From c64eb653766f4cdf4e6b2daf2247f809a9caa161 Mon Sep 17 00:00:00 2001 From: Kovid Goyal Date: Wed, 4 May 2022 07:20:58 +0530 Subject: [PATCH] Covers for mit tech review and seminar magazine --- recipes/mit_technology_review.recipe | 6 +++--- recipes/seminar_magazine.recipe | 3 ++- 2 files changed, 5 insertions(+), 4 deletions(-) diff --git a/recipes/mit_technology_review.recipe b/recipes/mit_technology_review.recipe index 50db7b8e08..a5c915d95d 100644 --- a/recipes/mit_technology_review.recipe +++ b/recipes/mit_technology_review.recipe @@ -50,11 +50,11 @@ class MitTechnologyReview(BasicNewsRecipe): def parse_index(self): soup = self.index_to_soup(self.INDEX) # find cover - self.cover_url = absurl(soup.find( + self.cover_url = soup.find( "div", attrs={"class":lambda name: name.startswith("magazineHero__image") if name else False}).find( "img", - src=True - )['src']) + src=True, attrs={"class":lambda x: x.startswith('image__img') if x else False} + )['src'] # parse articles current_articles = [] classNamePrefixes = ["magazineHero__letter--", "teaserItem__title", "teaserItem--aside__title"] diff --git a/recipes/seminar_magazine.recipe b/recipes/seminar_magazine.recipe index a963ddd928..38dcaeb185 100644 --- a/recipes/seminar_magazine.recipe +++ b/recipes/seminar_magazine.recipe @@ -14,11 +14,12 @@ class Seminar(BasicNewsRecipe): remove_javascript = True masthead_url = 'https://www.india-seminar.com/semlogo/semlogo_top_1.jpg' ignore_duplicate_articles = {'url'} + remove_attributes = ['style', 'height', 'width'] def get_cover_url(self): cover_url = None soup = self.index_to_soup('https://www.india-seminar.com/') - citem = soup.find('img', src=lambda x: x and x.endswith('cover.png')) + citem = soup.find('img', src=lambda x: x and 'covers' in x) if citem: cover_url = "https://www.india-seminar.com/" + citem['src'] return cover_url