1 from django.core import urlresolvers, paginator

     2 import urllib

     3 

     4 PING_URL = "http://www.google.com/webmasters/tools/ping"

     5 

     6 class SitemapNotFound(Exception):

     7     pass

     8 

     9 def ping_google(sitemap_url=None, ping_url=PING_URL):

    10     """

    11     Alerts Google that the sitemap for the current site has been updated.

    12     If sitemap_url is provided, it should be an absolute path to the sitemap

    13     for this site -- e.g., '/sitemap.xml'. If sitemap_url is not provided, this

    14     function will attempt to deduce it by using urlresolvers.reverse().

    15     """

    16     if sitemap_url is None:

    17         try:

    18             # First, try to get the "index" sitemap URL.

    19             sitemap_url = urlresolvers.reverse('django.contrib.sitemaps.views.index')

    20         except urlresolvers.NoReverseMatch:

    21             try:

    22                 # Next, try for the "global" sitemap URL.

    23                 sitemap_url = urlresolvers.reverse('django.contrib.sitemaps.views.sitemap')

    24             except urlresolvers.NoReverseMatch:

    25                 pass

    26 

    27     if sitemap_url is None:

    28         raise SitemapNotFound("You didn't provide a sitemap_url, and the sitemap URL couldn't be auto-detected.")

    29 

    30     from django.contrib.sites.models import Site

    31     current_site = Site.objects.get_current()

    32     url = "http://%s%s" % (current_site.domain, sitemap_url)

    33     params = urllib.urlencode({'sitemap':url})

    34     urllib.urlopen("%s?%s" % (ping_url, params))

    35 

    36 class Sitemap(object):

    37     # This limit is defined by Google. See the index documentation at

    38     # http://sitemaps.org/protocol.php#index.

    39     limit = 50000

    40 

    41     def __get(self, name, obj, default=None):

    42         try:

    43             attr = getattr(self, name)

    44         except AttributeError:

    45             return default

    46         if callable(attr):

    47             return attr(obj)

    48         return attr

    49 

    50     def items(self):

    51         return []

    52 

    53     def location(self, obj):

    54         return obj.get_absolute_url()

    55 

    56     def _get_paginator(self):

    57         if not hasattr(self, "_paginator"):

    58             self._paginator = paginator.Paginator(self.items(), self.limit)

    59         return self._paginator

    60     paginator = property(_get_paginator)

    61 

    62     def get_urls(self, page=1):

    63         from django.contrib.sites.models import Site

    64         current_site = Site.objects.get_current()

    65         urls = []

    66         for item in self.paginator.page(page).object_list:

    67             loc = "http://%s%s" % (current_site.domain, self.__get('location', item))

    68             url_info = {

    69                 'location':   loc,

    70                 'lastmod':    self.__get('lastmod', item, None),

    71                 'changefreq': self.__get('changefreq', item, None),

    72                 'priority':   self.__get('priority', item, None)

    73             }

    74             urls.append(url_info)

    75         return urls

    76 

    77 class FlatPageSitemap(Sitemap):

    78     def items(self):

    79         from django.contrib.sites.models import Site

    80         current_site = Site.objects.get_current()

    81         return current_site.flatpage_set.all()

    82 

    83 class GenericSitemap(Sitemap):

    84     priority = None

    85     changefreq = None

    86 

    87     def __init__(self, info_dict, priority=None, changefreq=None):

    88         self.queryset = info_dict['queryset']

    89         self.date_field = info_dict.get('date_field', None)

    90         self.priority = priority

    91         self.changefreq = changefreq

    92 

    93     def items(self):

    94         # Make sure to return a clone; we don't want premature evaluation.

    95         return self.queryset.filter()

    96 

    97     def lastmod(self, item):

    98         if self.date_field is not None:

    99             return getattr(item, self.date_field)

   100         return None