv(k) == key

+            ]

+

+    def set_all(self, key, values):

+        """

+        Remove the old values for a key and add new ones.

+        """

+        key_kconv = self._kconv(key)

+

+        new_fields = []

+        for field in self.fields:

+            if self._kconv(field[0]) == key_kconv:

+                if values:

+                    new_fields.append(

+                        (key, values.pop(0))

+                    )

+            else:

+                new_fields.append(field)

+        while values:

+            new_fields.append(

+                (key, values.pop(0))

+            )

+        self.fields = tuple(new_fields)

+

+    def add(self, key, value):

+        self.insert(len(self.fields), key, value)

+

+    def insert(self, index, key, value):

+        item = (key, value)

+        self.fields = self.fields[:index] + (item,) + self.fields[index:]

+

+    def keys(self, multi=False):

+        return (

+            k

+            for k, _ in self.items(multi)

+        )

+

+    def values(self, multi=False):

+        return (

+            v

+            for _, v in self.items(multi)

+        )

+

+    def items(self, multi=False):

+        if multi:

+            return self.fields

+        else:

+            return super(MultiDict, self).items()

+

+    def to_dict(self):

+        d = {}

+        for key in self:

+            values = self.get_all(key)

+            if len(values) == 1:

+                d[key] = values[0]

+            else:

+                d[key] = values

+        return d

+

+    def get_state(self):

+        return self.fields

+

+    def set_state(self, state):

+        self.fields = tuple(tuple(x) for x in state)

+

+    @classmethod

+    def from_state(cls, state):

+        return cls(tuple(x) for x in state)

diff --git a/netlib/utils.py b/netlib/utils.py
index be2701a0..7499f71f 100644
--- a/netlib/utils.py
+++ b/netlib/utils.py
@@ -51,17 +51,6 @@ def always_bytes(unicode_or_bytes, *encode_args):
     return unicode_or_bytes
 
 
-def always_byte_args(*encode_args):
-    """Decorator that transparently encodes all arguments passed as unicode"""
-    def decorator(fun):
-        def _fun(*args, **kwargs):
-            args = [always_bytes(arg, *encode_args) for arg in args]
-            kwargs = {k: always_bytes(v, *encode_args) for k, v in six.iteritems(kwargs)}
-            return fun(*args, **kwargs)
-        return _fun
-    return decorator
-
-
 def native(s, *encoding_opts):
     """
     Convert :py:class:`bytes` or :py:class:`unicode` to the native