pyLibrary 0.1.14070

Library of Wonderful Things

A library of wonderful things!

Motivation

Python is a Little Crufty

Python is awesome now, but it was originally a procedural language invented before functional semantics, before OO, and even before the discovery of vowels. As a consequence there are many procedures that alter their own parameters, or have disemvoweled names. This library puts a facade over these relics of the past and uses convention to name methods.

Advanced List Comprehensions are Awesome

My career has been permeated with the need to write ETL scripts. I have preferred SQL for it’s succinctness when dealing with data transformations, and it makes Python’s list comprehensions look limited in comparison. I have added a few more relational operators, and provided procedures to handle common complex list comprehensions.

Nones are Important

ETL inherently deals with diverse data and missing data. This leaves many opportunities for None to take the place of that missing data.

Heavy use of list comprehensions requires a significantly different strategy when dealing with Nulls/Nones. The Pythonic strategy of raising exceptions on invalid access is unacceptable inside the context of list operations because we want to process all good data and not exit early. Furthermore, even the data with Nones can go through partial processing.

A significant portion of this library implements familiar set/list operations while dealing with Nones gracefully. To do this, None is interpreted as not-applicable rather than unknown/missing. The former allows operations to assume the None can be ignored in the context of the operation, and this assumption turns out to be mostly correct in the context of ETL. The latter interpretation appears more in statistics. A good example is util.collections.AND:

AND(True, None, True) == True  # Assuming None means "not applicable"
AND(True, None, True) == None  # Assuming None means "unknown"

JSON is Everywhere

This library is also designed to query into hierarchical data, like JSON. Hierarchical queries are made easy by manipulating references to paths into the JSON trees, and further allowing set comprehensions to use those paths to deconstruct, or reconstruct trees succinctly. The dot “.“ operator is used to specify paths, and is so prevalent that the literal dot must be escaped (like so: “\.“). JSON, and the paths into it, are an excellent source of Nones.

Structured Logging integrated with Chained Exceptions is Good

Exception handling and logging are undeniably linked. There are many instances where exceptions are raised and must be logged, except when they are appropriately handled. The greatness of exception handling semantics comes from decoupling the cause from the solution, but this is at odds with clean logging - which couples raising and handling to make appropriate decisions about what to ultimately emit to the log. For this reason, the logging module is responsible for collecting the trace and context, raising the exception, and then deducing if there is something that will handle it (so it can be ignored), or if it really must be logged.

This library also expects all log messages and exception messages to have named parameters so they can be stored in easy-to-digest JSON, which can be processed by downstream tools.

Windows 7 Install Instructions for Python

Python was really made for Linux, and installation will be easier there. Technically, Python works on Windows too, but there are a few gotchas you can avoid by following these instructions.

• Download Python 2.7 (32bit ONLY!!! Many native libs are 32 bit)

• Install Python at c:\Python27 (The space in the “Program Files” may screw up installs of native libs)

• Add to you path: c:\Python27;c:\Python27\scripts;

• Download http://python-distribute.org/distribute_setup.py

  CALL python distribute_setup.py
  CALL easy_install pip
  CALL easy_install virtualenv

• Many “Python Powered” native installs require a pointer to the python installation, but they have no idea where to look in 64bit windows. You must alter the registry (http://stackoverflow.com/questions/3652625/installing-setuptools-on-64-bit-windows):

  SET HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Python\PythonCore\2.7\InstallPath = "C:\Python27"

Installing pyLibrary

Once Python is installed, other Python packages are are much easier.

• Install from PyPi:

  pip install pyLibrary

Installing for Development

• Download from Github:

  git clone https://github.com/klahnakoski/pyLibrary.git

• Install requirements:

  python setup.py develop

Optional

pyLibrary does not use these, but maybe you will find them useful:

• Install MySqldb from http://sourceforge.net/projects/mysql-python/files/mysql-python/

• Install ujson from http://www.lfd.uci.edu/~gohlke/pythonlibs/#ujson