Kouhei Sutou
null+****@clear*****
Mon Jul 10 11:40:36 JST 2017
Kouhei Sutou 2017-07-10 11:40:36 +0900 (Mon, 10 Jul 2017) New Revision: d6abe4168142924eba495436c65afd7f534aefc6 https://github.com/ranguba/chupa-text-decomposer-html/commit/d6abe4168142924eba495436c65afd7f534aefc6 Message: Ignore topic-path element Modified files: lib/chupa-text/decomposers/html.rb test/test-html.rb Modified: lib/chupa-text/decomposers/html.rb (+3 -3) =================================================================== --- lib/chupa-text/decomposers/html.rb 2017-07-05 17:10:58 +0900 (9ccf9ed) +++ lib/chupa-text/decomposers/html.rb 2017-07-10 11:40:36 +0900 (f92d6f5) @@ -143,7 +143,7 @@ module ChupaText def header_element?(element, name, classes) case name - when "header", "nav" + when "header" return true end @@ -191,13 +191,13 @@ module ChupaText classes.each do |klass| case klass - when "nav", "menu" + when "nav", "menu", /\Atopic[-_]?path\z/i return true end end case element["id"] - when "nav", "menu" + when "nav", "menu", /\Atopic[-_]?path\z/i return true end Modified: test/test-html.rb (+34 -0) =================================================================== --- test/test-html.rb 2017-07-05 17:10:58 +0900 (6be657c) +++ test/test-html.rb 2017-07-10 11:40:36 +0900 (217587a) @@ -436,6 +436,23 @@ class TestHTML < Test::Unit::TestCase decompose(@data)) end + def test_topic_path_class + @data.body = <<-HTML +<html> + <body> + Before + <div class="topic-path">topic-path</div> + <div class="topic_path">topic_path</div> + <div class="topicpath">topicpath</div> + <div class="TopicPath">TopicPath</div> + After + </body> +</html> + HTML + assert_equal(["Before\nAfter"], + decompose(@data)) + end + def test_nav_id @data.body = <<-HTML <html> @@ -455,6 +472,23 @@ class TestHTML < Test::Unit::TestCase assert_equal(["BeforeAfter"], decompose(@data)) end + + def test_topic_path_id + @data.body = <<-HTML +<html> + <body> + Before + <div id="topic-path">topic-path</div> + <div id="topic_path">topic_path</div> + <div id="topicpath">topicpath</div> + <div id="TopicPath">TopicPath</div> + After + </body> +</html> + HTML + assert_equal(["Before\nAfter"], + decompose(@data)) + end end end end -------------- next part -------------- HTML����������������������������...Download