govspeak 10.6.5 → 10.8.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b55c7097bba63f4d11e694c6cd683e13dddb4e4e444b61c3035e313c7836b4e5
-  data.tar.gz: 1aa8619222200af1e482598ffc53573cd862ac406369f4e15fbd954d7c85d34a
+  metadata.gz: '0920f169885278f664642da1cb6d8441a856b175779b4ad701aa966051644d25'
+  data.tar.gz: f0924a2e6574dc0b8c79ca819195368696b20b3c1e1943c3db283b98d3fada34
 SHA512:
-  metadata.gz: 8c0718694838de516cbb9583e599883422b0605ff150022e670c31fbe9896a376abaa92174053ab49ecff4e3a3c71da49a18a5ef3151eb24868282a4e2acf754
-  data.tar.gz: cb7f5f0b6a29154d4d9d0675ca72fa29d9c30f4b42f989365c33df81e76d0577b9d2d96577c72fdb626a543763c82d16e09ff459b37fd9801062d172f563b9e1
+  metadata.gz: 75028eeb6446acc724c97b76f8f1ccbc04447e21c2016b89abfdb4154e18bea563b5176cb5ba014ef7c60579e07311ecae50a669999f4502253ce4951c25e810
+  data.tar.gz: 884b2da2acb61aaa721201b80c67d40d57a2cc0ce517b1ff15114d8e530f37de35ab11473b4dca946ad34c8537e14eaaae9f7bc57d629228565f8f06cae4285c

data/CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## 10.8.0
+
+* feature: Allow callers to specify which headers will be auto-numbered [PR #451](https://github.com/alphagov/govspeak/pull/451)
+
+## 10.7.0
+
+* feature: Add auto-numbered headers option [PR #449](https://github.com/alphagov/govspeak/pull/449)
+* fix: When finding/returning headers, filter links from header titles [PR #439](https://github.com/alphagov/govspeak/pull/439)
+
 ## 10.6.5
 
 * Update dependencies

data/README.md

@@ -22,7 +22,7 @@ then create a new document
 
 Some additional steps or considerations are needed to ensure changes to govspeak cascade across GOV.UK in a holistic way.
 
-Once govspeak has been updated and version incremented then: 
+Once govspeak has been updated and version incremented then:
 - [`govuk_publishing_components` govspeak](https://components.publishing.service.gov.uk/component-guide/govspeak) will also need updating to reflect your most recent change.
 - [Publishing apps](https://docs.publishing.service.gov.uk/apps.html) (including but not limited to [content-publisher](https://github.com/alphagov/content-publisher) & [whitehall](https://github.com/alphagov/whitehall)) also use govspeak, these apps will need to be released with the new govspeak version present.
 
@@ -30,7 +30,7 @@ Also, consider if:
 - [whitehall](https://github.com/alphagov/whitehall) needs updating (as custom govspeak changes are present)
 - [govspeak-preview](https://github.com/alphagov/govspeak-preview) needs updating
 
-Any pages that use govspeak to generate Content will need to *republished* in order for the new changes to be reflected.  
+Any pages that use govspeak to generate Content will need to be *republished* in order for the new changes to be reflected.
 
 - Data Labs can help identify which pages need updating by [submitting a request](https://gov-uk.atlassian.net/wiki/spaces/GOVUK/pages/1860075525/GOV.UK+Data+Labs#Submitting-a-data-science-request) and [#govuk-2ndline](https://docs.publishing.service.gov.uk/manual/2nd-line.html) can help with republishing
 
@@ -48,6 +48,34 @@ with understanding how the content transforms from source through:
 
 In addition to the [standard Markdown syntax](http://daringfireball.net/projects/markdown/syntax "Markdown syntax"), we have added our own extensions.
 
+## Autonumbered headers
+
+This extension is off by default, can be turned on with
+
+    doc = Govspeak::Document.new "## Header\n### Sub-header\n#### Sub-sub-header", auto_numbered_headers: true
+
+Creates automatically numbered headers
+
+```html
+<h2>1 Header</h2>
+<h3>1.1 Sub-header</h3>
+<h4>1.1.1 Sub-sub-header</h4>
+```
+
+By default the numbering scheme will work for all headers apart from H1s, if you want to
+restrict this you can specify the levels you want in an array (default value is [2,3,4,5,6])
+
+    doc = Govspeak::Document.new "## Header\n### Sub-header\n#### Sub-sub-header", auto_numbered_headers: true, auto_numbered_header_levels: [2,3]
+
+Creates automatically numbered headers
+
+```html
+<h2>1 Header</h2>
+<h3>1.1 Sub-header</h3>
+<h4>Sub-sub-header</h4>
+```
+
+
 ## Callouts
 
 ### Information callouts

data/lib/govspeak/header_extractor.rb

@@ -20,11 +20,15 @@ module Govspeak
   private
 
     def id(element)
-      element.attr.fetch("id", generate_id(element.options[:raw_text]))
+      element.attr.fetch("id", generate_id(text_with_links(element)))
     end
 
     def build_header(element)
-      Header.new(element.options[:raw_text], element.options[:level], id(element))
+      Header.new(text_with_links(element), element.options[:level], id(element))
+    end
+
+    def text_with_links(element)
+      element.options[:raw_text].gsub(/\[(.+)\]\((.*)\)/, '\1')
     end
 
     def find_headers(parent)

data/lib/govspeak/post_processor.rb

@@ -160,6 +160,44 @@ module Govspeak
       end
     end
 
+    extension("use auto-numbered headers") do |document|
+      if govspeak_document.auto_numbered_headers
+        h2, h3, h4, h5, h6 = 0, 0, 0, 0, 0, 0
+
+        selector = govspeak_document.auto_numbered_header_levels.map { |i| "h#{i}" }.join(",")
+
+        document.css(selector).map do |el|
+          case el.name
+          when "h2"
+            h2 += 1
+            h3 = 0
+            h4 = 0
+            h5 = 0
+            h6 = 0
+            el.inner_html = "#{h2}. #{el.inner_html}"
+          when "h3"
+            h3 += 1
+            h4 = 0
+            h5 = 0
+            h6 = 0
+            el.inner_html = "#{h2}.#{h3} #{el.inner_html}"
+          when "h4"
+            h4 += 1
+            h5 = 0
+            h6 = 0
+            el.inner_html = "#{h2}.#{h3}.#{h4} #{el.inner_html}"
+          when "h5"
+            h5 += 1
+            h6 = 0
+            el.inner_html = "#{h2}.#{h3}.#{h4}.#{h5} #{el.inner_html}"
+          when "h6"
+            h6 += 1
+            el.inner_html = "#{h2}.#{h3}.#{h4}.#{h5}.#{h6} #{el.inner_html}"
+          end
+        end
+      end
+    end
+
     attr_reader :input, :govspeak_document
 
     def initialize(html, govspeak_document)

data/lib/govspeak/structured_header_extractor.rb

@@ -41,6 +41,8 @@ module Govspeak
         stack.push(header)
       end
 
+      add_auto_numbering(structured_headers, doc.auto_numbered_header_levels) if doc.auto_numbered_headers
+
       structured_headers
     end
 
@@ -96,6 +98,13 @@ module Govspeak
       @stack = []
     end
 
+    def add_auto_numbering(structured_headers, levels, prefix: "")
+      structured_headers.each.with_index(1) do |header, index|
+        header[:text] = "#{prefix}#{index}#{'.' if prefix == ''} #{header[:text]}" if levels.include?(header[:level])
+        add_auto_numbering(header[:headers], levels, prefix: "#{prefix}#{index}.")
+      end
+    end
+
   private
 
     attr_reader :doc, :stack, :structured_headers

data/lib/govspeak/version.rb

@@ -1,3 +1,3 @@
 module Govspeak
-  VERSION = "10.6.5".freeze
+  VERSION = "10.8.0".freeze
 end

data/lib/govspeak.rb

@@ -38,7 +38,7 @@ module Govspeak
     @extensions = []
 
     attr_accessor :images
-    attr_reader :attachments, :contacts, :links, :locale, :log_snapshots
+    attr_reader :attachments, :auto_numbered_headers, :auto_numbered_header_levels, :contacts, :links, :locale, :log_snapshots
 
     def self.to_html(source, options = {})
       new(source, options).to_html
@@ -55,11 +55,12 @@ module Govspeak
       @log_snapshots = options.fetch(:log_snapshots, false)
       log_snapshot("options", options)
       log_snapshot("source", @source)
-
       @images = options.delete(:images) || []
       @allowed_elements = options.delete(:allowed_elements) || []
       @allowed_image_hosts = options.delete(:allowed_image_hosts) || []
       @attachments = Array.wrap(options.delete(:attachments))
+      @auto_numbered_headers = options.fetch(:auto_numbered_headers, false)
+      @auto_numbered_header_levels = options.delete(:auto_numbered_header_levels) || [2, 3, 4, 5, 6]
       @links = Array.wrap(options.delete(:links))
       @contacts = Array.wrap(options.delete(:contacts))
       @locale = options.fetch(:locale, "en")

data/test/govspeak_structured_headers_test.rb

@@ -21,14 +21,16 @@ class GovspeakStructuredHeadersTest < Minitest::Test
 
 ### Sub heading 4.1
 
-#### Sub heading 4.1.1
+#### Sub sub heading 4.1.1
 
-##### Sub heading 4.1.1.1
+##### Sub sub sub heading 4.1.1.1
 
 ### Sub heading 4.2
 
 ## Heading 5
 
+### [Sub heading 5.1](https://www.example.com)
+
     )
   end
 
@@ -67,6 +69,11 @@ class GovspeakStructuredHeadersTest < Minitest::Test
     assert_equal "Sub heading 4.2", structured_headers[3].headers[1].text
   end
 
+  test "headers that are links are based on the link text not the link" do
+    assert_equal "Sub heading 5.1", structured_headers[4].headers[0].text
+    assert_equal "sub-heading-51", structured_headers[4].headers[0].id
+  end
+
   test "structured headers serialize to hashes recursively serializing sub headers" do
     serialized_headers = structured_headers[1].to_h
 
@@ -138,4 +145,46 @@ class GovspeakStructuredHeadersTest < Minitest::Test
   test "document with single h1 produces no headers" do
     assert_equal [], Govspeak::Document.new("# Heading\n").structured_headers
   end
+
+  test "auto-numbered headers are generated when the option is set on the document" do
+    doc = Govspeak::Document.new(document_body, auto_numbered_headers: true)
+
+    headers = doc.structured_headers
+
+    assert_equal "1. Heading 1", headers[0][:text]
+    assert_equal "2. Heading 2", headers[1][:text]
+    assert_equal "2.1 Sub heading 2.1", headers[1][:headers][0][:text]
+    assert_equal "2.2 Sub heading 2.2", headers[1][:headers][1][:text]
+    assert_equal "2.2.1 Sub sub heading 2.2.1", headers[1][:headers][1][:headers][0][:text]
+    assert_equal "2.3 Sub heading 2.3", headers[1][:headers][2][:text]
+    assert_equal "3. Heading 3", headers[2][:text]
+    assert_equal "4. Heading 4", headers[3][:text]
+    assert_equal "4.1 Sub heading 4.1", headers[3][:headers][0][:text]
+    assert_equal "4.1.1 Sub sub heading 4.1.1", headers[3][:headers][0][:headers][0][:text]
+    assert_equal "4.1.1.1 Sub sub sub heading 4.1.1.1", headers[3][:headers][0][:headers][0][:headers][0][:text]
+    assert_equal "4.2 Sub heading 4.2", headers[3][:headers][1][:text]
+    assert_equal "5. Heading 5", headers[4][:text]
+    assert_equal "5.1 Sub heading 5.1", headers[4][:headers][0][:text]
+  end
+
+  test "auto-numbered headers are restricted to appropriate levels when the option is set on the document" do
+    doc = Govspeak::Document.new(document_body, auto_numbered_headers: true, auto_numbered_header_levels: [2, 3])
+
+    headers = doc.structured_headers
+
+    assert_equal "1. Heading 1", headers[0][:text]
+    assert_equal "2. Heading 2", headers[1][:text]
+    assert_equal "2.1 Sub heading 2.1", headers[1][:headers][0][:text]
+    assert_equal "2.2 Sub heading 2.2", headers[1][:headers][1][:text]
+    assert_equal "Sub sub heading 2.2.1", headers[1][:headers][1][:headers][0][:text]
+    assert_equal "2.3 Sub heading 2.3", headers[1][:headers][2][:text]
+    assert_equal "3. Heading 3", headers[2][:text]
+    assert_equal "4. Heading 4", headers[3][:text]
+    assert_equal "4.1 Sub heading 4.1", headers[3][:headers][0][:text]
+    assert_equal "Sub sub heading 4.1.1", headers[3][:headers][0][:headers][0][:text]
+    assert_equal "Sub sub sub heading 4.1.1.1", headers[3][:headers][0][:headers][0][:headers][0][:text]
+    assert_equal "4.2 Sub heading 4.2", headers[3][:headers][1][:text]
+    assert_equal "5. Heading 5", headers[4][:text]
+    assert_equal "5.1 Sub heading 5.1", headers[4][:headers][0][:text]
+  end
 end

data/test/govspeak_test.rb

@@ -99,6 +99,50 @@ class GovspeakTest < Minitest::Test
     assert_equal "foo bar baz", doc.to_text
   end
 
+  test "adds auto-numbered headings when the auto_numbered_headings option is on" do
+    input = %(
+## H2 One
+
+### H3 One
+
+### H3 Two
+
+## H2 Two
+
+### H3 Three
+
+#### H4 One
+
+##### H5 One
+
+###### H6 One
+)
+    doc = Govspeak::Document.new(input, auto_numbered_headers: true)
+    assert_equal %(\n<h2 id="h2-one">1. H2 One</h2>\n\n<h3 id="h3-one">1.1 H3 One</h3>\n\n<h3 id="h3-two">1.2 H3 Two</h3>\n\n<h2 id="h2-two">2. H2 Two</h2>\n\n<h3 id="h3-three">2.1 H3 Three</h3>\n\n<h4 id="h4-one">2.1.1 H4 One</h4>\n\n<h5 id="h5-one">2.1.1.1 H5 One</h5>\n\n<h6 id="h6-one">2.1.1.1.1 H6 One</h6>\n), doc.to_html
+  end
+
+  test "restricts auto-numbered headings to specified levels when the auto_numbered_heading_levels option is set" do
+    input = %(
+## H2 One
+
+### H3 One
+
+### H3 Two
+
+## H2 Two
+
+### H3 Three
+
+#### H4 One
+
+##### H5 One
+
+###### H6 One
+)
+    doc = Govspeak::Document.new(input, auto_numbered_headers: true, auto_numbered_header_levels: [2, 3])
+    assert_equal %(\n<h2 id="h2-one">1. H2 One</h2>\n\n<h3 id="h3-one">1.1 H3 One</h3>\n\n<h3 id="h3-two">1.2 H3 Two</h3>\n\n<h2 id="h2-two">2. H2 Two</h2>\n\n<h3 id="h3-three">2.1 H3 Three</h3>\n\n<h4 id="h4-one">H4 One</h4>\n\n<h5 id="h5-one">H5 One</h5>\n\n<h6 id="h6-one">H6 One</h6>\n), doc.to_html
+  end
+
   test "trailing space after the address should not prevent parsing" do
     input = %($A
 123 Test Street

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: govspeak
 version: !ruby/object:Gem::Version
-  version: 10.6.5
+  version: 10.8.0
 platform: ruby
 authors:
 - GOV.UK Dev
@@ -18,7 +18,7 @@ dependencies:
         version: '6'
     - - "<"
       - !ruby/object:Gem::Version
-        version: 8.0.4
+        version: 8.1.2
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
@@ -28,7 +28,7 @@ dependencies:
         version: '6'
     - - "<"
       - !ruby/object:Gem::Version
-        version: 8.0.4
+        version: 8.1.2
 - !ruby/object:Gem::Dependency
   name: addressable
   requirement: !ruby/object:Gem::Requirement