The pickle module implements a fundamental, but powerful algorithm for serializing and de-serializing a Python object structure. “Pickling” is the process whereby a Python object hierarchy is converted into a byte stream, and “unpickling” is the inverse operation, whereby a byte stream is converted back into an object hierarchy. Pickling (and unpickling) is alternatively known as “serialization”, “marshalling,” [1] or “flattening”, however, to avoid confusion, the terms used here are “pickling” and “unpickling”..
The pickle module has an transparent optimizer (_pickle) written in C. It is used whenever available. Otherwise the pure Python implementation is used.
Python has a more primitive serialization module called marshal, but in general pickle should always be the preferred way to serialize Python objects. marshal exists primarily to support Python’s .pyc files.
The pickle module differs from marshal several significant ways:
The pickle module keeps track of the objects it has already serialized, so that later references to the same object won’t be serialized again. marshal doesn’t do this.
This has implications both for recursive objects and object sharing. Recursive objects are objects that contain references to themselves. These are not handled by marshal, and in fact, attempting to marshal recursive objects will crash your Python interpreter. Object sharing happens when there are multiple references to the same object in different places in the object hierarchy being serialized. pickle stores such objects only once, and ensures that all other references point to the master copy. Shared objects remain shared, which can be very important for mutable objects.
marshal cannot be used to serialize user-defined classes and their instances. pickle can save and restore class instances transparently, however the class definition must be importable and live in the same module as when the object was stored.
The marshal serialization format is not guaranteed to be portable across Python versions. Because its primary job in life is to support .pyc files, the Python implementers reserve the right to change the serialization format in non-backwards compatible ways should the need arise. The pickle serialization format is guaranteed to be backwards compatible across Python releases.
Warning
The pickle module is not intended to be secure against erroneous or maliciously constructed data. Never unpickle data received from an untrusted or unauthenticated source.
Note that serialization is a more primitive notion than persistence; although pickle reads and writes file objects, it does not handle the issue of naming persistent objects, nor the (even more complicated) issue of concurrent access to persistent objects. The pickle module can transform a complex object into a byte stream and it can transform the byte stream into an object with the same internal structure. Perhaps the most obvious thing to do with these byte streams is to write them onto a file, but it is also conceivable to send them across a network or store them in a database. The module shelve provides a simple interface to pickle and unpickle objects on DBM-style database files.
The data format used by pickle is Python-specific. This has the advantage that there are no restrictions imposed by external standards such as XDR (which can’t represent pointer sharing); however it means that non-Python programs may not be able to reconstruct pickled Python objects.
By default, the pickle data format uses a compact binary representation. The module pickletools contains tools for analyzing data streams generated by pickle.
There are currently 4 different protocols which can be used for pickling.
Refer to PEP 307 for information about improvements brought by protocol 2. See pickletools‘s source code for extensive comments about opcodes used by pickle protocols.
To serialize an object hierarchy, you first create a pickler, then you call the pickler’s dump() method. To de-serialize a data stream, you first create an unpickler, then you call the unpickler’s load() method. The pickle module provides the following constant:
Note
Be sure to always open pickle files created with protocols >= 1 in binary mode. For the old ASCII-based pickle protocol 0 you can use either text mode or binary mode as long as you stay consistent.
A pickle file written with protocol 0 in binary mode will contain lone linefeeds as line terminators and therefore will look “funny” when viewed in Notepad or other editors which do not support this format.
The pickle module provides the following functions to make the pickling process more convenient:
Write a pickled representation of obj to the open file object file. This is equivalent to Pickler(file, protocol).dump(obj).
The optional protocol argument tells the pickler to use the given protocol; supported protocols are 0, 1, 2, 3. The default protocol is 3; a backward-incompatible protocol designed for Python 3.0.
Specifying a negative protocol version selects the highest protocol version supported. The higher the protocol used, the more recent the version of Python needed to read the pickle produced.
The file argument must have a write() method that accepts a single bytes argument. It can thus be a file object opened for binary writing, a io.BytesIO instance, or any other custom object that meets this interface.
Return the pickled representation of the object as a bytes object, instead of writing it to a file.
The optional protocol argument tells the pickler to use the given protocol; supported protocols are 0, 1, 2, 3. The default protocol is 3; a backward-incompatible protocol designed for Python 3.0.
Specifying a negative protocol version selects the highest protocol version supported. The higher the protocol used, the more recent the version of Python needed to read the pickle produced.
Read a pickled object representation from the open file object file and return the reconstituted object hierarchy specified therein. This is equivalent to Unpickler(file).load().
The protocol version of the pickle is detected automatically, so no protocol argument is needed. Bytes past the pickled object’s representation are ignored.
The argument file must have two methods, a read() method that takes an integer argument, and a readline() method that requires no arguments. Both methods should return bytes. Thus file can be a binary file object opened for reading, a BytesIO object, or any other custom object that meets this interface.
Optional keyword arguments are encoding and errors, which are used to decode 8-bit string instances pickled by Python 2.x. These default to ‘ASCII’ and ‘strict’, respectively.
Read a pickled object hierarchy from a bytes object and return the reconstituted object hierarchy specified therein
The protocol version of the pickle is detected automatically, so no protocol argument is needed. Bytes past the pickled object’s representation are ignored.
Optional keyword arguments are encoding and errors, which are used to decode 8-bit string instances pickled by Python 2.x. These default to ‘ASCII’ and ‘strict’, respectively.
The pickle module defines three exceptions:
Error raised when an unpicklable object is encountered by Pickler. It inherits PickleError.
Refer to What can be pickled and unpickled? to learn what kinds of objects can be pickled.
Error raised when there a problem unpickling an object, such as a data corruption or a security violation. It inherits PickleError.
Note that other exceptions may also be raised during unpickling, including (but not necessarily limited to) AttributeError, EOFError, ImportError, and IndexError.
The pickle module exports two classes, Pickler and Unpickler:
This takes a binary file for writing a pickle data stream.
The optional protocol argument tells the pickler to use the given protocol; supported protocols are 0, 1, 2, 3. The default protocol is 3; a backward-incompatible protocol designed for Python 3.0.
Specifying a negative protocol version selects the highest protocol version supported. The higher the protocol used, the more recent the version of Python needed to read the pickle produced.
The file argument must have a write() method that accepts a single bytes argument. It can thus be a file object opened for binary writing, a io.BytesIO instance, or any other custom object that meets this interface.
Do nothing by default. This exists so a subclass can override it.
If persistent_id() returns None, obj is pickled as usual. Any other value causes Pickler to emit the returned value as a persistent ID for obj. The meaning of this persistent ID should be defined by Unpickler.persistent_load(). Note that the value returned by persistent_id() cannot itself have a persistent ID.
See Persistence of External Objects for details and examples of uses.
Deprecated. Enable fast mode if set to a true value. The fast mode disables the usage of memo, therefore speeding the pickling process by not generating superfluous PUT opcodes. It should not be used with self-referential objects, doing otherwise will cause Pickler to recurse infinitely.
Use pickletools.optimize() if you need more compact pickles.
It is possible to make multiple calls to the dump() method of the same Pickler instance. These must then be matched to the same number of calls to the load() method of the corresponding Unpickler instance. If the same object is pickled by multiple dump() calls, the load() will all yield references to the same object.
Please note, this is intended for pickling multiple objects without intervening modifications to the objects or their parts. If you modify an object and then pickle it again using the same Pickler instance, the object is not pickled again — a reference to it is pickled and the Unpickler will return the old value, not the modified one.
This takes a binary file for reading a pickle data stream.
The protocol version of the pickle is detected automatically, so no protocol argument is needed.
The argument file must have two methods, a read() method that takes an integer argument, and a readline() method that requires no arguments. Both methods should return bytes. Thus file can be a binary file object opened for reading, a BytesIO object, or any other custom object that meets this interface.
Optional keyword arguments are encoding and errors, which are used to decode 8-bit string instances pickled by Python 2.x. These default to ‘ASCII’ and ‘strict’, respectively.
Raise an UnpickingError by default.
If defined, persistent_load() should return the object specified by the persistent ID pid. If an invalid persistent ID is encountered, an UnpickingError should be raised.
See Persistence of External Objects for details and examples of uses.
Import module if necessary and return the object called name from it, where the module and name arguments are str objects. Note, unlike its name suggests, find_class() is also used for finding functions.
Subclasses may override this to gain control over what type of objects and how they can be loaded, potentially reducing security risks. Refer to Restricting Globals for details.
The following types can be pickled:
Attempts to pickle unpicklable objects will raise the PicklingError exception; when this happens, an unspecified number of bytes may have already been written to the underlying file. Trying to pickle a highly recursive data structure may exceed the maximum recursion depth, a RuntimeError will be raised in this case. You can carefully raise this limit with sys.setrecursionlimit().
Note that functions (built-in and user-defined) are pickled by “fully qualified” name reference, not by value. This means that only the function name is pickled, along with the name of module the function is defined in. Neither the function’s code, nor any of its function attributes are pickled. Thus the defining module must be importable in the unpickling environment, and the module must contain the named object, otherwise an exception will be raised. [2]
Similarly, classes are pickled by named reference, so the same restrictions in the unpickling environment apply. Note that none of the class’s code or data is pickled, so in the following example the class attribute attr is not restored in the unpickling environment:
class Foo:
attr = 'A class attribute'
picklestring = pickle.dumps(Foo)
These restrictions are why picklable functions and classes must be defined in the top level of a module.
Similarly, when class instances are pickled, their class’s code and data are not pickled along with them. Only the instance data are pickled. This is done on purpose, so you can fix bugs in a class or add methods to the class and still load objects that were created with an earlier version of the class. If you plan to have long-lived objects that will see many versions of a class, it may be worthwhile to put a version number in the objects so that suitable conversions can be made by the class’s __setstate__() method.
In this section, we describe the general mechanisms available to you to define, customize, and control how class instances are pickled and unpickled.
In most cases, no additional code is needed to make instances picklable. By default, pickle will retrieve the class and the attributes of an instance via introspection. When a class instance is unpickled, its __init__() method is usually not invoked. The default behaviour first creates an uninitialized instance and then restores the saved attributes. The following code shows an implementation of this behaviour:
def save(obj):
return (obj.__class__, obj.__dict__)
def load(cls, attributes):
obj = cls.__new__(cls)
obj.__dict__.update(attributes)
return obj
Classes can alter the default behaviour by providing one or severals special methods. In protocol 2 and newer, classes that implements the __getnewargs__() method can dictate the values passed to the __new__() method upon unpickling. This is often needed for classes whose __new__() method requires arguments.
Classes can further influence how their instances are pickled; if the class defines the method __getstate__(), it is called and the returned object is pickled as the contents for the instance, instead of the contents of the instance’s dictionary. If the __getstate__() method is absent, the instance’s __dict__ is pickled as usual.
Upon unpickling, if the class defines __setstate__(), it is called with the unpickled state. In that case, there is no requirement for the state object to be a dictionary. Otherwise, the pickled state must be a dictionary and its items are assigned to the new instance’s dictionary.
Note
If __getstate__() returns a false value, the __setstate__() method will not be called.
Refer to the section Handling Stateful Objects for more information about how to use the methods __getstate__() and __setstate__().
As we shall see, pickle does not use directly the methods described above. In fact, these methods are part of the copy protocol which implements the __reduce__() special method. The copy protocol provides a unified interface for retrieving the data necessary for pickling and copying objects. [3]
Although powerful, implementing __reduce__() directly in your classes is error prone. For this reason, class designers should use the high-level interface (i.e., __getnewargs__(), __getstate__() and __setstate__()) whenever possible. We will show however cases where using __reduce__() is the only option or leads to more efficient pickling or both.
The interface is currently defined as follow. The __reduce__() method takes no argument and shall return either a string or preferably a tuple (the returned object is often refered as the “reduce value”).
If a string is returned, the string should be interpreted as the name of a global variable. It should be the object’s local name relative to its module; the pickle module searches the module namespace to determine the object’s module. This behaviour is typically useful for singletons.
When a tuple is returned, it must be between two and five items long. Optional items can either be omitted, or None can be provided as their value. The semantics of each item are in order:
Alternatively, a __reduce_ex__() method may be defined. The only difference is this method should take a single integer argument, the protocol version. When defined, pickle will prefer it over the __reduce__() method. In addition, __reduce__() automatically becomes a synonym for the extended version. The main use for this method is to provide backwards-compatible reduce values for older Python releases.
For the benefit of object persistence, the pickle module supports the notion of a reference to an object outside the pickled data stream. Such objects are referenced by a persistent ID, which should be either a string of alphanumeric characters (for protocol 0) [4] or just an arbitrary object (for any newer protocol).
The resolution of such persistent IDs is not defined by the pickle module; it will delegate this resolution to the user defined methods on the pickler and unpickler, persistent_id() and persistent_load() respectively.
To pickle objects that have an external persistent id, the pickler must have a custom persistent_id() method that takes an object as an argument and returns either None or the persistent id for that object. When None is returned, the pickler simply pickles the object as normal. When a persistent ID string is returned, the pickler will pickle that object, along with a marker so that the unpickler will recognize it as a persistent ID.
To unpickle external objects, the unpickler must have a custom persistent_load() method that takes a persistent ID object and returns the referenced object.
Here is a comprehensive example presenting how persistent ID can be used to pickle external objects by reference.
# Simple example presenting how persistent ID can be used to pickle
# external objects by reference.
import pickle
import sqlite3
from collections import namedtuple
# Simple class representing a record in our database.
MemoRecord = namedtuple("MemoRecord", "key, task")
class DBPickler(pickle.Pickler):
def persistent_id(self, obj):
# Instead of pickling MemoRecord as a regular class instance, we emit a
# persistent ID.
if isinstance(obj, MemoRecord):
# Here, our persistent ID is simply a tuple, containing a tag and a
# key, which refers to a specific record in the database.
return ("MemoRecord", obj.key)
else:
# If obj does not have a persistent ID, return None. This means obj
# needs to be pickled as usual.
return None
class DBUnpickler(pickle.Unpickler):
def __init__(self, file, connection):
super().__init__(file)
self.connection = connection
def persistent_load(self, pid):
# This method is invoked whenever a persistent ID is encountered.
# Here, pid is the tuple returned by DBPickler.
cursor = self.connection.cursor()
type_tag, key_id = pid
if type_tag == "MemoRecord":
# Fetch the referenced record from the database and return it.
cursor.execute("SELECT * FROM memos WHERE key=?", (str(key_id),))
key, task = cursor.fetchone()
return MemoRecord(key, task)
else:
# Always raises an error if you cannot return the correct object.
# Otherwise, the unpickler will think None is the object referenced
# by the persistent ID.
raise pickle.UnpicklingError("unsupported persistent object")
def main():
import io, pprint
# Initialize and populate our database.
conn = sqlite3.connect(":memory:")
cursor = conn.cursor()
cursor.execute("CREATE TABLE memos(key INTEGER PRIMARY KEY, task TEXT)")
tasks = (
'give food to fish',
'prepare group meeting',
'fight with a zebra',
)
for task in tasks:
cursor.execute("INSERT INTO memos VALUES(NULL, ?)", (task,))
# Fetch the records to be pickled.
cursor.execute("SELECT * FROM memos")
memos = [MemoRecord(key, task) for key, task in cursor]
# Save the records using our custom DBPickler.
file = io.BytesIO()
DBPickler(file).dump(memos)
print("Pickled records:")
pprint.pprint(memos)
# Update a record, just for good measure.
cursor.execute("UPDATE memos SET task='learn italian' WHERE key=1")
# Load the records from the pickle data stream.
file.seek(0)
memos = DBUnpickler(file, conn).load()
print("Unpickled records:")
pprint.pprint(memos)
if __name__ == '__main__':
main()
Here’s an example that shows how to modify pickling behavior for a class. The TextReader class opens a text file, and returns the line number and line contents each time its readline() method is called. If a TextReader instance is pickled, all attributes except the file object member are saved. When the instance is unpickled, the file is reopened, and reading resumes from the last location. The __setstate__() and __getstate__() methods are used to implement this behavior.
class TextReader:
"""Print and number lines in a text file."""
def __init__(self, filename):
self.filename = filename
self.file = open(filename)
self.lineno = 0
def readline(self):
self.lineno += 1
line = self.file.readline()
if not line:
return None
if line.endswith("\n"):
line = line[:-1]
return "%i: %s" % (self.lineno, line)
def __getstate__(self):
# Copy the object's state from self.__dict__ which contains
# all our instance attributes. Always use the dict.copy()
# method to avoid modifying the original state.
state = self.__dict__.copy()
# Remove the unpicklable entries.
del state['file']
return state
def __setstate__(self, state):
# Restore instance attributes (i.e., filename and lineno).
self.__dict__.update(state)
# Restore the previously opened file's state. To do so, we need to
# reopen it and read from it until the line count is restored.
file = open(self.filename)
for _ in range(self.lineno):
file.readline()
# Finally, save the file.
self.file = file
A sample usage might be something like this:
>>> reader = TextReader("hello.txt")
>>> reader.readline()
'1: Hello world!'
>>> reader.readline()
'2: I am line number two.'
>>> new_reader = pickle.loads(pickle.dumps(reader))
>>> new_reader.readline()
'3: Goodbye!'
By default, unpickling will import any class or function that it finds in the pickle data. For many applications, this behaviour is unacceptable as it permits the unpickler to import and invoke arbitrary code. Just consider what this hand-crafted pickle data stream does when loaded:
>>> import pickle
>>> pickle.loads(b"cos\nsystem\n(S'echo hello world'\ntR.")
hello world
0
In this example, the unpickler imports the os.system() function and then apply the string argument “echo hello world”. Although this example is inoffensive, it is not difficult to imagine one that could damage your system.
For this reason, you may want to control what gets unpickled by customizing Unpickler.find_class(). Unlike its name suggests, find_class() is called whenever a global (i.e., a class or a function) is requested. Thus it is possible to either forbid completely globals or restrict them to a safe subset.
Here is an example of an unpickler allowing only few safe classes from the builtins module to be loaded:
import builtins
import io
import pickle
safe_builtins = {
'range',
'complex',
'set',
'frozenset',
'slice',
}
class RestrictedUnpickler(pickle.Unpickler):
def find_class(self, module, name):
# Only allow safe classes from builtins.
if module == "builtins" and name in safe_builtins:
return getattr(builtins, name)
# Forbid everything else.
raise pickle.UnpicklingError("global '%s.%s' is forbidden" %
(module, name))
def restricted_loads(s):
"""Helper function analogous to pickle.loads()."""
return RestrictedUnpickler(io.BytesIO(s)).load()
A sample usage of our unpickler working has intended:
>>> restricted_loads(pickle.dumps([1, 2, range(15)]))
[1, 2, range(0, 15)]
>>> restricted_loads(b"cos\nsystem\n(S'echo hello world'\ntR.")
Traceback (most recent call last):
...
pickle.UnpicklingError: global 'os.system' is forbidden
>>> restricted_loads(b'cbuiltins\neval\n'
... b'(S\'getattr(__import__("os"), "system")'
... b'("echo hello world")\'\ntR.')
Traceback (most recent call last):
...
pickle.UnpicklingError: global 'builtins.eval' is forbidden
As our examples shows, you have to be careful with what you allow to be unpickled. Therefore if security is a concern, you may want to consider alternatives such as the marshalling API in xmlrpc.client or third-party solutions.
For the simplest code, use the dump() and load() functions. Note that a self-referencing list is pickled and restored correctly.
import pickle
data1 = {'a': [1, 2.0, 3, 4+6j],
'b': ("string", "string using Unicode features \u0394"),
'c': None}
selfref_list = [1, 2, 3]
selfref_list.append(selfref_list)
output = open('data.pkl', 'wb')
# Pickle dictionary using protocol 2.
pickle.dump(data1, output, 2)
# Pickle the list using the highest protocol available.
pickle.dump(selfref_list, output, -1)
output.close()
The following example reads the resulting pickled data. When reading a pickle-containing file, you should open the file in binary mode because you can’t be sure if the ASCII or binary format was used.
import pprint, pickle
pkl_file = open('data.pkl', 'rb')
data1 = pickle.load(pkl_file)
pprint.pprint(data1)
data2 = pickle.load(pkl_file)
pprint.pprint(data2)
pkl_file.close()
See also
Footnotes
| [1] | Don’t confuse this with the marshal module |
| [2] | The exception raised will likely be an ImportError or an AttributeError but it could be something else. |
| [3] | The copy module uses this protocol for shallow and deep copying operations. |
| [4] | The limitation on alphanumeric characters is due to the fact the persistent IDs, in protocol 0, are delimited by the newline character. Therefore if any kind of newline characters occurs in persistent IDs, the resulting pickle will become unreadable. |