X-Git-Url: https://git.openstreetmap.org./osqa.git/blobdiff_plain/3ababc1ec667dd7bb0148d97770816462228a3fa..4f54d1c4dfe4b2b6c62601212c27c175fde3d5b0:/forum/sitemap.py diff --git a/forum/sitemap.py b/forum/sitemap.py index edb3220..0ba045b 100644 --- a/forum/sitemap.py +++ b/forum/sitemap.py @@ -1,28 +1,54 @@ +import re + from django.contrib.sitemaps import Sitemap from forum.models import Question +from forum.settings import QUESTIONS_SITEMAP_LIMIT, QUESTIONS_SITEMAP_CHANGEFREQ from django.conf import settings from django.http import HttpResponse, Http404 from django.template import loader from django.core import urlresolvers +from django.core.urlresolvers import get_script_prefix from django.utils.encoding import smart_str from django.core.paginator import EmptyPage, PageNotAnInteger def index(request, sitemaps): sites = [] - for section, site in sitemaps.items(): - if callable(site): - pages = site().paginator.num_pages - else: - pages = site.paginator.num_pages - sitemap_url = urlresolvers.reverse('forum.sitemap.sitemap', kwargs={'section': section}) - sites.append('%s%s' % (settings.APP_URL, sitemap_url)) - if pages > 1: - for page in range(2, pages+1): - sites.append('%s%s?p=%s' % (settings.APP_URL, sitemap_url, page)) + for section in sitemaps.keys(): + sitemap_url = urlresolvers.reverse('sitemap_section_index', prefix='/', kwargs={'section': section}) + + # Replace double forward slashes with single ones + final_url = '%s%s' % (settings.APP_URL, sitemap_url) + final_url = re.sub("/+", "/", final_url) + final_url = final_url.replace('http:/', 'http://') + final_url = final_url.replace('https:/', 'https://') + + sites.append(final_url) + xml = loader.render_to_string('sitemap_index.xml', {'sitemaps': sites}) - return HttpResponse(xml, mimetype='application/xml') + return HttpResponse(xml, content_type='application/xml') + +def sitemap_section_index(request, section, sitemaps): + try: + sitemap = sitemaps[section]() + except KeyError: + raise Http404("Sitemap doesn't exist") + + paginator = sitemap.paginator + + locations = [] + + for page in paginator.page_range: + location = urlresolvers.reverse('sitemap_section_page', prefix='/', kwargs={ 'page' : page, 'section' : section }) + location = '%s%s' % (settings.APP_URL, location) + location = re.sub("/+", "/", location) + location = location.replace('http:/', 'http://') + location = location.replace('https:/', 'https://') + locations.append(location) + + xml = loader.render_to_string('sitemap_section_index.xml', { 'locations' : locations, }) + return HttpResponse(xml, content_type='application/xml') -def sitemap(request, sitemaps, section=None): +def sitemap(request, sitemaps, section=None, page=1): maps, urls = [], [] if section is not None: if section not in sitemaps: @@ -30,7 +56,6 @@ def sitemap(request, sitemaps, section=None): maps.append(sitemaps[section]) else: maps = sitemaps.values() - page = request.GET.get("p", 1) for site in maps: try: @@ -43,13 +68,14 @@ def sitemap(request, sitemaps, section=None): except PageNotAnInteger: raise Http404("No page '%s'" % page) xml = smart_str(loader.render_to_string('sitemap.xml', {'urlset': urls})) - return HttpResponse(xml, mimetype='application/xml') + return HttpResponse(xml, content_type='application/xml') class OsqaSitemap(Sitemap): - changefreq = 'daily' + limit = QUESTIONS_SITEMAP_LIMIT + changefreq = QUESTIONS_SITEMAP_CHANGEFREQ priority = 0.5 def items(self): - return Question.objects.filter_state(deleted=False) + return Question.objects.filter_state(deleted=False).order_by('id') def lastmod(self, obj): return obj.last_activity_at @@ -69,7 +95,9 @@ class OsqaSitemap(Sitemap): def get_urls(self, page=1): urls = [] for item in self.paginator.page(page).object_list: - loc = "%s%s" % (settings.APP_URL, self.__get('location', item)) + root_relative_url = self.__get('location', item) + relative_url = root_relative_url[len(get_script_prefix()):] + loc = "%s/%s" % (settings.APP_URL, relative_url) url_info = { 'location': loc, 'lastmod': self.__get('lastmod', item, None), @@ -77,4 +105,4 @@ class OsqaSitemap(Sitemap): 'priority': self.__get('priority', item, None) } urls.append(url_info) - return urls \ No newline at end of file + return urls