【python】遍历log查找符合关键字的log

需求

遍历logfile文件夹，除去pass目录下的文件都查一遍

要求：log文件名要求以"log_"开头，类型为txt文件

遍历log文件，key_value_XXX.txt文件中包含的字符串，都存到result_开头的txt中

代码

# 0_log_auto_analysis.bat

@echo off
for /F "tokens=1,2 delims=#" %%a in ('"prompt #$H#$E# & echo on & for %%b in (1) do rem"') do (
  set "DEL=%%a"
)


@echo ------------------- python -----------------------
python auto_log_analysis.py


@echo ------------------- EXIT ---------------------------
echo=
pause
exit

::----------- delay_ms sub-function (unit : 10 ms) --------------------
:delay_ms
@echo off
if "%1"=="" goto :eof
set delay_msTime=%1
set TotalTime=0
set NowTime=%time%
:: read init time, format : 13:01:05.95
:delay_ms_continue
set /a minute1=1%NowTime:~3,2%-100
set /a second1=1%NowTime:~-5,2%%NowTime:~-2%0-100000
set NowTime=%time%
set /a minute2=1%NowTime:~3,2%-100
set /a second2=1%NowTime:~-5,2%%NowTime:~-2%0-100000
set /a TotalTime+=(%minute2%-%minute1%+60)%%60*60000+%second2%-%second1%
if %TotalTime% lss %delay_msTime% goto delay_ms_continue
goto :eof
:ColorText
@echo off
<nul set /p ".=%DEL%" > "%~2"
findstr /v /a:%1 /R "^$" "%~2" nul
del "%~2" > nul 2>&1
goto :eof

# auto_log_analysis.py
import os

def get_front_2_last (line, front, last):
        begin=line.find(front)+len(front)
        end=line.find(last)
        ss=line[begin:end]
        # print(ss)
        return ss

def apply_file (log_name, log_key):
        file_log=open(log_name,encoding='utf-8',errors='gbk')
        # file_key_value=open("log1.txt",errors='ignore')

        file_key_value=open(log_key)
        #file_key_result=open("key_result_usecase.txt","a")#use w just save 1 char

        result_name=log_name.replace('log_','result_')
        print(result_name)
        print(log_name)
        # result_name="log_file/result_log1.txt"
        file_key_result=open(result_name,"w")#use w just save 1 char
        file_key_result.truncate()              #delete now contect

        line_log=file_log.readline()

        while line_log:
                #func2{
                file_key_value.seek(0, 0)
                key_value=file_key_value.readline()
                while key_value:
                        key=key_value.strip()
                        result_a=key in line_log
                        if result_a == True :
                                break
                        key_value=file_key_value.readline()
                        
                temp="adev_open_input_stream:" in line_log
                if temp == True :
                        print("xxx")
                        file_key_result.write("\n")

                if result_a == True :
                        file_key_result.write(line_log)     #copy one line

                line_log=file_log.readline()  #read the next line
                #func2}

        file_log.close
        file_key_value.close
        file_key_result.close

        return



# func get java file path
def get_file_path(path, file_type, func, log_key):
        for root, dirs, files in os.walk(path):
                for f in files:
                        # ignore "pass" dir
                        a = "pass" in root
                        if a == True :
                                continue

                        temp1=0
                        temp2=0
                        temp=0
                        temp1=file_type in f    # file type is txt
                        temp2="log_" in f       # file is log not result
                        temp=temp1+temp2
                        if temp == 2 :
                                # print('file_path: ',file_path)
                                file_path=root+'/'+f
                                print(file_path)
                                print('file_name: ',f)
                                apply_file(file_path, log_key)
        return

dir_path="logfile"
file_type_a=".txt"
function_a="apply_file"
log_key="cmd/key_value_time.txt"

get_file_path(dir_path, file_type_a, function_a,log_key)

cmd目录下存放关键字txt，可以修改代码中的log_key选择。

logfile文件中有pass文件夹，用于存放过往的文件，pass文件夹不会被遍历
log_开头的为你需要处理的log
result_打头的是生成的结果