From 6c27c5385fad149780a65dd52045c2671c12577d Mon Sep 17 00:00:00 2001 From: Jackson Harper Date: Thu, 13 Apr 2023 12:10:07 +0800 Subject: [PATCH] Remove footnote links in wikipedia articles --- packages/content-handler/src/websites/wikipedia-handler.ts | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/packages/content-handler/src/websites/wikipedia-handler.ts b/packages/content-handler/src/websites/wikipedia-handler.ts index 37005a18a..db872ba27 100644 --- a/packages/content-handler/src/websites/wikipedia-handler.ts +++ b/packages/content-handler/src/websites/wikipedia-handler.ts @@ -13,6 +13,10 @@ export class WikipediaHandler extends ContentHandler { async preParse(url: string, dom: Document): Promise { // This removes the [edit] anchors from wikipedia pages dom.querySelectorAll('.mw-editsection').forEach((e) => e.remove()) + + // Remove footnotes + dom.querySelectorAll('sup[class="reference"]').forEach((e) => e.remove()) + // this removes the sidebar dom.querySelector('.infobox')?.remove() return Promise.resolve(dom)