Skip to main content
Warning: You are using the test version of PyPI. This is a pre-production deployment of Warehouse. Changes made here affect the production instance of TestPyPI (
Help us improve Python packaging - Donate today!

A make-like build utility entirely in python.

Project Description

About pake

pake is a make like python build utility where tasks, dependencies and build commands can be expressed entirely in python, similar to ruby rake.

pake supports automatic file/directory change detection when dealing with task inputs and outputs, and also parallel builds.

pake requires python3.5+

This readme contains only a little information about how to use pake, please checkout the latest documentation at the links above for an expansive overview on how pake works and how pakefiles should be written.


Note: pake is Alpha and likely to change some.

To install the latest release use:

sudo pip3 install python-pake --upgrade

If you want to install the development branch you can use:

sudo pip3 install git+git://

Example project using pake

I am using libasm_io to help test pake and have included a pakefile build along side the makefiles in that project.

Writing Basic Tasks

Here’s a contrived pake demo which demonstrates how tasks are written:

import pake

# Tasks are registered the the pake.Pake object
# returned by pake's initialization call, using the task decorator.

pk = pake.init()

# Try to grab a command line define,
# in particular the value of -D CC=.. if
# it has been passed on the command line.
# CC will default to gcc in this case
# you can also use the syntax: pk["CC"] to
# attempt to get the defines value, if it is not
# defined then it will return None.

CC = pk.get_define("CC", "gcc")

# If you just have a single input/output, there is no
# need to pass a list to the tasks inputs/outputs

@pk.task(i="foo/foo.c", o="foo/foo.o")
def foo(ctx):
    # Execute a program (gcc) and print its stdout/stderr to the tasks output., '-c', ctx.inputs, '-o', ctx.outputs)

# Pake can handle file change detection with multiple inputs
# and outputs. If the amount of inputs is different from
# the amount of outputs, the task is considered to be out
# of date if any input file is newer than any output file.
# When the amount of inputs is equal to the amount of outputs,
# pake will compare each input to its corresponding output
# and collect out of date input/outputs into ctx.outdated_inputs
# and ctx.outdated_outputs respectively.  ctx.outdated_pairs
# can be used to get a generator over (input, output) pairs,
# it is shorthand for zip(ctx.outdated_inputs, ctx.outdated_outputs)
@pk.task(i=pake.glob("bar/*.c"), o=pake.pattern('bar/%.o'))
def bar(ctx):

    # zip together the outdated inputs and outputs, since they
    # correspond to each other, this iterates of a sequence of python
    # tuple objects in the form ("input", "output")

    for i, o in ctx.outdated_pairs:, '-c', i, '-o', o)

# This task depends on the foo and bar tasks, as
# specified with the decorators leading parameters,
# And only outputs "bin/baz" by taking the input "main.c"
# and linking it to the object files produced in the other tasks.

# Documentation strings can be viewed by running 'pake -ti' in
# the directory the pakefile exists in, it will list all documented
# tasks with their python doc strings.
# The pake.FileHelper class can be used to preform basic file
# system operations while printing to the tasks output information
# about what said operation is doing.
@pk.task(foo, bar, o="bin/baz", i="main.c")
def baz(ctx):
    """Use this to build baz"""

    file_helper = pake.FileHelper(ctx)

    # Create a bin directory, this won't complain if it exists already

    # Execute gcc with, using the list argument form
    # instead of a string, this allows easily concatenating all the
    # immediate dependencies outputs to the command line arguments
    # ctx.dependency_outputs contains a list of all outputs that this
    # tasks immediate dependencies produce
    #, '-o', ctx.outputs, ctx.inputs, ctx.dependency_outputs)

def clean(ctx):
    """Clean binaries"""

    file_helper = pake.FileHelper(ctx)

    # Clean up using a the FileHelper object
    # Remove any bin directory, this wont complain if "bin"
    # does not exist.

    # Glob remove object files from the foo and bar directories

# Run pake, the default task that will be executed when
# none are specified will be 'baz'. the tasks parameter
# is optional, if it is not specified then you will have to specify
# which tasks need to be run on the command line., tasks=baz)

Output from command pake:

===== Executing task: "bar"
gcc -c bar/bar.c -o bar/bar.o
===== Executing task: "foo"
gcc -c foo/foo.c -o foo/foo.o
===== Executing task: "baz"
Created Directory(s): "bin"
gcc -o bin/baz main.c foo/foo.o bar/bar.o

Output from command pake clean:

===== Executing task: "clean"
Removed Directory(s): "bin"
Glob Removed Files: "foo/*.o"
Glob Removed Files: "bar/*.o"

Parallelism Inside Tasks

Work can be submitted to the threadpool pake is running it’s tasks on in order to achieve a predictable level of parallelism that is limited by the –jobs command line argument or the jobs parameter of

import pake


@pk.task(i=pake.glob('src/*.c'), o=pake.pattern('obj/%.o'))
def build_c(ctx):

    # Start multitasking

    with ctx.multitask() as mt:
        for i, o in ctx.outdated_pairs:
            # Submit a work function with arguments to the threadpool

            mt.submit(, ['gcc', '-c', i, '-o', o])

@pk.task(build_c, i=pake.glob('obj/*.o'), o='main')
def build(ctx):'gcc', ctx.inputs, '-o', ctx.outputs), tasks=build)

Running Sub Pakefiles

Pake is able to run itself through the use of pake.TaskContext.subpake or even pake.subpake.

pake.TaskContext.subpake is preferred because it handles writing program output to the tasks output queue in a synchronized manner when multiple jobs are running.

A pake.TaskContext is passed into the single argument of each task function.

Defines can be exported to pakefiles ran with the subpake functions using pake.export.

pake.subpake and pake.TaskContext.subpake use the –stdin-defines option of pake to pass exported define values into the new process instance, which means you can overwrite your exported define values with -D/–define in the subpake command arguments if you need to.

Export / Subpake Example:

import pake

pk = pake.init()

# Try to get the CC define from the command line,
# default to 'gcc'.

CC = pk.get_define('CC', 'gcc')

# Export the CC variable's value to all invocations
# of pake.subpake or ctx.subpake as a define that can be
# retrieved with pk.get_define()
pake.export('CC', CC)

# You can also export lists, dictionaries sets and tuples,
# as long as they only contain literal values.
# Literal values being: strings, integers, floats; and
# other lists, dicts, sets and tuples (if they only contain literals)

pake.export('CC_FLAGS', ['-Wextra', '-Wall'])

# Nesting works with composite literals,
# as long as everything is a pure literal or something
# that str()'s into a literal.

                {'bad' :
                     ['person', ['if', {'you', 'do'}, ('this',) ]]

# This export will be overrode in the next call
pake.export('OVERRIDE_ME', False)

# Execute outside of a task, by default the stdout/stderr
# of the subscript goes to this scripts stdout.  The file
# object to which stdout gets written to can be specified
# with pake.subpake(..., stdout=(file))

# This command also demonstrates that you can override
# your exports using the -D/--define option

pake.subpake('sometasks/', 'dotasks', '-D', 'OVERRIDE_ME=True')

# This task does not depend on anything or have any inputs/outputs
# it will basically only run if you explicitly specify it as a default
# task in, or specify it on the command line

def my_phony_task(ctx):
    # Arguments are passed in a variadic parameter...

    # Specify that the "foo" task is to be ran.
    # The scripts output is written to this tasks output queue

    ctx.subpake('library/', 'foo')

# Run this pake script, with a default task of 'my_phony_task', tasks=my_phony_task)

Output from the example above:

*** enter subpake[1]:
pake[1]: Entering Directory "(REST OF PATH...)/paketest/sometasks"
===== Executing Task: "dotasks"
Do Tasks
pake[1]: Exiting Directory "(REST OF PATH...)/paketest/sometasks"
*** exit subpake[1]:
===== Executing Task: "my_phony_task"
*** enter subpake[1]:
pake[1]: Entering Directory "(REST OF PATH...)/paketest/library"
===== Executing Task: "foo"
pake[1]: Exiting Directory "(REST OF PATH...)/paketest/library"
*** exit subpake[1]:

Running pake

cd your_pakefile_directory

# Run pake with up to 10 tasks running in parallel

pake -j 10

pake will look for “” or “pakefile” in the current directory and run it.

Or you can specify one or more files to run with -f/–file. The switch does not have multiple arguments, but it can be used more than once to specify multiple files.

For example:

pake -f foo

pake -f -f foo

You can also specify multiple tasks, but do not rely on unrelated tasks being executed in any specific order because they won’t be. If there is a specific order you need your tasks to execute in, the one that comes first should be declared a dependency of the one that comes second, then the second task should be specified to run.

When running parallel builds, leaf dependencies will start executing pretty much simultaneously, and non related tasks that have a dependency chain may execute in parallel.

pake task unrelated_task order_independent_phony

Command Line Options

usage: pake [-h] [-v] [-D DEFINE] [-j JOBS] [--stdin-defines] [-n]
            [-C DIRECTORY] [-t] [-ti] [-f FILE]
            [tasks [tasks ...]]

positional arguments:
  tasks                 Build tasks.

optional arguments:
  -h, --help            show this help message and exit
  -v, --version         show program's version number and exit
  -D DEFINE, --define DEFINE
                        Add defined value.
  -j JOBS, --jobs JOBS  Max number of parallel jobs. Using this option enables
                        unrelated tasks to run in parallel with a max of N
                        tasks running at a time.
  --stdin-defines       Read defines from a Python Dictionary piped into
                        stdin. Defines read with this option can be
                        overwritten by defines specified on the command line
                        with -D/--define.
  -n, --dry-run         Use to preform a dry run, lists all tasks that will be
                        executed in the next actual invocation.
                        Change directory before executing.
  -t, --show-tasks      List all task names.
  -ti, --show-task-info
                        List all tasks along side their doc string. Only tasks
                        with doc strings present will be shown.
  -f FILE, --file FILE  Pakefile path(s). This switch can be used more than
                        once, all specified pakefiles will be executed in
                        order with the current directory as the working
                        directory (unless -C is specified).
Release History

Release History

This version
History Node

History Node

Download Files

Download Files

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

File Name & Checksum SHA256 Checksum Help Version File Type Upload Date
python-pake- (51.5 kB) Copy SHA256 Checksum SHA256 Source Jun 16, 2017

Supported By

WebFaction WebFaction Technical Writing Elastic Elastic Search Pingdom Pingdom Monitoring Dyn Dyn DNS Sentry Sentry Error Logging CloudAMQP CloudAMQP RabbitMQ Heroku Heroku PaaS Kabu Creative Kabu Creative UX & Design Fastly Fastly CDN DigiCert DigiCert EV Certificate Rackspace Rackspace Cloud Servers DreamHost DreamHost Log Hosting