''' www.philstar.com ''' import time from calibre.web.feeds.recipes import BasicNewsRecipe class BanatNews(BasicNewsRecipe): title = 'Banat News' custom_title = "Banat News - " + time.strftime('%d %b %Y %I:%M %p') __author__ = 'jde' __date__ = '31 May 2012' __version__ = '1.0' description = 'Banat News is a daily Cebuano-language newspaper based in Cebu, Philippines - philstar.com is a Philippine news and entertainment portal for the Filipino global community. It is the online presence of the STAR Group of Publications, a leading publisher of newspapers and magazines in the Philippines.' # noqa language = 'ceb' publisher = 'The Philippine STAR' category = 'news, Philippines' tags = 'news, Philippines' cover_url = 'http://www.philstar.com/images/logo_Banat.jpg' masthead_url = 'http://www.philstar.com/images/logo_Banat.jpg' oldest_article = 1.5 # days max_articles_per_feed = 25 simultaneous_downloads = 10 publication_type = 'newspaper' timefmt = ' [%a, %d %b %Y %I:%M %p]' no_stylesheets = True use_embedded_content = False encoding = None recursions = 0 needs_subscription = False remove_javascript = True remove_empty_feeds = True auto_cleanup = False remove_tags = [dict(name='img', attrs={'id': 'Image1'}) # Logo # Section (Headlines, Nation, Metro, ...) # Comments # View Comments # Zoom , dict(name='span', attrs={'id': 'ControlArticle1_LabelHeader'}), dict(name='a', attrs={'id': 'ControlArticle1_FormView1_hlComments'}), dict(name='img', attrs={'src': 'images/post-comments.jpg'}), dict(name='a', attrs={'id': 'ControlArticle1_FormView1_ControlPhotoAndCaption1_hlImageCaption'}) # noqa ] conversion_options = {'title': custom_title, 'comments': description, 'tags': tags, 'language': language, 'publisher': publisher, 'authors': publisher, 'smarten_punctuation': True } feeds = [ ('Balita' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=101'), ('Opinyon' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=102'), ('Kalingawan' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=104'), ('Showbiz' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=62'), ('Palaro' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=103'), ('Imong Kapalaran' , 'http://rss.philstar.com/Rss.aspx?publicationSubCategoryId=105') ] # process the printer friendly version of article def print_version(self, url): return url.replace('/Article', '/ArticlePrinterFriendly') # obtain title from printer friendly version of article; avoiding # add_toc_thumbnail changing title when article has image def populate_article_metadata(self, article, soup, first): article.title = soup.find( 'span', {'id': 'ControlArticle1_FormView1_ArticleHeaderLabel'}).contents[0].strip()