Pickle Module : Pickling(Serializing) and unpickling(De-serializing) of Python Objects

The pickle module in Python serializes objects into byte stream and de-serializes the byte stream back into Python objects. The example uses the dump() method to pickle a Python object and uses the load() method to unpickle it.

Pickle Updated 26 Nov 2018 1 min read

Overview

  • Python objects can be serialized into binary form and deserialized back to Python objects using the methods and classes available from the Python module pickle.
  • The process of converting a Python object into binary form is called Pickling.
  • The process of converting the binary stream back to a Python object is called Unpickling.
  • Pickled objects can be stored in a disk file and unpickled back as objects.
  • The Python module pickle provides the methods, classes, constants and exceptions to be used during the pickling process.
  • While the dump() variants pickle the Python object hierarchies the load() variants unpickle the bytes back into Python object hierarchies.

Pickling and unpickling in Python

Example

In the Python example code below, the method pickle.dump() is used to convert a Python object type Part into a byte stream. The pickle.load() is used to convert the byte stream back to a Python object.

python
import pickle

# Class definition of a part

class Part:

    number = 0

    name   = ""

    def __init__(self, number, name):

        self.number  = number

        self.name    = name

    def identify(self):

        print("Part Number: %d"%(self.number))

        print("Part Name: %s"%(self.name))

# Instantiate a part

part1=Part(111, "Spindle");

# Create a file to store the pickled object

objectRepository = open("ObjectFile.picl", "wb");

# Pickle/serialize the python object and store the bytes into a binary file

pickle.dump(part1, objectRepository, protocol=pickle.HIGHEST_PROTOCOL)

objectRepository.close()

# Unpickle/de-serialize the python object and print the attributes of the object

objectRepository    = open("ObjectFile.picl", "rb",);

reconstructedObject = pickle.load(objectRepository)

# Print object attributes

reconstructedObject.identify()

Output

Output
Part Number: 111

Part Name: Spindle