def main(): sub_index_urls = set(['http://fitnhit.com/education/','http://fitnhit.com/entertainment/','http://fitnhit.com/health/','http://fitnhit.com/news/','http://fitnhit.com/technology/']) site_url = 'http://fitnhit.com/' article_patterns = ['^http://fitnhit.com/.*/\d{5}'] title = 'FitNHit' for sub_index_url in sub_index_urls: siterule = SiteRule.objects.get(sub_index_url=sub_index_url) print siterule.title siterule.title = title siterule.site_url = site_url siterule.article_patterns = article_patterns siterule.save() rules = Site() rules.title = title rules.url = site_url rules.article_patterns = article_patterns rules.save() sites = Site.objects(title='Tribune') if sites: site = sites[0] site.title = 'The Express Tribune' site.save() sites = Site.objects(title='Tribune India') if sites: site = sites[0] site.title = 'The Tribune' site.save() articles = Article.objects(title='Tribune') for article in articles: article.title = 'The Express Tribune' article.save() articles = Article.objects(title='Tribune India') for article in articles: article.title = 'The Tribune' article.save()
def get(self, request, name, sub_name, *args, **kwargs): page = request.GET.get('page') page = int(page) if page else 0 articles = Article.objects(source=sub_name, site_name=name).order_by('id').skip(page * COUNT).limit(COUNT) return Response([ArticleProfileSerializer(article).data for article in articles])