google.appengine.api.memcache package

Submodules

Module contents

Memcache API.

Provides memcached-alike API to application developers to store data in memory when reliable storage via the DataStore API isn’t required and higher performance is desired.

class google.appengine.api.memcache.Client(servers=None, debug=0, pickleProtocol=2, pickler=<built-in function Pickler>, unpickler=<built-in function Unpickler>, pload=None, pid=None, make_sync_call=None, _app_id=None)[source]

Bases: object

Memcache client object, through which one invokes all memcache operations.

Several methods are no-ops to retain source-level compatibility with the existing popular Python memcache library.

Any method that takes a ‘key’ argument will accept that key as a string (unicode or not) or a tuple of (hash_value, string) where the hash_value, normally used for sharding onto a memcache instance, is instead ignored, as Google App Engine deals with the sharding transparently. Keys in memcache are just bytes, without a specified encoding. All such methods may raise TypeError if provided a bogus key value and a ValueError if the key is too large.

Any method that takes a ‘value’ argument will accept as that value any string (unicode or not), int, long, or pickle-able Python object, including all native types. You’ll get back from the cache the same type that you originally put in.

The Client class is not thread-safe with respect to the gets(), cas() and cas_multi() methods (and other compare-and-set-related methods). Therefore, Client objects should not be used by more than one thread for CAS purposes. Note that the global Client for the module-level functions is okay because it does not expose any of the CAS methods.

add(key, value, time=0, min_compress_len=0, namespace=None)[source]

Sets a key’s value, iff item is not already in memcache.

Args:

key: Key to set. See docs on Client for details. value: Value to set. Any type. If complex, will be pickled. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

min_compress_len: Ignored option for compatibility. namespace: a string specifying an optional namespace to use in

the request.
Returns:
True if added. False on error.
add_multi(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None)[source]

Set multiple keys’ values iff items are not already in memcache.

Args:

mapping: Dictionary of keys to values. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

key_prefix: Prefix for to prepend to all keys. min_compress_len: Unimplemented compatibility option. namespace: a string specifying an optional namespace to use in

the request.
Returns:
A list of keys whose values were NOT set because they did not already exist in memcache. On total success, this list should be empty.
add_multi_async(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None, rpc=None)[source]

Async version of add_multi() – note different return value.

Returns:
See _set_multi_async_with_policy().
cas(key, value, time=0, min_compress_len=0, namespace=None)[source]

Compare-And-Set update.

This requires that the key has previously been successfully fetched with gets() or get(..., for_cas=True), and that no changes have been made to the key since that fetch. Typical usage is:

key = ... client = memcache.Client() value = client.gets(key) # OR client.get(key, for_cas=True) <updated value> ok = client.cas(key, value)

If two processes run similar code, the first one calling cas() will succeed (ok == True), while the second one will fail (ok == False). This can be used to detect race conditions.

NOTE: some state (the CAS id) is stored on the Client object for each key ever used with gets(). To prevent ever-increasing memory usage, you must use a Client object when using cas(), and the lifetime of your Client object should be limited to that of one incoming HTTP request. You cannot use the global-function-based API.

Args:

key: Key to set. See docs on Client for details. value: The new value. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

min_compress_len: Ignored option for compatibility. namespace: a string specifying an optional namespace to use in

the request.
Returns:
True if updated. False on RPC error or if the CAS id didn’t match.
cas_multi(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None)[source]

Compare-And-Set update for multiple keys.

See cas() docstring for an explanation.

Args:

mapping: Dictionary of keys to values. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

key_prefix: Prefix for to prepend to all keys. min_compress_len: Unimplemented compatibility option. namespace: a string specifying an optional namespace to use in

the request.
Returns:
A list of keys whose values were NOT set because the compare failed. On total success, this list should be empty.
cas_multi_async(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None, rpc=None)[source]

Async version of cas_multi() – note different return value.

Returns:
See _set_multi_async_with_policy().
cas_reset()[source]

Clear the remembered CAS ids.

debuglog()[source]

Logging function for debugging information.

This is purely a compatibility method. In Google App Engine, it’s a no-op.

decr(key, delta=1, namespace=None, initial_value=None)[source]

Atomically decrements a key’s value.

Internally, the value is a unsigned 64-bit integer. Memcache caps decrementing below zero to zero.

The key must already exist in the cache to be decremented. See docs on incr() for details.

Args:
key: Key to decrement. If an iterable collection, each one of the keys
will be offset. See Client’s docstring for details.
delta: Non-negative integer value (int or long) to decrement key by,
defaulting to 1.
namespace: a string specifying an optional namespace to use in
the request.
initial_value: initial value to put in the cache, if it doesn’t
already exist. The default value, None, will not create a cache entry if it doesn’t already exist.
Returns:

If key was a single value, the new long integer value, or None if key was not in the cache, could not be decremented for any other reason, or a network/RPC/server error occurred.

If key was an iterable collection, a dictionary will be returned mapping supplied keys to values, with the values having the same meaning as the singular return value of this method.

Raises:
ValueError: If number is negative. TypeError: If delta isn’t an int or long.
decr_async(key, delta=1, namespace=None, initial_value=None, rpc=None)[source]

Async version of decr().

Returns:
A UserRPC instance whose get_result() method returns the same kind of value as decr() returns.
delete(key, seconds=0, namespace=None)[source]

Deletes a key from memcache.

Args:

key: Key to delete. See docs on Client for detils. seconds: Optional number of seconds to make deleted items ‘locked’

for ‘add’ operations. Value can be a delta from current time (up to 1 month), or an absolute Unix epoch time. Defaults to 0, which means items can be immediately added. With or without this option, a ‘set’ operation will always work. Float values will be rounded up to the nearest whole second.
namespace: a string specifying an optional namespace to use in
the request.
Returns:
DELETE_NETWORK_FAILURE (0) on network failure, DELETE_ITEM_MISSING (1) if the server tried to delete the item but didn’t have it, or DELETE_SUCCESSFUL (2) if the item was actually deleted. This can be used as a boolean value, where a network failure is the only bad condition.
delete_multi(keys, seconds=0, key_prefix='', namespace=None)[source]

Delete multiple keys at once.

Args:

keys: List of keys to delete. seconds: Optional number of seconds to make deleted items ‘locked’

for ‘add’ operations. Value can be a delta from current time (up to 1 month), or an absolute Unix epoch time. Defaults to 0, which means items can be immediately added. With or without this option, a ‘set’ operation will always work. Float values will be rounded up to the nearest whole second.
key_prefix: Prefix to put on all keys when sending specified
keys to memcache. See docs for get_multi() and set_multi().
namespace: a string specifying an optional namespace to use in
the request.
Returns:
True if all operations completed successfully. False if one or more failed to complete.
delete_multi_async(keys, seconds=0, key_prefix='', namespace=None, rpc=None)[source]

Async version of delete_multi() – note different return value.

Returns:
A UserRPC instance whose get_result() method returns None if there was a network error, or a list of status values otherwise, where each status corresponds to a key and is either DELETE_SUCCESSFUL, DELETE_ITEM_MISSING, or DELETE_NETWORK_FAILURE (see delete() docstring for details).
disconnect_all()[source]

Closes all connections to memcache servers.

This is purely a compatibility method. In Google App Engine, it’s a no-op.

flush_all()[source]

Deletes everything in memcache.

Returns:
True on success, False on RPC or server error.
flush_all_async(rpc=None)[source]

Async version of flush_all().

Returns:
A UserRPC instance whose get_result() method returns True on success, False on RPC or server error.
forget_dead_hosts()[source]

Resets all servers to the alive status.

This is purely a compatibility method. In Google App Engine, it’s a no-op.

get(key, namespace=None, for_cas=False)[source]

Looks up a single key in memcache.

If you have multiple items to load, though, it’s much more efficient to use get_multi() instead, which loads them in one bulk operation, reducing the networking latency that’d otherwise be required to do many serialized get() operations.

Args:
key: The key in memcache to look up. See docs on Client
for details of format.
namespace: a string specifying an optional namespace to use in
the request.
for_cas: If True, request and store CAS ids on the client (see
cas() operation below).
Returns:
The value of the key, if found in memcache, else None.
get_multi(keys, key_prefix='', namespace=None, for_cas=False)[source]

Looks up multiple keys from memcache in one operation.

This is the recommended way to do bulk loads.

Args:
keys: List of keys to look up. Keys may be strings or
tuples of (hash_value, string). Google App Engine does the sharding and hashing automatically, though, so the hash value is ignored. To memcache, keys are just series of bytes, and not in any particular encoding.
key_prefix: Prefix to prepend to all keys when talking to the server;
not included in the returned dictionary.
namespace: a string specifying an optional namespace to use in
the request.

for_cas: If True, request and store CAS ids on the client.

Returns:
A dictionary of the keys and values that were present in memcache. Even if the key_prefix was specified, that key_prefix won’t be on the keys in the returned dictionary.
get_multi_async(keys, key_prefix='', namespace=None, for_cas=False, rpc=None)[source]

Async version of get_multi().

Returns:
A UserRPC instance whose get_result() method returns {} if there was a network error, otherwise a dict just like get_multi() returns.
get_stats()[source]

Gets memcache statistics for this application.

All of these statistics may reset due to various transient conditions. They provide the best information available at the time of being called.

Returns:

Dictionary mapping statistic names to associated values. Statistics and their associated meanings:

hits: Number of cache get requests resulting in a cache hit. misses: Number of cache get requests resulting in a cache miss. byte_hits: Sum of bytes transferred on get requests. Rolls over to

zero on overflow.

items: Number of key/value pairs in the cache. bytes: Total size of all items in the cache. oldest_item_age: How long in seconds since the oldest item in the

cache was accessed. Effectively, this indicates how long a new item will survive in the cache without being accessed. This is _not_ the amount of time that has elapsed since the item was created.

On error, returns None.

get_stats_async(rpc=None)[source]

Async version of get_stats().

Returns:
A UserRPC instance whose get_result() method returns None if there was a network error, otherwise a dict just like get_stats() returns.
gets(key, namespace=None)[source]

An alias for get(..., for_cas=True).

incr(key, delta=1, namespace=None, initial_value=None)[source]

Atomically increments a key’s value.

Internally, the value is a unsigned 64-bit integer. Memcache doesn’t check 64-bit overflows. The value, if too large, will wrap around.

Unless an initial_value is specified, the key must already exist in the cache to be incremented. To initialize a counter, either specify initial_value or set() it to the initial value, as an ASCII decimal integer. Future get()s of the key, post-increment, will still be an ASCII decimal value.

Args:
key: Key to increment. If an iterable collection, each one of the keys
will be offset. See Client’s docstring for details.
delta: Non-negative integer value (int or long) to increment key by,
defaulting to 1.
namespace: a string specifying an optional namespace to use in
the request.
initial_value: initial value to put in the cache, if it doesn’t
already exist. The default value, None, will not create a cache entry if it doesn’t already exist.
Returns:

If key was a single value, the new long integer value, or None if key was not in the cache, could not be incremented for any other reason, or a network/RPC/server error occurred.

If key was an iterable collection, a dictionary will be returned mapping supplied keys to values, with the values having the same meaning as the singular return value of this method.

Raises:
ValueError: If number is negative. TypeError: If delta isn’t an int or long.
incr_async(key, delta=1, namespace=None, initial_value=None, rpc=None)[source]

Async version of incr().

Returns:
A UserRPC instance whose get_result() method returns the same kind of value as incr() returns.
offset_multi(mapping, key_prefix='', namespace=None, initial_value=None)[source]

Offsets multiple keys by a delta, incrementing and decrementing in batch.

Args:
mapping: Dictionary mapping keys to deltas (positive or negative integers)
to apply to each corresponding key.

key_prefix: Prefix for to prepend to all keys. initial_value: Initial value to put in the cache, if it doesn’t

already exist. The default value, None, will not create a cache entry if it doesn’t already exist.
namespace: A string specifying an optional namespace to use in
the request.
Returns:
Dictionary mapping input keys to new integer values. The new value will be None if an error occurs, the key does not already exist, or the value was not an integer type. The values will wrap-around at unsigned 64-bit integer-maximum and underflow will be floored at zero.
offset_multi_async(mapping, key_prefix='', namespace=None, initial_value=None, rpc=None)[source]

Async version of offset_multi().

Returns:
A UserRPC instance whose get_result() method returns a dict just like offset_multi() returns.
replace(key, value, time=0, min_compress_len=0, namespace=None)[source]

Replaces a key’s value, failing if item isn’t already in memcache.

Args:

key: Key to set. See docs on Client for details. value: Value to set. Any type. If complex, will be pickled. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

min_compress_len: Ignored option for compatibility. namespace: a string specifying an optional namespace to use in

the request.
Returns:
True if replaced. False on RPC error or cache miss.
replace_multi(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None)[source]

Replace multiple keys’ values, failing if the items aren’t in memcache.

Args:

mapping: Dictionary of keys to values. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

key_prefix: Prefix for to prepend to all keys. min_compress_len: Unimplemented compatibility option. namespace: a string specifying an optional namespace to use in

the request.
Returns:
A list of keys whose values were NOT set because they already existed in memcache. On total success, this list should be empty.
replace_multi_async(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None, rpc=None)[source]

Async version of replace_multi() – note different return value.

Returns:
See _set_multi_async_with_policy().
set(key, value, time=0, min_compress_len=0, namespace=None)[source]

Sets a key’s value, regardless of previous contents in cache.

Unlike add() and replace(), this method always sets (or overwrites) the value in memcache, regardless of previous contents.

Args:

key: Key to set. See docs on Client for details. value: Value to set. Any type. If complex, will be pickled. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

min_compress_len: Ignored option for compatibility. namespace: a string specifying an optional namespace to use in

the request.
Returns:
True if set. False on error.
set_multi(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None)[source]

Set multiple keys’ values at once, regardless of previous contents.

Args:

mapping: Dictionary of keys to values. time: Optional expiration time, either relative number of seconds

from current time (up to 1 month), or an absolute Unix epoch time. By default, items never expire, though items may be evicted due to memory pressure. Float values will be rounded up to the nearest whole second.

key_prefix: Prefix for to prepend to all keys. min_compress_len: Unimplemented compatibility option. namespace: a string specifying an optional namespace to use in

the request.
Returns:
A list of keys whose values were NOT set. On total success, this list should be empty.
set_multi_async(mapping, time=0, key_prefix='', min_compress_len=0, namespace=None, rpc=None)[source]

Async version of set_multi() – note different return value.

Returns:
See _set_multi_async_with_policy().
set_servers(servers)[source]

Sets the pool of memcache servers used by the client.

This is purely a compatibility method. In Google App Engine, it’s a no-op.

google.appengine.api.memcache.create_rpc(deadline=None, callback=None)[source]

Creates an RPC object for use with the memcache API.

Args:
deadline: Optional deadline in seconds for the operation; the default
is a system-specific deadline (typically 5 seconds).

callback: Optional callable to invoke on completion.

Returns:
An apiproxy_stub_map.UserRPC object specialized for this service.
google.appengine.api.memcache.setup_client(client_obj)[source]

Sets the Client object instance to use for all module-level methods.

Use this method if you want to have customer persistent_id() or persistent_load() functions associated with your client.

NOTE: We don’t expose the _async methods as functions; they’re too obscure; and we don’t expose gets(), cas() and cas_multi() because they maintain state on the client object.

Args:
client_obj: Instance of the memcache.Client object.