From e709b8da0d685d3cc48d430a9761896094f67d72 Mon Sep 17 00:00:00 2001 From: ThibG Date: Mon, 17 Dec 2018 19:19:45 +0100 Subject: Ignore low-confidence CharlockHolmes guesses when parsing link cards (#9510) * Add failing test for windows-1251 link cards * Ignore low-confidence CharlockHolmes guesses Fixes #9466 * Fix no method error when charlock holmes cannot detect charset --- app/services/fetch_link_card_service.rb | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'app/services') diff --git a/app/services/fetch_link_card_service.rb b/app/services/fetch_link_card_service.rb index 38c578de2..7979c312e 100644 --- a/app/services/fetch_link_card_service.rb +++ b/app/services/fetch_link_card_service.rb @@ -137,7 +137,8 @@ class FetchLinkCardService < BaseService detector.strip_tags = true guess = detector.detect(@html, @html_charset) - page = Nokogiri::HTML(@html, nil, guess&.fetch(:encoding, nil)) + encoding = guess&.fetch(:confidence, 0).to_i > 60 ? guess&.fetch(:encoding, nil) : nil + page = Nokogiri::HTML(@html, nil, encoding) player_url = meta_property(page, 'twitter:player') if player_url && !bad_url?(Addressable::URI.parse(player_url)) -- cgit