1. pickle — Python object serialization¶

1.1. Relationship to other Python modules¶

The pickle module has an optimized cousin called the cPickle module. As its name implies, cPickle is written in C, so it can be up to 1000 times faster than pickle. However it does not support subclassing of the Pickler() and Unpickler() classes, because in cPickle these are functions, not classes. Most applications have no need for this functionality, and can benefit from the improved performance of cPickle. Other than that, the interfaces of the two modules are nearly identical; the common interface is described in this manual and differences are pointed out where necessary. In the following discussions, we use the term “pickle” to collectively describe the pickle and cPickle modules.

The data streams the two modules produce are guaranteed to be interchangeable.

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 in 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.

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.

1.2. Data stream format¶

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 printable ASCII representation. This is slightly more voluminous than a binary representation. The big advantage of using printable ASCII (and of some other characteristics of pickle‘s representation) is that for debugging or recovery purposes it is possible for a human to read the pickled file with a standard text editor.

There are currently 3 different protocols which can be used for pickling.

• Protocol version 0 is the original ASCII protocol and is backwards compatible with earlier versions of Python.
• Protocol version 1 is the old binary format which is also compatible with earlier versions of Python.
• Protocol version 2 was introduced in Python 2.3. It provides much more efficient pickling of new-style classes.

Refer to PEP 307 for more information.

If a protocol is not specified, protocol 0 is used. If protocol is specified as a negative value or HIGHEST_PROTOCOL, the highest protocol version available will be used.

A binary format, which is slightly more efficient, can be chosen by specifying a protocol version >= 1.

1.3. Usage¶

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:

pickle.HIGHEST_PROTOCOL¶

The highest protocol version available. This value can be passed as a protocol value.

New in version 2.3.

The pickle module provides the following functions to make the pickling process more convenient:

pickle.dump(obj, file[, protocol])[source]¶

Write a pickled representation of obj to the open file object file. This is equivalent to Pickler(file, protocol).dump(obj).

If the protocol parameter is omitted, protocol 0 is used. If protocol is specified as a negative value or HIGHEST_PROTOCOL, the highest protocol version will be used.

Changed in version 2.3: Introduced the protocol parameter.

file must have a write() method that accepts a single string argument. It can thus be a file object opened for writing, a StringIO object, or any other custom object that meets this interface.

pickle.load(file)[source]¶

Read a string from the open file object file and interpret it as a pickle data stream, reconstructing and returning the original object hierarchy. This is equivalent to Unpickler(file).load().

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 a string. Thus file can be a file object opened for reading, a StringIO object, or any other custom object that meets this interface.

This function automatically determines whether the data stream was written in binary mode or not.

pickle.dumps(obj[, protocol])[source]¶

Return the pickled representation of the object as a string, instead of writing it to a file.

If the protocol parameter is omitted, protocol 0 is used. If protocol is specified as a negative value or HIGHEST_PROTOCOL, the highest protocol version will be used.

Changed in version 2.3: The protocol parameter was added.

pickle.loads(string)[source]¶

Read a pickled object hierarchy from a string. Characters in the string past the pickled object’s representation are ignored.

The pickle module also defines three exceptions:

exception pickle.PickleError[source]¶

A common base class for the other exceptions defined below. This inherits from Exception.

exception pickle.PicklingError[source]¶

This exception is raised when an unpicklable object is passed to the dump() method.

exception pickle.UnpicklingError[source]¶

This exception is raised when there is a problem unpickling an object. Note that other exceptions may also be raised during unpickling, including (but not necessarily limited to) AttributeError, EOFError, ImportError, and IndexError.

The pickle module also exports two callables [2], Pickler and Unpickler:

class pickle.Pickler(file[, protocol])[source]¶

This takes a file-like object to which it will write a pickle data stream.

If the protocol parameter is omitted, protocol 0 is used. If protocol is specified as a negative value or HIGHEST_PROTOCOL, the highest protocol version will be used.

Changed in version 2.3: Introduced the protocol parameter.

file must have a write() method that accepts a single string argument. It can thus be an open file object, a StringIO object, or any other custom object that meets this interface.

Pickler objects define one (or two) public methods:

dump(obj)[source]¶

Write a pickled representation of obj to the open file object given in the constructor. Either the binary or ASCII format will be used, depending on the value of the protocol argument passed to the constructor.

clear_memo()[source]¶

Clears the pickler’s “memo”. The memo is the data structure that remembers which objects the pickler has already seen, so that shared or recursive objects pickled by reference and not by value. This method is useful when re-using picklers.

Note

Prior to Python 2.3, clear_memo() was only available on the picklers created by cPickle. In the pickle module, picklers have an instance variable called memo which is a Python dictionary. So to clear the memo for a pickle module pickler, you could do the following:

mypickler.memo.clear()

Code that does not need to support older versions of Python should simply use clear_memo().

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. [3]

Unpickler objects are defined as:

class pickle.Unpickler(file)[source]¶

This takes a file-like object from which it will read a pickle data stream. This class automatically determines whether the data stream was written in binary mode or not, so it does not need a flag as in the Pickler factory.

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 a string. Thus file can be a file object opened for reading, a StringIO object, or any other custom object that meets this interface.

Unpickler objects have one (or two) public methods:

load()[source]¶

Read a pickled object representation from the open file object given in the constructor, and return the reconstituted object hierarchy specified therein.

This method automatically determines whether the data stream was written in binary mode or not.

noload()¶

This is just like load() except that it doesn’t actually create any objects. This is useful primarily for finding what’s called “persistent ids” that may be referenced in a pickle data stream. See section The pickle protocol below for more details.

Note: the noload() method is currently only available on Unpickler objects created with the cPickle module. pickle module Unpicklers do not have the noload() method.

1.4. What can be pickled and unpickled?¶

The following types can be pickled:

• None, True, and False
• integers, long integers, floating point numbers, complex numbers
• normal and Unicode strings
• tuples, lists, sets, and dictionaries containing only picklable objects
• functions defined at the top level of a module
• built-in functions defined at the top level of a module
• classes that are defined at the top level of a module
• instances of such classes whose __dict__ or the result of calling __getstate__() is picklable (see section The pickle protocol for details).

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 the 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. [4]

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 attr'

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.

1.5. The pickle protocol¶

This section describes the “pickling protocol” that defines the interface between the pickler/unpickler and the objects that are being serialized. This protocol provides a standard way for you to define, customize, and control how your objects are serialized and de-serialized. The description in this section doesn’t cover specific customizations that you can employ to make the unpickling environment slightly safer from untrusted pickle data streams; see section Subclassing Unpicklers for more details.

import pickle
from cStringIO import StringIO

src = StringIO()
p = pickle.Pickler(src)

def persistent_id(obj):
    if hasattr(obj, 'x'):
        return 'the value %d' % obj.x
    else:
        return None

p.persistent_id = persistent_id

class Integer:
    def __init__(self, x):
        self.x = x
    def __str__(self):
        return 'My name is integer %d' % self.x

i = Integer(7)
print i
p.dump(i)

datastream = src.getvalue()
print repr(datastream)
dst = StringIO(datastream)

up = pickle.Unpickler(dst)

class FancyInteger(Integer):
    def __str__(self):
        return 'I am the integer %d' % self.x

def persistent_load(persid):
    if persid.startswith('the value '):
        value = int(persid.split()[2])
        return FancyInteger(value)
    else:
        raise pickle.UnpicklingError, 'Invalid persistent id'

up.persistent_load = persistent_load

j = up.load()
print j

1.6. Subclassing Unpicklers¶

By default, unpickling will import any class that it finds in the pickle data. You can control exactly what gets unpickled and what gets called by customizing your unpickler. Unfortunately, exactly how you do this is different depending on whether you’re using pickle or cPickle. [9]

In the pickle module, you need to derive a subclass from Unpickler, overriding the load_global() method. load_global() should read two lines from the pickle data stream where the first line will the name of the module containing the class and the second line will be the name of the instance’s class. It then looks up the class, possibly importing the module and digging out the attribute, then it appends what it finds to the unpickler’s stack. Later on, this class will be assigned to the __class__ attribute of an empty class, as a way of magically creating an instance without calling its class’s __init__(). Your job (should you choose to accept it), would be to have load_global() push onto the unpickler’s stack, a known safe version of any class you deem safe to unpickle. It is up to you to produce such a class. Or you could raise an error if you want to disallow all unpickling of instances. If this sounds like a hack, you’re right. Refer to the source code to make this work.

Things are a little cleaner with cPickle, but not by much. To control what gets unpickled, you can set the unpickler’s find_global attribute to a function or None. If it is None then any attempts to unpickle instances will raise an UnpicklingError. If it is a function, then it should accept a module name and a class name, and return the corresponding class object. It is responsible for looking up the class and performing any necessary imports, and it may raise an error to prevent instances of the class from being unpickled.

The moral of the story is that you should be really careful about the source of the strings your application unpickles.

1.7. Example¶

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', u'Unicode string'),
         'c': None}

selfref_list = [1, 2, 3]
selfref_list.append(selfref_list)

output = open('data.pkl', 'wb')

# Pickle dictionary using protocol 0.
pickle.dump(data1, output)

# 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()

Here’s a larger 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.

#!/usr/local/bin/python

class TextReader:
    """Print and number lines in a text file."""
    def __init__(self, file):
        self.file = file
        self.fh = open(file)
        self.lineno = 0

    def readline(self):
        self.lineno = self.lineno + 1
        line = self.fh.readline()
        if not line:
            return None
        if line.endswith("\n"):
            line = line[:-1]
        return "%d: %s" % (self.lineno, line)

    def __getstate__(self):
        odict = self.__dict__.copy() # copy the dict since we change it
        del odict['fh']              # remove filehandle entry
        return odict

    def __setstate__(self, dict):
        fh = open(dict['file'])      # reopen file
        count = dict['lineno']       # read from file...
        while count:                 # until line count is restored
            fh.readline()
            count = count - 1
        self.__dict__.update(dict)   # update attributes
        self.fh = fh                 # save the file object

A sample usage might be something like this:

>>> import TextReader
>>> obj = TextReader.TextReader("TextReader.py")
>>> obj.readline()
'1: #!/usr/local/bin/python'
>>> obj.readline()
'2: '
>>> obj.readline()
'3: class TextReader:'
>>> import pickle
>>> pickle.dump(obj, open('save.p', 'wb'))

If you want to see that pickle works across Python processes, start another Python session, before continuing. What follows can happen from either the same process or a new process.

>>> import pickle
>>> reader = pickle.load(open('save.p', 'rb'))
>>> reader.readline()
'4:     """Print and number lines in a text file."""'

Footnotes