

skywalker

Here below is a list of tools in skywalker.

checkpoint

checkpoint(key, argvals=False, tempdir=False, refresh=False, verbose=True)

Decorator to checkpoint the output of a function to hdf5 files. Add @skywalker.checkpoint(filename) before a function and the output will be stored to file and computed only if necessary. Deeply inspired by the ediblepickle module. Example:

def test_checkpoint():
    import time
    @skywalker.checkpoint('checkpoint',argvals=True)
    def long_calculation(x,arg=10):
        time.sleep(5)
        return x,arg

    print(long_calculation(2,arg=10))
    print(long_calculation(2,arg=10))
    print(long_calculation(1,arg=20))

Options are:

• argvals. If True all args will be listed in the file name. If False (default), none of them will. If, e.g., [0,3] only the first and the fourth arg/kwarg will be listed.
• refresh. If True, disable checkpointing
• tempdir. If True, store in ./tmp

dontprint

dontprint(self)

A context manager to suppress all printouts, both stdout and stderr. Full credit goes to randlet. Example:

def test_dontprint():
    def message():
        print("Function is printing")

    print("Main is printing")
    message()
    with skywalker.dontprint():
        message()

plot

plot(function)

This is a decorator to handle various matplotlib options, including saving the file to pdf. Just add @skywalker.plot to a function that returns a matplotlib figure object. If a list of figure objects is returned, save a single pdf with many pages. Example:

@skywalker.plot
def test_plot():
    import matplotlib.pyplot as plt
    x=range(100)
    y=range(100)
    fig = plt.figure(figsize=(6,6))
    ax=fig.add_axes([0,0,1,1])
    ax.plot(x,y)
    return fig

processify

processify(func)

Decorator to spawn a new process every time a function is called. Full credit for this incredibly nice piece of code goes to schlamar; for a nice description see here.

def test_processify():

    @skywalker.processify
    def tricky():
        return os.getpid()

    print(os.getpid(), tricky(), tricky())

singleton

singleton(*args, **kwargs)

Decorator to implement the singleton pattern in Python. A single instance of the decorated class can exist at any time. If multiple instances are initiated, identical pointers are returned. Here I use the singleton_decorator module.

def test_singleton():

    @skywalker.singleton
    class simple(object):
        def simple(self,x):
            return x

    s1=simple()
    s2=simple()
    print(s1,s2, s1==s2)

timer

timer(function)

Decorator to print a function execution time to screen. Example:

@skywalker.timer
def test_timer():
    x = range(int(1e7))