From feee3e943dbd6178ff08153d48423ef6ee09bb01 Mon Sep 17 00:00:00 2001 From: Adrian Breiding Date: Mon, 13 Jan 2025 20:04:06 +0100 Subject: [PATCH] remove unused imports --- src/fundus/publishers/es/la_vanguardia.py | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/src/fundus/publishers/es/la_vanguardia.py b/src/fundus/publishers/es/la_vanguardia.py index 3dc29d1b..72f83a7b 100644 --- a/src/fundus/publishers/es/la_vanguardia.py +++ b/src/fundus/publishers/es/la_vanguardia.py @@ -9,8 +9,6 @@ extract_article_body_with_selector, generic_author_parsing, generic_date_parsing, - generic_nodes_to_text, - generic_topic_parsing, image_extraction, ) @@ -18,11 +16,11 @@ class LaVanguardiaParser(ParserProxy): class V1(BaseParser): _paragraph_selector = XPath( - "//div[@class='article-modules']//p[@class='paragraph']|" + "//div[@class='article-modules']//p[@class='paragraph'] | " "//div[@class='widget' and not(@id)]//p[not(@class='creditos')]" ) _subheadline_selector = XPath( - "//div[@class='article-modules']//h3[@class='subtitle']|" + "//div[@class='article-modules']//h3[@class='subtitle'] | " "//div[@class='widget' and not(@id)]//h2|//span[@class='ubicacion']" ) _summary_selector = XPath("//h2[@class='epigraph']|//div[@id='slide-content-1']/p")