Detailed explanation of the use of pickle library

The use of the pickle library is mentioned in the article "Understanding what machine learning is through a simple example", and this article will further elaborate on it.

So why do we need the operations of serialization and deserialization?

1. Easy to store. The serialization process turns textual information into a stream of binary data. In this way, the information is easily stored in the hard disk. When a file needs to be read, the data is read from the hard disk and then deserialized to obtain the original data. When the Python program is running, some data such as strings, lists, and dictionaries are obtained. I want to save them for a long time for future use, instead of simply putting them in the memory and losing the data when the power is turned off. The Pickle module in the python module collection comes in handy, it can convert objects into a format that can be transmitted or stored.

2. The loads() function performs the same deserialization as the load() function. Instead of accepting a stream object and reading the serialized data from a file, it accepts a str object containing the serialized data and returns the object directly.

3. Code example:
   

4. [python] view plain copy
   #!/usr/bin/env python  
   # -*- coding: UTF-8 -*-  
   import cPickle as pickle   
   obj = 123, "abcdedf", ["ac", 123], {"key": "value", "key1": "value1"}  
   print obj# 输出：(123, 'abcdedf', ['ac', 123], {'key1': 'value1', 'key': 'value'})      
   # 序列化到文件    
   with open(r"d:\a.txt", "r+") as f:  
       pickle.dump(obj, f)      
   with open(r"d:\a.txt") as f:  
       print pickle.load(f)# 输出：(123, 'abcdedf', ['ac', 123], {'key1': 'value1', 'key': 'value'})    
   # 序列化到内存（字符串格式保存），然后对象可以以任何方式处理如通过网络传输    
   obj1 = pickle.dumps(obj)  
   print type(obj1)# 输出：<type &#39;str&#39;>    
   print obj1# 输出：python专用的存储格式    
     
   obj2 = pickle.loads(obj1)  
   print type(obj2)# 输出：<type &#39;tuple&#39;>    
   print obj2# 输出：(123, &#39;abcdedf&#39;, [&#39;ac&#39;, 123], {&#39;key1&#39;: &#39;value1&#39;, &#39;key&#39;: &#39;value&#39;})

2. Easy to transmit. When two processes communicate remotely, they can send various types of data to each other. No matter what type of data it is, it is transmitted over the network in the form of a binary sequence. The sender needs to convert this object into a byte sequence before it can be transmitted on the network; the receiver needs to restore the byte sequence into an object.

• Understand what machine learning is through a simple example

pickle is a standard module of the python language. The pickle library is included after installing python. No need to install it separately.
The pickle module implements basic data serialization and deserialization. Through the serialization operation of the pickle module, we can save the object information running in the program to a file and store it permanently; through the deserialization operation of the pickle module, we can create the object saved by the last program from the file.
In the official introduction, the English description of the serialization operation has several words, such as "serializing", "pickling", "serialization", "marshalling" or "flattening", etc., they all represent serialization mean. Correspondingly, there are many English words for deserialization operations, such as "de-serializing", "unpickling", "deserailization", etc. To avoid confusion, generally use "pickling"/"unpickling", or "serialization"/"deserailization".
pickle The module is serialized in binary form and saved to a file (the suffix of the saved file is ".pkl"), and cannot be opened directly for preview. Another serialization standard module of python, json, is human-readable and can be opened and viewed directly (for example, viewed in notepad++).

The pickle module has two main types of interfaces, namely serialization and deserialization.

is often used in this way:

[python] view plain copy
import cPickle as pickle  
pickle.dump(obj,f)  
pickle.dumps(obj,f)  
pickle.load(f)  
pickle.loads(f)

The serialization operation includes:

• pickle.dump()

• Deserialization operations include:

• Pickler(file, protocol).dump(obj)

• pickle.load()

• Unpickler(file).load()

2 序列化操作

2.1 序列化方法pickle.dump()

序列化的方法为 pickle.dump()，该方法的相关参数如下：
pickle.dump(obj, file, protocol=None,*,fix_imports=True)
该方法实现的是将序列化后的对象obj以二进制形式写入文件file中，进行保存。它的功能等同于 Pickler(file, protocol).dump(obj)。
关于参数file，有一点需要注意，必须是以二进制的形式进行操作（写入）。
参考前文的案例如下：

import picklewith open('svm_model_iris.pkl', 'wb') as f:
    pickle.dump(svm_classifier, f)

file为’svm_model_iris.pkl’，并且以二进制的形式（’wb’）写入。

关于参数protocol，一共有5中不同的类型，即（0,1,2,3,4）。（0,1,2）对应的是python早期的版本，（3,4）则是在python3之后的版本。
此外，参数可选 pickle.HIGHEST_PROTOCOL和pickle.DEFAULT_PROTOCOL。当前，python3.5版本中，pickle.HIGHEST_PROTOCOL的值为4，pickle.DEFAULT_PROTOCOL的值为3。当protocol参数为负数时，表示选择的参数是pickle.HIGHEST_PROTOCOL。
关于参数protocol，官方的详细介绍如下：

2.2 序列化方法pickle.dumps()

pickle.dumps()方法的参数如下：
pickle.dumps(obj, protocol=None,*,fix_imports=True)
pickle.dumps()方法跟pickle.dump()方法的区别在于，pickle.dumps()方法不需要写入文件中，它是直接返回一个序列化的bytes对象。

2.3 序列化方法Pickler(file, protocol).dump(obj)

pickle模块提供了序列化的面向对象的类方法，即 class pickle.Pickler(file, protocol=None,*,fix_imports=True),Pickler类有dump()方法。
Pickler(file, protocol).dump(obj) 实现的功能跟 pickle.dump() 是一样的。
关于Pickler类的其他method，请参考官方API。

插播一条硬广：技术文章转发太多，本文来自微信公众号：“Python数据之道”（ID：PyDataRoad）。

3 反序列化操作

3.1 反序列化方法pickle.load()

序列化的方法为 pickle.load()，该方法的相关参数如下：
pickle.load(file, *,fix_imports=True, encoding=”ASCII”. errors=”strict”)
该方法实现的是将序列化的对象从文件file中读取出来。它的功能等同于 Unpickler(file).load()。
关于参数file，有一点需要注意，必须是以二进制的形式进行操作（读取）。
参考前文的案例如下：

import picklewith open('svm_model_iris.pkl', 'rb') as f:
    model = pickle.load(f)

file为’svm_model_iris.pkl’，并且以二进制的形式（’rb’）读取。

读取的时候，参数protocol是自动选择的，load()方法中没有这个参数。

3.2 反序列化方法pickle.loads()

pickle.loads()方法的参数如下：
pickle.loads(bytes_object, *,fix_imports=True, encoding=”ASCII”. errors=”strict”)
pickle.loads()方法跟pickle.load()方法的区别在于，pickle.loads()方法是直接从bytes对象中读取序列化的信息，而非从文件中读取。

3.3 反序列化方法Unpickler(file).load()

pickle模块提供了反序列化的面向对象的类方法，即 class pickle.Unpickler(file, *,fix_imports=True, encoding="ASCII". errors="strict"),Pickler类有load()方法。
Unpickler(file).load() 实现的功能跟 pickle.load() 是一样的。
关于Unpickler类的其他method，请参考官方API。

4 那些类型可以进行序列化和反序列化操作

官方文档是这么介绍的，这里我就不进一步描述了。

写在后面

pickle模块还是比较实用的，当然，关于pickle模块，其实还有许多的信息可以去了解，想了解更多信息的童鞋，建议可以阅读下python官方的API文档（library文件）。

The above is the detailed content of Detailed explanation of the use of pickle library. For more information, please follow other related articles on the PHP Chinese website!