Internationalization for pytz timezones and territories
© 2014-2016 Thomas Khyn
Locale internationalization package. Translations for places, timezones …
Tested with the latest minor versions of Python 2 and 3.
Supported languages: English, French, German, Czech, Chinese (want to add yours?)
As you may have noticed, l18n is a contraction of i18n and l10n, namely ‘internationalisation’ and ‘localization’. It basically provides lazy translations for names used for localization purposes (e.g. places and timezones).
I started writing l18n when I was looking for translations for the pytz library. Indeed, on a multi-lingual site where users can select the timezone they are in, it’s much better if they can select in their language, as in some cases, the differences with the english name can be significant, hence the place to look for it when it’s sorted in alphabetical order.
And as I am lazy, I thought of a way to - almost - automatically fetch the translations from the CLDR (Unicode’s Common Locale Data Repository) database.
Integrating function to link timezone to country codes, there was no reason not to try and provide translations also for the latter. In the near future, I - or contributors - may also add currencies or measurement units fetched from the CLDR database …
To use l18n, you first need to install it. It works well with pip:
pip install l18n
Then, in your code:
>>> import l18n
l18n exposes several read-only dictionary-like objects:
is a mapping between all the timezones listed in pytz.all_timezones and human-friendly lazy versions of the translated name of the city in the current language (see Selecting the language below). For example, if the language is English:>>> l18n.tz_cities['Pacific/Easter'] L18NLazyString <Easter Island> >>> str(l18n.tz_cities['Pacific/Easter']) 'Easter Island'
In French, it would give:>>> str(l18n.tz_cities['Pacific/Easter']) 'Île de Pâques'
is a mapping between all the timezones listed in pytz.all_timezones and lazy versions of the timezones’ full names in the current language. For example:>>> str(l18n.tz_fullnames['Pacific/Easter']) 'Pacific/Easter Island' # or 'Pacifique/Île de Pâques' in French
It is interesting to note that for 3-components timezone names where the local state or territory appears in the city name, l18n cleverly strips this information so that it is not repeated:>>> str(l18n.tz_fullnames['America/North_Dakota/New_Salem']) 'America/North Dakota/New Salem'
indeed:>>> str(l18n.tz_cities['America/North_Dakota/New_salem']) 'New Salem, North Dakota'
is a mapping between the territory codes as defined in the CLDR and their localized names, lazily defined. For example:>>> str(l18n.territories['CZ']) 'Czech Republic' # or 'République Tchèque' in French
The values are translated each time they are evaluated, there is no caching. This means that the same L18NLazyString / L18NLazyStringsList instance can be used and produce 2 different outputs if you change the language between the evaluations.
The values in the above mentionned dictionaries can be overriden by your own translations. The dictionaries are not read-only and values can be added or removed at your convenience.
The fore-mentioned tz_cities, tz_fullnames and territories are not simple dictionaries and provide additional features.
When iterating over an L18NMap, the items, keys or values are yielded in alphabetical order in the currently selected language. For performance, the results are cached by language, so the sort is only performed once per language. Note that the values are still lazy objects that are evaluated only when rendered into a string.
It is possible to generate a new L18NMap from an existing one by using the subset method and passing an iterable of keys that need to be kept in the new mapping. Any cached sort is also used to generate the new cache, so that there is nothing to re-calculate in the new subset.
For example, one can generate a map of translations for pytz.common_timezones:
>>> common_cities = l18n.tz_cities.subset(pytz.common_timezones.keys())
By default, when importing l18n, the current default locale is used (via locale.getdefaultlocale()). If it is not the one you want or if you need to change it, it is rather easy:
>>> l18n.set_language('en') >>> str(l18n.tz_cities['Pacific/Easter']) 'Easter Island' >>> l18n.set_language('fr') >>> str(l18n.tz_cities['Pacific/Easter']) 'Île de Pâques'
And in case you want to disable translation and use raw default strings:
l18n also exposes a few functions that may be helpful in some cases:
returns a list of locations for the given country code, sorted in alphabetical order in the currently selected language
returns the country code from a given (untranslated) timezone
l18n’s main version number matches pytz’s version number. l18n 2014.10.X will be fully compatible with pytz 2014.10 whatever the value of X. Indeed, the primary aim is to keep l18n’s translation files consistent with pytz’s timezone names.
Before l18n 2016.6, the pytz version was pinned against the l18n version. Now, l18n YEAR.MONTH can now be used with any subsequent pytz version. However, note that there may be missing translations if the 2 versions are too different from each other. In that case, open an issue to request a new version of l18n to be published.
Great idea !! Have a look at CONTRIBUTE.rst.