diff --git a/recipes/epw.recipe b/recipes/epw.recipe index f57125664f..c3a35f4003 100644 --- a/recipes/epw.recipe +++ b/recipes/epw.recipe @@ -1,33 +1,47 @@ +__license__ = 'GPL v3' +__copyright__ = '2014-2015, Karthik , Krittika Goyal' + from calibre.web.feeds.news import BasicNewsRecipe +class EconomicAndPoliticalWeekly(BasicNewsRecipe): + title = 'Economic and Poitical Weekly' + __author__ = 'Karthik K, Krittika Goyal' + description = 'Economic and Political news from India' + publisher = 'epw.in' + category = 'news, finances, politics, India' + oldest_article = 7 + max_articles_per_feed = 100 + no_stylesheets = True + use_embedded_content = False + simultaneous_downloads = 1 + encoding = 'utf-8' + language = 'en_IN' + publication_type = 'newspaper' + #timefmt = '' + masthead_url = 'http://www.epw.in/system/files/epw_masthead.png' + extra_css = """ + body{font-family: Arial,Helvetica,sans-serif} + """ + conversion_options = {'comment' : description, + 'tags' : category, + 'publisher' : publisher, + 'language' : language + } + remove_tags_before = dict(name='h1', attrs={'class':'print-title'}) + remove_tags_after = dict(name='div', attrs={'class':'print-content'}) + remove_tags = [dict(name='div', attrs={'class':'terms'}), + dict(name='span', attrs={'class':'print-link'})] + feeds = [(u'Editorials', u'http://www.epw.in/taxonomy/term/1/feed'), + (u'Commentary', u'http://www.epw.in/taxonomy/term/3/feed'), + (u'Insight', u'http://www.epw.in/taxonomy/term/14/feed'), + (u'Book Reviews', u'http://www.epw.in/taxonomy/term/4/feed'), + (u'Perspectives', u'http://www.epw.in/taxonomy/term/5/feed'), + (u'Special Articles', u'http://www.epw.in/taxonomy/term/6/feed'), + (u'Discussion', u'http://www.epw.in/taxonomy/term/7/feed'), + (u'Web Exclusives', u'http://www.epw.in/taxonomy/term/11087/feed')] -class Politics(BasicNewsRecipe): - title = u'Economic and Political Weekly' - description = 'Economic and Political news from India' - language = 'en_IN' - __author__ = 'Krittika Goyal' - oldest_article = 15 # days - max_articles_per_feed = 20 - use_embedded_content = False + def print_version(self, url): + return url.replace('http://www.epw.in', 'http://www.epw.in/print') - no_stylesheets = True - auto_cleanup = True - - feeds = [ - -('Editorials', ' http://www.epw.in/feed/editorials.xml'), - -('Commentry', ' http://www.epw.in/feed/commentary.xml'), - -('Insight', ' http://www.epw.in/feed/insight.xml'), - -('Book Reviews', ' http://www.epw.in/feed/book-reviews.xml'), - -('Perspectives', ' http://www.epw.in/feed/perspectives.xml'), - -('Special Articles', ' http://www.epw.in/feed/special-articles.xml'), - -('Discussion', ' http://www.epw.in/feed/discussion.xml'), - -('Web Exclusives', ' http://www.epw.in/feed/web-exclusives.xml'), -] + def postprocess_html(self, soup, first_fetch): + return self.adeify_images(soup)