govspeak 6.5.3 → 6.5.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 753ae3784f7f2ea1afcba0ab048130ea81a12d18e6b83297c0044a029b9a7f96
-  data.tar.gz: 66b4612a2f7f821e94b5e0b5bc1a2471e40cc171506337f4c76dd3c95c087921
+  metadata.gz: 26ef4d3bfb3ed9dcf07d44a1428e815928a4fa1ee94f388de829bb999b3cd79c
+  data.tar.gz: c2982f083911dcaf38a0883d48e9ca302a3b514b7bc05b99b827830c02db884d
 SHA512:
-  metadata.gz: 979a24087e5278d5c5c045f02bbd65bd2b9265ba255af4d00de3697d44af9024330973d477177aacf117ac3cfa5210c209bc545117d4a1731e9b0b2cdfd4fbb9
-  data.tar.gz: b6e1a3e1d72f7c83b747cdc204b32efc7824dbea550cd9bdc3d965cd685952ff1dee44066dd9d3215c04f050abe8647b30584a3fca7d43c181a3d2eed2ae4b9b
+  metadata.gz: f75532f9905ae86df812be38a837d60abe1007cad72e8d886c95019f2563734a9e136368b74c1d1499d41bc8c83b2c3a41f415128c93c7025c4f4aacb9765e6f
+  data.tar.gz: e6ef52ca611f2b3e74e77a3330ddaf602385eb90685b3839c8bd28f4bdf06c21652c50cff4a85fc2b157e8e009e11c5ce9c7c439a92d5abdf78821725fbaba2c

data/CHANGELOG.md

@@ -1,3 +1,23 @@
+## 6.5.8
+
+* Customise footnote markup for accessibility (PR#192)
+
+## 6.5.7
+
+* Update GOV.UK Publishing components to version to 23.0.0 or greater
+
+## 6.5.6
+
+* Update Kramdown version to 2.3.0 or greater
+
+## 6.5.5
+
+* Prevent links in table headers being stripped (PR#187)
+
+## 6.5.4
+
+* Require Sanitize to be at least 5.2.1 due to https://nvd.nist.gov/vuln/detail/CVE-2020-4054
+
 ## 6.5.3
 
 * Use button component for buttons (PR#176)

data/Rakefile

@@ -1,15 +1,18 @@
 require "rake"
 require "rake/testtask"
+require "rubocop/rake_task"
 require "bundler"
 
 Bundler::GemHelper.install_tasks
 
+RuboCop::RakeTask.new
+
 desc "Run basic tests"
-Rake::TestTask.new("test") { |t|
+Rake::TestTask.new("test") do |t|
   t.libs << "test"
   t.pattern = "test/*_test.rb"
   t.verbose = true
   t.warning = true
-}
+end
 
-task default: [:test]
+task default: %i[test rubocop]

data/lib/govspeak.rb

@@ -57,7 +57,9 @@ module Govspeak
       @links = Array.wrap(options.delete(:links))
       @contacts = Array.wrap(options.delete(:contacts))
       @locale = options.fetch(:locale, "en")
-      @options = { input: PARSER_CLASS_NAME, sanitize: true }.merge(options)
+      @options = { input: PARSER_CLASS_NAME,
+                   sanitize: true,
+                   syntax_highlighter: nil }.merge(options)
       @options[:entity_output] = :symbolic
     end
 
@@ -108,9 +110,9 @@ module Govspeak
       source = Govspeak::BlockquoteExtraQuoteRemover.remove(source)
       source = remove_forbidden_characters(source)
       self.class.extensions.each do |_, regexp, block|
-        source.gsub!(regexp) {
+        source.gsub!(regexp) do
           instance_exec(*Regexp.last_match.captures, &block)
-        }
+        end
       end
       source
     end
@@ -127,20 +129,20 @@ module Govspeak
     end
 
     def self.surrounded_by(open, close = nil)
-      open = Regexp::escape(open)
+      open = Regexp.escape(open)
       if close
-        close = Regexp::escape(close)
-        %r+(?:\r|\n|^)#{open}(.*?)#{close} *(\r|\n|$)?+m
+        close = Regexp.escape(close)
+        %r{(?:\r|\n|^)#{open}(.*?)#{close} *(\r|\n|$)?}m
       else
-        %r+(?:\r|\n|^)#{open}(.*?)#{open}? *(\r|\n|$)+m
+        %r{(?:\r|\n|^)#{open}(.*?)#{open}? *(\r|\n|$)}m
       end
     end
 
     def self.wrap_with_div(class_name, character, parser = Kramdown::Document)
-      extension(class_name, surrounded_by(character)) { |body|
+      extension(class_name, surrounded_by(character)) do |body|
         content = parser ? parser.new("#{body.strip}\n").to_html : body.strip
-        %{\n<div class="#{class_name}">\n#{content}</div>\n}
-      }
+        %(\n<div class="#{class_name}">\n#{content}</div>\n)
+      end
     end
 
     def insert_strong_inside_p(body, parser = Govspeak::Document)
@@ -160,7 +162,7 @@ module Govspeak
         \s*  # any whitespace between opening bracket and link
       {\/button} # match ending bracket
       (?:\r|\n|$) # non-capturing match to make sure end of line and linebreak
-    }x) { |attributes, text, href|
+    }x) do |attributes, text, href|
       button_classes = "govuk-button"
       /cross-domain-tracking:(?<cross_domain_tracking>.[^\s*]+)/ =~ attributes
       data_attribute = ""
@@ -173,36 +175,36 @@ module Govspeak
       text = text.strip
       href = href.strip
 
-      %{\n<a role="button" class="#{button_classes}" href="#{href}" #{data_attribute}>#{text}</a>\n}
-    }
+      %(\n<a role="button" class="#{button_classes}" href="#{href}" #{data_attribute}>#{text}</a>\n)
+    end
 
-    extension("highlight-answer") { |body|
-      %{\n\n<div class="highlight-answer">
-#{Govspeak::Document.new(body.strip).to_html}</div>\n}
-    }
+    extension("highlight-answer") do |body|
+      %(\n\n<div class="highlight-answer">
+#{Govspeak::Document.new(body.strip).to_html}</div>\n)
+    end
 
-    extension("stat-headline", %r${stat-headline}(.*?){/stat-headline}$m) { |body|
-      %{\n\n<aside class="stat-headline">
-#{Govspeak::Document.new(body.strip).to_html}</aside>\n}
-    }
+    extension("stat-headline", %r${stat-headline}(.*?){/stat-headline}$m) do |body|
+      %(\n\n<aside class="stat-headline">
+#{Govspeak::Document.new(body.strip).to_html}</aside>\n)
+    end
 
     # FIXME: these surrounded_by arguments look dodgy
-    extension("external", surrounded_by("x[", ")x")) { |body|
+    extension("external", surrounded_by("x[", ")x")) do |body|
       Kramdown::Document.new("[#{body.strip}){:rel='external'}").to_html
-    }
+    end
 
-    extension("informational", surrounded_by("^")) { |body|
-      %{\n\n<div role="note" aria-label="Information" class="application-notice info-notice">
-#{Govspeak::Document.new(body.strip).to_html}</div>\n}
-    }
+    extension("informational", surrounded_by("^")) do |body|
+      %(\n\n<div role="note" aria-label="Information" class="application-notice info-notice">
+#{Govspeak::Document.new(body.strip).to_html}</div>\n)
+    end
 
-    extension("important", surrounded_by("@")) { |body|
-      %{\n\n<div role="note" aria-label="Important" class="advisory">#{insert_strong_inside_p(body)}</div>\n}
-    }
+    extension("important", surrounded_by("@")) do |body|
+      %(\n\n<div role="note" aria-label="Important" class="advisory">#{insert_strong_inside_p(body)}</div>\n)
+    end
 
-    extension("helpful", surrounded_by("%")) { |body|
-      %{\n\n<div role="note" aria-label="Warning" class="application-notice help-notice">\n#{Govspeak::Document.new(body.strip).to_html}</div>\n}
-    }
+    extension("helpful", surrounded_by("%")) do |body|
+      %(\n\n<div role="note" aria-label="Warning" class="application-notice help-notice">\n#{Govspeak::Document.new(body.strip).to_html}</div>\n)
+    end
 
     extension("barchart", /{barchart(.*?)}/) do |captures|
       stacked = ".mc-stacked" if captures.include? "stacked"
@@ -210,13 +212,13 @@ module Govspeak
       negative = ".mc-negative" if captures.include? "negative"
 
       [
-       "{:",
-       ".js-barchart-table",
-       stacked,
-       compact,
-       negative,
-       ".mc-auto-outdent",
-       "}",
+        "{:",
+        ".js-barchart-table",
+        stacked,
+        compact,
+        negative,
+        ".mc-auto-outdent",
+        "}",
       ].join(" ")
     end
 
@@ -234,12 +236,12 @@ module Govspeak
 
       attachment = AttachmentPresenter.new(attachment)
 
-      span_id = attachment.id ? %{ id="attachment_#{attachment.id}"} : ""
+      span_id = attachment.id ? %( id="attachment_#{attachment.id}") : ""
       # new lines inside our title cause problems with govspeak rendering as this is expected to be on one line.
       title = (attachment.title || "").tr("\n", " ")
       link = attachment.link(title, attachment.url)
       attributes = attachment.attachment_attributes.empty? ? "" : " (#{attachment.attachment_attributes})"
-      %{<span#{span_id} class="attachment-inline">#{link}#{attributes}</span>}
+      %(<span#{span_id} class="attachment-inline">#{link}#{attributes}</span>)
     end
 
     # DEPRECATED: use 'Image:image-id' instead
@@ -261,10 +263,10 @@ module Govspeak
     #
     # This issue is not considered a bug by kramdown: https://github.com/gettalong/kramdown/issues/191
     def render_image(image)
-      id_attr = image.id ? %{ id="attachment_#{image.id}"} : ""
+      id_attr = image.id ? %( id="attachment_#{image.id}") : ""
       lines = []
-      lines << %{<figure#{id_attr} class="image embedded">}
-      lines << %{<div class="img"><img src="#{encode(image.url)}" alt="#{encode(image.alt_text)}"></div>}
+      lines << %(<figure#{id_attr} class="image embedded">)
+      lines << %(<div class="img"><img src="#{encode(image.url)}" alt="#{encode(image.alt_text)}"></div>)
       lines << image.figcaption_html if image.figcaption?
       lines << "</figure>"
       lines.join
@@ -279,9 +281,9 @@ module Govspeak
     wrap_with_div("example", "$E", Govspeak::Document)
     wrap_with_div("call-to-action", "$CTA", Govspeak::Document)
 
-    extension("address", surrounded_by("$A")) { |body|
-      %{\n<div class="address"><div class="adr org fn"><p>\n#{body.sub("\n", '').gsub("\n", '<br />')}\n</p></div></div>\n}
-    }
+    extension("address", surrounded_by("$A")) do |body|
+      %(\n<div class="address"><div class="adr org fn"><p>\n#{body.sub("\n", '').gsub("\n", '<br />')}\n</p></div></div>\n)
+    end
 
     extension("legislative list", /#{NEW_PARAGRAPH_LOOKBEHIND}\$LegislativeList\s*$(.*?)\$EndLegislativeList/m) do |body|
       Govspeak::KramdownOverrides.with_kramdown_ordered_lists_disabled do
@@ -295,9 +297,9 @@ module Govspeak
 
     extension("numbered list", /^[ \t]*((s\d+\.\s.*(?:\n|$))+)/) do |body|
       body.gsub!(/s(\d+)\.\s(.*)(?:\n|$)/) do
-        "<li>#{Govspeak::Document.new($2.strip).to_html}</li>\n"
+        "<li>#{Govspeak::Document.new(Regexp.last_match(2).strip).to_html}</li>\n"
       end
-      %{<ol class="steps">\n#{body}</ol>}
+      %(<ol class="steps">\n#{body}</ol>)
     end
 
     def self.devolved_options
@@ -362,13 +364,13 @@ module Govspeak
     extension("Attachment", /#{NEW_PARAGRAPH_LOOKBEHIND}\[Attachment:\s*(.*?)\s*\]/) do |attachment_id|
       next "" if attachments.none? { |a| a[:id] == attachment_id }
 
-      %{<govspeak-embed-attachment id="#{attachment_id}"></govspeak-embed-attachment>}
+      %(<govspeak-embed-attachment id="#{attachment_id}"></govspeak-embed-attachment>)
     end
 
     extension("AttachmentLink", /\[AttachmentLink:\s*(.*?)\s*\]/) do |attachment_id|
       next "" if attachments.none? { |a| a[:id] == attachment_id }
 
-      %{<govspeak-embed-attachment-link id="#{attachment_id}"></govspeak-embed-attachment-link>}
+      %(<govspeak-embed-attachment-link id="#{attachment_id}"></govspeak-embed-attachment-link>)
     end
 
   private

data/lib/govspeak/html_sanitizer.rb

@@ -54,12 +54,12 @@ class Govspeak::HtmlSanitizer
       elements: Sanitize::Config::RELAXED[:elements] + %w[govspeak-embed-attachment govspeak-embed-attachment-link svg path],
       attributes: {
         :all => Sanitize::Config::RELAXED[:attributes][:all] + %w[role aria-label],
-        "a"  => Sanitize::Config::RELAXED[:attributes]["a"] + [:data],
+        "a" => Sanitize::Config::RELAXED[:attributes]["a"] + [:data],
         "svg" => Sanitize::Config::RELAXED[:attributes][:all] + %w[xmlns width height viewbox focusable],
         "path" => Sanitize::Config::RELAXED[:attributes][:all] + %w[fill d],
         "div" => [:data],
-        "th"  => Sanitize::Config::RELAXED[:attributes]["th"] + %w[style],
-        "td"  => Sanitize::Config::RELAXED[:attributes]["td"] + %w[style],
+        "th" => Sanitize::Config::RELAXED[:attributes]["th"] + %w[style],
+        "td" => Sanitize::Config::RELAXED[:attributes]["td"] + %w[style],
         "govspeak-embed-attachment" => %w[content-id],
       },
     )

data/lib/govspeak/kramdown_overrides.rb

@@ -1,9 +1,9 @@
 module Govspeak
   module KramdownOverrides
-    # This depends on two internal parts of Kramdown.
+    # This depends on two internal parts of Kramdown.
     # 1. Parser registry (kramdown/parser/kramdown.rb#define_parser)
     # 2. Kramdown list regexes (kramdown/parser/kramdown/list.rb)
-    # Updating the Kramdown gem therefore also means updating this file to to
+    # Updating the Kramdown gem therefore also means updating this file to to
     # match Kramdown's internals.
 
     def self.with_kramdown_ordered_lists_disabled

data/lib/govspeak/link_extractor.rb

@@ -14,9 +14,9 @@ module Govspeak
     attr_reader :document, :website_root
 
     def extract_links
-      document_anchors.
-        map { |link| extract_href_from_link(link) }.
-        reject(&:blank?)
+      document_anchors
+        .map { |link| extract_href_from_link(link) }
+        .reject(&:blank?)
     end
 
     def extract_href_from_link(link)

data/lib/govspeak/post_processor.rb

@@ -92,11 +92,13 @@ module Govspeak
       end
 
       document.css(":not(thead) tr td:first-child").map do |el|
-        if el.content.match?(/^#($|\s.*$)/)
-          el.content = el.content.gsub(/^#($|\s)/, "") # Replace '# ' and '#', but not '#Word'.
-          el.name = "th" if el.content.present? # This also prevents a `th` with nothing inside it; a `td` is preferable.
-          el[:scope] = "row" if el.content.present? # `scope` shouldn't be used if there's nothing in the table heading.
-        end
+        next unless el.content.match?(/^#($|\s.*$)/)
+
+        # Replace '# ' and '#', but not '#Word'.
+        # This runs on the first child of the element to preserve any links
+        el.children.first.content = el.children.first.content.gsub(/^#($|\s)/, "")
+        el.name = "th" if el.content.present? # This also prevents a `th` with nothing inside it; a `td` is preferable.
+        el[:scope] = "row" if el.content.present? # `scope` shouldn't be used if there's nothing in the table heading.
       end
     end
 
@@ -118,6 +120,16 @@ module Govspeak
       end
     end
 
+    extension("use custom footnotes") do |document|
+      document.css("a.footnote").map do |el|
+        footnote_number = el[:href].gsub(/\D/, "")
+        el.content = "[footnote #{footnote_number}]"
+      end
+      document.css("[role='doc-backlink']").map do |el|
+        el.content = "[go to where this is referenced]"
+      end
+    end
+
     attr_reader :input, :govspeak_document
 
     def initialize(html, govspeak_document)

data/lib/govspeak/presenters/attachment_presenter.rb

@@ -55,46 +55,46 @@ module Govspeak
 
     def humanized_content_type(file_extension)
       file_extension_vs_humanized_content_type = {
-        "chm"  => file_abbr_tag("CHM", "Microsoft Compiled HTML Help"),
-        "csv"  => file_abbr_tag("CSV", "Comma-separated Values"),
+        "chm" => file_abbr_tag("CHM", "Microsoft Compiled HTML Help"),
+        "csv" => file_abbr_tag("CSV", "Comma-separated Values"),
         "diff" => file_abbr_tag("DIFF", "Plain text differences"),
-        "doc"  => MS_WORD_DOCUMENT_HUMANIZED_CONTENT_TYPE,
+        "doc" => MS_WORD_DOCUMENT_HUMANIZED_CONTENT_TYPE,
         "docx" => MS_WORD_DOCUMENT_HUMANIZED_CONTENT_TYPE,
-        "dot"  => file_abbr_tag("DOT", "MS Word Document Template"),
-        "dxf"  => file_abbr_tag("DXF", "AutoCAD Drawing Exchange Format"),
-        "eps"  => file_abbr_tag("EPS", "Encapsulated PostScript"),
-        "gif"  => file_abbr_tag("GIF", "Graphics Interchange Format"),
-        "gml"  => file_abbr_tag("GML", "Geography Markup Language"),
+        "dot" => file_abbr_tag("DOT", "MS Word Document Template"),
+        "dxf" => file_abbr_tag("DXF", "AutoCAD Drawing Exchange Format"),
+        "eps" => file_abbr_tag("EPS", "Encapsulated PostScript"),
+        "gif" => file_abbr_tag("GIF", "Graphics Interchange Format"),
+        "gml" => file_abbr_tag("GML", "Geography Markup Language"),
         "html" => file_abbr_tag("HTML", "Hypertext Markup Language"),
         "ics" => file_abbr_tag("ICS", "iCalendar file"),
-        "jpg"  => "JPEG",
-        "odp"  => file_abbr_tag("ODP", "OpenDocument Presentation"),
-        "ods"  => file_abbr_tag("ODS", "OpenDocument Spreadsheet"),
-        "odt"  => file_abbr_tag("ODT", "OpenDocument Text document"),
-        "pdf"  => file_abbr_tag("PDF", "Portable Document Format"),
-        "png"  => file_abbr_tag("PNG", "Portable Network Graphic"),
-        "ppt"  => MS_POWERPOINT_PRESENTATION_HUMANIZED_CONTENT_TYPE,
+        "jpg" => "JPEG",
+        "odp" => file_abbr_tag("ODP", "OpenDocument Presentation"),
+        "ods" => file_abbr_tag("ODS", "OpenDocument Spreadsheet"),
+        "odt" => file_abbr_tag("ODT", "OpenDocument Text document"),
+        "pdf" => file_abbr_tag("PDF", "Portable Document Format"),
+        "png" => file_abbr_tag("PNG", "Portable Network Graphic"),
+        "ppt" => MS_POWERPOINT_PRESENTATION_HUMANIZED_CONTENT_TYPE,
         "pptx" => MS_POWERPOINT_PRESENTATION_HUMANIZED_CONTENT_TYPE,
-        "ps"   => file_abbr_tag("PS", "PostScript"),
-        "rdf"  => file_abbr_tag("RDF", "Resource Description Framework"),
-        "rtf"  => file_abbr_tag("RTF", "Rich Text Format"),
-        "sch"  => file_abbr_tag("SCH", "XML based Schematic"),
-        "txt"  => "Plain text",
+        "ps" => file_abbr_tag("PS", "PostScript"),
+        "rdf" => file_abbr_tag("RDF", "Resource Description Framework"),
+        "rtf" => file_abbr_tag("RTF", "Rich Text Format"),
+        "sch" => file_abbr_tag("SCH", "XML based Schematic"),
+        "txt" => "Plain text",
         "wsdl" => file_abbr_tag("WSDL", "Web Services Description Language"),
-        "xls"  => MS_EXCEL_SPREADSHEET_HUMANIZED_CONTENT_TYPE,
+        "xls" => MS_EXCEL_SPREADSHEET_HUMANIZED_CONTENT_TYPE,
         "xlsm" => file_abbr_tag("XLSM", "MS Excel Macro-Enabled Workbook"),
         "xlsx" => MS_EXCEL_SPREADSHEET_HUMANIZED_CONTENT_TYPE,
-        "xlt"  => file_abbr_tag("XLT", "MS Excel Spreadsheet Template"),
-        "xsd"  => file_abbr_tag("XSD", "XML Schema"),
+        "xlt" => file_abbr_tag("XLT", "MS Excel Spreadsheet Template"),
+        "xsd" => file_abbr_tag("XSD", "XML Schema"),
         "xslt" => file_abbr_tag("XSLT", "Extensible Stylesheet Language Transformation"),
-        "zip"  => file_abbr_tag("ZIP", "Zip archive"),
+        "zip" => file_abbr_tag("ZIP", "Zip archive"),
       }
       file_extension_vs_humanized_content_type.fetch(file_extension.to_s.downcase, "")
     end
 
     def link(body, url, options = {})
-      options_str = options.map { |k, v| %{#{encode(k)}="#{encode(v)}"} }.join(" ")
-      %{<a href="#{encode(url)}" #{options_str}>#{body}</a>}
+      options_str = options.map { |k, v| %(#{encode(k)}="#{encode(v)}") }.join(" ")
+      %(<a href="#{encode(url)}" #{options_str}>#{body}</a>)
     end
 
   private

data/lib/govspeak/presenters/image_presenter.rb

@@ -33,8 +33,8 @@ module Govspeak
     def figcaption_html
       lines = []
       lines << "<figcaption>"
-      lines << %{<p>#{caption}</p>} if caption.present?
-      lines << %{<p>#{I18n.t('govspeak.image.figure.credit', credit: credit)}</p>} if credit.present?
+      lines << %(<p>#{caption}</p>) if caption.present?
+      lines << %(<p>#{I18n.t('govspeak.image.figure.credit', credit: credit)}</p>) if credit.present?
       lines << "</figcaption>"
       lines.join
     end

data/lib/govspeak/structured_header_extractor.rb

@@ -48,9 +48,9 @@ module Govspeak
     private :doc, :stack, :structured_headers
 
     def headers_list
-      @headers_list ||= doc.headers.map { |h|
+      @headers_list ||= doc.headers.map do |h|
         StructuredHeader.new(h.text, h.level, h.id, [])
-      }
+      end
     end
 
     def add_top_level(header)

data/lib/govspeak/version.rb

@@ -1,3 +1,3 @@
 module Govspeak
-  VERSION = "6.5.3".freeze
+  VERSION = "6.5.8".freeze
 end

data/lib/kramdown/parser/govuk.rb

@@ -9,7 +9,7 @@ module Kramdown
       end
     end
 
-    define(:document_domains, Object, %w{www.gov.uk}, <<~DESCRIPTION) do |val|
+    define(:document_domains, Object, %w[www.gov.uk], <<~DESCRIPTION) do |val|
       Defines the domains which are considered local to the document
 
       Default: www.gov.uk
@@ -21,10 +21,10 @@ module Kramdown
 
   module Parser
     class Govuk < Kramdown::Parser::Kramdown
-      CUSTOM_INLINE_ELEMENTS = %w(govspeak-embed-attachment-link).freeze
+      CUSTOM_INLINE_ELEMENTS = %w[govspeak-embed-attachment-link].freeze
 
       def initialize(source, options)
-        @document_domains = options[:document_domains] || %w(www.gov.uk)
+        @document_domains = options[:document_domains] || %w[www.gov.uk]
         super
       end
 
@@ -35,11 +35,10 @@ module Kramdown
             unless host.nil? || @document_domains.compact.include?(host)
               element.attr["rel"] = "external"
             end
-          # rubocop:disable Lint/SuppressedException
           rescue Addressable::URI::InvalidURIError
             # it's safe to ignore these very *specific* exceptions
           end
-          # rubocop:enable Lint/SuppressedException
+
         end
         super
       end