Skip to main content

A python script to run commands in parallel (think xargs and GNU parallel) without intermixing the output.

Project description

pyargs is a python cli program to run a series of scripts or commands using multiple set or arguments provided from a file or stdin and to use more than one process to achieve some parallel execution.

The name is deliberately a play on xargs as pyargs is a simplified alternative to xargs.

The motivation for this project was to get parallel execution of commands where the output from each command invocation is kept in a contigous block rather than being intermixed.

To make for easy replacement of xargs in existing scripts I used the same general usage pattern as xargs and kept of few of the option short names the same.

I should offer an apology to the python community. This is my first python effort so there is probably a lot in this small project the is no very *pythonic*.


usage: [-h] [-v] [--in INFILE_PATH] [--out OUTFILE_PATH]
                    [-n NARGS] [-P NPROCS] [--stream] [--debug] [--mark]
                    cmd [cmd_args [cmd_args ...]]

Runs multiple instances of a command in parallel with different arguments.
Think xargs.

positional arguments:
  cmd                   Command to execute
  cmd_args              Arguments for command to be used for every execution.
                        If any of these are options like -c you might have to
                        enclose them in quotes.

optional arguments:
  -h, --help            show this help message and exit
  -v, --version         Prints the version number.
  --in INFILE_PATH      Path to input file, each line has arguments for
                        command. If not provided uses stdin.
  --out OUTFILE_PATH    Path to output from all commands go to this file path.
                        If not provided stdout.
  -n NARGS, --nargs NARGS
                        Number of args to be found on each line of infile,
                        default = 1.
  -P NPROCS, --nprocs NPROCS
                        Number of parallel process, default = 1.
  --stream              Treat input as a single string rather than a series of
                        line for the purposes of tokenizing into arguments
  --debug               Prints out the command to be executed rather than
                        execute the command, to help problem solve
  --mark, -m            Put markers in the output to make visible the start
                        and output of each command.
  --lines, -L           Send the output line by line rather than keep output
                        frm each execution together.


Using pip

pip install pyargs

Alternatively download or clone the github repo and from the project directory

python install


Pythonic testing is via

python test

There are only two test cases each of which reside in the tests/ These tests cases execute shell scripts





This script executes multiple instances of ./tests/ using pyargs with arguments provided from tests/test_args. The output is piped into, and inspected by tests/ which parses that output and verifies that it is as expected.


Pipes the output form into to ensure that the previous test is using valid test data.

Command options - a bit more detail

The options --debug, --mark, --lines maybe require a little more explanation.

The script ./tests/ demonstrates the --line and --mark options, and ./tests/ demonstrates the --mark option without --lines.

  • –debug

    When set pyargs does not execute the commands but rather outputs the full command that would have been executed. This enables a user to see how pyargs has interpreted its options and input. This option can be helpful in debugging commands that fail.

  • –lines

    The original motivation for pyargs was to keep all the output from a single command invocation in a single contiguous block. However this may not always be necessary so this option allows or requires that pyargs will print each line of output from command invocations as soon as possible without waiting for the command to complete. This means that lines from different command invocations can be intermixed. Though note that concurrent output is still coordinated to ensure that lines do not corrupt each other.

    In order that each line of output can be attributed to the command that created it, in this mode, each output line is prefixed with the command string of the command that caused the output.

  • –mark

    Sometimes it is difficult to be sure that the output from different command invocations have not intermixed (this is when –list is NOT set), particularly when many commands are being executed and each command generates a lot of output.

    To assist users examine such a situation the `--mark options is provided.

    When --mark is set pyargs will modify the output in the following manner:

    • just before the execution of a command instance starts pyargs will output a string like

      MARK: <the command string to be executed> ===================
    • the output from each command invocation will be bracketed (that is have a additional marker line before and after the actual command output). This lines will look like this:

      START OUTPUT[<command string>]
      ...... output lines in here
      END OUTPUT[<command string>]

      These lines (between and including START and END) should be contiguous and should be the output from only one command and that command should be the one identified in the START and END lines (which or course should be the same command). If any of this is not the case you have found a bug in pyargs.


The two scripts tests/ and tests/ demonstrate the usage of pyargs.

Note that both these examples attempt to contact hosts/urls that do not exist and will hence timeout. Hence the output include error messages.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

pyargs-0.11.0.tar.gz (19.3 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page