Skip to main content

Pieshell is a Python shell environment that combines the expressiveness of shell pipelines with the power of python iterators. It can be used both as an interactive shell and as an ordinary python module replacing e.g. subprocess.Popen

Project description

Python application


Pieshell is a Python shell environment that combines the expressiveness of shell pipelines with the power of python iterators.

It can be used in two major ways:

  • As an interactive shell replacing e.g. bash
  • As an ordinary python module replacing e.g. subprocess.Popen

Table of contents


$ pip install pieshell

As a shell

Executing basic commands

To start pieshell in interactive mode, just run the command pieshell:

$ pieshell

The interactive pieshell environment supports all normal python syntax.

140:/home/redhog/Projects/beta/pieshell >>> print 3+4

In addition, you can run programs just like in any shell by writing their names

140:/home/redhog/Projects/beta/pieshell >>> ls
build deps dist LICENSE.txt pieshell pieshell.egg-info

Parameters to programs however have to be given as proper python strings within parenthesis, like a python function call

140:/home/redhog/Projects/beta/pieshell >>> ls("-a")
. .. build deps dist .git .gitignore LICENSE.txt pieshell

Piping the standard output of one command to the standard input of another works just like in bash

140:/home/redhog/Projects/beta/pieshell >>> ls("-a") | grep("-e", ".py")

Changing directory is done using the command cd:

140:/home/redhog/Projects/beta/pieshell >>> cd("..")
140:/home/redhog/Projects/beta >>> 

Full syntax for command lines

To execute commands that require a path, for example ones in the current directory, or commands with a dot in their names

140:/home/redhog/Projects/beta/pieshell >>> _("./", "--help")
Common commands: (see '--help-commands' for more)

The underscore represents the virtual root command that has no parameters, not even a command name. In general, there are two equivalent syntaxes for parameters: as function parameter strings, and as attribute names. The two syntaxes can be mixed freely. All of the following are equivalent:

_("foo", "bar", "fie")"bar", "fie")

Example usage:


In addition to these two generic syntaxes, there are two more specialized syntaxes for options:

The function call syntax also supports named parameters, which are converted into "--name=value" pairs. Note that the order can not be guaranteed as named parameters are sent around as dictionaries inside python:


Short options, like -U above do not actually need quotes, and can be specified inside or outside the function call syntax:

git.diff(-U, -w)

Notes for programs with wierd parameter syntax, like find: find does not use the standard double minus (--) before long options, and takes the option value as a separate argument, rather than separating the name and value with = like most programs. The above special syntax for short options actually cover this use case too:

find(".", -name, "*.py")


Standard out and standard in of a pipeline can be redirected to a file by piping to or from a string (the filename). As a special case, None is a short hand for "/dev/null"

140:/home/redhog/Projects/beta/pieshell >>> ls | "foo"

140:/home/redhog/Projects/beta/pieshell >>> "foo" | cat

140:/home/redhog/Projects/beta/pieshell >>> ls | None

Redirects can also be made with a more explicit syntax that allows redirecting other file descriptors than stdin and stdout:

139:/home/redhog/Projects/beta/pieshell >>> cat |
  Redirect("stdin", "foo") | Redirect("stdout", "bar")

The constructor for redirect takes the following arguments:

Redirect(fd, source, flag=None, mode=0777)

fd can be either an int, or one of "stdin", "stdout" and "stderr. source is either a string filename, or an int file descriptor. flag and mode have the same semantics as for Flags do not have to be given for stdin, stdout and stderr / fd 0, 1 and 2 and defaults to os.O_RDONLY or os.O_RDONLY | os.O_CREAT.

Interfacing between python functions and shell commands

Shell commands are first class python objects, and their input and output can be interacted with easily from python in the form of iterators. Iterating over a shell command iterates over the lines of its standard out

140:/home/redhog/Projects/beta/pieshell >>> list(ls("-a"))
['.', '..', 'build', 'deps', 'dist', '.git', '.gitignore',
 'LICENSE.txt', 'pieshell', 'pieshell.egg-info', '',
 '', '']
140:/home/redhog/Projects/beta/pieshell >>> for x in ls("-a"):
...   if x.endswith('.py'):
...      print x

Piping an iterator into a shell command, sends its items as lines to the standard in of the shell command

140:/home/redhog/Projects/beta/pieshell >>> list(["foo", "", ""] |
  grep("-e", ".py"))
['', '']
140:/home/redhog/Projects/beta/pieshell >>> def foo():
...     yield "hello"
...     yield "world"
140:/home/redhog/Projects/beta/pieshell >>> foo() | cat

In addtion, iterators and pipelines may be used as arguments to commands and will be seen by the command as a filename, which when opened and read from will produce the output of that iterator as lines, or the standard output of the pipeline.

140:/home/redhog/Projects/beta/pieshell >>> list(cat(["foo", "bar"]))
['foo', 'bar']
140:/home/redhog/Projects/beta/pieshell >>> list(cat(["foo", "bar"] | cat))
['foo', 'bar']

Environment variables

Environment variables are available directly in the shell as variables, together with any local python variables. In addition, they are available in the dictionary exports.

140:/home/redhog/Projects/beta/pieshell >>> LANG

Assigning to the name of an already exported environment variable updates the value of that variable.

140:/home/redhog/Projects/beta/pieshell >>> LANG = "sv_SE.UTF-8"
140:/home/redhog/Projects/beta/pieshell >>> exports["LANG"]

Assigning to a variable name not already used as an environment variable creates a local python variable.

140:/home/redhog/Projects/beta/pieshell >>> foo = "hello"
140:/home/redhog/Projects/beta/pieshell >>> "foo" in exports
140:/home/redhog/Projects/beta/pieshell >>> foo

To export a new variable, you have to assign it in the exports dictionary.

140:/home/redhog/Projects/beta/pieshell >>> exports["bar"] = "world"
140:/home/redhog/Projects/beta/pieshell >>> bar

Argument expansion

All parameter strings in commands are subject to expansion unless wrapped in a call to R(), e.g. R("my * string * here").

  • "~" and "~username" are expanded using os.path.expanduser()

  • Variable expansion is done using the python % operator on python variables as well as environment variables.

  • Pattern matching is done using glob.glob()


A running pipeline is represented by a RunningPipeline instance. This object is returned by the and Pipeline.run_interactive() methods. In interactive shell mode the RunningPipeline instance for the last executed pipeline is available in the last_pipeline variable.

A RunningPipeline instance can be used to extract events and statuses of the processes involved in the pipeline:

  • RunningPipeline.processes is a list of RunningItem instances, each representing an external process or a python function.

  • RunningPipeline.failed_processes is a list of RunningItem instances for those processes in the pipeline that have failed (returned a non-zero exit status).

  • RunningPipeline.pipeline is a (deep) copy of the original pipeline object, with additional run status added, e.g. links to processes, exit status etc.

  • RunningPipeline.wait() waits for all processes in the pipeline to terminate.

A RunningItem instance represents an external process or a python function:

  • RunningItem.cmd points to the part of the RunningPipeline.pipeline structure that gave rise to this process.

  • RunningItem.is_running is True if the process is still running.

  • RunningItem.is_failed is True if the process has failed somehow (process with non-zero exit status, function threw an exception).

  • RunningItem.output_content contains a dictionary of the output of any STRING redirection for the process with the file descriptors as keys.

  • RunningProcess.iohandler.last_event contains a dictionary of the members of the last event from the process. The members have the same names and meaning as the members of the signalfd_siginfo struct, see "man signalfd" for details.

  • If psutil is installed, a psutil.Process() instance is available as RunningProcess.details, and most of its members and methods are available directly as members and methods on the RunningProcess instance, e.g. RunningProcess.environ(), RunningProcess.exe().

Job control

A pipeline can be started in the background by appending &True, or &None to do the same and also redirect stdout to /dev/null.

last_pipeline can be used to access the backgrounded pipeline.

A running pipeline can be stopped by hitting CTRL-Z. A stopped pipeline can be restarted in the background with any of


or in the foreground with

fg fg(last_pipeline) last_pipeline.wait()

Error handling

When a pipeline fails, e.g. by one of the involved processes exiting with a non-zero status, RunningPipeline.wait() and Pipeline.run_interactive() will throw a PipelineFailed exception after all processes have exited.

  • PipelineFailed.pipeline holds a reference to the RunningPipeline instance that generated the exception.

If a pipeline is interrupted with CTRL-C, a PipelineInterrupted is raised.

  • PipelineInterrupted.pipeline holds a reference to the RunningPipeline instance.

If you want to catch errors in a script, you can use normal Python exception handling:

except PipelineFailed as e:


Bash provides the command source to run the content of a bash script inside the current shell, effectively letting an external script update the environment variables of the running shell. This functionality is often used for setting up local development environments, like virtualenv.

Pieshell provides a builtin to emulate this functionality, with bash scripts:

>>> bashsource("")

will run in a bash shell followed by declare -x. It parses the output of declare -x and updates exports accordingly. As a special case

>>> bashsource()

will do the same, but without running any script first, esentially just using whatever variables are set up by your .bashrc or .profile.

As a python module

>>> from pieshell import *

All functionality available in the interactive shell is available when using pieshell as an ordinary python module. However, a slighly more cumbersome syntax is required.

In particular, shell commands can not be run just by writing their names. Instead, they have to be accessed as members of the "env" object:

>>> list("-a") | env.grep("-e", "io"))
["", "iterio.pyc"]

Commands are also not run with standard out to the screen when simply printed using the repr() function but must instead be used as iterators as is done above using the list() function.

A pipeline can be run in the background (with input/output to the screen) using

>>>> status = ~("-a") | env.grep("-e", "io"))

or in the forground (waiting until it exists)

>>>> status = +("-a") | env.grep("-e", "io"))

The env object holds the current working directory, which can be changed with


You can also create multiple environments and use them siumultaneously, even within the same pipeline

>>> env2 = env()

Environment variables

Environment variables are available as a dictionary in env._exports.

Argument expansion

Variable expansion is only done on environment variables, as there is no way for pieshell to find out about the right scope to do variable lookups in in any given situation.

Pysh modules

In addition to being able to use pieshell code in ordinary python modules using this slightly more verbose syntax, pieshell supports importing modules named modulename.pysh rather than Pysh modules support the full syntax of the interactive pieshell console. Pysh modules can be imported using the standard import syntax as soon as pieshell itself has been imported, and from the interactive pieshell.


When running pieshell in interactive mode it executes ~/.config/pieshell at startup if it exists. This file can be used to configure the interactive environment the same way ~/.bashrc can be used to configure the bash shell. For example it can be used to load python modules, execute shell pipelines or set environment variables. An example config file is supplied in contrib/cofig.


While pieshell lets you pipe to and from ordinary python functions, they don't offer the same syntax and tab-completion as external commands (e.g. 'myfunction.arg1.arg2(name=value)'), they can't modify the environment or do fancy redirects. Builtin commands provide all of this, at the cost of a slightly clumsier syntax:

class MyMagicBuiltin(pieshell.Builtin):
    """More magic to the people
    name = "magic"

    def _run(self, redirects, sess, indentation = ""):
        # redirects is an instance of pieshell.Redirects
        # sess is an opaque data structure that must be passed to
        # any call to _run() you do yourself from this method (or
        # any function it calls).
        # indentation is a string containing only whitespace, to
        # be prepended to any debug printing lines you print.
        # Returns a list of instances of some pieshell.RunningItem
        # subclass

        self._cmd = self._env.find(
            ".", "-name", "%s.txt" % self._arg[1]) | self._env.tac
        return self._cmd._run(redirects, sess, indentation)

    # Optional for tab completion
    def __dir__(self):
        return ["light", "dark"]

External tools

A short list of tools that might be usefull together with this project:

  • psutil - python api for getting ps / top style information about a process
  • ReRedirect - redirect io for an already running process
  • Reptyr - move a running process to a new controlling terminal
  • Deptyr - forward output for a new process to another controlling terminal

Unit tests

To run the unit- and integration tests

pip install nose2
nose2 -s tests


Pieshell copyright 2016 Egil Möller

Pieshell is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU Lesser General Public License along with this program. If not, see

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

pieshell-0.1.7.tar.gz (37.3 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