What's the best way to parse command line arguments?

Viewed 309174

What's the easiest, tersest, and most flexible method or library for parsing Python command line arguments?

15 Answers

This answer suggests optparse which is appropriate for older Python versions. For Python 2.7 and above, argparse replaces optparse. See this answer for more information.

As other people pointed out, you are better off going with optparse over getopt. getopt is pretty much a one-to-one mapping of the standard getopt(3) C library functions, and not very easy to use.

optparse, while being a bit more verbose, is much better structured and simpler to extend later on.

Here's a typical line to add an option to your parser:

parser.add_option('-q', '--query',
            action="store", dest="query",
            help="query string", default="spam")

It pretty much speaks for itself; at processing time, it will accept -q or --query as options, store the argument in an attribute called query and has a default value if you don't specify it. It is also self-documenting in that you declare the help argument (which will be used when run with -h/--help) right there with the option.

Usually you parse your arguments with:

options, args = parser.parse_args()

This will, by default, parse the standard arguments passed to the script (sys.argv[1:])

options.query will then be set to the value you passed to the script.

You create a parser simply by doing

parser = optparse.OptionParser()

These are all the basics you need. Here's a complete Python script that shows this:

import optparse

parser = optparse.OptionParser()

parser.add_option('-q', '--query',
    action="store", dest="query",
    help="query string", default="spam")

options, args = parser.parse_args()

print 'Query string:', options.query

5 lines of python that show you the basics.

Save it in sample.py, and run it once with

python sample.py

and once with

python sample.py --query myquery

Beyond that, you will find that optparse is very easy to extend. In one of my projects, I created a Command class which allows you to nest subcommands in a command tree easily. It uses optparse heavily to chain commands together. It's not something I can easily explain in a few lines, but feel free to browse around in my repository for the main class, as well as a class that uses it and the option parser

The new hip way is argparse for these reasons. argparse > optparse > getopt

update: As of py2.7 argparse is part of the standard library and optparse is deprecated.

Pretty much everybody is using getopt

Here is the example code for the doc :

import getopt, sys

def main():
    try:
        opts, args = getopt.getopt(sys.argv[1:], "ho:v", ["help", "output="])
    except getopt.GetoptError:
        # print help information and exit:
        usage()
        sys.exit(2)
    output = None
    verbose = False
    for o, a in opts:
        if o == "-v":
            verbose = True
        if o in ("-h", "--help"):
            usage()
            sys.exit()
        if o in ("-o", "--output"):
            output = a

So in a word, here is how it works.

You've got two types of options. Those who are receiving arguments, and those who are just like switches.

sys.argv is pretty much your char** argv in C. Like in C you skip the first element which is the name of your program and parse only the arguments : sys.argv[1:]

Getopt.getopt will parse it according to the rule you give in argument.

"ho:v" here describes the short arguments : -ONELETTER. The : means that -o accepts one argument.

Finally ["help", "output="] describes long arguments ( --MORETHANONELETTER ). The = after output once again means that output accepts one arguments.

The result is a list of couple (option,argument)

If an option doesn't accept any argument (like --help here) the arg part is an empty string. You then usually want to loop on this list and test the option name as in the example.

I hope this helped you.

Use optparse which comes with the standard library. For example:

#!/usr/bin/env python
import optparse

def main():
  p = optparse.OptionParser()
  p.add_option('--person', '-p', default="world")
  options, arguments = p.parse_args()
  print 'Hello %s' % options.person

if __name__ == '__main__':
  main()

Source: Using Python to create UNIX command line tools

However as of Python 2.7 optparse is deprecated, see: Why use argparse rather than optparse?

Just in case you might need to, this may help if you need to grab unicode arguments on Win32 (2K, XP etc):


from ctypes import *

def wmain(argc, argv):
    print argc
    for i in argv:
        print i
    return 0

def startup():
    size = c_int()
    ptr = windll.shell32.CommandLineToArgvW(windll.kernel32.GetCommandLineW(), byref(size))
    ref = c_wchar_p * size.value
    raw = ref.from_address(ptr)
    args = [arg for arg in raw]
    windll.kernel32.LocalFree(ptr)
    exit(wmain(len(args), args))
startup()

Argparse code can be longer than actual implementation code!

That's a problem I find with most popular argument parsing options is that if your parameters are only modest, the code to document them becomes disproportionately large to the benefit they provide.

A relative new-comer to the argument parsing scene (I think) is plac.

It makes some acknowledged trade-offs with argparse, but uses inline documentation and wraps simply around main() type function function:

def main(excel_file_path: "Path to input training file.",
     excel_sheet_name:"Name of the excel sheet containing training data including columns 'Label' and 'Description'.",
     existing_model_path: "Path to an existing model to refine."=None,
     batch_size_start: "The smallest size of any minibatch."=10.,
     batch_size_stop:  "The largest size of any minibatch."=250.,
     batch_size_step:  "The step for increase in minibatch size."=1.002,
     batch_test_steps: "Flag.  If True, show minibatch steps."=False):
"Train a Spacy (http://spacy.io/) text classification model with gold document and label data until the model nears convergence (LOSS < 0.5)."

    pass # Implementation code goes here!

if __name__ == '__main__':
    import plac; plac.call(main)

I prefer optparse to getopt. It's very declarative: you tell it the names of the options and the effects they should have (e.g., setting a boolean field), and it hands you back a dictionary populated according to your specifications.

http://docs.python.org/lib/module-optparse.html

I think the best way for larger projects is optparse, but if you are looking for an easy way, maybe http://werkzeug.pocoo.org/documentation/script is something for you.

from werkzeug import script

# actions go here
def action_foo(name=""):
    """action foo does foo"""
    pass

def action_bar(id=0, title="default title"):
    """action bar does bar"""
    pass

if __name__ == '__main__':
    script.run()

So basically every function action_* is exposed to the command line and a nice help message is generated for free.

python foo.py 
usage: foo.py <action> [<options>]
       foo.py --help

actions:
  bar:
    action bar does bar

    --id                          integer   0
    --title                       string    default title

  foo:
    action foo does foo

    --name                        string

I extended Erco's approach to allow for required positional arguments and for optional arguments. These should precede the -d, -v etc. arguments.

Positional and optional arguments can be retrieved with PosArg(i) and OptArg(i, default) respectively. When an optional argument is found the start position of searching for options (e.g. -i) is moved 1 ahead to avoid causing an 'unexpected' fatal.

import os,sys


def HelpAndExit():
    print("<<your help output goes here>>")
    sys.exit(1)

def Fatal(msg):
    sys.stderr.write("%s: %s\n" % (os.path.basename(sys.argv[0]), msg))
    sys.exit(1)

def NextArg(i):
    '''Return the next command line argument (if there is one)'''
    if ((i+1) >= len(sys.argv)):
        Fatal("'%s' expected an argument" % sys.argv[i])
    return(1, sys.argv[i+1])

def PosArg(i):
    '''Return positional argument'''
    if i >= len(sys.argv):
        Fatal("'%s' expected an argument" % sys.argv[i])
    return sys.argv[i]

def OptArg(i, default):
    '''Return optional argument (if there is one)'''
    if i >= len(sys.argv):
        Fatal("'%s' expected an argument" % sys.argv[i])
    if sys.argv[i][:1] != '-':
        return True, sys.argv[i]
    else:
        return False, default


### MAIN
if __name__=='__main__':

    verbose = 0
    debug   = 0
    infile  = "infile"
    outfile = "outfile"
    options_start = 3

    # --- Parse two positional parameters ---
    n1 = int(PosArg(1))
    n2 = int(PosArg(2))

    # --- Parse an optional parameters ---
    present, a3 = OptArg(3,50)
    n3 = int(a3)
    options_start += int(present)

    # --- Parse rest of command line ---
    skip = 0
    for i in range(options_start, len(sys.argv)):
        if not skip:
            if   sys.argv[i][:2] == "-d": debug ^= 1
            elif sys.argv[i][:2] == "-v": verbose ^= 1
            elif sys.argv[i][:2] == "-i": (skip,infile)  = NextArg(i)
            elif sys.argv[i][:2] == "-o": (skip,outfile) = NextArg(i)
            elif sys.argv[i][:2] == "-h": HelpAndExit()
            elif sys.argv[i][:1] == "-":  Fatal("'%s' unknown argument" % sys.argv[i])
            else:                         Fatal("'%s' unexpected" % sys.argv[i])
        else: skip = 0

    print("Number 1 = %d" % n1)
    print("Number 2 = %d" % n2)
    print("Number 3 = %d" % n3)
    print("Debug    = %d" % debug)
    print("verbose  = %d" % verbose)
    print("infile   = %s" % infile)
    print("outfile  = %s" % outfile) 
Related