Update internal link extraction to filter by class 'interlink-script'

2025-07-14 12:46:34 +02:00
parent 34093ebeca
commit 79a41a0637
1 changed files with 3 additions and 1 deletions
--- a/action_script/internal_link.py
+++ b/action_script/internal_link.py
@@ -49,7 +49,9 @@ def get_internal_links_pointing_to_pages(sitemap_url):
            page_response.raise_for_status()
            page_soup = BeautifulSoup(page_response.content, "html.parser")

-            for link in page_soup.find_all("a", href=True):
+            for link in page_soup.find_all(
+                "a", {"class": "interlink-script"}, href=True
+            ):
                href = link.get("href")
                absolute_href = urljoin(page_url, href)
                parsed_absolute_href = urlparse(absolute_href)