Little known and/or useful Python features for fun & profit by Larry Weya - Co-founder and Software Engineer at Ona Labs Kenya - Video Game Development Hobbyist

Why I love Python Productivity, think less do more. How? ● REPL (Read-Eval-Print loop)/Interactive shell ● Rapid prototyping ● Mature package management ● Inbuilt testing framework ● Pretty code == readable code == maintainable code

1. (virtualenv) Isolated working copy of Python. Why? Run different python installations on the same system, each with its own set of packages. Invaluable in any python dev machine. virtualenv - http://www.virtualenv.org/en/latest/ virtualenvwrapper - http://virtualenvwrapper.readthedocs.org/en/latest/index. html Make using virtualenv even easier. ● Central location of all virtual environments ● switch virtual environments with the `workon` command, includes TAB completion

2. setuptools Packaging for Python. Build, distribute and install python packages. ● Create distributable python .egg ● Create distributable Python sources ● Upload packages to PyPi ● Automatically find and download your packages dependencies on installation (by yourself or 3rd parties) ● … http://pythonhosted.org/setuptools/setuptools.html Python package skeleton: $> easy_install PasteScript $> paster create basic_package

setuptools Development Mode Make the python source directories available to a python installation. Changes made to your sources are immediately available in the Python installation. ProjectB (depends on ProjectA) (virtualenv) for Project B ProjectA

setuptools Wrapper script generation Generate command line tools from you python functions. mymodule/utils.py import sys def setup_database(argv=sys.argv): create_and_populate_database() return 0 def upgrade_database(argv=sys.argv): run_db_migrations() return 0 setup.py from setuptools import setup, find_packages setup( … entry_points = { 'console_scripts': [ 'setup_db = mymodule.utils: setup_database' ] } … # creates /bin/setup_db $> setup_db

IPython “Command shell for interactive computing” - Wikipedia ● TAB completion (on imports, function names …) ● Command history across sessions ● Object introspection - func_name? ● Magic methods ○ %timeit ○ %run ○ %cd ○ … ○ %debug - after an exception ○ %pdb - toggle dropping into debugger on exceptions

try..except..else Language Features try: o = get_db_object(pk) parent_object = get_object(o.parent_id) except ObjectNotFound: log(“Object with pk: {} not found”.format(pk)) try: db_object = get_db_object(pk) except ObjectNotFound: log(“Object with pk: {} not found”.format(pk)) else: parent_object = get_object(o.parent_id) ● Don’t catch the wrong exceptions ● Readability - “do this only when the exception is not raised” vs

the ternary operator Language Features Javascript: var val = is_valid?1:0; Python val = 1 if is_valid else 0

Language Features List Comprehensions kids = [{‘name’: ‘Billy’, ‘age’: 2}, {‘name’: ‘Bob’, ‘age’: 5}] ages = [] for kid in kids: ages.append(kid[‘age’]) do_something(ages) do_something([kid['age'] for kid in kids]) ● Clear and concise syntax when working with lists (and other iterables) ● Readability? Not when it gets too complex

Language Features Generators x = [n for n in range(1000000000)] x = (n for n in range(1000000000)) Pros ● No extra storage is created ● You can then iterate over only part of the list without the extra memory overhead ● Cons: ● You can only iterate over the result once vs

Language Features List Stepping values = range(10) third = [] for idx, val in enumerate(values): if val % 3 == 0 third.append(val) values = range(10) # get every 3rd item result = values[::3] ● READABILITY vs

Language Features Debugging Regular Expressions (Experimental) import re re.compile(r"^hell(o)", re.DEBUG) at at_beginning literal 104 literal 101 literal 108 literal 108 subpattern 1 literal 111

Language Features mymodule/utils.py def failing_function(): import ipdb; ipdb.set_trace() result = call_some_other_function() # s to step into result += 1 # n to step over return result Debugging pdb/ipdb Python Debugger Cheatsheet - http://nblock.org/2011/11/15/pdb-cheatsheet/

Language Features >>> func_is_even = lambda n: n % 2 == 0 >>> func_is_even(3) False >>> filter(func_is_even, range(10)) [0, 2, 4, 6, 8] Lambdas/Anonymous Functions

Additional References ● Ona Data Platform - https://github.com/onaio/onadata, http://ona.io/ ● PyCharm IDE - http://www.jetbrains.com/pycharm/ ● Blender 3D - http://www.blender.org/