Douglas Soares de Andrad

# coding: utf-8

import locale

def isorted(items, key):
    """
    Improved sorted (Locale aware sorted)

    Sorts a list of dictionaries/objects by key/attribute taking locales in
    consideration.

    Example of use:
    >>> data = [{'name': u'John Doe', 'age': 20},
    ...         {'name': u'Jos� da Silva', 'age': 25},
    ...         {'name': u'Almir Sarter', 'age': 35},
    ...         {'name': u'Solim�es', 'age': 40},
    ...         {'name': u'Al�rio', 'age': 10}]

    >>> names = isorted(data, 'name')
    >>> for item in names:
    ...     print repr(item['name'])
    u'Al\\xc3\\xadrio'
    u'Almir Sarter'
    u'John Doe'
    u'Jos\\xc3\\xa9 da Silva'
    u'Solim\\xc3\\xb5es'
    """

    # set the locale to be used if we have a locale set in the computer
    default_locale = locale.getdefaultlocale()

    if None not in default_locale:
        locale.setlocale(locale.LC_ALL, ".".join(default_locale))

    # Next, we sort the dictionary list, using locale-aware comparation and
    # using a lambda expression to get unicode values with a trick:
    # If the element is a dictionary it will use the dictionary get method,
    # if it is a dictionary like object (django querysets) it will use the
    # getattr builtin method to get the attribute (object.attribute).
    # IMPORTANT: It will rise an exception if the key does not exists.
    return sorted(items,
                  cmp=locale.strcoll,
                  key=lambda x: unicode(x.get(key) if isinstance(x,dict) \
                                                   else getattr(x,key)))

if __name__ == "__main__":
    import doctest, sys
    doctest.testmod(sys.modules[__name__])