twitter-text-kow 1.3.1.0

data/lib/twitter-text/autolink.rb

@@ -0,0 +1,455 @@
+# Copyright 2018 Twitter, Inc.
+# Licensed under the Apache License, Version 2.0
+# http://www.apache.org/licenses/LICENSE-2.0
+
+# encoding: utf-8
+
+require 'set'
+require 'twitter-text/hash_helper'
+
+module Twitter
+  module TwitterText
+    # A module for including Tweet auto-linking in a class. The primary use of this is for helpers/views so they can auto-link
+    # usernames, lists, hashtags and URLs.
+    module Autolink extend self
+      # Default CSS class for auto-linked lists
+      DEFAULT_LIST_CLASS = "tweet-url list-slug".freeze
+      # Default CSS class for auto-linked usernames
+      DEFAULT_USERNAME_CLASS = "tweet-url username".freeze
+      # Default CSS class for auto-linked hashtags
+      DEFAULT_HASHTAG_CLASS = "tweet-url hashtag".freeze
+      # Default CSS class for auto-linked cashtags
+      DEFAULT_CASHTAG_CLASS = "tweet-url cashtag".freeze
+
+      # Default URL base for auto-linked usernames
+      DEFAULT_USERNAME_URL_BASE = "https://twitter.com/".freeze
+      # Default URL base for auto-linked lists
+      DEFAULT_LIST_URL_BASE = "https://twitter.com/".freeze
+      # Default URL base for auto-linked hashtags
+      DEFAULT_HASHTAG_URL_BASE = "https://twitter.com/search?q=%23".freeze
+      # Default URL base for auto-linked cashtags
+      DEFAULT_CASHTAG_URL_BASE = "https://twitter.com/search?q=%24".freeze
+
+      # Default attributes for invisible span tag
+      DEFAULT_INVISIBLE_TAG_ATTRS = "style='position:absolute;left:-9999px;'".freeze
+
+      DEFAULT_OPTIONS = {
+        :list_class     => DEFAULT_LIST_CLASS,
+        :username_class => DEFAULT_USERNAME_CLASS,
+        :hashtag_class  => DEFAULT_HASHTAG_CLASS,
+        :cashtag_class  => DEFAULT_CASHTAG_CLASS,
+
+        :username_url_base => DEFAULT_USERNAME_URL_BASE,
+        :list_url_base     => DEFAULT_LIST_URL_BASE,
+        :hashtag_url_base  => DEFAULT_HASHTAG_URL_BASE,
+        :cashtag_url_base  => DEFAULT_CASHTAG_URL_BASE,
+
+        :invisible_tag_attrs => DEFAULT_INVISIBLE_TAG_ATTRS
+      }.freeze
+
+      def auto_link_with_json(text, json, options = {})
+        # concatenate entities
+        entities = json.values().flatten()
+
+        # map JSON entity to twitter-text entity
+        # be careful not to alter arguments received
+        entities.map! do |entity|
+          entity = HashHelper.symbolize_keys(entity)
+          # hashtag
+          entity[:hashtag] = entity[:text] if entity[:text]
+          entity
+        end
+
+        auto_link_entities(text, entities, options)
+      end
+
+      def auto_link_entities(text, entities, options = {}, &block)
+        return text if entities.empty?
+
+        # NOTE deprecate these attributes not options keys in options hash, then use html_attrs
+        options = DEFAULT_OPTIONS.merge(options)
+        options[:html_attrs] = extract_html_attrs_from_options!(options)
+        options[:html_attrs][:rel] ||= "nofollow" unless options[:suppress_no_follow]
+        options[:html_attrs][:target] = "_blank" if options[:target_blank] == true
+
+        Twitter::TwitterText::Rewriter.rewrite_entities(text.dup, entities) do |entity, chars|
+          if entity[:url]
+            link_to_url(entity, chars, options, &block)
+          elsif entity[:hashtag]
+            link_to_hashtag(entity, chars, options, &block)
+          elsif entity[:screen_name]
+            link_to_screen_name(entity, chars, options, &block)
+          elsif entity[:cashtag]
+            link_to_cashtag(entity, chars, options, &block)
+          elsif entity[:emoji]
+            entity[:emoji]
+          end
+        end
+      end
+
+      # Add <tt><a></a></tt> tags around the usernames, lists, hashtags and URLs in the provided <tt>text</tt>.
+      # The <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt> hash:
+      # Also any elements in the <tt>options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      #
+      # <tt>:url_class</tt>::      class to add to url <tt><a></tt> tags
+      # <tt>:list_class</tt>::     class to add to list <tt><a></tt> tags
+      # <tt>:username_class</tt>:: class to add to username <tt><a></tt> tags
+      # <tt>:hashtag_class</tt>::  class to add to hashtag <tt><a></tt> tags
+      # <tt>:cashtag_class</tt>::  class to add to cashtag <tt><a></tt> tags
+      # <tt>:username_url_base</tt>::  the value for <tt>href</tt> attribute on username links. The <tt>@username</tt> (minus the <tt>@</tt>) will be appended at the end of this.
+      # <tt>:list_url_base</tt>::      the value for <tt>href</tt> attribute on list links. The <tt>@username/list</tt> (minus the <tt>@</tt>) will be appended at the end of this.
+      # <tt>:hashtag_url_base</tt>::   the value for <tt>href</tt> attribute on hashtag links. The <tt>#hashtag</tt> (minus the <tt>#</tt>) will be appended at the end of this.
+      # <tt>:cashtag_url_base</tt>::   the value for <tt>href</tt> attribute on cashtag links. The <tt>$cashtag</tt> (minus the <tt>$</tt>) will be appended at the end of this.
+      # <tt>:invisible_tag_attrs</tt>::   HTML attribute to add to invisible span tags
+      # <tt>:username_include_symbol</tt>:: place the <tt>@</tt> symbol within username and list links
+      # <tt>:suppress_lists</tt>::          disable auto-linking to lists
+      # <tt>:suppress_no_follow</tt>::      do not add <tt>rel="nofollow"</tt> to auto-linked items
+      # <tt>:symbol_tag</tt>::          tag to apply around symbol (@, #, $) in username / hashtag / cashtag links
+      # <tt>:text_with_symbol_tag</tt>::          tag to apply around text part in username / hashtag / cashtag links
+      # <tt>:url_target</tt>::     the value for <tt>target</tt> attribute on URL links.
+      # <tt>:target_blank</tt>:: adds <tt>target="_blank"</tt> to all auto_linked items username / hashtag / cashtag links / urls
+      # <tt>:link_attribute_block</tt>::     function to modify the attributes of a link based on the entity. called with |entity, attributes| params, and should modify the attributes hash.
+      # <tt>:link_text_block</tt>::     function to modify the text of a link based on the entity. called with |entity, text| params, and should return a modified text.
+      def auto_link(text, options = {}, &block)
+        auto_link_entities(text, Extractor.extract_entities_with_indices(text, :extract_url_without_protocol => false), options, &block)
+      end
+
+      # Add <tt><a></a></tt> tags around the usernames and lists in the provided <tt>text</tt>. The
+      # <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt> hash.
+      # Also any elements in the <tt>options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      #
+      # <tt>:list_class</tt>::     class to add to list <tt><a></tt> tags
+      # <tt>:username_class</tt>:: class to add to username <tt><a></tt> tags
+      # <tt>:username_url_base</tt>:: the value for <tt>href</tt> attribute on username links. The <tt>@username</tt> (minus the <tt>@</tt>) will be appended at the end of this.
+      # <tt>:list_url_base</tt>::     the value for <tt>href</tt> attribute on list links. The <tt>@username/list</tt> (minus the <tt>@</tt>) will be appended at the end of this.
+      # <tt>:username_include_symbol</tt>:: place the <tt>@</tt> symbol within username and list links
+      # <tt>:suppress_lists</tt>::          disable auto-linking to lists
+      # <tt>:suppress_no_follow</tt>::      do not add <tt>rel="nofollow"</tt> to auto-linked items
+      # <tt>:symbol_tag</tt>::          tag to apply around symbol (@, #, $) in username / hashtag / cashtag links
+      # <tt>:text_with_symbol_tag</tt>::          tag to apply around text part in username / hashtag / cashtag links
+      # <tt>:link_attribute_block</tt>::     function to modify the attributes of a link based on the entity. called with |entity, attributes| params, and should modify the attributes hash.
+      # <tt>:link_text_block</tt>::     function to modify the text of a link based on the entity. called with |entity, text| params, and should return a modified text.
+      def auto_link_usernames_or_lists(text, options = {}, &block) # :yields: list_or_username
+        auto_link_entities(text, Extractor.extract_mentions_or_lists_with_indices(text), options, &block)
+      end
+
+      # Add <tt><a></a></tt> tags around the hashtags in the provided <tt>text</tt>.
+      # The <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt> hash.
+      # Also any elements in the <tt>options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      #
+      # <tt>:hashtag_class</tt>:: class to add to hashtag <tt><a></tt> tags
+      # <tt>:hashtag_url_base</tt>:: the value for <tt>href</tt> attribute. The hashtag text (minus the <tt>#</tt>) will be appended at the end of this.
+      # <tt>:suppress_no_follow</tt>:: do not add <tt>rel="nofollow"</tt> to auto-linked items
+      # <tt>:symbol_tag</tt>::          tag to apply around symbol (@, #, $) in username / hashtag / cashtag links
+      # <tt>:text_with_symbol_tag</tt>::          tag to apply around text part in username / hashtag / cashtag links
+      # <tt>:link_attribute_block</tt>::     function to modify the attributes of a link based on the entity. called with |entity, attributes| params, and should modify the attributes hash.
+      # <tt>:link_text_block</tt>::     function to modify the text of a link based on the entity. called with |entity, text| params, and should return a modified text.
+      def auto_link_hashtags(text, options = {}, &block)  # :yields: hashtag_text
+        auto_link_entities(text, Extractor.extract_hashtags_with_indices(text), options, &block)
+      end
+
+      # Add <tt><a></a></tt> tags around the cashtags in the provided <tt>text</tt>.
+      # The <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt> hash.
+      # Also any elements in the <tt>options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      #
+      # <tt>:cashtag_class</tt>:: class to add to cashtag <tt><a></tt> tags
+      # <tt>:cashtag_url_base</tt>:: the value for <tt>href</tt> attribute. The cashtag text (minus the <tt>$</tt>) will be appended at the end of this.
+      # <tt>:suppress_no_follow</tt>:: do not add <tt>rel="nofollow"</tt> to auto-linked items
+      # <tt>:symbol_tag</tt>::          tag to apply around symbol (@, #, $) in username / hashtag / cashtag links
+      # <tt>:text_with_symbol_tag</tt>::          tag to apply around text part in username / hashtag / cashtag links
+      # <tt>:link_attribute_block</tt>::     function to modify the attributes of a link based on the entity. called with |entity, attributes| params, and should modify the attributes hash.
+      # <tt>:link_text_block</tt>::     function to modify the text of a link based on the entity. called with |entity, text| params, and should return a modified text.
+      def auto_link_cashtags(text, options = {}, &block)  # :yields: cashtag_text
+        auto_link_entities(text, Extractor.extract_cashtags_with_indices(text), options, &block)
+      end
+
+      # Add <tt><a></a></tt> tags around the URLs in the provided <tt>text</tt>.
+      # The <tt><a></tt> tags can be controlled with the following entries in the <tt>options</tt> hash.
+      # Also any elements in the <tt>options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      #
+      # <tt>:url_class</tt>::     class to add to url <tt><a></tt> tags
+      # <tt>:invisible_tag_attrs</tt>::   HTML attribute to add to invisible span tags
+      # <tt>:suppress_no_follow</tt>:: do not add <tt>rel="nofollow"</tt> to auto-linked items
+      # <tt>:symbol_tag</tt>::          tag to apply around symbol (@, #, $) in username / hashtag / cashtag links
+      # <tt>:text_with_symbol_tag</tt>::          tag to apply around text part in username / hashtag / cashtag links
+      # <tt>:url_target</tt>::     the value for <tt>target</tt> attribute on URL links.
+      # <tt>:link_attribute_block</tt>::     function to modify the attributes of a link based on the entity. called with |entity, attributes| params, and should modify the attributes hash.
+      # <tt>:link_text_block</tt>::     function to modify the text of a link based on the entity. called with |entity, text| params, and should return a modified text.
+      def auto_link_urls(text, options = {}, &block)
+        auto_link_entities(text, Extractor.extract_urls_with_indices(text, :extract_url_without_protocol => false), options, &block)
+      end
+
+      # These methods are deprecated, will be removed in future.
+      extend Deprecation
+
+      # <b>Deprecated</b>: Please use auto_link_urls instead.
+      # Add <tt><a></a></tt> tags around the URLs in the provided <tt>text</tt>.
+      # Any elements in the <tt>href_options</tt> hash will be converted to HTML attributes
+      # and place in the <tt><a></tt> tag.
+      # Unless <tt>href_options</tt> contains <tt>:suppress_no_follow</tt>
+      # the <tt>rel="nofollow"</tt> attribute will be added.
+      alias :auto_link_urls_custom :auto_link_urls
+      deprecate :auto_link_urls_custom, :auto_link_urls
+
+      private
+
+      HTML_ENTITIES = {
+        '&' => '&amp;',
+        '>' => '&gt;',
+        '<' => '&lt;',
+        '"' => '&quot;',
+        "'" => '&#39;'
+      }
+
+      def html_escape(text)
+        text && text.to_s.gsub(/[&"'><]/) do |character|
+          HTML_ENTITIES[character]
+        end
+      end
+
+      # NOTE We will make this private in future.
+      public :html_escape
+
+      # Options which should not be passed as HTML attributes
+      OPTIONS_NOT_ATTRIBUTES = Set.new([
+                                         :url_class, :list_class, :username_class, :hashtag_class, :cashtag_class,
+                                         :username_url_base, :list_url_base, :hashtag_url_base, :cashtag_url_base,
+                                         :username_url_block, :list_url_block, :hashtag_url_block, :cashtag_url_block, :link_url_block,
+                                         :username_include_symbol, :suppress_lists, :suppress_no_follow, :url_entities,
+                                         :invisible_tag_attrs, :symbol_tag, :text_with_symbol_tag, :url_target, :target_blank,
+                                         :link_attribute_block, :link_text_block
+                                       ]).freeze
+
+      def extract_html_attrs_from_options!(options)
+        html_attrs = {}
+        options.reject! do |key, value|
+          unless OPTIONS_NOT_ATTRIBUTES.include?(key)
+            html_attrs[key] = value
+            true
+          end
+        end
+        html_attrs
+      end
+
+      def url_entities_hash(url_entities)
+        (url_entities || {}).inject({}) do |entities, entity|
+          # be careful not to alter arguments received
+          _entity = HashHelper.symbolize_keys(entity)
+          entities[_entity[:url]] = _entity
+          entities
+        end
+      end
+
+      def link_to_url(entity, chars, options = {})
+        url = entity[:url]
+
+        href = if options[:link_url_block]
+                 options[:link_url_block].call(url)
+               else
+                 url
+               end
+
+        # NOTE auto link to urls do not use any default values and options
+        # like url_class but use suppress_no_follow.
+        html_attrs = options[:html_attrs].dup
+        html_attrs[:class] = options[:url_class] if options.key?(:url_class)
+
+        # add target attribute only if :url_target is specified
+        html_attrs[:target] = options[:url_target] if options.key?(:url_target)
+
+        url_entities = url_entities_hash(options[:url_entities])
+
+        # use entity from urlEntities if available
+        url_entity = url_entities[url] || entity
+        link_text = if url_entity[:display_url]
+                      html_attrs[:title] ||= url_entity[:expanded_url]
+                      link_url_with_entity(url_entity, options)
+                    else
+                      html_escape(url)
+                    end
+
+        link_to_text(entity, link_text, href, html_attrs, options)
+      end
+
+      def link_url_with_entity(entity, options)
+        display_url = entity[:display_url]
+        expanded_url = entity[:expanded_url]
+        invisible_tag_attrs = options[:invisible_tag_attrs] || DEFAULT_INVISIBLE_TAG_ATTRS
+
+        # Goal: If a user copies and pastes a tweet containing t.co'ed link, the resulting paste
+        # should contain the full original URL (expanded_url), not the display URL.
+        #
+        # Method: Whenever possible, we actually emit HTML that contains expanded_url, and use
+        # font-size:0 to hide those parts that should not be displayed (because they are not part of display_url).
+        # Elements with font-size:0 get copied even though they are not visible.
+        # Note that display:none doesn't work here. Elements with display:none don't get copied.
+        #
+        # Additionally, we want to *display* ellipses, but we don't want them copied.  To make this happen we
+        # wrap the ellipses in a tco-ellipsis class and provide an onCopy handler that sets display:none on
+        # everything with the tco-ellipsis class.
+        #
+        # Exception: pic.twitter.com images, for which expandedUrl = "https://twitter.com/username/status/1234/photo/1
+        # For those URLs, display_url is not a substring of expanded_url, so we don't do anything special to render the elided parts.
+        # For a pic.twitter.com URL, the only elided part will be the "https://", so this is fine.
+        display_url_sans_ellipses = display_url.gsub("…", "")
+
+        if expanded_url.include?(display_url_sans_ellipses)
+          before_display_url, after_display_url = expanded_url.split(display_url_sans_ellipses, 2)
+          preceding_ellipsis = /\A…/.match(display_url).to_s
+          following_ellipsis = /…\z/.match(display_url).to_s
+
+          # As an example: The user tweets "hi http://longdomainname.com/foo"
+          # This gets shortened to "hi http://t.co/xyzabc", with display_url = "…nname.com/foo"
+          # This will get rendered as:
+          # <span class='tco-ellipsis'> <!-- This stuff should get displayed but not copied -->
+          #   …
+          #   <!-- There's a chance the onCopy event handler might not fire. In case that happens,
+          #        we include an &nbsp; here so that the … doesn't bump up against the URL and ruin it.
+          #        The &nbsp; is inside the tco-ellipsis span so that when the onCopy handler *does*
+          #        fire, it doesn't get copied.  Otherwise the copied text would have two spaces in a row,
+          #        e.g. "hi  http://longdomainname.com/foo".
+          #   <span style='font-size:0'>&nbsp;</span>
+          # </span>
+          # <span style='font-size:0'>  <!-- This stuff should get copied but not displayed -->
+          #   http://longdomai
+          # </span>
+          # <span class='js-display-url'> <!-- This stuff should get displayed *and* copied -->
+          #   nname.com/foo
+          # </span>
+          # <span class='tco-ellipsis'> <!-- This stuff should get displayed but not copied -->
+          #   <span style='font-size:0'>&nbsp;</span>
+          #   …
+          # </span>
+          %(<span class="tco-ellipsis">#{preceding_ellipsis}<span #{invisible_tag_attrs}>&nbsp;</span></span>) <<
+            %(<span #{invisible_tag_attrs}>#{html_escape(before_display_url)}</span>) <<
+            %(<span class="js-display-url">#{html_escape(display_url_sans_ellipses)}</span>) <<
+            %(<span #{invisible_tag_attrs}>#{html_escape(after_display_url)}</span>) <<
+            %(<span class="tco-ellipsis"><span #{invisible_tag_attrs}>&nbsp;</span>#{following_ellipsis}</span>)
+        else
+          html_escape(display_url)
+        end
+      end
+
+      def link_to_hashtag(entity, chars, options = {})
+        hash = chars[entity[:indices].first]
+        hashtag = entity[:hashtag]
+        hashtag = yield(hashtag) if block_given?
+        hashtag_class = options[:hashtag_class].to_s
+
+        if hashtag.match Twitter::TwitterText::Regex::REGEXEN[:rtl_chars]
+          hashtag_class += ' rtl'
+        end
+
+        href = if options[:hashtag_url_block]
+                 options[:hashtag_url_block].call(hashtag)
+               else
+                 "#{options[:hashtag_url_base]}#{hashtag}"
+               end
+
+        html_attrs = {
+          :class => hashtag_class,
+          # FIXME As our conformance test, hash in title should be half-width,
+          # this should be bug of conformance data.
+          :title => "##{hashtag}"
+        }.merge(options[:html_attrs])
+
+        link_to_text_with_symbol(entity, hash, hashtag, href, html_attrs, options)
+      end
+
+      def link_to_cashtag(entity, chars, options = {})
+        dollar = chars[entity[:indices].first]
+        cashtag = entity[:cashtag]
+        cashtag = yield(cashtag) if block_given?
+
+        href = if options[:cashtag_url_block]
+                 options[:cashtag_url_block].call(cashtag)
+               else
+                 "#{options[:cashtag_url_base]}#{cashtag}"
+               end
+
+        html_attrs = {
+          :class => "#{options[:cashtag_class]}",
+          :title => "$#{cashtag}"
+        }.merge(options[:html_attrs])
+
+        link_to_text_with_symbol(entity, dollar, cashtag, href, html_attrs, options)
+      end
+
+      def link_to_screen_name(entity, chars, options = {})
+        name  = "#{entity[:screen_name]}#{entity[:list_slug]}"
+
+        chunk = name.dup
+        chunk = yield(chunk) if block_given?
+
+        at = chars[entity[:indices].first]
+
+        html_attrs = options[:html_attrs].dup
+
+        if entity[:list_slug] && !entity[:list_slug].empty? && !options[:suppress_lists]
+          href = if options[:list_url_block]
+                   options[:list_url_block].call(name)
+                 else
+                   "#{options[:list_url_base]}#{name}"
+                 end
+          html_attrs[:class] ||= "#{options[:list_class]}"
+        else
+          href = if options[:username_url_block]
+                   options[:username_url_block].call(chunk)
+                 else
+                   "#{options[:username_url_base]}#{name}"
+                 end
+          html_attrs[:class] ||= "#{options[:username_class]}"
+        end
+
+        link_to_text_with_symbol(entity, at, chunk, href, html_attrs, options)
+      end
+
+      def link_to_text_with_symbol(entity, symbol, text, href, attributes = {}, options = {})
+        tagged_symbol = options[:symbol_tag] ? "<#{options[:symbol_tag]}>#{symbol}</#{options[:symbol_tag]}>" : symbol
+        text = html_escape(text)
+        tagged_text = options[:text_with_symbol_tag] ? "<#{options[:text_with_symbol_tag]}>#{text}</#{options[:text_with_symbol_tag]}>" : text
+        if options[:username_include_symbol] || symbol !~ Twitter::TwitterText::Regex::REGEXEN[:at_signs]
+          "#{link_to_text(entity, tagged_symbol + tagged_text, href, attributes, options)}"
+        else
+          "#{tagged_symbol}#{link_to_text(entity, tagged_text, href, attributes, options)}"
+        end
+      end
+
+      def link_to_text(entity, text, href, attributes = {}, options = {})
+        attributes[:href] = href
+        options[:link_attribute_block].call(entity, attributes) if options[:link_attribute_block]
+        text = options[:link_text_block].call(entity, text) if options[:link_text_block]
+        %(<a#{tag_attrs(attributes)}>#{text}</a>)
+      end
+
+      BOOLEAN_ATTRIBUTES = Set.new([:disabled, :readonly, :multiple, :checked]).freeze
+
+      def tag_attrs(attributes)
+        attributes.keys.sort_by{|k| k.to_s}.inject("") do |attrs, key|
+          value = attributes[key]
+
+          if BOOLEAN_ATTRIBUTES.include?(key)
+            value = value ? key : nil
+          end
+
+          unless value.nil?
+            value = case value
+                    when Array
+                      value.compact.join(" ")
+                    else
+                      value
+                    end
+            attrs << %( #{html_escape(key)}="#{html_escape(value)}")
+          end
+
+          attrs
+        end
+      end
+    end
+  end
+end

data/lib/twitter-text/configuration.rb

@@ -0,0 +1,68 @@
+# Copyright 2018 Twitter, Inc.
+# Licensed under the Apache License, Version 2.0
+# http://www.apache.org/licenses/LICENSE-2.0
+
+# encoding: UTF-8
+
+module Twitter
+  module TwitterText
+    class Configuration
+      require 'json'
+
+      PARSER_VERSION_CLASSIC = "v1"
+      PARSER_VERSION_WEIGHTED = "v2"
+      PARSER_VERSION_EMOJI_PARSING = "v3"
+
+      PARSER_VERSION_DEFAULT = PARSER_VERSION_WEIGHTED
+
+      class << self
+        attr_accessor :default_configuration
+      end
+
+      attr_reader :version, :max_weighted_tweet_length, :scale
+      attr_reader :default_weight, :transformed_url_length, :ranges
+      attr_reader :emoji_parsing_enabled
+
+      CONFIG_V1 = File.join(
+        File.expand_path('../../../config', __FILE__), # project root
+        "#{PARSER_VERSION_CLASSIC}.json"
+      )
+
+      CONFIG_V2 = File.join(
+        File.expand_path('../../../config', __FILE__), # project root
+        "#{PARSER_VERSION_WEIGHTED}.json"
+      )
+
+      CONFIG_V3 = File.join(
+        File.expand_path('../../../config', __FILE__), # project root
+        "#{PARSER_VERSION_EMOJI_PARSING}.json"
+      )
+
+      def self.parse_string(string, options = {})
+        JSON.parse(string, options.merge(symbolize_names: true))
+      end
+
+      def self.parse_file(filename)
+        string = File.open(filename, 'rb') { |f| f.read }
+        parse_string(string)
+      end
+
+      def self.configuration_from_file(filename)
+        config = parse_file(filename)
+        config ? self.new(config) : nil
+      end
+
+      def initialize(config = {})
+        @version = config[:version]
+        @max_weighted_tweet_length = config[:maxWeightedTweetLength]
+        @scale = config[:scale]
+        @default_weight = config[:defaultWeight]
+        @transformed_url_length = config[:transformedURLLength]
+        @emoji_parsing_enabled = config[:emojiParsingEnabled]
+        @ranges = config[:ranges].map { |range| Twitter::TwitterText::WeightedRange.new(range) } if config.key?(:ranges) && config[:ranges].is_a?(Array)
+      end
+
+      self.default_configuration = self.configuration_from_file(CONFIG_V3)
+    end
+  end
+end

data/lib/twitter-text/deprecation.rb

@@ -0,0 +1,21 @@
+# Copyright 2018 Twitter, Inc.
+# Licensed under the Apache License, Version 2.0
+# http://www.apache.org/licenses/LICENSE-2.0
+
+module Twitter
+  module TwitterText
+    module Deprecation
+      def deprecate(method, new_method = nil)
+        deprecated_method = :"deprecated_#{method}"
+        message = "Deprecation: `#{method}` is deprecated."
+        message << " Please use `#{new_method}` instead." if new_method
+
+        alias_method(deprecated_method, method)
+        define_method method do |*args, &block|
+          warn message unless $TESTING
+          send(deprecated_method, *args, &block)
+        end
+      end
+    end
+  end
+end

data/lib/twitter-text/emoji_regex.rb

@@ -0,0 +1,27 @@
+#
+#  emoji_regex.rb
+#
+#  Copyright © 2018 Twitter. All rights reserved.
+#
+# DO NOT MODIFY THIS FILE -- it is generated for twitter-text automatically
+
+# encoding: utf-8
+
+module Twitter
+  module TwitterText
+    class Regex
+      class Emoji
+        REGEXEN = {} # :nodoc:
+
+        # This regex pattern matches a single emoji
+        REGEXEN[:valid_emoji] = %r{
+    (?:\u{01f468}\u{01f3fb}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fc}-\u{01f3ff}]|\u{01f468}\u{01f3fc}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}\u{01f3fd}-\u{01f3ff}]|\u{01f468}\u{01f3fd}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}\u{01f3fc}\u{01f3fe}\u{01f3ff}]|\u{01f468}\u{01f3fe}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}-\u{01f3fd}\u{01f3ff}]|\u{01f468}\u{01f3ff}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}-\u{01f3fe}]|\u{01f469}\u{01f3fb}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fc}-\u{01f3ff}]|\u{01f469}\u{01f3fb}\u200d\u{01f91d}\u200d\u{01f469}[\u{01f3fc}-\u{01f3ff}]|\u{01f469}\u{01f3fc}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}\u{01f3fd}-\u{01f3ff}]|\u{01f469}\u{01f3fc}\u200d\u{01f91d}\u200d\u{01f469}[\u{01f3fb}\u{01f3fd}-\u{01f3ff}]|\u{01f469}\u{01f3fd}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}\u{01f3fc}\u{01f3fe}\u{01f3ff}]|\u{01f469}\u{01f3fd}\u200d\u{01f91d}\u200d\u{01f469}[\u{01f3fb}\u{01f3fc}\u{01f3fe}\u{01f3ff}]|\u{01f469}\u{01f3fe}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}-\u{01f3fd}\u{01f3ff}]|\u{01f469}\u{01f3fe}\u200d\u{01f91d}\u200d\u{01f469}[\u{01f3fb}-\u{01f3fd}\u{01f3ff}]|\u{01f469}\u{01f3ff}\u200d\u{01f91d}\u200d\u{01f468}[\u{01f3fb}-\u{01f3fe}]|\u{01f469}\u{01f3ff}\u200d\u{01f91d}\u200d\u{01f469}[\u{01f3fb}-\u{01f3fe}]|\u{01f9d1}\u{01f3fb}\u200d\u{01f91d}\u200d\u{01f9d1}[\u{01f3fb}-\u{01f3ff}]|\u{01f9d1}\u{01f3fc}\u200d\u{01f91d}\u200d\u{01f9d1}[\u{01f3fb}-\u{01f3ff}]|\u{01f9d1}\u{01f3fd}\u200d\u{01f91d}\u200d\u{01f9d1}[\u{01f3fb}-\u{01f3ff}]|\u{01f9d1}\u{01f3fe}\u200d\u{01f91d}\u200d\u{01f9d1}[\u{01f3fb}-\u{01f3ff}]|\u{01f9d1}\u{01f3ff}\u200d\u{01f91d}\u200d\u{01f9d1}[\u{01f3fb}-\u{01f3ff}]|\u{01f9d1}\u200d\u{01f91d}\u200d\u{01f9d1}|\u{01f46b}[\u{01f3fb}-\u{01f3ff}]|\u{01f46c}[\u{01f3fb}-\u{01f3ff}]|\u{01f46d}[\u{01f3fb}-\u{01f3ff}]|[\u{01f46b}-\u{01f46d}])|[\u{01f468}\u{01f469}\u{01f9d1}][\u{01f3fb}-\u{01f3ff}]?\u200d(?:\u2695\ufe0f|\u2696\ufe0f|\u2708\ufe0f|[\u{01f33e}\u{01f373}\u{01f393}\u{01f3a4}\u{01f3a8}\u{01f3eb}\u{01f3ed}\u{01f4bb}\u{01f4bc}\u{01f527}\u{01f52c}\u{01f680}\u{01f692}\u{01f9af}-\u{01f9b3}\u{01f9bc}\u{01f9bd}])|[\u26f9\u{01f3cb}\u{01f3cc}\u{01f574}\u{01f575}](?:[\ufe0f\u{01f3fb}-\u{01f3ff}]\u200d[\u2640\u2642]\ufe0f)|[\u{01f3c3}\u{01f3c4}\u{01f3ca}\u{01f46e}\u{01f471}\u{01f473}\u{01f477}\u{01f481}\u{01f482}\u{01f486}\u{01f487}\u{01f645}-\u{01f647}\u{01f64b}\u{01f64d}\u{01f64e}\u{01f6a3}\u{01f6b4}-\u{01f6b6}\u{01f926}\u{01f935}\u{01f937}-\u{01f939}\u{01f93d}\u{01f93e}\u{01f9b8}\u{01f9b9}\u{01f9cd}-\u{01f9cf}\u{01f9d6}-\u{01f9dd}][\u{01f3fb}-\u{01f3ff}]?\u200d[\u2640\u2642]\ufe0f|(?:\u{01f468}\u200d\u2764\ufe0f\u200d\u{01f48b}\u200d\u{01f468}|\u{01f469}\u200d\u2764\ufe0f\u200d\u{01f48b}\u200d[\u{01f468}\u{01f469}]|\u{01f468}\u200d\u{01f468}\u200d\u{01f466}\u200d\u{01f466}|\u{01f468}\u200d\u{01f468}\u200d\u{01f467}\u200d[\u{01f466}\u{01f467}]|\u{01f468}\u200d\u{01f469}\u200d\u{01f466}\u200d\u{01f466}|\u{01f468}\u200d\u{01f469}\u200d\u{01f467}\u200d[\u{01f466}\u{01f467}]|\u{01f469}\u200d\u{01f469}\u200d\u{01f466}\u200d\u{01f466}|\u{01f469}\u200d\u{01f469}\u200d\u{01f467}\u200d[\u{01f466}\u{01f467}]|\u{01f468}\u200d\u2764\ufe0f\u200d\u{01f468}|\u{01f469}\u200d\u2764\ufe0f\u200d[\u{01f468}\u{01f469}]|\u{01f3f3}\ufe0f\u200d\u26a7\ufe0f|\u{01f468}\u200d\u{01f466}\u200d\u{01f466}|\u{01f468}\u200d\u{01f467}\u200d[\u{01f466}\u{01f467}]|\u{01f468}\u200d\u{01f468}\u200d[\u{01f466}\u{01f467}]|\u{01f468}\u200d\u{01f469}\u200d[\u{01f466}\u{01f467}]|\u{01f469}\u200d\u{01f466}\u200d\u{01f466}|\u{01f469}\u200d\u{01f467}\u200d[\u{01f466}\u{01f467}]|\u{01f469}\u200d\u{01f469}\u200d[\u{01f466}\u{01f467}]|\u{01f3f3}\ufe0f\u200d\u{01f308}|\u{01f3f4}\u200d\u2620\ufe0f|\u{01f46f}\u200d\u2640\ufe0f|\u{01f46f}\u200d\u2642\ufe0f|\u{01f93c}\u200d\u2640\ufe0f|\u{01f93c}\u200d\u2642\ufe0f|\u{01f9de}\u200d\u2640\ufe0f|\u{01f9de}\u200d\u2642\ufe0f|\u{01f9df}\u200d\u2640\ufe0f|\u{01f9df}\u200d\u2642\ufe0f|\u{01f415}\u200d\u{01f9ba}|\u{01f441}\u200d\u{01f5e8}|\u{01f468}\u200d[\u{01f466}\u{01f467}]|\u{01f469}\u200d[\u{01f466}\u{01f467}])|[#*0-9]\ufe0f?\u20e3|(?:[©®\u2122\u265f]\ufe0f)|[\u203c\u2049\u2139\u2194-\u2199\u21a9\u21aa\u231a\u231b\u2328\u23cf\u23ed-\u23ef\u23f1\u23f2\u23f8-\u23fa\u24c2\u25aa\u25ab\u25b6\u25c0\u25fb-\u25fe\u2600-\u2604\u260e\u2611\u2614\u2615\u2618\u2620\u2622\u2623\u2626\u262a\u262e\u262f\u2638-\u263a\u2640\u2642\u2648-\u2653\u2660\u2663\u2665\u2666\u2668\u267b\u267f\u2692-\u2697\u2699\u269b\u269c\u26a0\u26a1\u26a7\u26aa\u26ab\u26b0\u26b1\u26bd\u26be\u26c4\u26c5\u26c8\u26cf\u26d1\u26d3\u26d4\u26e9\u26ea\u26f0-\u26f5\u26f8\u26fa\u26fd\u2702\u2708\u2709\u270f\u2712\u2714\u2716\u271d\u2721\u2733\u2734\u2744\u2747\u2757\u2763\u2764\u27a1\u2934\u2935\u2b05-\u2b07\u2b1b\u2b1c\u2b50\u2b55\u3030\u303d\u3297\u3299\u{01f004}\u{01f170}\u{01f171}\u{01f17e}\u{01f17f}\u{01f202}\u{01f21a}\u{01f22f}\u{01f237}\u{01f321}\u{01f324}-\u{01f32c}\u{01f336}\u{01f37d}\u{01f396}\u{01f397}\u{01f399}-\u{01f39b}\u{01f39e}\u{01f39f}\u{01f3cd}\u{01f3ce}\u{01f3d4}-\u{01f3df}\u{01f3f3}\u{01f3f5}\u{01f3f7}\u{01f43f}\u{01f441}\u{01f4fd}\u{01f549}\u{01f54a}\u{01f56f}\u{01f570}\u{01f573}\u{01f576}-\u{01f579}\u{01f587}\u{01f58a}-\u{01f58d}\u{01f5a5}\u{01f5a8}\u{01f5b1}\u{01f5b2}\u{01f5bc}\u{01f5c2}-\u{01f5c4}\u{01f5d1}-\u{01f5d3}\u{01f5dc}-\u{01f5de}\u{01f5e1}\u{01f5e3}\u{01f5e8}\u{01f5ef}\u{01f5f3}\u{01f5fa}\u{01f6cb}\u{01f6cd}-\u{01f6cf}\u{01f6e0}-\u{01f6e5}\u{01f6e9}\u{01f6f0}\u{01f6f3}](?:\ufe0f|(?!\ufe0e))|(?:[\u261d\u26f7\u26f9\u270c\u270d\u{01f3cb}\u{01f3cc}\u{01f574}\u{01f575}\u{01f590}](?:\ufe0f|(?!\ufe0e))|[\u270a\u270b\u{01f385}\u{01f3c2}-\u{01f3c4}\u{01f3c7}\u{01f3ca}\u{01f442}\u{01f443}\u{01f446}-\u{01f450}\u{01f466}-\u{01f469}\u{01f46e}\u{01f470}-\u{01f478}\u{01f47c}\u{01f481}-\u{01f483}\u{01f485}-\u{01f487}\u{01f4aa}\u{01f57a}\u{01f595}\u{01f596}\u{01f645}-\u{01f647}\u{01f64b}-\u{01f64f}\u{01f6a3}\u{01f6b4}-\u{01f6b6}\u{01f6c0}\u{01f6cc}\u{01f90f}\u{01f918}-\u{01f91c}\u{01f91e}\u{01f91f}\u{01f926}\u{01f930}-\u{01f939}\u{01f93d}\u{01f93e}\u{01f9b5}\u{01f9b6}\u{01f9b8}\u{01f9b9}\u{01f9bb}\u{01f9cd}-\u{01f9cf}\u{01f9d1}-\u{01f9dd}])[\u{01f3fb}-\u{01f3ff}]?|(?:\u{01f3f4}\u{0e0067}\u{0e0062}\u{0e0065}\u{0e006e}\u{0e0067}\u{0e007f}|\u{01f3f4}\u{0e0067}\u{0e0062}\u{0e0073}\u{0e0063}\u{0e0074}\u{0e007f}|\u{01f3f4}\u{0e0067}\u{0e0062}\u{0e0077}\u{0e006c}\u{0e0073}\u{0e007f}|\u{01f1e6}[\u{01f1e8}-\u{01f1ec}\u{01f1ee}\u{01f1f1}\u{01f1f2}\u{01f1f4}\u{01f1f6}-\u{01f1fa}\u{01f1fc}\u{01f1fd}\u{01f1ff}]|\u{01f1e7}[\u{01f1e6}\u{01f1e7}\u{01f1e9}-\u{01f1ef}\u{01f1f1}-\u{01f1f4}\u{01f1f6}-\u{01f1f9}\u{01f1fb}\u{01f1fc}\u{01f1fe}\u{01f1ff}]|\u{01f1e8}[\u{01f1e6}\u{01f1e8}\u{01f1e9}\u{01f1eb}-\u{01f1ee}\u{01f1f0}-\u{01f1f5}\u{01f1f7}\u{01f1fa}-\u{01f1ff}]|\u{01f1e9}[\u{01f1ea}\u{01f1ec}\u{01f1ef}\u{01f1f0}\u{01f1f2}\u{01f1f4}\u{01f1ff}]|\u{01f1ea}[\u{01f1e6}\u{01f1e8}\u{01f1ea}\u{01f1ec}\u{01f1ed}\u{01f1f7}-\u{01f1fa}]|\u{01f1eb}[\u{01f1ee}-\u{01f1f0}\u{01f1f2}\u{01f1f4}\u{01f1f7}]|\u{01f1ec}[\u{01f1e6}\u{01f1e7}\u{01f1e9}-\u{01f1ee}\u{01f1f1}-\u{01f1f3}\u{01f1f5}-\u{01f1fa}\u{01f1fc}\u{01f1fe}]|\u{01f1ed}[\u{01f1f0}\u{01f1f2}\u{01f1f3}\u{01f1f7}\u{01f1f9}\u{01f1fa}]|\u{01f1ee}[\u{01f1e8}-\u{01f1ea}\u{01f1f1}-\u{01f1f4}\u{01f1f6}-\u{01f1f9}]|\u{01f1ef}[\u{01f1ea}\u{01f1f2}\u{01f1f4}\u{01f1f5}]|\u{01f1f0}[\u{01f1ea}\u{01f1ec}-\u{01f1ee}\u{01f1f2}\u{01f1f3}\u{01f1f5}\u{01f1f7}\u{01f1fc}\u{01f1fe}\u{01f1ff}]|\u{01f1f1}[\u{01f1e6}-\u{01f1e8}\u{01f1ee}\u{01f1f0}\u{01f1f7}-\u{01f1fb}\u{01f1fe}]|\u{01f1f2}[\u{01f1e6}\u{01f1e8}-\u{01f1ed}\u{01f1f0}-\u{01f1ff}]|\u{01f1f3}[\u{01f1e6}\u{01f1e8}\u{01f1ea}-\u{01f1ec}\u{01f1ee}\u{01f1f1}\u{01f1f4}\u{01f1f5}\u{01f1f7}\u{01f1fa}\u{01f1ff}]|\u{01f1f4}\u{01f1f2}|\u{01f1f5}[\u{01f1e6}\u{01f1ea}-\u{01f1ed}\u{01f1f0}-\u{01f1f3}\u{01f1f7}-\u{01f1f9}\u{01f1fc}\u{01f1fe}]|\u{01f1f6}\u{01f1e6}|\u{01f1f7}[\u{01f1ea}\u{01f1f4}\u{01f1f8}\u{01f1fa}\u{01f1fc}]|\u{01f1f8}[\u{01f1e6}-\u{01f1ea}\u{01f1ec}-\u{01f1f4}\u{01f1f7}-\u{01f1f9}\u{01f1fb}\u{01f1fd}-\u{01f1ff}]|\u{01f1f9}[\u{01f1e6}\u{01f1e8}\u{01f1e9}\u{01f1eb}-\u{01f1ed}\u{01f1ef}-\u{01f1f4}\u{01f1f7}\u{01f1f9}\u{01f1fb}\u{01f1fc}\u{01f1ff}]|\u{01f1fa}[\u{01f1e6}\u{01f1ec}\u{01f1f2}\u{01f1f3}\u{01f1f8}\u{01f1fe}\u{01f1ff}]|\u{01f1fb}[\u{01f1e6}\u{01f1e8}\u{01f1ea}\u{01f1ec}\u{01f1ee}\u{01f1f3}\u{01f1fa}]|\u{01f1fc}[\u{01f1eb}\u{01f1f8}]|\u{01f1fd}\u{01f1f0}|\u{01f1fe}[\u{01f1ea}\u{01f1f9}]|\u{01f1ff}[\u{01f1e6}\u{01f1f2}\u{01f1fc}]|[\u23e9-\u23ec\u23f0\u23f3\u267e\u26ce\u2705\u2728\u274c\u274e\u2753-\u2755\u2795-\u2797\u27b0\u27bf\ue50a\u{01f0cf}\u{01f18e}\u{01f191}-\u{01f19a}\u{01f1e6}-\u{01f1ff}\u{01f201}\u{01f232}-\u{01f236}\u{01f238}-\u{01f23a}\u{01f250}\u{01f251}\u{01f300}-\u{01f320}\u{01f32d}-\u{01f335}\u{01f337}-\u{01f37c}\u{01f37e}-\u{01f384}\u{01f386}-\u{01f393}\u{01f3a0}-\u{01f3c1}\u{01f3c5}\u{01f3c6}\u{01f3c8}\u{01f3c9}\u{01f3cf}-\u{01f3d3}\u{01f3e0}-\u{01f3f0}\u{01f3f4}\u{01f3f8}-\u{01f43e}\u{01f440}\u{01f444}\u{01f445}\u{01f451}-\u{01f465}\u{01f46a}\u{01f46f}\u{01f479}-\u{01f47b}\u{01f47d}-\u{01f480}\u{01f484}\u{01f488}-\u{01f4a9}\u{01f4ab}-\u{01f4fc}\u{01f4ff}-\u{01f53d}\u{01f54b}-\u{01f54e}\u{01f550}-\u{01f567}\u{01f5a4}\u{01f5fb}-\u{01f644}\u{01f648}-\u{01f64a}\u{01f680}-\u{01f6a2}\u{01f6a4}-\u{01f6b3}\u{01f6b7}-\u{01f6bf}\u{01f6c1}-\u{01f6c5}\u{01f6d0}-\u{01f6d2}\u{01f6d5}\u{01f6eb}\u{01f6ec}\u{01f6f4}-\u{01f6fa}\u{01f7e0}-\u{01f7eb}\u{01f90d}\u{01f90e}\u{01f910}-\u{01f917}\u{01f91d}\u{01f920}-\u{01f925}\u{01f927}-\u{01f92f}\u{01f93a}\u{01f93c}\u{01f93f}-\u{01f945}\u{01f947}-\u{01f971}\u{01f973}-\u{01f976}\u{01f97a}-\u{01f9a2}\u{01f9a5}-\u{01f9aa}\u{01f9ae}-\u{01f9b4}\u{01f9b7}\u{01f9ba}\u{01f9bc}-\u{01f9ca}\u{01f9d0}\u{01f9de}-\u{01f9ff}\u{01fa70}-\u{01fa73}\u{01fa78}-\u{01fa7a}\u{01fa80}-\u{01fa82}\u{01fa90}-\u{01fa95}])|\ufe0f
+  }iox
+
+        def self.[](key)
+          REGEXEN[key]
+        end
+      end
+    end
+  end
+end