Re: [Maposmatic-dev] [PATCH ocitysmap] First generic code for Arabic index generation (from Bassem JARKAS)

[Thomas Petazzoni]

Pulled, thanks.

Le Fri,  8 Jan 2010 18:17:35 +0100,
David MENTRE <address@hidden> a écrit :

> This is a first attempt at building an index for Arabic language.
> Prefixes have been provided by Bassem JARKAS. The regexp might need
> some tweaking.
> 
> Do not forget to do "sudo locale-gen --lang ar" on Ubuntu machines.
> ---
>  ocitysmap/i18n.py |   57
> +++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 files
> changed, 57 insertions(+), 0 deletions(-)
> 
> diff --git a/ocitysmap/i18n.py b/ocitysmap/i18n.py
> index 4554cff..a8ed22c 100644
> --- a/ocitysmap/i18n.py
> +++ b/ocitysmap/i18n.py
> @@ -309,6 +309,45 @@ class i18n_pt_br_generic(i18n):
>      def first_letter_equal(self, a, b):
>          return self._upper_unaccent_string(a) ==
> self._upper_unaccent_string(b) 
> +class i18n_ar_generic(i18n):
> +    APPELLATIONS = [ u"شارع", u"طريق", u"زقاق",
> u"نهج", u"جادة",
> +                     u"ممر", u"حارة",
> +                     u"كوبري", u"كوبرى", u"جسر",
> u"مطلع", u"منزل",
> +                     u"Ù…Ù_رق", u"ملÙ_", u"تقاطع",
> +                     u"ساحل",
> +                     u"ميدان", u"ساحة", u"دوار" ]
> +
> +    DETERMINANTS = [ u" ال", u"" ]
> +
> +    SPACE_REDUCE = re.compile(r"\s+")
> +    PREFIX_REGEXP =
> re.compile(r"^(?P<prefix>(%s)(%s)?)\s?(?P<name>.+)" %
> +                                    ("|".join(APPELLATIONS),
> +                                     "|".join(DETERMINANTS)),
> re.IGNORECASE
> +                                                                 |
> re.UNICODE) +
> +    # for IndexPageGenerator._upper_unaccent_string
> +    A_ACCENT = re.compile(ur"[اإآ]", re.IGNORECASE | re.UNICODE)
> +
> +    def __init__(self, language, locale_path):
> +        self.language = str(language)
> +        _install_language(language, locale_path)
> +
> +    def _upper_unaccent_string(self, s):
> +        s = self.A_ACCENT.sub("Ø£", s)
> +        return s.upper()
> +
> +    def language_code(self):
> +        return self.language
> +
> +    def user_readable_street(self, name):
> +        name = name.strip()
> +        name = self.SPACE_REDUCE.sub(" ", name)
> +        name = self.PREFIX_REGEXP.sub(r"\g<name> (\g<prefix>)", name)
> +        return name
> +
> +    def first_letter_equal(self, a, b):
> +        return self._upper_unaccent_string(a) ==
> self._upper_unaccent_string(b) +
>  class i18n_generic(i18n):
>      def __init__(self, language, locale_path):
>          self.language = str(language)
> @@ -359,6 +398,24 @@ language_class_map = {
>      'ca_ES.UTF-8': i18n_ca_generic,
>      'pt_BR.UTF-8': i18n_pt_br_generic,
>      'da_DK.UTF-8': i18n_generic,
> +    'ar_AE.UTF-8': i18n_ar_generic,
> +    'ar_BH.UTF-8': i18n_ar_generic,
> +    'ar_DZ.UTF-8': i18n_ar_generic,
> +    'ar_EG.UTF-8': i18n_ar_generic,
> +    'ar_IN': i18n_ar_generic,
> +    'ar_IQ.UTF-8': i18n_ar_generic,
> +    'ar_JO.UTF-8': i18n_ar_generic,
> +    'ar_KW.UTF-8': i18n_ar_generic,
> +    'ar_LB.UTF-8': i18n_ar_generic,
> +    'ar_LY.UTF-8': i18n_ar_generic,
> +    'ar_MA.UTF-8': i18n_ar_generic,
> +    'ar_OM.UTF-8': i18n_ar_generic,
> +    'ar_QA.UTF-8': i18n_ar_generic,
> +    'ar_SA.UTF-8': i18n_ar_generic,
> +    'ar_SD.UTF-8': i18n_ar_generic,
> +    'ar_SY.UTF-8': i18n_ar_generic,
> +    'ar_TN.UTF-8': i18n_ar_generic,
> +    'ar_YE.UTF-8': i18n_ar_generic,
>  }
>  
>  def install_translation(locale_name, locale_path):


-- 
Thomas Petazzoni                         http://thomas.enix.org
Promouvoir et défendre le Logiciel Libre http://www.april.org
Logiciels Libres à Toulouse              http://www.toulibre.org

signature.asc
Description: PGP signature