找文件用的Python代码片段

摘要

python 很方便

但是用的时候还是要有手.

索性弄个片段, 可以只用脚 去找文件了

代码片段

# coding=utf-8
# python=2.7

import os as os
import re as re

# NOTE: 仅用于类型注释
try:
    import typing as typing
except:
    pass

class FilePathProcessor(object):
    u"""用于处理文件的类"""

    @classmethod
    def CheckIfDirExist (cls, dirPath):
        # type: (str)->None
        u"""判断文件夹是否存在, 否则抛出异常"""
        
        if not os.path.exists(dirPath):
            raise Exception(u"不存在文件夹, 或不是文件夹 : {}".format(dirPath))

    @classmethod
    def GetAllFileWithExt (cls, dirPath, extFormat, bSingleDir=True):
        # type: (str, str, bool) -> typing.Iterable[str]
        u"""遍历文件夹以寻找特定格式的文件"""

        # NOTE: 判断文件夹是否存在, 否则抛出异常
        cls.CheckIfDirExist (dirPath)
        
        # NOTE: 小写后缀扩展格式
        lowerExtFormat = extFormat.lower()
        
        # NOTE: 返回一个 路径生成器
        if bSingleDir: # NOTE 单层模式
            for fileName in os.listdir(dirPath):
                tempFilePath = os.path.join(dirPath,fileName)
                # NOTE: 非目录 && 以extFormat结尾
                if (not os.path.isdir(tempFilePath)) and (fileName.lower().endswith(lowerExtFormat)):
                    yield tempFilePath
        else : # NOTE 遍历所有子目录
            for dirPath,dirNames,fileNames in os.walk(dirPath):
                for fileName in fileNames:
                    # NOTE: 以extFormat结尾
                    if fileName.lower().endswith(lowerExtFormat):
                        yield os.path.join(dirPath, fileName)

    @classmethod
    def GetAllFileRegexWithExt (cls, dirPath, extFormat, bSingleDir=True, bUseRegex=False, regexRule=""):
        # type: (str, str, bool, bool, str) -> typing.Iterable[str]
        u"""遍历文件夹以寻找特定格式的文件, 且可选是否包含特定的Regex"""

        pathGenerator = cls.GetAllFileWithExt(dirPath,extFormat,bSingleDir)
        if not bUseRegex: # NOTE 不使用正则表达式
            for filePath in pathGenerator:
                yield filePath
        else : # NOTE 使用正则表达式
            regexMatchRule = re.compile(regexRule,re.I)
            for filePath in pathGenerator:
                regexMatched = regexMatchRule.search(os.path.basename(filePath))
                if regexMatched != None :
                    yield filePath


# NOTE: Easy CV!
if __name__ == "__main__":
    targetDirPath = os.environ["userprofile"]
    targetExtFormat = ".fbx"
    bUseSingleDir = False
    bUseRegex = True
    regexRule = r"^anim.*?"
    filePathGenerator = FilePathProcessor.GetAllFileRegexWithExt(
                        targetDirPath,
                        targetExtFormat,
                        bUseSingleDir,
                        bUseRegex,
                        regexRule)

    for filePath in filePathGenerator :
        print(filePath)