pathlib~更优雅地操作路径~

最近在做图像相关的算法，经常有对文件路径的一些条件遍历操作，发现最好使的还是下面这行神奇代码。

from pathlib import Path
imgs = [str(x) for x in Path('./dataset/images/').rglob('*.jp*g') if 'checkpoint' not in str(x)]

这行代码主要使用的是pathlib库，相比于传统的os模块的函数式文件操作，使用pathlib模块来操作文件系统更加简单，可读性更强。

pathlib库的核心类是pathlib.Path，所有功能都可以通过调用这个类的属性和方法进行实现，非常的面向对象。

公众号算法美食屋后台回复关键词：pathlib，获取本文jupyter notebook 源代码

一，Path构造

from pathlib import Path

#指定路径
mypath = Path('../demo.py')
print(mypath)

../demo.py

#当前路径
cwd = Path.cwd() 
print(str(cwd))

/Users/liangyun2/CodeFiles/machine_learning_AK47/1_Feature_Engineering

#home路径
home = Path.home() 
print(str(home))

/Users/liangyun2

二，Path属性

mypath = Path.cwd()

mypath.name  #文件名

'1_Feature_Engineering'

mypath.suffix  #后缀

''

mypath.parent #

PosixPath('/Users/liangyun2/CodeFiles/machine_learning_AK47')

list(mypath.parents)

[PosixPath('/Users/liangyun2/CodeFiles/machine_learning_AK47'),
 PosixPath('/Users/liangyun2/CodeFiles'),
 PosixPath('/Users/liangyun2'),
 PosixPath('/Users'),
 PosixPath('/')]

三，Path遍历

mypath = Path.cwd().parent

file = mypath/'1_Feature_Engineering'/'demo.py'
print(file)

/Users/liangyun2/CodeFiles/machine_learning_AK47/1_Feature_Engineering/demo.py

file.exists()

True

for sub in mypath.iterdir():
    print(sub)

/Users/liangyun2/CodeFiles/machine_learning_AK47/.DS_Store
/Users/liangyun2/CodeFiles/machine_learning_AK47/3_Tune_Tools
/Users/liangyun2/CodeFiles/machine_learning_AK47/6_CV
/Users/liangyun2/CodeFiles/machine_learning_AK47/8_RL
/Users/liangyun2/CodeFiles/machine_learning_AK47/README.ipynb
/Users/liangyun2/CodeFiles/machine_learning_AK47/utils
/Users/liangyun2/CodeFiles/machine_learning_AK47/models
/Users/liangyun2/CodeFiles/machine_learning_AK47/2_Base_Models
/Users/liangyun2/CodeFiles/machine_learning_AK47/push_to_github.ipynb
/Users/liangyun2/CodeFiles/machine_learning_AK47/.gitignore
/Users/liangyun2/CodeFiles/machine_learning_AK47/.ipynb_checkpoints
/Users/liangyun2/CodeFiles/machine_learning_AK47/1_Feature_Engineering
/Users/liangyun2/CodeFiles/machine_learning_AK47/.git
/Users/liangyun2/CodeFiles/machine_learning_AK47/data
/Users/liangyun2/CodeFiles/machine_learning_AK47/7_NLP
/Users/liangyun2/CodeFiles/machine_learning_AK47/5_Integrated_Examples
/Users/liangyun2/CodeFiles/machine_learning_AK47/4_Ensemble_Tools

for sub in mypath.glob('*'):
    print(sub)

/Users/liangyun2/CodeFiles/machine_learning_AK47/.DS_Store
/Users/liangyun2/CodeFiles/machine_learning_AK47/3_Tune_Tools
/Users/liangyun2/CodeFiles/machine_learning_AK47/6_CV
/Users/liangyun2/CodeFiles/machine_learning_AK47/8_RL
/Users/liangyun2/CodeFiles/machine_learning_AK47/README.ipynb
/Users/liangyun2/CodeFiles/machine_learning_AK47/utils
/Users/liangyun2/CodeFiles/machine_learning_AK47/models
/Users/liangyun2/CodeFiles/machine_learning_AK47/2_Base_Models
/Users/liangyun2/CodeFiles/machine_learning_AK47/push_to_github.ipynb
/Users/liangyun2/CodeFiles/machine_learning_AK47/.gitignore
/Users/liangyun2/CodeFiles/machine_learning_AK47/.ipynb_checkpoints
/Users/liangyun2/CodeFiles/machine_learning_AK47/1_Feature_Engineering
/Users/liangyun2/CodeFiles/machine_learning_AK47/.git
/Users/liangyun2/CodeFiles/machine_learning_AK47/data
/Users/liangyun2/CodeFiles/machine_learning_AK47/7_NLP
/Users/liangyun2/CodeFiles/machine_learning_AK47/5_Integrated_Examples
/Users/liangyun2/CodeFiles/machine_learning_AK47/4_Ensemble_Tools

for sub in mypath.glob('*.ipynb'):
    print(sub)

/Users/liangyun2/CodeFiles/machine_learning_AK47/README.ipynb
/Users/liangyun2/CodeFiles/machine_learning_AK47/push_to_github.ipynb

files = [x for x in mypath.rglob('*') if x.is_file()]
len(files)

9670

四，Path操作

path = Path.cwd()/'test.txt'

if not path.exists():
    path.touch() #创建文件
    
#读写内容
path.write_text("hello world！\n你好，中国！\n")
path.read_text()

'hello world！\n你好，中国！\n'

#增量形式写
with path.open('a')  as f:
    f.write('你好，北京')

path.read_text()

'hello world！\n你好，中国！\n你好，北京'

#重命名
path = path.rename('hello.txt')

# %load hello.txt
hello world！
你好，中国！
你好，北京

公众号算法美食屋后台回复关键词：pathlib，获取本文jupyter notebook 源代码。