Peler

: / opt / cpanel-ccs / twistedcaldav /
Filename : memcacheclient.py
back
#!/usr/bin/env python
from __future__ import print_function

"""
client module for memcached (memory cache daemon)

Overview
========

See U{the MemCached homepage<http://www.danga.com/memcached>} for more about memcached.

Usage summary
=============

This should give you a feel for how this module operates::

    import memcacheclient
    mc = memcacheclient.Client(['127.0.0.1:11211'], debug=0)

    mc.set("some_key", "Some value")
    value = mc.get("some_key")

    mc.set("another_key", 3)
    mc.delete("another_key")

    mc.set("key", "1")   # note that the key used for incr/decr must be a string.
    mc.incr("key")
    mc.decr("key")

The standard way to use memcache with a database is like this::

    key = derive_key(obj)
    obj = mc.get(key)
    if not obj:
        obj = backend_api.get(...)
        mc.set(obj)

    # we now have obj, and future passes through this code
    # will use the object from the cache.

Detailed Documentation
======================

More detailed documentation is available in the L{Client} class.
"""

import sys
import socket
import time
import os
import re
import types

from twext.python.log import Logger

from twistedcaldav.config import config

log = Logger()


try:
    import cPickle as pickle
except ImportError:
    import pickle

try:
    from zlib import compress, decompress
    _supports_compress = True
except ImportError:
    _supports_compress = False
    # quickly define a decompress just in case we recv compressed data.

    def decompress(val):
        raise _Error("received compressed data but I don't support compession (import error)")

try:
    from cStringIO import StringIO
except ImportError:
    from StringIO import StringIO

from binascii import crc32   # zlib version is not cross-platform
serverHashFunction = crc32

__author__ = "Evan Martin <martine@danga.com>"
__version__ = "1.44"
__copyright__ = "Copyright (C) 2003 Danga Interactive"
__license__ = "Python"

SERVER_MAX_KEY_LENGTH = 250
#  Storing values larger than 1MB requires recompiling memcached.  If you do,
#  this value can be changed by doing "memcacheclient.SERVER_MAX_VALUE_LENGTH = N"
#  after importing this module.
SERVER_MAX_VALUE_LENGTH = 1024 * 1024


class _Error(Exception):
    pass


class MemcacheError(_Error):
    """
    Memcache connection error
    """


class NotFoundError(MemcacheError):
    """
    NOT_FOUND error
    """


class TokenMismatchError(MemcacheError):
    """
    Check-and-set token mismatch
    """

try:
    # Only exists in Python 2.4+
    from threading import local
except ImportError:
    # TODO:  add the pure-python local implementation
    class local(object):
        pass


class ClientFactory(object):

    # unit tests should set this to True to enable the fake test cache
    allowTestCache = False

    @classmethod
    def getClient(
        cls, servers, debug=0, pickleProtocol=0,
        pickler=pickle.Pickler, unpickler=pickle.Unpickler,
        pload=None, pid=None
    ):

        if cls.allowTestCache:
            return TestClient(
                servers, debug=debug,
                pickleProtocol=pickleProtocol, pickler=pickler,
                unpickler=unpickler, pload=pload, pid=pid)
        elif config.Memcached.Pools.Default.ClientEnabled:
            return Client(
                servers, debug=debug, pickleProtocol=pickleProtocol,
                pickler=pickler, unpickler=unpickler, pload=pload, pid=pid)
        else:
            return None


class Client(local):
    """
    Object representing a pool of memcache servers.

    See L{memcache} for an overview.

    In all cases where a key is used, the key can be either:
        1. A simple hashable type (string, integer, etc.).
        2. A tuple of C{(hashvalue, key)}.  This is useful if you want to avoid
        making this module calculate a hash value.  You may prefer, for
        example, to keep all of a given user's objects on the same memcache
        server, so you could use the user's unique id as the hash value.

    @group Setup: __init__, set_servers, forget_dead_hosts, disconnect_all, debuglog
    @group Insertion: set, add, replace, set_multi
    @group Retrieval: get, get_multi
    @group Integers: incr, decr
    @group Removal: delete, delete_multi
    @sort: __init__, set_servers, forget_dead_hosts, disconnect_all, debuglog,\
           set, set_multi, add, replace, get, get_multi, incr, decr, delete, delete_multi
    """
    _FLAG_PICKLE = 1 << 0
    _FLAG_INTEGER = 1 << 1
    _FLAG_LONG = 1 << 2
    _FLAG_COMPRESSED = 1 << 3

    _SERVER_RETRIES = 10  # how many times to try finding a free server.

    # exceptions for Client
    class MemcachedKeyError(Exception):
        pass

    class MemcachedKeyLengthError(MemcachedKeyError):
        pass

    class MemcachedKeyCharacterError(MemcachedKeyError):
        pass

    class MemcachedKeyNoneError(MemcachedKeyError):
        pass

    class MemcachedKeyTypeError(MemcachedKeyError):
        pass

    class MemcachedStringEncodingError(Exception):
        pass

    def __init__(self, servers, debug=0, pickleProtocol=0,
                 pickler=pickle.Pickler, unpickler=pickle.Unpickler,
                 pload=None, pid=None):
        """
        Create a new Client object with the given list of servers.

        @param servers: C{servers} is passed to L{set_servers}.
        @param debug: whether to display error messages when a server can't be
        contacted.
        @param pickleProtocol: number to mandate protocol used by (c)Pickle.
        @param pickler: optional override of default Pickler to allow subclassing.
        @param unpickler: optional override of default Unpickler to allow subclassing.
        @param pload: optional persistent_load function to call on pickle loading.
        Useful for cPickle since subclassing isn't allowed.
        @param pid: optional persistent_id function to call on pickle storing.
        Useful for cPickle since subclassing isn't allowed.
        """
        local.__init__(self)
        self.set_servers(servers)
        self.debug = debug
        self.stats = {}

        # Allow users to modify pickling/unpickling behavior
        self.pickleProtocol = pickleProtocol
        self.pickler = pickler
        self.unpickler = unpickler
        self.persistent_load = pload
        self.persistent_id = pid

        #  figure out the pickler style
        file = StringIO()
        try:
            pickler = self.pickler(file, protocol=self.pickleProtocol)
            self.picklerIsKeyword = True
        except TypeError:
            self.picklerIsKeyword = False

    def set_servers(self, servers):
        """
        Set the pool of servers used by this client.

        @param servers: an array of servers.
        Servers can be passed in two forms:
            1. Strings of the form C{"host:port"}, which implies a default weight of 1.
            2. Tuples of the form C{("host:port", weight)}, where C{weight} is
            an integer weight value.
        """
        self.servers = [_Host(s, self.debuglog) for s in servers]
        self._init_buckets()

    def get_stats(self):
        '''Get statistics from each of the servers.

        @return: A list of tuples ( server_identifier, stats_dictionary ).
            The dictionary contains a number of name/value pairs specifying
            the name of the status field and the string value associated with
            it.  The values are not converted from strings.
        '''
        data = []
        for s in self.servers:
            if not s.connect():
                continue
            if s.family == socket.AF_INET:
                name = '%s:%s (%s)' % (s.ip, s.port, s.weight)
            else:
                name = 'unix:%s (%s)' % (s.address, s.weight)
            s.send_cmd('stats')
            serverData = {}
            data.append((name, serverData))
            readline = s.readline
            while 1:
                line = readline()
                if not line or line.strip() == 'END':
                    break
                stats = line.split(' ', 2)
                serverData[stats[1]] = stats[2]

        return(data)

    def get_slabs(self):
        data = []
        for s in self.servers:
            if not s.connect():
                continue
            if s.family == socket.AF_INET:
                name = '%s:%s (%s)' % (s.ip, s.port, s.weight)
            else:
                name = 'unix:%s (%s)' % (s.address, s.weight)
            serverData = {}
            data.append((name, serverData))
            s.send_cmd('stats items')
            readline = s.readline
            while 1:
                line = readline()
                if not line or line.strip() == 'END':
                    break
                item = line.split(' ', 2)
                # 0 = STAT, 1 = ITEM, 2 = Value
                slab = item[1].split(':', 2)
                # 0 = items, 1 = Slab #, 2 = Name
                if slab[1] not in serverData:
                    serverData[slab[1]] = {}
                serverData[slab[1]][slab[2]] = item[2]
        return data

    def flush_all(self):
        'Expire all data currently in the memcache servers.'
        for s in self.servers:
            if not s.connect():
                continue
            s.send_cmd('flush_all')
            s.expect("OK")

    def debuglog(self, str):
        if self.debug:
            sys.stderr.write("MemCached: %s\n" % str)

    def _statlog(self, func):
        if func not in self.stats:
            self.stats[func] = 1
        else:
            self.stats[func] += 1

    def forget_dead_hosts(self):
        """
        Reset every host in the pool to an "alive" state.
        """
        for s in self.servers:
            s.deaduntil = 0

    def _init_buckets(self):
        self.buckets = []
        for server in self.servers:
            for _ignroe_i in range(server.weight):
                self.buckets.append(server)

    def _get_server(self, key):
        if isinstance(key, tuple):
            serverhash, key = key
        else:
            serverhash = serverHashFunction(key)

        for i in range(Client._SERVER_RETRIES):
            server = self.buckets[serverhash % len(self.buckets)]
            if server.connect():
                # print("(using server %s)" % server, end="")
                return server, key
            serverhash = serverHashFunction(str(serverhash) + str(i))
        log.error("Memcacheclient _get_server( ) failed to connect")
        return None, None

    def disconnect_all(self):
        for s in self.servers:
            s.close_socket()

    def delete_multi(self, keys, time=0, key_prefix=''):
        '''
        Delete multiple keys in the memcache doing just one query.

        >>> notset_keys = mc.set_multi({'key1' : 'val1', 'key2' : 'val2'})
        >>> mc.get_multi(['key1', 'key2']) == {'key1' : 'val1', 'key2' : 'val2'}
        1
        >>> mc.delete_multi(['key1', 'key2'])
        1
        >>> mc.get_multi(['key1', 'key2']) == {}
        1


        This method is recommended over iterated regular L{delete}s as it reduces total latency, since
        your app doesn't have to wait for each round-trip of L{delete} before sending
        the next one.

        @param keys: An iterable of keys to clear
        @param time: number of seconds any subsequent set / update commands should fail. Defaults to 0 for no delay.
        @param key_prefix:  Optional string to prepend to each key when sending to memcache.
            See docs for L{get_multi} and L{set_multi}.

        @return: 1 if no failure in communication with any memcacheds.
        @rtype: int

        '''

        self._statlog('delete_multi')

        server_keys, _ignore_prefixed_to_orig_key = self._map_and_prefix_keys(keys, key_prefix)

        # send out all requests on each server before reading anything
        dead_servers = []

        rc = 1
        for server in server_keys.iterkeys():
            bigcmd = []
            write = bigcmd.append
            if time != None:
                for key in server_keys[server]:  # These are mangled keys
                    write("delete %s %d\r\n" % (key, time))
            else:
                for key in server_keys[server]:  # These are mangled keys
                    write("delete %s\r\n" % key)
            try:
                server.send_cmds(''.join(bigcmd))
            except socket.error, msg:
                rc = 0
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
                dead_servers.append(server)

        # if any servers died on the way, don't expect them to respond.
        for server in dead_servers:
            del server_keys[server]

        for server, keys in server_keys.iteritems():
            try:
                for key in keys:
                    server.expect("DELETED")
            except socket.error, msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
                rc = 0
        return rc

    def delete(self, key, time=0):
        '''Deletes a key from the memcache.

        @return: Nonzero on success.
        @param time: number of seconds any subsequent set / update commands should fail. Defaults to 0 for no delay.
        @rtype: int
        '''
        check_key(key)
        server, key = self._get_server(key)
        if not server:
            return 0
        self._statlog('delete')
        if time != None:
            cmd = "delete %s %d" % (key, time)
        else:
            cmd = "delete %s" % key

        try:
            server.send_cmd(cmd)
            server.expect("DELETED")
        except socket.error, msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            server.mark_dead(msg)
            return 0
        return 1

    def incr(self, key, delta=1):
        """
        Sends a command to the server to atomically increment the value for C{key} by
        C{delta}, or by 1 if C{delta} is unspecified.  Returns None if C{key} doesn't
        exist on server, otherwise it returns the new value after incrementing.

        Note that the value for C{key} must already exist in the memcache, and it
        must be the string representation of an integer.

        >>> mc.set("counter", "20")  # returns 1, indicating success
        1
        >>> mc.incr("counter")
        21
        >>> mc.incr("counter")
        22

        Overflow on server is not checked.  Be aware of values approaching
        2**32.  See L{decr}.

        @param delta: Integer amount to increment by (should be zero or greater).
        @return: New value after incrementing.
        @rtype: int
        """
        return self._incrdecr("incr", key, delta)

    def decr(self, key, delta=1):
        """
        Like L{incr}, but decrements.  Unlike L{incr}, underflow is checked and
        new values are capped at 0.  If server value is 1, a decrement of 2
        returns 0, not -1.

        @param delta: Integer amount to decrement by (should be zero or greater).
        @return: New value after decrementing.
        @rtype: int
        """
        return self._incrdecr("decr", key, delta)

    def _incrdecr(self, cmd, key, delta):
        check_key(key)
        server, key = self._get_server(key)
        if not server:
            return 0
        self._statlog(cmd)
        cmd = "%s %s %d" % (cmd, key, delta)
        try:
            server.send_cmd(cmd)
            line = server.readline()
            return int(line)
        except socket.error, msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            server.mark_dead(msg)
            return None

    def add(self, key, val, time=0, min_compress_len=0):
        '''
        Add new key with value.

        Like L{set}, but only stores in memcache if the key doesn't already exist.

        @return: Nonzero on success.
        @rtype: int
        '''
        return self._set("add", key, val, time, min_compress_len)

    def append(self, key, val, time=0, min_compress_len=0):
        '''Append the value to the end of the existing key's value.

        Only stores in memcache if key already exists.
        Also see L{prepend}.

        @return: Nonzero on success.
        @rtype: int
        '''
        return self._set("append", key, val, time, min_compress_len)

    def prepend(self, key, val, time=0, min_compress_len=0):
        '''Prepend the value to the beginning of the existing key's value.

        Only stores in memcache if key already exists.
        Also see L{append}.

        @return: Nonzero on success.
        @rtype: int
        '''
        return self._set("prepend", key, val, time, min_compress_len)

    def replace(self, key, val, time=0, min_compress_len=0):
        '''Replace existing key with value.

        Like L{set}, but only stores in memcache if the key already exists.
        The opposite of L{add}.

        @return: Nonzero on success.
        @rtype: int
        '''
        return self._set("replace", key, val, time, min_compress_len)

    def set(self, key, val, time=0, min_compress_len=0, token=None):
        '''Unconditionally sets a key to a given value in the memcache.

        The C{key} can optionally be an tuple, with the first element
        being the server hash value and the second being the key.
        If you want to avoid making this module calculate a hash value.
        You may prefer, for example, to keep all of a given user's objects
        on the same memcache server, so you could use the user's unique
        id as the hash value.

        @return: Nonzero on success.
        @rtype: int
        @param time: Tells memcached the time which this value should expire, either
        as a delta number of seconds, or an absolute unix time-since-the-epoch
        value. See the memcached protocol docs section "Storage Commands"
        for more info on <exptime>. We default to 0 == cache forever.
        @param min_compress_len: The threshold length to kick in auto-compression
        of the value using the zlib.compress() routine. If the value being cached is
        a string, then the length of the string is measured, else if the value is an
    , then the length of the pickle result is measured. If the resulting
        attempt at compression yeilds a larger string than the input, then it is
        discarded. For backwards compatability, this parameter defaults to 0,
        indicating don't ever try to compress.
        '''
        return self._set("set", key, val, time, min_compress_len, token=token)

    def _map_and_prefix_keys(self, key_iterable, key_prefix):
        """Compute the mapping of server (_Host instance) -> list of keys to stuff onto that server, as well as the mapping of
        prefixed key -> original key.


        """
        # Check it just once ...
        key_extra_len = len(key_prefix)
        if key_prefix:
            check_key(key_prefix)

        # server (_Host) -> list of unprefixed server keys in mapping
        server_keys = {}

        prefixed_to_orig_key = {}
        # build up a list for each server of all the keys we want.
        for orig_key in key_iterable:
            if isinstance(orig_key, tuple):
                # Tuple of hashvalue, key ala _get_server(). Caller is essentially telling us what server to stuff this on.
                # Ensure call to _get_server gets a Tuple as well.
                str_orig_key = str(orig_key[1])
                server, key = self._get_server((orig_key[0], key_prefix + str_orig_key))  # Gotta pre-mangle key before hashing to a server. Returns the mangled key.
            else:
                str_orig_key = str(orig_key)  # set_multi supports int / long keys.
                server, key = self._get_server(key_prefix + str_orig_key)

            # Now check to make sure key length is proper ...
            check_key(str_orig_key, key_extra_len=key_extra_len)

            if not server:
                continue

            if server not in server_keys:
                server_keys[server] = []
            server_keys[server].append(key)
            prefixed_to_orig_key[key] = orig_key

        return (server_keys, prefixed_to_orig_key)

    def set_multi(self, mapping, time=0, key_prefix='', min_compress_len=0):
        '''
        Sets multiple keys in the memcache doing just one query.

        >>> notset_keys = mc.set_multi({'key1' : 'val1', 'key2' : 'val2'})
        >>> mc.get_multi(['key1', 'key2']) == {'key1' : 'val1', 'key2' : 'val2'}
        1


        This method is recommended over regular L{set} as it lowers the number of
        total packets flying around your network, reducing total latency, since
        your app doesn't have to wait for each round-trip of L{set} before sending
        the next one.

        @param mapping: A dict of key/value pairs to set.
        @param time: Tells memcached the time which this value should expire, either
        as a delta number of seconds, or an absolute unix time-since-the-epoch
        value. See the memcached protocol docs section "Storage Commands"
        for more info on <exptime>. We default to 0 == cache forever.
        @param key_prefix:  Optional string to prepend to each key when sending to memcache. Allows you to efficiently stuff these keys into a pseudo-namespace in memcache:
            >>> notset_keys = mc.set_multi({'key1' : 'val1', 'key2' : 'val2'}, key_prefix='subspace_')
            >>> len(notset_keys) == 0
            True
            >>> mc.get_multi(['subspace_key1', 'subspace_key2']) == {'subspace_key1' : 'val1', 'subspace_key2' : 'val2'}
            True

            Causes key 'subspace_key1' and 'subspace_key2' to be set. Useful in conjunction with a higher-level layer which applies namespaces to data in memcache.
            In this case, the return result would be the list of notset original keys, prefix not applied.

        @param min_compress_len: The threshold length to kick in auto-compression
        of the value using the zlib.compress() routine. If the value being cached is
        a string, then the length of the string is measured, else if the value is an
        object, then the length of the pickle result is measured. If the resulting
        attempt at compression yeilds a larger string than the input, then it is
        discarded. For backwards compatability, this parameter defaults to 0,
        indicating don't ever try to compress.
        @return: List of keys which failed to be stored [ memcache out of memory, etc. ].
        @rtype: list

        '''

        self._statlog('set_multi')

        server_keys, prefixed_to_orig_key = self._map_and_prefix_keys(mapping.iterkeys(), key_prefix)

        # send out all requests on each server before reading anything
        dead_servers = []

        for server in server_keys.iterkeys():
            bigcmd = []
            write = bigcmd.append
            try:
                for key in server_keys[server]:  # These are mangled keys
                    store_info = self._val_to_store_info(mapping[prefixed_to_orig_key[key]], min_compress_len)
                    write("set %s %d %d %d\r\n%s\r\n" % (key, store_info[0], time, store_info[1], store_info[2]))
                server.send_cmds(''.join(bigcmd))
            except socket.error, msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
                dead_servers.append(server)

        # if any servers died on the way, don't expect them to respond.
        for server in dead_servers:
            del server_keys[server]

        #  short-circuit if there are no servers, just return all keys
        if not server_keys:
            return(mapping.keys())

        notstored = []  # original keys.
        for server, keys in server_keys.iteritems():
            try:
                for key in keys:
                    line = server.readline()
                    if line == 'STORED':
                        continue
                    else:
                        notstored.append(prefixed_to_orig_key[key])  # un-mangle.
            except (_Error, socket.error), msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
        return notstored

    def _val_to_store_info(self, val, min_compress_len):
        """
           Transform val to a storable representation, returning a tuple of the flags, the length of the new value, and the new value itself.
        """
        flags = 0
        if isinstance(val, str):
            pass
        elif isinstance(val, int):
            flags |= Client._FLAG_INTEGER
            val = "%d" % val
            # force no attempt to compress this silly string.
            min_compress_len = 0
        elif isinstance(val, long):
            flags |= Client._FLAG_LONG
            val = "%d" % val
            # force no attempt to compress this silly string.
            min_compress_len = 0
        else:
            flags |= Client._FLAG_PICKLE
            file = StringIO()
            if self.picklerIsKeyword:
                pickler = self.pickler(file, protocol=self.pickleProtocol)
            else:
                pickler = self.pickler(file, self.pickleProtocol)
            if self.persistent_id:
                pickler.persistent_id = self.persistent_id
            pickler.dump(val)
            val = file.getvalue()

        lv = len(val)
        # We should try to compress if min_compress_len > 0 and we could
        # import zlib and this string is longer than our min threshold.
        if min_compress_len and _supports_compress and lv > min_compress_len:
            comp_val = compress(val)
            # Only retain the result if the compression result is smaller
            # than the original.
            if len(comp_val) < lv:
                flags |= Client._FLAG_COMPRESSED
                val = comp_val

        #  silently do not store if value length exceeds maximum
        if len(val) >= SERVER_MAX_VALUE_LENGTH:
            return(0)

        return (flags, len(val), val)

    def _set(self, cmd, key, val, time, min_compress_len=0, token=None):
        check_key(key)
        server, key = self._get_server(key)
        if not server:
            raise MemcacheError("Memcache connection error")

        self._statlog(cmd)

        store_info = self._val_to_store_info(val, min_compress_len)
        if not store_info:
            return(0)

        if token is not None:
            cmd = "cas"
            fullcmd = "cas %s %d %d %d %s\r\n%s" % (key, store_info[0], time, store_info[1], token, store_info[2])
        else:
            fullcmd = "%s %s %d %d %d\r\n%s" % (cmd, key, store_info[0], time, store_info[1], store_info[2])
        try:
            server.send_cmd(fullcmd)
            result = server.expect("STORED")

            if (result == "STORED"):
                return True

            if (result == "NOT_FOUND"):
                raise NotFoundError(key)

            if token and result == "EXISTS":
                log.debug("Memcacheclient check-and-set failed")
                raise TokenMismatchError(key)

            log.error("Memcacheclient {cmd} command failed with result ({r})", cmd=cmd, r=result)

            return False

        except socket.error, msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            server.mark_dead(msg)
        return 0

    def get(self, key):
        '''Retrieves a key from the memcache.

        @return: The value or None.
        '''
        check_key(key)
        server, key = self._get_server(key)
        if not server:
            raise MemcacheError("Memcache connection error")

        self._statlog('get')

        try:
            server.send_cmd("get %s" % key)
            rkey, flags, rlen, = self._expectvalue(server)
            if not rkey:
                return None
            value = self._recv_value(server, flags, rlen)
            server.expect("END")
        except (_Error, socket.error), msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            server.mark_dead(msg)
            raise MemcacheError("Memcache connection error")
        return value

    def gets(self, key):
        '''Retrieves a key from the memcache.

        @return: The value or None.
        '''
        check_key(key)
        server, key = self._get_server(key)
        if not server:
            raise MemcacheError("Memcache connection error")

        self._statlog('get')

        try:
            server.send_cmd("gets %s" % key)
            rkey, flags, rlen, cas_token = self._expectvalue_cas(server)
            if not rkey:
                return (None, None)
            value = self._recv_value(server, flags, rlen)
            server.expect("END")
        except (_Error, socket.error), msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            server.mark_dead(msg)
            raise MemcacheError("Memcache connection error")
        return (value, cas_token)

    def get_multi(self, keys, key_prefix=''):
        '''
        Retrieves multiple keys from the memcache doing just one query.

        >>> success = mc.set("foo", "bar")
        >>> success = mc.set("baz", 42)
        >>> mc.get_multi(["foo", "baz", "foobar"]) == {"foo": "bar", "baz": 42}
        1
        >>> mc.set_multi({'k1' : 1, 'k2' : 2}, key_prefix='pfx_') == []
        1

        This looks up keys 'pfx_k1', 'pfx_k2', ... . Returned dict will just have unprefixed keys 'k1', 'k2'.
        >>> mc.get_multi(['k1', 'k2', 'nonexist'], key_prefix='pfx_') == {'k1' : 1, 'k2' : 2}
        1

        get_mult [ and L{set_multi} ] can take str()-ables like ints / longs as keys too. Such as your db pri key fields.
        They're rotored through str() before being passed off to memcache, with or without the use of a key_prefix.
        In this mode, the key_prefix could be a table name, and the key itself a db primary key number.

        >>> mc.set_multi({42: 'douglass adams', 46 : 'and 2 just ahead of me'}, key_prefix='numkeys_') == []
        1
        >>> mc.get_multi([46, 42], key_prefix='numkeys_') == {42: 'douglass adams', 46 : 'and 2 just ahead of me'}
        1

        This method is recommended over regular L{get} as it lowers the number of
        total packets flying around your network, reducing total latency, since
        your app doesn't have to wait for each round-trip of L{get} before sending
        the next one.

        See also L{set_multi}.

        @param keys: An array of keys.
        @param key_prefix: A string to prefix each key when we communicate with memcache.
            Facilitates pseudo-namespaces within memcache. Returned dictionary keys will not have this prefix.
        @return:  A dictionary of key/value pairs that were available. If key_prefix was provided, the keys in the retured dictionary will not have it present.

        '''

        self._statlog('get_multi')

        server_keys, prefixed_to_orig_key = self._map_and_prefix_keys(keys, key_prefix)

        # send out all requests on each server before reading anything
        dead_servers = []
        for server in server_keys.iterkeys():
            try:
                server.send_cmd("get %s" % " ".join(server_keys[server]))
            except socket.error, msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
                dead_servers.append(server)

        # if any servers died on the way, don't expect them to respond.
        for server in dead_servers:
            del server_keys[server]

        retvals = {}
        for server in server_keys.iterkeys():
            try:
                line = server.readline()
                while line and line != 'END':
                    rkey, flags, rlen = self._expectvalue(server, line)
                    #  Bo Yang reports that this can sometimes be None
                    if rkey is not None:
                        val = self._recv_value(server, flags, rlen)
                        try:
                            retvals[prefixed_to_orig_key[rkey]] = val   # un-prefix returned key.
                        except KeyError:
                            pass
                    line = server.readline()
            except (_Error, socket.error), msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
        return retvals

    def gets_multi(self, keys, key_prefix=''):
        '''
        Retrieves multiple keys from the memcache doing just one query.
        See also L{gets} and L{get_multi}.
        '''

        self._statlog('gets_multi')

        server_keys, prefixed_to_orig_key = self._map_and_prefix_keys(keys, key_prefix)

        # send out all requests on each server before reading anything
        dead_servers = []
        for server in server_keys.iterkeys():
            try:
                server.send_cmd("gets %s" % " ".join(server_keys[server]))
            except socket.error, msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
                dead_servers.append(server)

        # if any servers died on the way, don't expect them to respond.
        for server in dead_servers:
            del server_keys[server]

        retvals = {}
        for server in server_keys.iterkeys():
            try:
                line = server.readline()
                while line and line != 'END':
                    rkey, flags, rlen, cas_token = self._expectvalue_cas(server, line)
                    #  Bo Yang reports that this can sometimes be None
                    if rkey is not None:
                        val = self._recv_value(server, flags, rlen)
                        try:
                            retvals[prefixed_to_orig_key[rkey]] = (val, cas_token)   # un-prefix returned key.
                        except KeyError:
                            pass
                    line = server.readline()
            except (_Error, socket.error), msg:
                if isinstance(msg, tuple):
                    msg = msg[1]
                server.mark_dead(msg)
        return retvals

    def _expectvalue(self, server, line=None):
        if not line:
            line = server.readline()

        if line[:5] == 'VALUE':
            _ignore_resp, rkey, flags, len = line.split()
            flags = int(flags)
            rlen = int(len)
            return (rkey, flags, rlen)
        else:
            return (None, None, None)

    def _expectvalue_cas(self, server, line=None):
        if not line:
            line = server.readline()

        if line[:5] == 'VALUE':
            _ignore_resp, rkey, flags, len, rtoken = line.split()
            flags = int(flags)
            rlen = int(len)
            return (rkey, flags, rlen, rtoken)
        else:
            return (None, None, None, None)

    def _recv_value(self, server, flags, rlen):
        rlen += 2  # include \r\n
        buf = server.recv(rlen)
        if len(buf) != rlen:
            raise _Error("received %d bytes when expecting %d" % (len(buf), rlen))

        if len(buf) == rlen:
            buf = buf[:-2]  # strip \r\n

        if flags & Client._FLAG_COMPRESSED:
            buf = decompress(buf)

        if flags == 0 or flags == Client._FLAG_COMPRESSED:
            # Either a bare string or a compressed string now decompressed...
            val = buf
        elif flags & Client._FLAG_INTEGER:
            val = int(buf)
        elif flags & Client._FLAG_LONG:
            val = long(buf)
        elif flags & Client._FLAG_PICKLE:
            try:
                file = StringIO(buf)
                unpickler = self.unpickler(file)
                if self.persistent_load:
                    unpickler.persistent_load = self.persistent_load
                val = unpickler.load()
            except Exception, e:
                self.debuglog('Pickle error: %s\n' % e)
                val = None
        else:
            self.debuglog("unknown flags on get: %x\n" % flags)

        return val


class TestClient(Client):
    """
    Fake memcache client for unit tests

    """

    def __init__(self, servers, debug=0, pickleProtocol=0,
                 pickler=pickle.Pickler, unpickler=pickle.Unpickler,
                 pload=None, pid=None):

        local.__init__(self)

        super(TestClient, self).__init__(
            servers, debug=debug,
            pickleProtocol=pickleProtocol, pickler=pickler, unpickler=unpickler,
            pload=pload, pid=pid)

        self.data = {}
        self.token = 0

    def get_stats(self):
        raise NotImplementedError()

    def get_slabs(self):
        raise NotImplementedError()

    def flush_all(self):
        raise NotImplementedError()

    def forget_dead_hosts(self):
        raise NotImplementedError()

    def delete_multi(self, keys, time=0, key_prefix=''):
        '''
        Delete multiple keys in the memcache doing just one query.

        >>> notset_keys = mc.set_multi({'key1' : 'val1', 'key2' : 'val2'})
        >>> mc.get_multi(['key1', 'key2']) == {'key1' : 'val1', 'key2' : 'val2'}
        1
        >>> mc.delete_multi(['key1', 'key2'])
        1
        >>> mc.get_multi(['key1', 'key2']) == {}
        1
        '''

        self._statlog('delete_multi')
        for key in keys:
            key = key_prefix + key
            del self.data[key]
        return 1

    def delete(self, key, time=0):
        '''Deletes a key from the memcache.

        @return: Nonzero on success.
        @param time: number of seconds any subsequent set / update commands should fail. Defaults to 0 for no delay.
        @rtype: int
        '''
        check_key(key)
        del self.data[key]
        return 1

    def incr(self, key, delta=1):
        raise NotImplementedError()

    def decr(self, key, delta=1):
        raise NotImplementedError()

    def add(self, key, val, time=0, min_compress_len=0):
        raise NotImplementedError()

    def append(self, key, val, time=0, min_compress_len=0):
        raise NotImplementedError()

    def prepend(self, key, val, time=0, min_compress_len=0):
        raise NotImplementedError()

    def replace(self, key, val, time=0, min_compress_len=0):
        raise NotImplementedError()

    def set(self, key, val, time=0, min_compress_len=0, token=None):
        self._statlog('set')
        return self._set("set", key, val, time, min_compress_len, token=token)

    def set_multi(self, mapping, time=0, key_prefix='', min_compress_len=0):
        self._statlog('set_multi')
        for key, val in mapping.iteritems():
            key = key_prefix + key
            self._set("set", key, val, time, min_compress_len)
        return []

    def _set(self, cmd, key, val, time, min_compress_len=0, token=None):
        check_key(key)
        self._statlog(cmd)

        serialized = pickle.dumps(val, pickle.HIGHEST_PROTOCOL)

        if token is not None:
            if key in self.data:
                _ignore_stored_val, stored_token = self.data[key]
                if token != stored_token:
                    raise TokenMismatchError(key)

        self.data[key] = (serialized, str(self.token))
        self.token += 1

        return True

    def get(self, key):
        check_key(key)

        self._statlog('get')
        if key in self.data:
            stored_val, _ignore_stored_token = self.data[key]
            val = pickle.loads(stored_val)
            return val
        return None

    def gets(self, key):
        check_key(key)
        if key in self.data:
            stored_val, stored_token = self.data[key]
            val = pickle.loads(stored_val)
            return (val, stored_token)
        return (None, None)

    def get_multi(self, keys, key_prefix=''):
        self._statlog('get_multi')

        results = {}
        for key in keys:
            key = key_prefix + key
            val = self.get(key)
            results[key] = val
        return results

    def gets_multi(self, keys, key_prefix=''):
        self._statlog('gets_multi')
        results = {}
        for key in keys:
            key = key_prefix + key
            result = self.gets(key)
            if result[1] is not None:
                results[key] = result
        return results


class _Host:
    _DEAD_RETRY = 1  # number of seconds before retrying a dead server.
    _SOCKET_TIMEOUT = 3  # number of seconds before sockets timeout.

    def __init__(self, host, debugfunc=None):
        if isinstance(host, types.TupleType):
            host, self.weight = host
        else:
            self.weight = 1

        #  parse the connection string
        m = re.match(r'^(?P<proto>unix):(?P<path>.*)$', host)
        if not m:
            m = re.match(
                r'^(?P<proto>inet):'
                r'(?P<host>[^:]+)(:(?P<port>[0-9]+))?$', host)
        if not m:
            m = re.match(r'^(?P<host>[^:]+):(?P<port>[0-9]+)$', host)
        if not m:
            raise ValueError('Unable to parse connection string: "%s"' % host)

        hostData = m.groupdict()
        if hostData.get('proto') == 'unix':
            self.family = socket.AF_UNIX
            self.address = hostData['path']
        else:
            self.family = socket.AF_INET
            self.ip = hostData['host']
            self.port = int(hostData.get('port', 11211))
            self.address = (self.ip, self.port)

        if not debugfunc:
            debugfunc = lambda x: x
        self.debuglog = debugfunc

        self.deaduntil = 0
        self.socket = None

        self.buffer = ''

    def _check_dead(self):
        if self.deaduntil and self.deaduntil > time.time():
            return 1
        self.deaduntil = 0
        return 0

    def connect(self):
        if self._get_socket():
            return 1
        return 0

    def mark_dead(self, reason):
        log.error("Memcacheclient socket marked dead ({r})", r=reason)
        self.debuglog("MemCache: %s: %s.  Marking dead." % (self, reason))
        self.deaduntil = time.time() + _Host._DEAD_RETRY
        self.close_socket()

    def _get_socket(self):
        if self._check_dead():
            log.error("Memcacheclient _get_socket() found dead socket")
            return None
        if self.socket:
            return self.socket
        s = socket.socket(self.family, socket.SOCK_STREAM)
        if hasattr(s, 'settimeout'):
            s.settimeout(self._SOCKET_TIMEOUT)
        try:
            s.connect(self.address)
        except socket.timeout, msg:
            log.error("Memcacheclient _get_socket() connection timed out ({m})", m=msg)
            self.mark_dead("connect: %s" % msg)
            return None
        except socket.error, msg:
            if isinstance(msg, tuple):
                msg = msg[1]
            log.error("Memcacheclient _get_socket() connection error ({m})", m=msg)
            self.mark_dead("connect: %s" % msg)
            return None
        self.socket = s
        self.buffer = ''
        return s

    def close_socket(self):
        if self.socket:
            self.socket.close()
            self.socket = None

    def send_cmd(self, cmd):
        self.socket.sendall(cmd + '\r\n')

    def send_cmds(self, cmds):
        """ cmds already has trailing \r\n's applied """
        self.socket.sendall(cmds)

    def readline(self):
        buf = self.buffer
        recv = self.socket.recv
        while True:
            index = buf.find('\r\n')
            if index >= 0:
                break
            data = recv(4096)
            if not data:
                self.mark_dead('Connection closed while reading from %s' % repr(self))
                break
            buf += data
        if index >= 0:
            self.buffer = buf[index + 2:]
            buf = buf[:index]
        else:
            self.buffer = ''
        return buf

    def expect(self, text):
        line = self.readline()
        if line != text:
            self.debuglog("while expecting '%s', got unexpected response '%s'" % (text, line))
        return line

    def recv(self, rlen):
        self_socket_recv = self.socket.recv
        buf = self.buffer
        while len(buf) < rlen:
            foo = self_socket_recv(4096)
            buf += foo
            if len(foo) == 0:
                raise _Error, (
                    'Read %d bytes, expecting %d, '
                    'read returned 0 length bytes' % (len(buf), rlen))
        self.buffer = buf[rlen:]
        return buf[:rlen]

    def __str__(self):
        d = ''
        if self.deaduntil:
            d = " (dead until %d)" % self.deaduntil

        if self.family == socket.AF_INET:
            return "inet:%s:%d%s" % (self.address[0], self.address[1], d)
        else:
            return "unix:%s%s" % (self.address, d)


def check_key(key, key_extra_len=0):
    """Checks sanity of key.  Fails if:
        Key length is > SERVER_MAX_KEY_LENGTH (Raises MemcachedKeyLength).
        Contains control characters  (Raises MemcachedKeyCharacterError).
        Is not a string (Raises MemcachedStringEncodingError)
        Is an unicode string (Raises MemcachedStringEncodingError)
        Is not a string (Raises MemcachedKeyError)
        Is None (Raises MemcachedKeyError)
    """

    return  # Short-circuit this expensive method

    if isinstance(key, tuple):
        key = key[1]
    if not key:
        raise Client.MemcachedKeyNoneError, ("Key is None")
    if isinstance(key, unicode):
        raise Client.MemcachedStringEncodingError, (
            "Keys must be str()'s, not "
            "unicode.  Convert your unicode strings using "
            "mystring.encode(charset)!")
    if not isinstance(key, str):
        raise Client.MemcachedKeyTypeError, ("Key must be str()'s")

    if isinstance(key, basestring):
        if len(key) + key_extra_len > SERVER_MAX_KEY_LENGTH:
            raise Client.MemcachedKeyLengthError, (
                "Key length is > %s"
                % SERVER_MAX_KEY_LENGTH)
        for char in key:
            if ord(char) < 32 or ord(char) == 127:
                raise Client.MemcachedKeyCharacterError, "Control characters not allowed"


def _doctest():
    import doctest
    from twistedcaldav import memcacheclient
    servers = ["127.0.0.1:11211"]
    mc = Client(servers, debug=1)
    globs = {"mc": mc}
    return doctest.testmod(memcacheclient, globs=globs)

if __name__ == "__main__":
    print("Testing docstrings...")
    _doctest()
    print("Running tests:")
    print
    serverList = [["127.0.0.1:11211"]]
    if '--do-unix' in sys.argv:
        serverList.append([os.path.join(os.getcwd(), 'memcached.socket')])

    for servers in serverList:
        mc = Client(servers, debug=1)

        def to_s(val):
            if not isinstance(val, types.StringTypes):
                return "%s (%s)" % (val, type(val))
            return "%s" % val

        def test_setget(key, val):
            print("Testing set/get {'%s': %s} ..." % (to_s(key), to_s(val)), end="")
            mc.set(key, val)
            newval = mc.get(key)
            if newval == val:
                print("OK")
                return 1
            else:
                print("FAIL")
                return 0

        class FooStruct:

            def __init__(self):
                self.bar = "baz"

            def __str__(self):
                return "A FooStruct"

            def __eq__(self, other):
                if isinstance(other, FooStruct):
                    return self.bar == other.bar
                return 0

        test_setget("a_string", "some random string")
        test_setget("an_integer", 42)
        if test_setget("long", long(1 << 30)):
            print("Testing delete ...", end="")
            if mc.delete("long"):
                print("OK")
            else:
                print("FAIL")
        print("Testing get_multi ...", end="")
        print(mc.get_multi(["a_string", "an_integer"]))

        print("Testing get(unknown value) ...", end="")
        print(to_s(mc.get("unknown_value")))

        f = FooStruct()
        test_setget("foostruct", f)

        print("Testing incr ...", end="")
        x = mc.incr("an_integer", 1)
        if x == 43:
            print("OK")
        else:
            print("FAIL")

        print("Testing decr ...", end="")
        x = mc.decr("an_integer", 1)
        if x == 42:
            print("OK")
        else:
            print("FAIL")

        # sanity tests
        print("Testing sending spaces...", end="")
        try:
            x = mc.set("this has spaces", 1)
        except Client.MemcachedKeyCharacterError, msg:
            print("OK")
        else:
            print("FAIL")

        print("Testing sending control characters...", end="")
        try:
            x = mc.set("this\x10has\x11control characters\x02", 1)
        except Client.MemcachedKeyCharacterError, msg:
            print("OK")
        else:
            print("FAIL")

        print("Testing using insanely long key...", end="")
        try:
            x = mc.set('a' * SERVER_MAX_KEY_LENGTH + 'aaaa', 1)
        except Client.MemcachedKeyLengthError, msg:
            print("OK")
        else:
            print("FAIL")

        print("Testing sending a unicode-string key...", end="")
        try:
            x = mc.set(u'keyhere', 1)
        except Client.MemcachedStringEncodingError, msg:
            print("OK", end="")
        else:
            print("FAIL", end="")
        try:
            x = mc.set((u'a' * SERVER_MAX_KEY_LENGTH).encode('utf-8'), 1)
        except:
            print("FAIL", end="")
        else:
            print("OK", end="")
        s = pickle.loads('V\\u4f1a\np0\n.')
        try:
            x = mc.set((s * SERVER_MAX_KEY_LENGTH).encode('utf-8'), 1)
        except Client.MemcachedKeyLengthError:
            print("OK")
        else:
            print("FAIL")

        print("Testing using a value larger than the memcached value limit...", end="")
        x = mc.set('keyhere', 'a' * SERVER_MAX_VALUE_LENGTH)
        if mc.get('keyhere') == None:
            print("OK", end="")
        else:
            print("FAIL", end="")
        x = mc.set('keyhere', 'a' * SERVER_MAX_VALUE_LENGTH + 'aaa')
        if mc.get('keyhere') == None:
            print("OK")
        else:
            print("FAIL")

        print("Testing set_multi() with no memcacheds running", end="")
        mc.disconnect_all()
        errors = mc.set_multi({'keyhere': 'a', 'keythere': 'b'})
        if errors != []:
            print("FAIL")
        else:
            print("OK")

        print("Testing delete_multi() with no memcacheds running", end="")
        mc.disconnect_all()
        ret = mc.delete_multi({'keyhere': 'a', 'keythere': 'b'})
        if ret != 1:
            print("FAIL")
        else:
            print("OK")

# vim: ts=4 sw=4 et :