传统Python语言的主要控制结构是for循环。然而,需要注意的是for循环在Pandas中不常用,因此Python中for循环的有效执行并不适用于Pandas模式。一些常见控制结构如下。
- for循环
- while循环
- if/else语句
- try/except语句
- 生成器表达式
- 列表推导式
- 模式匹配
所有的程序最终都需要一种控制执行流的方式。本节介绍一些控制执行流的技术。
01 for循环
for循环是Python的一种最基本的控制结构。使用for循环的一种常见模式是使用range函数生成数值范围,然后对其进行迭代。
- res = range(3)
- print(list(res))
- #输出:[0, 1, 2]
- for i in range(3):
- print(i)
- '''输出:
- 0
- 1
- 2
- '''
- for循环列表
使用for循环的另一种常见模式是对列表进行迭代。
- martial_arts = ["Sambo","Muay Thai","BJJ"]
- for martial_art in martial_arts:
- print(f"{ martial_art} has influenced\
- modern mixed martial arts")
- '''输出:
- Sambo has influenced modern mixed martial arts
- Muay Thai has influenced modern mixed martial arts
- BJJ has influenced modern mixed martial arts
- '''
02 while循环
while循环是一种条件有效就会重复执行的循环方式。while循环的常见用途是创建无限循环。在本示例中,while循环用于过滤函数,该函数返回两种攻击类型中的一种。
- def attacks():
- list_of_attacks = ["lower_body", "lower_body",
- "upper_body"]
- print("There are a total of {lenlist_of_attacks)}\
- attacks coming!")
- for attack in list_of_ attacks:
- yield attack
- attack = attacks()
- count = 0
- while next(attack) == "lower_body":
- count +=1
- print(f"crossing legs to prevent attack #{count}")
- else:
- count += 1
- print(f"This is not lower body attack, \
- I will cross my arms for# count}")
- '''输出:
- There are a total of 3 attacks coming!
- crossing legs to prevent attack #1
- crossing legs to prevent attack #2
- This is not a lower body attack, I will cross my arms for #3
- '''
03 if/else语句
if/else语句是一条在判断之间进行分支的常见语句。在本示例中,if/elif用于匹配分支。如果没有匹配项,则执行最后一条else语句。
- def recommended_attack(position):
- """Recommends an attack based on the position"""
- if position == "full_guard":
- print(f"Try an armbar attack")
- elif position == "half_guard":
- print(f"Try a kimura attack")
- elif position == "fu1l_mount":
- print(f"Try an arm triangle")
- else:
- print(f"You're on your own, \
- there is no suggestion for an attack")
- recommended_attack("full_guard")#输出:Try an armbar attack
- recommended_attack("z_guard")
- #输出:You're on your own, there is no suggestion for an attack
04 生成器表达式
生成器表达式建立在yield语句的概念上,它允许对序列进行惰性求值。生成器表达式的益处是,在实际求值计算前不会对任何内容进行求值或将其放入内存。这就是下面的示例可以在生成的无限随机攻击序列中执行的原因。
在生成器管道中,诸如 “arm_triangle”的小写攻击被转换为“ARM_TRIANGLE”,接下来删除其中的下划线,得到“ARM TRIANGLE”。
- def lazy_return_random_attacks():
- """Yield attacks each time"""
- import random
- attacks = {"kimura": "upper_body",
- "straight_ankle_lock": "lower_body",
- "arm_triangle": "upper_body",
- "keylock": "upper_body",
- "knee_bar": "lower_body"}
- while True:
- random_attack random.choices(list(attacks.keys()))
- yield random attack
- #Make all attacks appear as Upper Case
- upper_case_attacks = \
- (attack.pop().upper() for attack in \
- lazy_return_random_attacks())
- next(upper-case_attacks)
- #输出:ARM-TRIANGLE
- ## Generator Pipeline: One expression chains into the next
- #Make all attacks appear as Upper Case
- upper-case_attacks =\
- (attack. pop().upper() for attack in\
- lazy_return_random_attacks())
- #remove the underscore
- remove underscore =\
- (attack.split("_")for attack in\
- upper-case_attacks)
- #create a new phrase
- new_attack_phrase =\
- (" ".join(phrase) for phrase in\
- remove_underscore)
- next(new_attack_phrase)
- #输出:'STRAIGHT ANKLE LOCK'
- for number in range(10):
- print(next(new_attack_phrase))
- '''输出:
- KIMURA
- KEYLOCK
- STRAIGHT ANKLE LOCK
- '''
05 列表推导式
语法上列表推导式与生成器表达式类似,然而直接对比它们,会发现列表推导式是在内存中求值。此外,列表推导式是优化的C代码,可以认为这是对传统for循环的重大改进。
- martial_arts = ["Sambo", "Muay Thai", "BJJ"]
- new_phrases [f"mixed Martial Arts is influenced by \
- (martial_art)" for martial_art in martial_arts]
- print(new_phrases)
- ['Mixed Martial Arts is influenced by Sambo', \
- 'Mixed Martial Arts is influenced by Muay Thai', \
- 'Mixed Martial Arts is influenced by BJJ']
06 中级主题
有了这些基础知识后,重要的是不仅要了解如何创建代码,还要了解如何创建可维护的代码。创建可维护代码的一种方法是创建一个库,另一种方法是使用已经安装的第三方库编写的代码。其总体思想是最小化和分解复杂性。
- 使用Python编写库
使用Python编写库非常重要,之后将该库导入项目无须很长时间。下面这些示例是编写库的基础知识:在存储库中有一个名为funclib的文件夹,其中有一个_init_ .py文件。要创建库,在该目录中需要有一个包含函数的模块。
首先创建一个文件。
- touch funclib/funcmod.py
然后在该文件中创建一个函数。
- """This is a simple module"""
- def list_of_belts_in_bjj():
- """Returns a list of the belts in Brazilian jiu-jitsu"""
- belts= ["white", "blue", "purple", "brown", "black"]
- return belts
- import sys;sys.path.append("..")
- from funclib import funcmod
- funcmod.list_of_belts_in-bjj()
- #输出:['white', 'blue', 'purple', 'brown', 'black']
- 导入库
如果库是上面的目录,则可以用Jupyter添加sys.path.append方法来将库导入。接下来,使用前面创建的文件夹/文件名/函数名的命名空间导入模块。
- 安装第三方库
可使用pip install命令安装第三方库。请注意,conda命令(
https://conda.io/docs/user-guide/tasks/manage-pkgs.html)是pip命令的可选替代命令。如果使用conda命令,那么pip命令也会工作得很好,因为pip是virtualenv虚拟环境的替代品,但它也能直接安装软件包。
安装pandas包。
- pip install pandas
另外,还可使用requirements.txt文件安装包。
- > ca requirements.txt
- pylint
- pytest
- pytest-cov
- click
- jupyter
- nbval
- > pip install -r requirements.txt
下面是在Jupyter Notebook中使用小型库的示例。值得指出的是,在Jupyter Notebook中创建程序代码组成的巨型蜘蛛网很容易,而且非常简单的解决方法就是创建一些库,然后测试并导入这些库。
- """This is a simple module"""
- import pandas as pd
- def list_of_belts_in_bjj():
- """Returns a list of the belts in Brazilian jiu-jitsu"""
- belts = ["white", "blue", "purple", "brown", "black"]
- return belts
- def count_belts():
- """Uses Pandas to count number of belts"""
- belts = list_of_belts_in_bjj()
- df = pd.Dataframe(belts)
- res = df.count()
- count = res.values.tolist()[0]
- return count
- from funclib.funcmod import count_belts
- print(count_belts())
- #输出:5
- 类
可在Jupyter Notebook中重复使用类并与类进行交互。最简单的类类型就是一个名称,类的定义形式如下。
- class Competitor: pass
该类可实例化为多个对象。
- class Competitor: pass
- conor = Competitor()
- conor.name = "Conor McGregor"
- conor.age = 29
- conor.weight = 155
- nate = Competitor()
- nate.name = "Nate Diaz"
- nate.age = 30
- nate.weight = 170
- def print_competitor _age(object):
- """Print out age statistics about a competitor"""
- print(f"{object.name} is {object.age} years old")
- print_competitor_age(nate)
- #输出:Nate Diaz is 30 years old
- print_competitor_age(conor)
- #输出:Conor McGregor is 29 years old
- 类和函数的区别
类和函数的主要区别包括:
- 函数更容易解释。
- 函数(典型情况下)只在函数内部具有状态,而类在函数外部保持不变的状态。
- 类能以复杂性为代价提供更高级别的抽象。