Metamake is a dead-simple task-based automation tool written in Python.
Metamake is a simple way to define common tasks and execute those tasks by name, similar to Rake. Metamake is not a dependency-tracking build tool like Make, ant, qmake, SCons, Visual Studio, or XCode. Metamake is used with these build tools to orchestrate complex builds that work in a cross-platform fashion.
Using Metamake is as easy as creating a ‘Makefile.py’ in your project directory:
from metamake import task, shell, path @task def build(): """builds the widget""" shell("qmake proj.pro -o Makefile.proj && make -f Makefile.proj") path("src/headers").copytree("dist/include")
On the commandline, you can then type metamake ls to see a listing of all Metamake tasks defined in your Makefile.py, with their docstrings helpfully listed to describe the purpose of that task.
Please file any bugs against the Google Code project @ http://code.google.com/p/metamake/issues/list
To organize some of your tasks, you may want to namespace them:
from metamake import task, namespace @namespace def mynamespace(): @task def foo(): """do something""" pass @task def bar(): """do something else""" pass
Then, when you list these tasks on the commandline, you will see:
mynamespace:foo # do something mynamespace:bar # do something else
To make things easier on newcomers to Metamake, a “Makefile” can be created in your working directory that contains a bootstrapped version of Metamake inside. This allows anybody to build your project without needing Metamake to be installed. With the bootstrapped Makefile, you can type make <args> to achieve the same effect as metamake <args>. Whenever you update Metamake on your system, these bootstrapped Makefiles will be updated automatically next time you execute Metamake for that project. You should commit these Makefiles to your repository so that other people can check out your project and build it without installing Metamake. To create a Makefile bootstrap:
from metamake import task, shell, bootstrap bootstrap("Makefile") # ...
What if you already have a Makefile that you are using for other purposes? That’s easy to solve, simply by specifying a different filename:
from metamake import task, shell, bootstrap bootstrap("Makefile.meta") # ...
Metamake allows you to define commandline flags that can be passed into your build process. These flags will work regardless of whether you use the ‘metamake’ tool or the bootstrapped Makefile:
from metamake import task, Flag Flag("cleanfirst").explain("set this flag to 'true' to do a clean build") if Flag("cleanfirst").given: print "flag was given" if Flag("cleanfirst").value == "true": print "flag value was True"
When you execute metamake ls on the commandline, you will see these flags listed underneath all of the task definitions, with the explanation that you provided as documentation.
From the first example, asking if the value is “true” or “false” is a bit unweildy for a boolean value. This gets even worse for flags that take a restricted set of values, for example a range of integers. The explain method can take a few more parameters to help you out here:
from metamake import task, Flag Flag("debuglevel").explain("set the debug level", convert=int, allow=[1,2,3,4]) if Flag("debuglevel").value == 3: print "flag was converted to the integer 3"
You can also grab the Flag instance instead of always referring to the string name:
from metamake import task, Flag cleanflag = Flag("cleanfirst") debugflag = Flag("debuglevel") if cleanflag.given and debugflag.value == 3: print "we just checked the flag instances"
Jason Orendorff’s excellent path.py library unifies all of the cross-platform Python filesystem manipulations under a single object called path:
from metamake import task, path @task def build(): """builds the widget""" path("dist/include").makedirs() path("src/widget").copytree("dist/include/widget") for header_file in path("src/gadget").listdir("*.h"): header_file.copyfile("dist/include/gadget/%s" % header_file.basename())
Metamake extends Jason’s library by providing console logging for file operations. This makes it easy to see the manipulations that are happening to your filesystem on the commandline.