wovnrb 1.1.0 → 2.0.0

data/lib/wovnrb/lang.rb

@@ -1,51 +1,49 @@
-# -*- encoding: UTF-8 -*-
-
 require 'addressable'
 
 module Wovnrb
   class Lang
     LANG = {
-      #http://msdn.microsoft.com/en-us/library/hh456380.aspx
-      'ar' => {name: 'العربية',           code: 'ar',     en: 'Arabic'},
-      'eu' => {name: 'Euskara',           code: 'eu',     en: 'Basque'},
-      'bn' => {name: 'বাংলা ভাষা',             code: 'bn',     en: 'Bengali'},
-      'bg' => {name: 'Български',         code: 'bg',     en: 'Bulgarian'},
-      'ca' => {name: 'Català',            code: 'ca',     en: 'Catalan'},
-      'zh-CHS' => {name: '简体中文',      code: 'zh-CHS', en: 'Simp Chinese'},
-      'zh-CHT' => {name: '繁體中文',      code: 'zh-CHT', en: 'Trad Chinese'},
-      'da' => {name: 'Dansk',             code: 'da',     en: 'Danish'},
-      'nl' => {name: 'Nederlands',        code: 'nl',     en: 'Dutch'},
-      'en' => {name: 'English',           code: 'en',     en: 'English'},
-      'fi' => {name: 'Suomi',             code: 'fi',     en: 'Finnish'},
-      'fr' => {name: 'Français',          code: 'fr',     en: 'French'},
-      'gl' => {name: 'Galego',            code: 'gl',     en: 'Galician'},
-      'de' => {name: 'Deutsch',           code: 'de',     en: 'German'},
-      'el' => {name: 'Ελληνικά',          code: 'el',     en: 'Greek'},
-      'he' => {name: 'עברית',             code: 'he',     en: 'Hebrew'},
-      'hu' => {name: 'Magyar',            code: 'hu',     en: 'Hungarian'},
-      'id' => {name: 'Bahasa Indonesia',  code: 'id',     en: 'Indonesian'},
-      'it' => {name: 'Italiano',          code: 'it',     en: 'Italian'},
-      'ja' => {name: '日本語',            code: 'ja',     en: 'Japanese'},
-      'ko' => {name: '한국어',            code: 'ko',     en: 'Korean'},
-      'lv' =>{name: 'Latviešu',          code: 'lv',     en: 'Latvian'},
-      'ms' => {name: 'Bahasa Melayu',     code: 'ms',     en: 'Malay'},
+      # http://msdn.microsoft.com/en-us/library/hh456380.aspx
+      'ar' => { name: 'العربية',           code: 'ar',     en: 'Arabic' },
+      'eu' => { name: 'Euskara',           code: 'eu',     en: 'Basque' },
+      'bn' => { name: 'বাংলা ভাষা',             code: 'bn',     en: 'Bengali' },
+      'bg' => { name: 'Български',         code: 'bg',     en: 'Bulgarian' },
+      'ca' => { name: 'Català',            code: 'ca',     en: 'Catalan' },
+      'zh-CHS' => { name: '简体中文',      code: 'zh-CHS', en: 'Simp Chinese' },
+      'zh-CHT' => { name: '繁體中文',      code: 'zh-CHT', en: 'Trad Chinese' },
+      'da' => { name: 'Dansk',             code: 'da',     en: 'Danish' },
+      'nl' => { name: 'Nederlands',        code: 'nl',     en: 'Dutch' },
+      'en' => { name: 'English',           code: 'en',     en: 'English' },
+      'fi' => { name: 'Suomi',             code: 'fi',     en: 'Finnish' },
+      'fr' => { name: 'Français',          code: 'fr',     en: 'French' },
+      'gl' => { name: 'Galego',            code: 'gl',     en: 'Galician' },
+      'de' => { name: 'Deutsch',           code: 'de',     en: 'German' },
+      'el' => { name: 'Ελληνικά',          code: 'el',     en: 'Greek' },
+      'he' => { name: 'עברית',             code: 'he',     en: 'Hebrew' },
+      'hu' => { name: 'Magyar',            code: 'hu',     en: 'Hungarian' },
+      'id' => { name: 'Bahasa Indonesia',  code: 'id',     en: 'Indonesian' },
+      'it' => { name: 'Italiano',          code: 'it',     en: 'Italian' },
+      'ja' => { name: '日本語',            code: 'ja',     en: 'Japanese' },
+      'ko' => { name: '한국어',            code: 'ko',     en: 'Korean' },
+      'lv' => { name: 'Latviešu',          code: 'lv',     en: 'Latvian' },
+      'ms' => { name: 'Bahasa Melayu',     code: 'ms',     en: 'Malay' },
       'my' => { name: 'ဗမာစာ',             code: 'my',     en: 'Burmese' },
-      'ne' => {name: 'नेपाली भाषा',            code: 'ne',     en: 'Nepali'},
-      'fa' => {name: 'زبان_فارسی',        code: 'fa',     en: 'Persian'},
-      'no' => {name: 'Norsk',             code: 'no',     en: 'Norwegian'},
-      'pl' => {name: 'Polski',            code: 'pl',     en: 'Polish'},
-      'pt' => {name: 'Português',         code: 'pt',     en: 'Portuguese'},
-      'ru' => {name: 'Русский',           code: 'ru',     en: 'Russian'},
-      'es' => {name: 'Español',           code: 'es',     en: 'Spanish'},
-      'sw' => {name: 'Kiswahili',         code: 'sw',     en: 'Swahili'},
-      'sv' => {name: 'Svensk',            code: 'sv',     en: 'Swedish'},
-      'th' => {name: 'ภาษาไทย',           code: 'th',     en: 'Thai'},
-      'hi' => {name: 'हिन्दी',               code: 'hi',     en: 'Hindi'},
-      'tr' => {name: 'Türkçe',            code: 'tr',     en: 'Turkish'},
-      'uk' => {name: 'Українська',        code: 'uk',     en: 'Ukrainian'},
-      'ur' => {name: 'اردو',              code: 'ur',     en: 'Urdu'},
-      'vi' => {name: 'Tiếng Việt',        code: 'vi',     en: 'Vietnamese'},
-    }
+      'ne' => { name: 'नेपाली भाषा',            code: 'ne',     en: 'Nepali' },
+      'fa' => { name: 'زبان_فارسی',        code: 'fa',     en: 'Persian' },
+      'no' => { name: 'Norsk',             code: 'no',     en: 'Norwegian' },
+      'pl' => { name: 'Polski',            code: 'pl',     en: 'Polish' },
+      'pt' => { name: 'Português',         code: 'pt',     en: 'Portuguese' },
+      'ru' => { name: 'Русский',           code: 'ru',     en: 'Russian' },
+      'es' => { name: 'Español',           code: 'es',     en: 'Spanish' },
+      'sw' => { name: 'Kiswahili',         code: 'sw',     en: 'Swahili' },
+      'sv' => { name: 'Svensk',            code: 'sv',     en: 'Swedish' },
+      'th' => { name: 'ภาษาไทย',           code: 'th',     en: 'Thai' },
+      'hi' => { name: 'हिन्दी',               code: 'hi',     en: 'Hindi' },
+      'tr' => { name: 'Türkçe',            code: 'tr',     en: 'Turkish' },
+      'uk' => { name: 'Українська',        code: 'uk',     en: 'Ukrainian' },
+      'ur' => { name: 'اردو',              code: 'ur',     en: 'Urdu' },
+      'vi' => { name: 'Tiếng Việt',        code: 'vi',     en: 'Vietnamese' }
+    }.freeze
 
     # Provides the ISO639-1 code for a given lang code.
     # Source: https://support.google.com/webmasters/answer/189077?hl=en
@@ -54,35 +52,33 @@ module Wovnrb
     #
     # @return [String] The ISO639-1 code of the language.
     def self.iso_639_1_normalization(lang_code)
-      return lang_code.sub(/zh-CHT/i, 'zh-Hant').sub(/zh-CHS/i, 'zh-Hans')
+      lang_code.sub(/zh-CHT/i, 'zh-Hant').sub(/zh-CHS/i, 'zh-Hans')
     end
 
     def self.get_code(lang_name)
       return nil if lang_name.nil?
       return lang_name if LANG[lang_name]
+
       custom_lang_aliases = Store.instance.settings['custom_lang_aliases']
       custom_lang = LANG[custom_lang_aliases.invert[lang_name]]
       return custom_lang[:code] if custom_lang
-      LANG.each do |k, l|
-        if lang_name.downcase == l[:name].downcase || lang_name.downcase == l[:en].downcase || lang_name.downcase == l[:code].downcase
-          return l[:code]
-        end
+
+      LANG.each do |_k, l|
+        return l[:code] if lang_name.casecmp(l[:name]).zero? || lang_name.casecmp(l[:en]).zero? || lang_name.casecmp(l[:code]).zero?
       end
-      return nil
+      nil
     end
 
     def self.get_lang(lang)
       lang_code = get_code(lang)
-      return LANG[lang_code]
+      LANG[lang_code]
     end
 
     def initialize(lang_name)
       @lang_code = Lang.get_code(lang_name)
     end
 
-    def lang_code
-      @lang_code
-    end
+    attr_reader :lang_code
 
     # Adds language code to URL in "href" variable by "pattern" variable and own @lang_code.
     #  When @lang_code is 'ja', add_lang_code('https://wovn.io', 'path', headers) returns 'https://wovn.io/ja/'.
@@ -94,6 +90,7 @@ module Wovnrb
     # @return [String]                 URL added langauge code.
     def add_lang_code(href, pattern, headers)
       return href if href =~ /^(#.*)?$/
+
       code_to_add = Store.instance.settings['custom_lang_aliases'][@lang_code] || @lang_code
       # absolute links
       new_href = href
@@ -109,83 +106,54 @@ module Wovnrb
         # DNS names are case insensitive
         if uri.host.downcase === headers.host.downcase
           case pattern
-            when 'subdomain'
-              sub_d = href.match(/\/\/([^\.]*)\./)[1]
-              sub_code = Lang.get_code(sub_d)
-              if sub_code && sub_code.downcase == code_to_add.downcase
-                new_href = href.sub(Regexp.new(code_to_add, 'i'), code_to_add.downcase)
-              else
-                new_href = href.sub(/(\/\/)([^\.]*)/, '\1' + code_to_add.downcase + '.' + '\2')
-              end
-            when 'query'
-              new_href = add_query_lang_code(href, code_to_add)
-            else # path
-              new_href = href.sub(/([^\.]*\.[^\/]*)(\/|$)/, '\1/' + code_to_add + '/')
+          when 'subdomain'
+            sub_d = href.match(/\/\/([^\.]*)\./)[1]
+            sub_code = Lang.get_code(sub_d)
+            new_href = if sub_code && sub_code.casecmp(code_to_add).zero?
+                         href.sub(Regexp.new(code_to_add, 'i'), code_to_add.downcase)
+                       else
+                         href.sub(/(\/\/)([^\.]*)/, '\1' + code_to_add.downcase + '.' + '\2')
+                       end
+          when 'query'
+            new_href = add_query_lang_code(href, code_to_add)
+          else # path
+            new_href = href.sub(/([^\.]*\.[^\/]*)(\/|$)/, '\1/' + code_to_add + '/')
           end
         end
       elsif href
         case pattern
-          when 'subdomain'
-            lang_url = headers.protocol + '://' + code_to_add.downcase + '.' + headers.host
+        when 'subdomain'
+          lang_url = headers.protocol + '://' + code_to_add.downcase + '.' + headers.host
+          current_dir = headers.pathname.sub(/[^\/]*\.[^\.]{2,6}$/, '')
+          new_href = if href =~ /^\.\..*$/
+                       # ../path
+                       lang_url + '/' + href.gsub(/^\.\.\//, '')
+                     elsif href =~ /^\..*$/
+                       # ./path
+                       lang_url + current_dir + '/' + href.gsub(/^\.\//, '')
+                     elsif href =~ /^\/.*$/
+                       # /path
+                       lang_url + href
+                     else
+                       # path
+                       lang_url + current_dir + '/' + href
+                     end
+        when 'query'
+          new_href = add_query_lang_code(href, code_to_add)
+        else # path
+          if href =~ /^\//
+            new_href = '/' + code_to_add + href
+          else
             current_dir = headers.pathname.sub(/[^\/]*\.[^\.]{2,6}$/, '')
-            if href =~ /^\.\..*$/
-              # ../path
-              new_href = lang_url + '/' + href.gsub(/^\.\.\//, '')
-            elsif href =~ /^\..*$/
-              # ./path
-              new_href = lang_url + current_dir + '/' + href.gsub(/^\.\//, '')
-            elsif href =~ /^\/.*$/
-              # /path
-              new_href = lang_url + href
-            else
-              # path
-              new_href = lang_url + current_dir + '/' + href
-            end
-          when 'query'
-            new_href = add_query_lang_code(href, code_to_add)
-          else # path
-            if href =~ /^\//
-              new_href = '/' + code_to_add + href
-            else
-              current_dir = headers.pathname.sub(/[^\/]*\.[^\.]{2,6}$/, '')
-              current_dir = '/' if current_dir == ''
-              new_href = '/' + code_to_add + current_dir + href
-            end
+            current_dir = '/' if current_dir == ''
+            new_href = '/' + code_to_add + current_dir + href
+          end
         end
       end
 
       new_href
     end
 
-    def switch_dom_lang(dom, store, values, url, headers)
-      replace_dom_values(dom, values, store, url, headers)
-
-      if dom.html?
-        # INSERT LANGUAGE METALINKS
-        parent_node = dom.at_css('head') || dom.at_css('body') || dom.at_css('html')
-        published_langs = get_langs(values)
-        all_langs = published_langs.add(store.settings['default_lang'])
-        all_langs.each do |l|
-          insert_node = Nokogiri::XML::Node.new('link', dom)
-          insert_node['rel'] = 'alternate'
-          insert_node['hreflang'] = Lang::iso_639_1_normalization(l)
-          insert_node['href'] = headers.redirect_location(l)
-          parent_node.add_child(insert_node)
-        end
-
-        # set lang property on HTML tag
-        if dom.at_css('html') || dom.at_css('HTML')
-          (dom.at_css('html') || dom.at_css('HTML')).set_attribute('lang', Lang::iso_639_1_normalization(@lang_code))
-        end
-      end
-
-      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(save_with: 0).gsub(/href="([^"]*)"/) { |m| "href=\"#{index_href[$1] || $1}\"" }
-    end
-
     private
 
     def index_href_for_encoding_and_decoding(dom)
@@ -216,11 +184,11 @@ module Wovnrb
         replacers << LinkReplacer.new(store, pattern, headers)
       end
 
-      unless html_text_index.empty?
-        replacers << UnifiedValues::TextReplacer.new(store, html_text_index)
-      else
-        replacers << TextReplacer.new(store, text_index)
-      end
+      replacers << if html_text_index.empty?
+                     TextReplacer.new(store, text_index)
+                   else
+                     UnifiedValues::TextReplacer.new(store, html_text_index)
+                   end
       replacers << MetaReplacer.new(store, text_index, pattern, headers)
       replacers << InputReplacer.new(store, text_index)
       replacers << ImageReplacer.new(store, url, text_index, src_index, img_src_prefix, host_aliases)
@@ -233,8 +201,8 @@ module Wovnrb
 
     def get_langs(values)
       langs = Set.new
-      (values['text_vals'] || {}).merge(values['img_vals'] || {}).each do |key, index|
-        index.each do |l, val|
+      (values['text_vals'] || {}).merge(values['img_vals'] || {}).each do |_key, index|
+        index.each do |l, _val|
           langs.add(l)
         end
       end

data/lib/wovnrb/railtie.rb

@@ -1,20 +1,18 @@
 module Wovnrb
-
   class Railtie < Rails::Railtie
     initializer 'wovnrb.configure_rails_initialization' do |app|
       app.middleware.insert_before(0, Wovnrb::Interceptor)
-      #begin
+      # begin
       #  app.middleware.insert_before(Rack::Runtime, Wovnrb::Interceptor)
-      #rescue
+      # rescue
       #  app.middleware.insert_before(0, Wovnrb::Interceptor)
-      #end
+      # end
 
-      #if Rails.env.development? && config.respond_to?(:wovnrb)
+      # if Rails.env.development? && config.respond_to?(:wovnrb)
       #  config.after_initialize do
       #    config.wovnrb[:project_token] = User.first.short_token
       #  end
-      #end
+      # end
     end
   end
-
 end

data/lib/wovnrb/services/glob.rb

@@ -11,8 +11,8 @@ module Wovnrb
       sub_directories = pattern.split('/**', -1)
       regexp = sub_directories.map do |sub_dir|
         sub_dir.split('*', -1)
-          .map {|p| Regexp.escape(p)}
-          .join('[^/]*')
+               .map { |p| Regexp.escape(p) }
+               .join('[^/]*')
       end.join('(/[^/]*)*')
 
       @regexp = Regexp.new("^#{regexp}$")
@@ -22,4 +22,4 @@ module Wovnrb
       !@regexp.match(url).nil?
     end
   end
-end
+end

data/lib/wovnrb/services/html_converter.rb

@@ -0,0 +1,192 @@
+module Wovnrb
+  class HtmlConverter
+    def initialize(dom, store, headers)
+      @dom = dom
+      @headers = headers
+      @store = store
+    end
+
+    def build
+      transform_html
+      html
+    end
+
+    def build_api_compatible_html
+      marker = HtmlReplaceMarker.new
+      converted_html = replace_dom(marker)
+
+      [converted_html, marker]
+    end
+
+    private
+
+    def html
+      @dom.to_html(save_with: 0).strip
+    end
+
+    def transform_html
+      replace_snippet
+      replace_hreflangs
+      inject_lang_html_tag
+    end
+
+    def replace_snippet
+      strip_snippet
+      insert_snippet
+    end
+
+    def replace_dom(marker)
+      strip_snippet
+      strip_hreflangs if add_hreflang
+
+      @dom.traverse { |node| transform_node(node, marker) }
+
+      insert_snippet(true)
+      insert_hreflang_tags
+
+      html
+    end
+
+    def transform_node(node, marker)
+      strip_wovn_ignore(node, marker)
+      strip_custom_ignore(node, marker)
+      strip_form(node, marker)
+      strip_script(node, marker)
+    end
+
+    def strip_script(node, marker)
+      put_replace_marker(node, marker) if node.name.casecmp('script').zero?
+    end
+
+    def strip_form(node, marker)
+      if node.name.casecmp('form').zero?
+        put_replace_marker(node, marker)
+        return
+      end
+
+      if node.name.casecmp('input').zero? && node.get_attribute('type') == 'hidden'
+        original_text = node.get_attribute('value')
+        return if original_text.include?(HtmlReplaceMarker::KEY_PREFIX)
+
+        node.set_attribute('value', marker.add_value(original_text))
+      end
+    end
+
+    def strip_custom_ignore(node, marker)
+      classes = node.get_attribute('class')
+      return unless classes.present?
+
+      ignored_classes = @store.settings['ignore_class']
+      should_be_ignored = (ignored_classes & classes.split(' ')).present?
+
+      put_replace_marker(node, marker) if should_be_ignored
+    end
+
+    def strip_wovn_ignore(node, marker)
+      put_replace_marker(node, marker) if node && node.get_attribute('wovn-ignore')
+    end
+
+    def put_replace_marker(node, marker)
+      original_text = node.inner_text
+      return if original_text.include?(HtmlReplaceMarker::KEY_PREFIX)
+
+      node.inner_html = marker.add_comment_value(original_text)
+    end
+
+    def strip_hreflangs
+      supported_langs = @store.supported_langs
+      @dom.xpath('//link') do |node|
+        node.remove if node['hreflang'] && supported_langs.include?(Lang.iso_639_1_normalization(node['hreflang']))
+      end
+    end
+
+    def add_hreflang
+      !!(@store && @headers)
+    end
+
+    def inject_lang_html_tag
+      root = @dom.at_css('html')
+      return unless root
+
+      current_lang = @headers.lang_code
+      default_lang = @store.default_lang
+
+      if current_lang != default_lang
+        root['lang'] = current_lang
+      else
+        root.delete('lang')
+      end
+    end
+
+    def replace_hreflangs
+      strip_hreflang_tags
+      insert_hreflang_tags
+    end
+
+    def strip_hreflang_tags
+      @dom.xpath('//link').each do |node|
+        node.remove if node['hreflang'] && @store.supported_langs.include?(Lang.iso_639_1_normalization(node['hreflang']))
+      end
+    end
+
+    def insert_hreflang_tags
+      parent_node = @dom.at_css('head') || @dom.at_css('body') || @dom.at_css('html')
+      return unless parent_node
+
+      @store.supported_langs.each do |lang_code|
+        insert_node = Nokogiri::XML::Node.new('link', @dom)
+        insert_node['rel'] = 'alternate'
+        insert_node['hreflang'] = Lang.iso_639_1_normalization(lang_code)
+        insert_node['href'] = @headers.redirect_location(lang_code)
+
+        parent_node.add_child(insert_node.to_s)
+      end
+    end
+
+    # Remove wovn snippet code from dom
+    def strip_snippet
+      @dom.xpath('//script').each do |script_node|
+        script_node.remove if script_node['src'] && script_node['src'] =~ /^\/\/j.(dev-)?wovn.io(:3000)?\//
+      end
+    end
+
+    def insert_snippet(adds_backend_error_mark = true)
+      parent_node = @dom.at_css('head') || @dom.at_css('body') || @dom.at_css('html')
+      return unless parent_node
+
+      insert_node = Nokogiri::XML::Node.new('script', @dom)
+      insert_node['src'] = "//j.#{@store.wovn_host}/1"
+      insert_node['async'] = true
+      insert_node['data-wovnio'] = data_wovnio
+      insert_node['data-wovnio-type'] = 'fallback_snippet' if adds_backend_error_mark
+      # do this so that there will be a closing tag (better compatibility with browsers)
+      insert_node.content = ''
+
+      if !parent_node.children.empty?
+        parent_node.children.first.add_previous_sibling(insert_node)
+      else
+        parent_node.add_child(insert_node)
+      end
+    end
+
+    def data_wovnio
+      token = @store.settings['project_token']
+      current_lang = @headers.lang_code
+      default_lang = @store.settings['default_lang']
+      url_pattern = @store.settings['url_pattern']
+      lang_code_aliases_json = JSON.generate(@store.settings['custom_lang_aliases'])
+
+      CGI.escapeHTML(
+        [
+          "key=#{token}",
+          'backend=true',
+          "currentLang=#{current_lang}",
+          "defaultLang=#{default_lang}",
+          "urlPattern=#{url_pattern}",
+          "langCodeAliases=#{lang_code_aliases_json}",
+          "version=WOVN.rb_#{VERSION}"
+        ].join('&')
+      )
+    end
+  end
+end