Printing all shell commands fired from python scri

2019-07-25 17:22发布

I have a moderately large python script that executes a lot of shell commands from within itself. I would like to print all these commands to screen before executing it, so that I can trace the control flow of the python script.

Now, the script uses multiple commands to actually execute shell commands, including system() from os module, and call(), popen() and check_output() from subprocess module. What might be the easiest way to do this?

I was thinking of a wrapper function that prints the shell command argument before executing it, but I don't know how to write a generic one that can call the correct call/Popen or other command as per user discretion. And we also have to keep in mind that these calls take in different number and type of arguments.

Thanks!

3条回答
Anthone
2楼-- · 2019-07-25 17:58
import commands

commands = r'''find .'''
result = commands.getstatusoutput(command)[0]
print("Command: {}\nresult: {}".format(command,result))
查看更多
别忘想泡老子
3楼-- · 2019-07-25 18:04

Since you want to start a child process, but log your behavior, you will need to record child process stdout, stderr,

import subprocess as sp
import datetime
dt=datetime.datetime.now().strftime("%Y%m%d%H%M00")
ofd = open("res.sql","w+")
efd = open("log/cmd-{}.log".format(dt),"a+")

Suppose you are constructing a command (example, mysqldump) and you have the database, table, and credentials filenames (db, tbl, cnf) loaded. You want to print the command you are about to executed,

args = ["mysqldump","--defaults-extra-file="+cnf,"--single-transaction","--quick",db,tbl]
print " ".join(args)

Now, assume you have opened output and error files above (ofd, efd),

proc = sp.Popen(args, shell=False, stdin=sp.PIPE, stdout=ofd, stderr=efd)
stdout,stderr = proc.communicate
rc = proc.wait()
if rc>0: print "error",rc,"dbdump failed"
else: print "result",stdout,stderr

Remember to close ofd, efd.

查看更多
闹够了就滚
4楼-- · 2019-07-25 18:13

Build and install a decorator for the various calls you wish to intercept. Because even the "builtin" functions are first-class objects, you can replace them with logging versions:

import os

def log_wrap(f):
    def logging_wrapper(*args, **kwargs):
        print("Calling {} with args: {!r}, kwargs: {!r}".format(f.__name__, args, kwargs))
        return f(*args, **kwargs)

    return logging_wrapper

os.system("echo 'Hello, world!'")
os.system = log_wrap(os.system)
os.system("echo 'How do you like me now?!'")

This code, when run (python3) prints:

$ python test.py
Hello, world!
Calling system with args: ("echo 'How do you like me now?!'",), kwargs: {}
How do you like me now?!

Note that between the first and second calls to os.system, I replaced the os.system function with one that prints a log message before passing the arguments along to the original function. You can print your log message to a file, or (better yet) call the logging module, or invoke the pdb debugger, or whatever you like...

查看更多
登录 后发表回答