resolves an issue with tags used count that can get out of control if called from...

[osqa.git] / forum / sitemap.py

diff --git a/forum/sitemap.py b/forum/sitemap.py
index b53cb160728351124d3f9e7af4425396e2b2d9ee..0bf8538ce49b41c6f840d2325d75924f0c9fc0eb 100644 (file)
--- a/forum/sitemap.py
+++ b/forum/sitemap.py
@@ -1,9 +1,14 @@
+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
+from django.http import HttpResponse, Http404
 from django.template import loader
 from django.core import urlresolvers
+from django.utils.encoding import smart_str
+from django.core.paginator import EmptyPage, PageNotAnInteger
 
 def index(request, sitemaps):
     sites = []
@@ -12,19 +17,68 @@ def index(request, sitemaps):
             pages = site().paginator.num_pages
         else:
             pages = site.paginator.num_pages
-        sitemap_url = urlresolvers.reverse('django.contrib.sitemaps.views.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))
+        sitemap_url = urlresolvers.reverse('sitemap_section_index', 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')
 
+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', 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, mimetype='application/xml')
+
+def sitemap(request, sitemaps, section=None, page=1):
+    maps, urls = [], []
+    if section is not None:
+        if section not in sitemaps:
+            raise Http404("No sitemap available for section: %r" % section)
+        maps.append(sitemaps[section])
+    else:
+        maps = sitemaps.values()
+    
+    for site in maps:
+        try:
+            if callable(site):
+                urls.extend(site().get_urls(page=page))
+            else:
+                urls.extend(site.get_urls(page=page))
+        except EmptyPage:
+            raise Http404("Page %s empty" % page)
+        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')
+
 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
@@ -52,4 +106,4 @@ class OsqaSitemap(Sitemap):
                 'priority':   self.__get('priority', item, None)
             }
             urls.append(url_info)
-        return urls    
\ No newline at end of file
+        return urls