Skip to main content

Internationalization for pytz timezones and territories

Project description

© 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 (want to add yours?)

What is l18n?

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 …

How does it work?

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:

l18n.tz_cities

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'

l18n.tz_fullnames

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'

l18n.territories

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

Lazy mappings special features (v.2016.6.3 onwards)

The fore-mentioned tz_cities, tz_fullnames and territories are not simple dictionaries and provide additional features.

Sorting

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.

Subsets

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())

Selecting the language

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.set_language(None)

Utilities

l18n also exposes a few functions that may be helpful in some cases:

l18n.utils.get_country_tzs(country_code)

returns a list of locations for the given country code, sorted in alphabetical order in the currently selected language

l18n.utils.get_country_code_from_tz(timezone)

returns the country code from a given (untranslated) timezone

Versionning

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.

Want to add a language?

Great idea !! Have a look at CONTRIBUTE.rst.

Roadmap

  • Add supported languages

  • Add currencies and other stuff

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

l18n-2016.6.3.zip (46.8 kB view details)

Uploaded Source

File details

Details for the file l18n-2016.6.3.zip.

File metadata

  • Download URL: l18n-2016.6.3.zip
  • Upload date:
  • Size: 46.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No

File hashes

Hashes for l18n-2016.6.3.zip
Algorithm Hash digest
SHA256 c7e2941ea5729d56392027a0df191f1f3d2bac8c8da45bbd6b19f251a0c77a62
MD5 fdd7deed7d4698eae8b5edae58cae6f3
BLAKE2b-256 e99e9ee8855fc3cddd8410279686d98051ade4e14d5471a44740e3824f48ee34

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page