Rakefile

# frozen_string_literal: true

require 'erb'
require 'json'
require 'nokogiri'
require 'sqlite3'

# Nomad website URL
BASE_URL = 'https://www.nomadproject.io'

# This method attempts to resolve the location of a file on disk given a
# relative, or absolute path.
def resolve_file_path(anchor)
  return anchor if anchor.path.end_with?('index.html') && File.exist?("out#{anchor.path}")

  # Check if path is in list of redirects
  # This the initial source path
  redirect_source = anchor.path
  # This is the final destination
  redirect_final_dest = nil
  # Trigger to stop redirect evaluation
  stop_redirect_evaluation = false

  until stop_redirect_evaluation
    stop_redirect_evaluation = true

    next unless REDIRECTS.key?(redirect_source)

    redirect_source = REDIRECTS[redirect_source]
    redirect_final_dest = URI.parse(redirect_source)
    stop_redirect_evaluation = false
  end

  # Return path if redirect resolved to an external URL
  case redirect_final_dest.class.to_s
  when 'URI::HTTP', 'URI::HTTPS'
    return redirect_final_dest
  when 'URI::Generic'
    begin
      anchor = anchor.merge(redirect_final_dest)
    rescue URI::BadURIError
      anchor = redirect_final_dest
    end
  end

  # List of potential path extensions
  potential_extensions = [
    # change /path/ to path.html
    anchor.path.sub(%r{/$}, '.html'),
    # change /path to /path.html
    "#{anchor.path}.html",
    # /path/index.html
    "#{anchor.path}/index.html"
  ]

  potential_prefixes = [
    # relative
    '',
    # absolute
    '/'
  ]

  potential_prefixes.each do |prefix|
    potential_extensions.each do |suffix|
      test_path = "#{prefix}#{suffix}"
      if File.exist?("out#{test_path}") && File.file?("out#{test_path}")
        anchor.path = test_path
        return anchor
      end
    end
  end

  # Return the original path we were passed in.
  # Its possible this is a redirect which will be resolved later.
  anchor
end

# Handles creation and insertion of data into the docset's index
class Index
  attr_accessor :db

  def initialize(path)
    @db = SQLite3::Database.new path
  end

  def drop
    @db.execute <<-SQL
      DROP TABLE IF EXISTS searchIndex
    SQL
  end

  def create
    # Create the docset index
    # https://kapeli.com/docsets#createsqlite
    db.execute <<-SQL
      CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT)
    SQL
    db.execute <<-SQL
      CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path)
    SQL
  end

  def reset
    drop
    create
  end

  def insert(type, path)
    # Insert records into the index
    # https://kapeli.com/docsets#fillsqlite
    doc = Nokogiri::HTML(File.open(path).read)
    name = doc.title.sub(' | Nomad by HashiCorp', '').sub(/.*: (.*)/, '\\1')
    @db.execute <<-SQL, name: name, type: type, path: path
      INSERT OR IGNORE INTO searchIndex (name, type, path)
      VALUES(:name, :type, :path)
    SQL
  end
end

task default: %i[clean setup copy create_index package]

task :clean do
  rm_rf 'Nomad.docset'
end

task :setup do
  mkdir_p 'Nomad.docset/Contents/Resources/Documents'

  # Docset icon
  if File.exist?('out/img/favicons/favicon-16x16.png') &&
     File.exist?('out/img/favicons/favicon-32x32.png')
    FileUtils.cp('out/img/favicons/favicon-16x16.png',
                 'Nomad.docset/icon.png')
    FileUtils.cp('out/img/favicons/favicon-32x32.png',
                 'Nomad.docset/icon@2x.png')
  elsif File.exist? 'ui/public/favicon.png'
    cp 'ui/public/favicon.png', 'Nomad.docset/icon.png'
  end

  # Create Info.plist
  # https://kapeli.com/docsets#infoplist
  File.open('Nomad.docset/Contents/Info.plist', 'w') do |f|
    f.write <<~XML
      <?xml version="1.0" encoding="UTF-8"?>
      <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
      <plist version="1.0">
        <dict>
          <key>CFBundleIdentifier</key>
          <string>nomad</string>
          <key>CFBundleName</key>
          <string>Nomad</string>
          <key>DocSetPlatformFamily</key>
          <string>nomad</string>
          <key>isDashDocset</key>
          <true/>
          <key>DashDocSetFamily</key>
          <string>dashtoc</string>
          <key>dashIndexFilePath</key>
          <string>docs/index.html</string>
          <key>DashDocSetFallbackURL</key>
          <string>#{BASE_URL}</string>
        </dict>
      </plist>
    XML
  end
end

task :copy do
  # Read redirects file
  if File.exist?('redirects.json')
    redirect_file = File.read('redirects.json')
    REDIRECTS = JSON.parse(redirect_file)
  else
    REDIRECTS = {}
  end

  file_list = []
  Dir.chdir('out') { file_list = Dir.glob('**/*').sort }

  file_list.each do |path|
    source = "out/#{path}"
    target = "Nomad.docset/Contents/Resources/Documents/#{path}"

    # Determine relative path of current file
    begin
      source_path = URI.parse(BASE_URL + source.delete_prefix('out'))
    rescue URI::InvalidURIError
      # Skip if unable to parse URL
      next
    end

    if File.stat(source).directory?
      mkdir_p target
    elsif source.end_with?('.gz')
      next
    elsif source.end_with?('.html')
      doc = Nokogiri::HTML(File.open(source).read)

      # Remove following from document titles
      doc.title = doc.title.sub(' | Nomad by HashiCorp', '')
      doc.title = doc.title.sub(' - HTTP API', '')

      # Find section headings within document
      doc.xpath("//a[contains(@class, '__target-h')]").each do |e|
        a = Nokogiri::XML::Node.new 'a', doc

        # Remove '»' permalink
        e.previous.remove if e.previous.text == '»'

        # Obtain section name from element following anchor tag
        next_sibling = e.next_sibling

        section_name = next_sibling.text.strip

        # If first part of header is in a code block, grab the following text
        # as well
        section_name += next_sibling.next_sibling&.text.to_s if next_sibling.name == 'code'

        # Skip headings with the same name as the document title
        next if section_name == doc.title || section_name.empty?

        # Add dashAnchor links to generate per-page Table of Contents
        # https://kapeli.com/docsets#tableofcontents
        a['class'] = 'dashAnchor'
        a['name'] = format(
          '//apple_ref/cpp/%<type>s/%<name>s',
          # TODO: Modify type based on content being indexed
          # https://kapeli.com/docsets#supportedentrytypes
          type: 'Section',
          name: ERB::Util.url_encode(section_name)
        )
        e.previous = a
      end

      # Remove JavaScript tags on all documentation pages
      doc.xpath('//script').each do |script|
        script.remove if script.text != ''
      end

      # Find all links on page
      doc.xpath('//link').each do |link|
        # If the link is relative
        next unless link.attributes.key?('href') && link.attributes['href'].value.start_with?('/')

        # Remove CORS settings on link
        link.remove_attribute('crossorigin')

        # Do not preload content. Load it synchronously.
        if link.attributes['rel'].value == 'preload'

          as_value = link.attributes['as'].value

          case as_value
          when 'style'
            link.attributes['rel'].value = 'stylesheet'
          when 'script'
            link.attributes['rel'].value = 'script'
          end

          link.remove_attribute('as')
        end

        # Ensure script/stylesheet location is relative to the current file
        link_href = BASE_URL + link.attributes['href'].value
        link.attributes['href'].value = source_path.route_to(link_href).to_s
      end

      # Remove the following elements from doc pages
      doc.xpath("id('header')").each(&:remove)
      doc.xpath("//div[contains(@class, 'g-alert-banner')]").each(&:remove)
      doc.xpath("//div[contains(@class, 'g-mega-nav')]").each(&:remove)
      doc.xpath("//div[contains(@class, 'g-subnav')]").each(&:remove)
      doc.xpath("id('sidebar')").each(&:remove)
      doc.xpath("id('edit-this-page')").each(&:remove)
      doc.xpath("//footer[contains(@class, 'g-footer')]").each(&:remove)

      # Remove margin on headings
      doc.xpath('//div[@id="inner"]/*/h2').each do |e|
        e['style'] = 'margin-top: 0px'
      end

      # Fix links to other documents
      doc.xpath('//div[@id="inner"]//a').each do |e|
        # Skip anchor if it does not have an href attribute
        next unless e.attributes.key?('href')

        anchor = e.attributes['href']
        begin
          parsed_uri = URI.parse(anchor.value)
        rescue URI::InvalidURIError
          # Skip if unable to parse URL
          next
        end

        # Skip unknown URIs
        next unless [URI::Generic, URI::HTTP, URI::HTTPS].member? parsed_uri.class

        # Skip bookmark links
        next if parsed_uri.is_a?(URI::Generic) && parsed_uri.path.empty? && parsed_uri.fragment

        # Determine relative path of relative URI
        parsed_uri = resolve_file_path(parsed_uri)

        # Skip processing if URL ultimately redirected to an external site
        next if [URI::HTTP, URI::HTTPS].member?(parsed_uri.class) && !parsed_uri.host.include?('nomadproject.io')

        # Ensure that file exists
        exempt_urls = [
          '/docs/autoscaling/interals/checks',
          '/docs/job-specification/hcl2/functions/datetime/timestamp',
          '/docs/commands/license/put',
          '/docs/who-uses/noamd'
        ]
        unless File.exist?("out#{parsed_uri.path}")
          # Special case /api/ to rewrite to /api-docs/
          if parsed_uri.path.start_with?('/api/')
            parsed_uri.path.sub!(%r{/api/}, '/api-docs/')
          elsif exempt_urls.include?(parsed_uri.path)
            next
          end
        end

        # Convert relative URLs to absolute URLs
        if parsed_uri.is_a? URI::Generic
          relative_value = parsed_uri.path
          parsed_uri = URI.parse("#{BASE_URL}#{relative_value}")
        end

        anchor.value = source_path.route_to(parsed_uri).to_s
      end

      # Set width of documentation content to 100%
      doc.xpath("//div[contains(@role, 'main')]").each do |e|
        e['style'] = 'width: 100%'
      end

      # Write resultant file to Nomad.docset/ path
      File.open(target, 'w') { |f| f.write doc }
    else
      # If the file does not end .html, it must be an image or other asset
      # Copy it directly into the target directory
      cp source, target
    end
  end
end

task :create_index do
  index = Index.new('Nomad.docset/Contents/Resources/docSet.dsidx')
  index.reset

  Dir.chdir('Nomad.docset/Contents/Resources/Documents') do
    # api
    Dir.glob('api-docs/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Define', path
    end
    # docs
    Dir.glob('docs/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Guide', path
    end
    # docs/commands
    Dir.glob('docs/commands/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Command', path
    end
    # docs/configuration
    Dir.glob('docs/configuration/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/devices
    Dir.glob('docs/devices/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/drivers
    Dir.glob('docs/drivers/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/enterprise
    Dir.glob('docs/enterprise/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Environment', path
    end
    # docs/internals
    Dir.glob('docs/internals/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Instruction', path
    end
    # docs/job-specification
    Dir.glob('docs/job-specification/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/runtime
    Dir.glob('docs/runtime/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Environment', path
    end
    # docs/telemetry
    Dir.glob('docs/telemetry/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/vault-integration
    Dir.glob('docs/vault-integration/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Setting', path
    end
    # docs/integrations/
    Dir.glob('docs/integrations/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Guide', path
    end
    # getting-started
    Dir.glob('intro/getting-started/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Guide', path if path.end_with?('.html')
    end
    # vs
    Dir.glob('intro/vs/**/*')
       .find_all { |f| File.stat(f).file? }.each do |path|
      index.insert 'Guide', path if path.end_with?('.html')
    end
  end
end

task :package do
  sh "tar --exclude='.DS_Store' -cvzf Nomad.tgz Nomad.docset"
end