Rework slixmpp.jid’s JID classes to make them more efficient.

1 files changed, 120 insertions, 208 deletions

diff --git a/slixmpp/jid.py b/slixmpp/jid.py
index c03c5df7..abb8067a 100644
--- a/slixmpp/jid.py
+++ b/slixmpp/jid.py
@@ -11,12 +11,11 @@
     :license: MIT, see LICENSE for more details
 """
 
-from __future__ import unicode_literals
-
 import re
 import socket
 
 from copy import deepcopy
+from functools import lru_cache
 
 from slixmpp.stringprep import nodeprep, resourceprep, idna, StringprepError
 
@@ -30,22 +29,8 @@ JID_PATTERN = re.compile(
 )
 
 #: The set of escape sequences for the characters not allowed by nodeprep.
-JID_ESCAPE_SEQUENCES = set(['\\20', '\\22', '\\26', '\\27', '\\2f',
-                            '\\3a', '\\3c', '\\3e', '\\40', '\\5c'])
-
-#: A mapping of unallowed characters to their escape sequences. An escape
-#: sequence for '\' is also included since it must also be escaped in
-#: certain situations.
-JID_ESCAPE_TRANSFORMATIONS = {' ': '\\20',
-                              '"': '\\22',
-                              '&': '\\26',
-                              "'": '\\27',
-                              '/': '\\2f',
-                              ':': '\\3a',
-                              '<': '\\3c',
-                              '>': '\\3e',
-                              '@': '\\40',
-                              '\\': '\\5c'}
+JID_ESCAPE_SEQUENCES = {'\\20', '\\22', '\\26', '\\27', '\\2f',
+                        '\\3a', '\\3c', '\\3e', '\\40', '\\5c'}
 
 #: The reverse mapping of escape sequences to their original forms.
 JID_UNESCAPE_TRANSFORMATIONS = {'\\20': ' ',
@@ -60,6 +45,8 @@ JID_UNESCAPE_TRANSFORMATIONS = {'\\20': ' ',
                                 '\\5c': '\\'}
 
 
+# TODO: Find the best cache size for a standard usage.
+@lru_cache(maxsize=1024)
 def _parse_jid(data):
     """
     Parse string data into the node, domain, and resource
@@ -91,17 +78,19 @@ def _validate_node(node):
 
     :returns: The local portion of a JID, as validated by nodeprep.
     """
-    if node is not None:
-        try:
-            node = nodeprep(node)
-        except StringprepError:
-            raise InvalidJID('Nodeprep failed')
+    if node is None:
+        return ''
 
-        if not node:
-            raise InvalidJID('Localpart must not be 0 bytes')
-        if len(node) > 1023:
-            raise InvalidJID('Localpart must be less than 1024 bytes')
-        return node
+    try:
+        node = nodeprep(node)
+    except StringprepError:
+        raise InvalidJID('Nodeprep failed')
+
+    if not node:
+        raise InvalidJID('Localpart must not be 0 bytes')
+    if len(node) > 1023:
+        raise InvalidJID('Localpart must be less than 1024 bytes')
+    return node
 
 
 def _validate_domain(domain):
@@ -170,42 +159,19 @@ def _validate_resource(resource):
 
     :returns: The local portion of a JID, as validated by resourceprep.
     """
-    if resource is not None:
-        try:
-            resource = resourceprep(resource)
-        except StringprepError:
-            raise InvalidJID('Resourceprep failed')
-
-        if not resource:
-            raise InvalidJID('Resource must not be 0 bytes')
-        if len(resource) > 1023:
-            raise InvalidJID('Resource must be less than 1024 bytes')
-        return resource
-
-
-def _escape_node(node):
-    """Escape the local portion of a JID."""
-    result = []
+    if resource is None:
+        return ''
 
-    for i, char in enumerate(node):
-        if char == '\\':
-            if ''.join((node[i:i+3])) in JID_ESCAPE_SEQUENCES:
-                result.append('\\5c')
-                continue
-        result.append(char)
-
-    for i, char in enumerate(result):
-        if char != '\\':
-            result[i] = JID_ESCAPE_TRANSFORMATIONS.get(char, char)
-
-    escaped = ''.join(result)
-
-    if escaped.startswith('\\20') or escaped.endswith('\\20'):
-        raise InvalidJID('Escaped local part starts or ends with "\\20"')
-
-    _validate_node(escaped)
+    try:
+        resource = resourceprep(resource)
+    except StringprepError:
+        raise InvalidJID('Resourceprep failed')
 
-    return escaped
+    if not resource:
+        raise InvalidJID('Resource must not be 0 bytes')
+    if len(resource) > 1023:
+        raise InvalidJID('Resource must be less than 1024 bytes')
+    return resource
 
 
 def _unescape_node(node):
@@ -230,9 +196,7 @@ def _unescape_node(node):
             seq = seq[1:]
         else:
             unescaped.append(char)
-    unescaped = ''.join(unescaped)
-
-    return unescaped
+    return ''.join(unescaped)
 
 
 def _format_jid(local=None, domain=None, resource=None):
@@ -266,47 +230,47 @@ class InvalidJID(ValueError):
     """
 
 # pylint: disable=R0903
-class UnescapedJID(object):
+class UnescapedJID:
 
     """
     .. versionadded:: 1.1.10
     """
 
-    def __init__(self, local, domain, resource):
-        self._jid = (local, domain, resource)
+    __slots__ = ('_node', '_domain', '_resource')
 
-    # pylint: disable=R0911
-    def __getattr__(self, name):
+    def __init__(self, node, domain, resource):
+        self._node = node
+        self._domain = domain
+        self._resource = resource
+
+    def __getattribute__(self, name):
         """Retrieve the given JID component.
 
         :param name: one of: user, server, domain, resource,
                      full, or bare.
         """
         if name == 'resource':
-            return self._jid[2] or ''
-        elif name in ('user', 'username', 'local', 'node'):
-            return self._jid[0] or ''
-        elif name in ('server', 'domain', 'host'):
-            return self._jid[1] or ''
-        elif name in ('full', 'jid'):
-            return _format_jid(*self._jid)
-        elif name == 'bare':
-            return _format_jid(self._jid[0], self._jid[1])
-        elif name == '_jid':
-            return getattr(super(JID, self), '_jid')
-        else:
-            return None
+            return self._resource
+        if name in ('user', 'username', 'local', 'node'):
+            return self._node
+        if name in ('server', 'domain', 'host'):
+            return self._domain
+        if name in ('full', 'jid'):
+            return _format_jid(self._node, self._domain, self._resource)
+        if name == 'bare':
+            return _format_jid(self._node, self._domain)
+        return object.__getattribute__(self, name)
 
     def __str__(self):
         """Use the full JID as the string value."""
-        return _format_jid(*self._jid)
+        return _format_jid(self._node, self._domain, self._resource)
 
     def __repr__(self):
         """Use the full JID as the representation."""
-        return self.__str__()
+        return _format_jid(self._node, self._domain, self._resource)
 
 
-class JID(object):
+class JID:
 
     """
     A representation of a Jabber ID, or JID.
@@ -318,13 +282,13 @@ class JID(object):
     The JID is a full JID otherwise.
 
     **JID Properties:**
-        :jid: Alias for ``full``.
         :full: The string value of the full JID.
+        :jid: Alias for ``full``.
         :bare: The string value of the bare JID.
-        :user: The username portion of the JID.
-        :username: Alias for ``user``.
-        :local: Alias for ``user``.
-        :node: Alias for ``user``.
+        :node: The node portion of the JID.
+        :user: Alias for ``node``.
+        :local: Alias for ``node``.
+        :username: Alias for ``node``.
         :domain: The domain name portion of the JID.
         :server: Alias for ``domain``.
         :host: Alias for ``domain``.
@@ -332,49 +296,23 @@ class JID(object):
 
     :param string jid:
         A string of the form ``'[user@]domain[/resource]'``.
-    :param string local:
-        Optional. Specify the local, or username, portion
-        of the JID. If provided, it will override the local
-        value provided by the `jid` parameter. The given
-        local value will also be escaped if necessary.
-    :param string domain:
-        Optional. Specify the domain of the JID. If
-        provided, it will override the domain given by
-        the `jid` parameter.
-    :param string resource:
-        Optional. Specify the resource value of the JID.
-        If provided, it will override the domain given
-        by the `jid` parameter.
 
     :raises InvalidJID:
     """
 
-    # pylint: disable=W0212
-    def __init__(self, jid=None, **kwargs):
-        in_local = kwargs.get('local', None)
-        in_domain = kwargs.get('domain', None)
-        in_resource = kwargs.get('resource', None)
-        parts = None
-        if in_local or in_domain or in_resource:
-            parts = (in_local, in_domain, in_resource)
+    __slots__ = ('_node', '_domain', '_resource')
 
+    def __init__(self, jid=None):
         if not jid:
-            parsed_jid = (None, None, None)
+            self._node = ''
+            self._domain = ''
+            self._resource = ''
         elif not isinstance(jid, JID):
-            parsed_jid = _parse_jid(jid)
+            self._node, self._domain, self._resource = _parse_jid(jid)
         else:
-            parsed_jid = jid._jid
-
-        local, domain, resource = parsed_jid
-
-        if 'local' in kwargs:
-            local = _escape_node(in_local)
-        if 'domain' in kwargs:
-            domain = _validate_domain(in_domain)
-        if 'resource' in kwargs:
-            resource = _validate_resource(in_resource)
-
-        self._jid = (local, domain, resource)
+            self._node = jid._node
+            self._domain = jid._domain
+            self._resource = jid._resource
 
     def unescape(self):
         """Return an unescaped JID object.
@@ -387,151 +325,125 @@ class JID(object):
 
         .. versionadded:: 1.1.10
         """
-        return UnescapedJID(_unescape_node(self._jid[0]),
-                            self._jid[1],
-                            self._jid[2])
-
-    def regenerate(self):
-        """No-op
-
-        .. deprecated:: 1.1.10
-        """
-        pass
-
-    def reset(self, data):
-        """Start fresh from a new JID string.
-
-        :param string data: A string of the form ``'[user@]domain[/resource]'``.
-
-        .. deprecated:: 1.1.10
-        """
-        self._jid = JID(data)._jid
+        return UnescapedJID(_unescape_node(self._node),
+                            self._domain,
+                            self._resource)
 
     @property
-    def resource(self):
-        return self._jid[2] or ''
+    def node(self):
+        return self._node
 
     @property
     def user(self):
-        return self._jid[0] or ''
+        return self._node
 
     @property
     def local(self):
-        return self._jid[0] or ''
-
-    @property
-    def node(self):
-        return self._jid[0] or ''
+        return self._node
 
     @property
     def username(self):
-        return self._jid[0] or ''
+        return self._node
 
     @property
-    def bare(self):
-        return _format_jid(self._jid[0], self._jid[1])
+    def domain(self):
+        return self._domain
 
     @property
     def server(self):
-        return self._jid[1] or ''
-
-    @property
-    def domain(self):
-        return self._jid[1] or ''
+        return self._domain
 
     @property
     def host(self):
-        return self._jid[1] or ''
+        return self._domain
 
     @property
-    def full(self):
-        return _format_jid(*self._jid)
+    def bare(self):
+        return _format_jid(self._node, self._domain)
 
     @property
-    def jid(self):
-        return _format_jid(*self._jid)
+    def resource(self):
+        return self._resource
 
     @property
-    def bare(self):
-        return _format_jid(self._jid[0], self._jid[1])
+    def full(self):
+        return _format_jid(self._node, self._domain, self._resource)
 
+    @property
+    def jid(self):
+        return _format_jid(self._node, self._domain, self._resource)
 
-    @resource.setter
-    def resource(self, value):
-        self._jid = JID(self, resource=value)._jid
+    @node.setter
+    def node(self, value):
+        self._node = _validate_node(value)
 
     @user.setter
     def user(self, value):
-        self._jid = JID(self, local=value)._jid
-
-    @username.setter
-    def username(self, value):
-        self._jid = JID(self, local=value)._jid
+        self._node = _validate_node(value)
 
     @local.setter
     def local(self, value):
-        self._jid = JID(self, local=value)._jid
-
-    @node.setter
-    def node(self, value):
-        self._jid = JID(self, local=value)._jid
+        self._node = _validate_node(value)
 
-    @server.setter
-    def server(self, value):
-        self._jid = JID(self, domain=value)._jid
+    @username.setter
+    def username(self, value):
+        self._node = _validate_node(value)
 
     @domain.setter
     def domain(self, value):
-        self._jid = JID(self, domain=value)._jid
+        self._domain = _validate_domain(value)
+
+    @server.setter
+    def server(self, value):
+        self._domain = _validate_domain(value)
 
     @host.setter
     def host(self, value):
-        self._jid = JID(self, domain=value)._jid
+        self._domain = _validate_domain(value)
+
+    @bare.setter
+    def bare(self, value):
+        node, domain, resource = _parse_jid(value)
+        assert not resource
+        self._node = node
+        self._domain = domain
+
+    @resource.setter
+    def resource(self, value):
+        self._resource = _validate_resource(value)
 
     @full.setter
     def full(self, value):
-        self._jid = JID(value)._jid
+        self._node, self._domain, self._resource = _parse_jid(value)
 
     @jid.setter
     def jid(self, value):
-        self._jid = JID(value)._jid
-
-    @bare.setter
-    def bare(self, value):
-        parsed = JID(value)._jid
-        self._jid = (parsed[0], parsed[1], self._jid[2])
-
+        self._node, self._domain, self._resource = _parse_jid(value)
 
     def __str__(self):
         """Use the full JID as the string value."""
-        return _format_jid(*self._jid)
+        return _format_jid(self._node, self._domain, self._resource)
 
     def __repr__(self):
         """Use the full JID as the representation."""
-        return self.__str__()
+        return _format_jid(self._node, self._domain, self._resource)
 
     # pylint: disable=W0212
     def __eq__(self, other):
         """Two JIDs are equal if they have the same full JID value."""
         if isinstance(other, UnescapedJID):
             return False
+        if not isinstance(other, JID):
+            other = JID(other)
 
-        other = JID(other)
-        return self._jid == other._jid
+        return (self._node == other._node and
+                self._domain == other._domain and
+                self._resource == other._resource)
 
-    # pylint: disable=W0212
     def __ne__(self, other):
         """Two JIDs are considered unequal if they are not equal."""
         return not self == other
 
     def __hash__(self):
         """Hash a JID based on the string version of its full JID."""
-        return hash(self.__str__())
-
-    def __copy__(self):
-        """Generate a duplicate JID."""
-        return JID(self)
-
-    def __deepcopy__(self, memo):
-        """Generate a duplicate JID."""
-        return JID(deepcopy(str(self), memo))
+        return hash(_format_jid(self._node, self._domain, self._resource))