i18n Landscape

The i18n landscape in Python, as in many other languages, is centered around gettext. Various i18n libraries in the Python universe have to do with the static extraction of python strings from instrumented code to build templates for catalog templates.

In order to understand this, it is required to have a basic understanding of the various moving parts that make up the typical i18n implementation. What follows is a highly simplified description, glossing over a lot of details. This description falls squarely in the Lies to Children category.

• At the core of i18n is the translation catalog files. These are the .po files that one often sees in a locale/<LANG>/LC_MESSAGES or similar folder. This file format and folder structure carries over from the legacy gettext implementation that is used by more or less every language and across most platforms today.

• Each catalog file is associated with a specific locale and contains a set of msgids, each associated with a single msgstring, i.e., the string which represents the message in the language the catalog provides translations for.

• These catalogs are built from templates, which are .pot files. These files contain the same set of msgid, but with a blank msgstring.

• Various i18n tools, including python libraries like babel, focus on building these templates. They extract translatable strings from python code, usually by some form of static analysis, and assemble them into a template which can then be manually translated into each locale by i18n specific tools.

For someone just starting out on an i18n journey, the approach can seem unnecessarily dense and at times non-pythonic. This package was written to push all of the complexity into a single pythonic entity. This is a very subjective goal.