diff --git a/recipes/ap.recipe b/recipes/ap.recipe index b3f809fe33..aa7003e020 100644 --- a/recipes/ap.recipe +++ b/recipes/ap.recipe @@ -28,14 +28,14 @@ class AssociatedPress(BasicNewsRecipe): classes('StoryPage-lede-content Page-lead Page-byline-info RichTextStoryBody'), ] remove_tags = [ - classes('displayNone Advertisement HTMLModuleEnhancement videoSlide'), + classes('displayNone Advertisement HTMLModuleEnhancement AudioEnhancement'), dict( name=[ 'source', 'button', 'svg', 'template', - 'bsp-jw-player', + 'video', 'astro-island', 'iframe', 'document', @@ -64,7 +64,10 @@ class AssociatedPress(BasicNewsRecipe): return [('Articles', feeds)] def preprocess_html(self, soup): - for st in soup.findAll(**classes('CarouselSlide-infoDescription')): + for v in soup.findAll('bsp-jw-player', attrs={'poster': True}): + v.name = 'img' + v.attrs = {'src': v.get('poster', '')} + for st in soup.findAll(**classes('CarouselSlide-infoDescription Figure-caption')): if p := st.find('p'): p.name = 'span' for h in soup.findAll(['h2', 'h3']):