本地代码 搜索脚本 python实现

本来用find+grep可以搞定的，不过如果搜索多个路径和多个规则，写正则可能写过不来

find . -type f -name "*.py" | xargs grep "redis"

上面语句是在py文件中查找redis的字符。
 
 不过如果要在指定多个位置查找，可能要拼接几个管道，并且如果我要几个字符的关系是并集，就是多个关键字要在文本中同时出现，而且不一定在同一行，所以也不好写。
 
所以写了个python脚本，也方便在centos下运行

# -*- coding: utf-8 -*-

# @Time : 2021/4/14 1:46

# @File : search_string_in_folder.py

# @Author : Rocky C@www.30daydo.com



'''

搜索代码脚本

'''

import fire

import glob

import re



# TODO 用PYQT重写一个



PATH_LIST = [r'C:\git\\',r'C:\OtherGit\\',r'C:\OneDrive\viewed_code\\']

POST_FIX = 'py' # 后缀文件

# 关键词

WORDS=[]



EXCLUDE_PATH=[r'C:\OtherGit\cpython']



DEBUG = True



class FileSearcher:



    def __init__(self,kw):

        self.root_path_list = PATH_LIST

        self.default_coding ='utf-8'

        self.exception_handle_coding='gbk'

        self.kw=[]

        if not isinstance(kw,tuple):

            kw=(kw,)



        for k in kw:

            k=k.strip()

            self.kw.append(k)



    def search(self,file,encoding):

        match_dict = dict()



        for w in self.kw:

            match_dict.setdefault(w, False)



        line_number = 0

        line_list=list()

        with open(file, 'r', encoding=encoding) as fp:



            while 1:

                try:

                    line = fp.readline()



                except UnicodeDecodeError as e:



                    if DEBUG:

                        print(f'Error coding in file {file}')

                        print(e)



                    return None,None,None



                except Exception as e:

                    if DEBUG:

                        print(f'Error in file {file}')

                        print(e)

                    break



                if not line:

                    break



                line = line.strip()

                if not line:

                    continue



                for w in self.kw:

                    m=re.search(w,line,re.IGNORECASE)

                    if m:

                        match_dict.update({w:True})

                        line_list.append(line_number)



                line_number+=1



        return True,match_dict.copy(),line_list.copy()



    def print_match_result(self,file,line_list,encoding):



        with open(file, 'r', encoding=encoding) as fp:

            line_number = 0

            while 1:

                try:

                    line = fp.readline()

                except Exception as e:

                    if DEBUG:

                        print(f'Error in file {file}')

                        print(e)

                    break



                if not line:

                    break

                line=line.strip()



                if not line:

                    continue



                if line_number in line_list:

                    print(f'{file} :: {line_number} ====>\n {line[:50]}\n')



                line_number += 1



    def run(self):

        for path in self.root_path_list:



            search_path=path+'**/*.'+POST_FIX



            for file in glob.iglob(search_path,recursive=True):



                for ex_path in EXCLUDE_PATH:

                    ex_path=ex_path.replace('\\','')

                    temp_file=file.replace('\\','')

                    if ex_path in temp_file:

                        continue



                    use_encoding=self.default_coding

                    encode_proper,match_dict,line_list=self.search(file,use_encoding)



                    if not encode_proper:

                        use_encoding = self.exception_handle_coding

                        encode_proper,match_dict,line_list=self.search(file, use_encoding)



                    if match_dict is not None and len(match_dict)>0 and all(match_dict.values()):

                        # print(match_dict.values())

                        self.print_match_result(file,line_list,use_encoding)

                        # print(line_list)





def test_error_file():

    path=r'C:\git\CodePool\example-code\19-dyn-attr-prop\oscon\schedule2.py'

    with open(path,'r',encoding='utf8') as fp:

        while 1:

            x=fp.readline()

            if not x:

                break

            print(x)



def main(kw):

    app = FileSearcher(kw)

    app.run()



if __name__ == '__main__':

    fire.Fire(main)

运行： python main.py --kw=asyncio,gather