wovnrb 0.2.28 → 0.2.29

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: cbefcf5370f91d37b96f311755e68fce4666118a
-  data.tar.gz: df2d38a8b611b2cbc38c92f4d2036c4de8c0a660
+  metadata.gz: f15715b5f8a8d0d3bb58c2d8a86a915353d9cb7c
+  data.tar.gz: 0bb121940678dbc50aa0beb7f9b5839f35935844
 SHA512:
-  metadata.gz: fa17224c7706eda6e7fa399295c09900c8ed4e6a10ffd3a69e6e249952411b7081f08b6abd53c8b8896e65f2044c842087ae7ead72e42d6512d113a1dbb43c29
-  data.tar.gz: 66e899fe8363e92533ae207228a68601aff56d9cb8931b3d029d8f67cd2f85a35bbeecc895e99e89d2d358e0c80052a1fa8b675632c1216b9d16bb37634e4ec5
+  metadata.gz: c51823510261c6a2d97ec201285f9eda6f51927e883f017815b48ddaf5c989b7776f670bc4c23599bbdfa5be2a46d33b60e72717d274072725898965b5062645
+  data.tar.gz: 01e7f5d664b053574da8a78a86c363ad23556f4e7e443ffc05998ffb2ae1fc7feea1f706cb8b376920522aaff9b44085a08316757e1c2ff41510a0295c9cd958

data/lib/wovnrb.rb

@@ -95,16 +95,13 @@ module Wovnrb
         if ignore_all || !d.xpath('//html[@wovn-ignore]').empty? || is_amp_page?(d)
           ignore_all = true
           output = d.to_html.gsub(/href="([^"]*)"/) { |m| "href=\"#{URI.decode($1)}\"" }
+          put_back_noscripts!(output, noscripts)
           new_body.push(output)
           next
         end
 
         output = lang.switch_dom_lang(d, @store, values, url, headers)
-        # put back noscripts
-        noscripts.each_with_index do |noscript, index|
-          noscript_identifier = "<noscript wovn-id=\"#{index}\"></noscript>"
-          output.sub!(noscript_identifier, noscript)
-        end
+        put_back_noscripts!(output, noscripts)
         new_body.push(output)
       end
       body.close if body.respond_to?(:close)
@@ -113,6 +110,13 @@ module Wovnrb
 
     private
 
+    def put_back_noscripts!(output, noscripts)
+      noscripts.each_with_index do |noscript, index|
+        noscript_identifier = "<noscript wovn-id=\"#{index}\"></noscript>"
+        output.sub!(noscript_identifier, noscript)
+      end
+    end
+
     # Checks if a given HTML body is an Accelerated Mobile Page (AMP).
     # To do so, it looks at the required attributes for the HTML tag:
     # https://www.ampproject.org/docs/tutorials/create/basic_markup.

data/lib/wovnrb/lang.rb

@@ -89,7 +89,7 @@ module Wovnrb
         # in the future, perhaps validate url rather than using begin rescue
         # "#{url =~ /\// ? 'http:' : ''}#{url}" =~ URI::regexp
         begin
-          uri = URI(href)
+          uri = Addressable::URI.parse(href)
         rescue
           return new_href
         end
@@ -164,10 +164,25 @@ module Wovnrb
         (dom.at_css('html') || dom.at_css('HTML')).set_attribute('lang', Lang::iso_639_1_normalization(@lang_code))
       end
 
-      dom.to_html.gsub(/href="([^"]*)"/) { |m| "href=\"#{URI.decode($1)}\"" }
+      index_href = index_href_for_encoding_and_decoding(dom)
+      # NOTE: when we use `#to_html` with nokogiri, nokogiri encode all href.
+      #       but we want to keep original href as much as possible.
+      #       That's why we replace href with original href which added lang info by wovnrb like this after we used `to_html`
+      dom.to_html.gsub(/href="([^"]*)"/) { |m| "href=\"#{index_href[$1] || $1}\"" }
     end
 
     private
+
+    def index_href_for_encoding_and_decoding(dom)
+      result = {}
+      dom.xpath('//*[@href]').each do |a_tag|
+        url = a_tag['href']
+        encoded_url = Addressable::URI.parse(url).normalize.to_s
+        result[encoded_url] = url if encoded_url != url
+      end
+      result
+    end
+
     def replace_dom_values(dom, values, store, url, headers)
       text_index = values['text_vals'] || {}
       src_index = values['img_vals'] || {}

data/lib/wovnrb/version.rb

@@ -1,3 +1,3 @@
 module Wovnrb
-  VERSION = "0.2.28"
+  VERSION = "0.2.29"
 end

data/test/lib/lang_test.rb

@@ -583,5 +583,33 @@ module Wovnrb
       h = Wovnrb::Headers.new(Wovnrb.get_env('url' => 'http://favy.tips'), Wovnrb.get_settings('url_pattern' => 'subdomain', 'url_pattern_reg' => '^(?<lang>[^.]+).'))
       assert_equal("http://staging-fr.favy.tips/topics/50", lang.add_lang_code("/topics/50", 'subdomain', h))
     end
+
+    def test_switch_dom_lang_for_keeping_originl_encoding_decoding
+      lang = Lang.new('ja')
+      header = Wovnrb::Headers.new(Wovnrb.get_env('url' => 'http://page.com'), Wovnrb.get_settings)
+      html = <<-HTML
+        <head>
+          <link rel="stylesheet" type="text/css" href="%3Ca%3E%E3%81%B5%E3%82%86%3C%2Fa%3E/theme.css">
+          <link rel="stylesheet" type="text/css" href="テスト.css">
+        </head>
+        <ul>
+          <li><a href="http://page.com/%e5%ba%83%e5%b0%be">encoded japanese path</a></li>
+          <li><a href="http://page.com/テスト/DxA9J">no encoded japanese path</a></li>
+          <li><a href="http://page.com/#DnE897">decoded invalid path</a></li>
+          <li><a href="http://page.com/%23DnE897">encoded invalid path</a>
+        </ul>
+      HTML
+
+      dom = Wovnrb.to_dom(html)
+      url = header.url
+
+      swapped_body = lang.switch_dom_lang(dom, Store.instance, generate_values, url, header)
+      assert(swapped_body.include?('<link rel="stylesheet" type="text/css" href="%3Ca%3E%E3%81%B5%E3%82%86%3C%2Fa%3E/theme.css">'))
+      assert(swapped_body.include?('<link rel="stylesheet" type="text/css" href="テスト.css">'))
+      assert(swapped_body.include?('<a href="http://page.com/ja/%e5%ba%83%e5%b0%be">encoded japanese path</a>'))
+      assert(swapped_body.include?('<a href="http://page.com/ja/テスト/DxA9J">no encoded japanese path</a>'))
+      assert(swapped_body.include?('<a href="http://page.com/ja/#DnE897">decoded invalid path</a>'))
+      assert(swapped_body.include?('<a href="http://page.com/ja/%23DnE897">encoded invalid path</a>'))
+    end
   end
 end

data/test/lib/wovnrb_test.rb

@@ -112,6 +112,7 @@ class WovnrbTest < Minitest::Test
     headers = Wovnrb::Headers.new(Wovnrb.get_env('url' => 'http://page.com'), Wovnrb.get_settings('url_pattern' => 'subdomain', 'url_pattern_reg' => '^(?<lang>[^.]+).'))
     body = <<HTML
 <html amp>
+<head><noscript><style amp-boilerplate>body{-webkit-animation:none;-moz-animation:none;-ms-animation:none;animation:none}</style></noscript></head>
 <body>
   <h1>Mr. Belvedere Fan Club</h1>
   <div><p>Hello</p></div>
@@ -120,7 +121,10 @@ class WovnrbTest < Minitest::Test
 HTML
     expected_body = <<HTML
 <html amp="">
-<head><meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\"></head>
+<head>
+<meta http-equiv=\"Content-Type\" content=\"text/html; charset=UTF-8\">
+<noscript><style amp-boilerplate>body{-webkit-animation:none;-moz-animation:none;-ms-animation:none;animation:none}</style></noscript>
+</head>
 <body>
   <h1>Mr. Belvedere Fan Club</h1>
   <div><p>Hello</p></div>

data/test/test_helper.rb

@@ -94,7 +94,7 @@ module Wovnrb
       env['PATH_INFO'] = url.path
     end
 
-    return Rack::MockRequest.env_for(url, env.merge(options))
+    return Rack::MockRequest.env_for(url.to_s, env.merge(options))
   end
 
   def to_dom(html)

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: wovnrb
 version: !ruby/object:Gem::Version
-  version: 0.2.28
+  version: 0.2.29
 platform: ruby
 authors:
 - Jeff Sandford
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-11-10 00:00:00.000000000 Z
+date: 2017-12-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nokogumbo
@@ -423,3 +423,4 @@ test_files:
 - test/lib/wovnrb_test.rb
 - test/services/url_test.rb
 - test/test_helper.rb
+has_rdoc: