Use Html2Text to generate plaintext for keyword mutes. #236.

This allows us to match URLs inside link hrefs.
This commit is contained in:
David Yip 2018-02-10 10:32:14 -06:00
parent 9105b0c954
commit f1f67c46c5
No known key found for this signature in database
GPG key ID: 7DA0036508FCC0CC
2 changed files with 9 additions and 6 deletions

View file

@ -1,16 +1,12 @@
require 'htmlentities' require 'html2text'
class Glitch::KeywordMuteHelper class Glitch::KeywordMuteHelper
include ActionView::Helpers::SanitizeHelper
attr_reader :text_matcher attr_reader :text_matcher
attr_reader :tag_matcher attr_reader :tag_matcher
attr_reader :entity_decoder
def initialize(receiver_id) def initialize(receiver_id)
@text_matcher = Glitch::KeywordMute.text_matcher_for(receiver_id) @text_matcher = Glitch::KeywordMute.text_matcher_for(receiver_id)
@tag_matcher = Glitch::KeywordMute.tag_matcher_for(receiver_id) @tag_matcher = Glitch::KeywordMute.tag_matcher_for(receiver_id)
@entity_decoder = HTMLEntities.new
end end
def matches?(status) def matches?(status)
@ -26,6 +22,6 @@ class Glitch::KeywordMuteHelper
end end
def prepare_text(text) def prepare_text(text)
entity_decoder.decode(strip_tags(text)).tap { |x| puts x } Html2Text.convert(text)
end end
end end

View file

@ -39,5 +39,12 @@ RSpec.describe Glitch::KeywordMuteHelper do
expect(helper.matches?(status)).to be true expect(helper.matches?(status)).to be true
end end
it 'matches link hrefs in HTML text' do
status = Fabricate(:status, text: '<p><a href="https://example.com/it-was-milk">yep</a></p>')
Glitch::KeywordMute.create!(account: alice, keyword: 'milk')
expect(helper.matches?(status)).to be true
end
end end
end end