require 'set'

IGNORED_PAGES = Set['sitemap.xml']
LANGS = Set['en', 'pt', 'fr', 'eo'] # jp zh es de

module Jekyll
  class Linter < Generator
    safe true
    priority :high

    def insert_id(name, document)
      lang = document.data['lang']
      ref = document.data['ref']
      id = "#{name}:#{lang}:#{ref}"
      if @known_ids.include? id then
        raise "Duplicate ID found: '#{id}'"
      else
        @known_ids.add id
      end
    end

    def assert_unique_ids(site)
      @known_ids = Set[]
      all_documents(site) do |collection_name, document|
        insert_id collection_name, document
      end
    end

    def slugify(s)
      s.ljust(100)
        .gsub(/[\W]+/, ' ')
        .strip
        .gsub(/\s\s+/, '-')
        .downcase
        .gsub(' ', '-')
        .gsub('_', '-')
    end

    def assert(value, message)
      unless value
        raise message
      end
      value
    end

    def assert_field(document, field)
      f = document.data[field]
      raise "Undefined '#{field}' for #{document.path}" unless f
      f
    end

    COLLECTION_LAYOUTS = {
      'page' => 'page',
      'articles' => 'post',
      'pastebins' => 'post',
      'tils' => 'post',
      'slides' => 'slides',
      'podcasts' => 'cast',
      'screencasts' => 'cast'
    }

    def assert_frontmatter_fields(name, document)
      title = assert_field document, 'title'
      lang = assert_field document, 'lang'
      ref = assert_field document, 'ref'
      layout = assert_field document, 'layout'
      date = document.date.strftime('%Y-%m-%d') unless layout == 'page'
      slug = layout == 'page' ? ref : assert_field(document, 'slug')
      extension = name == 'slides' ? 'slides' : 'md'

      unless LANGS.member? lang
        raise "Invalid lang '#{lang}' in #{document.path}"
      end

      if COLLECTION_LAYOUTS[name] != layout
        raise "Layout mismatch: expected '#{COLLECTION_LAYOUTS[name]}', got '#{layout}' for #{document.path}"
      end

      if lang == 'en'
        unless ['index', 'root', 'tils'].include? ref
          if slugify(title) != ref then
            raise "#{ref} isn't a slug of the title.\nref:        '#{ref}'\ntitle slug: '#{slugify(title)}'"
            p slugify(title)
          end
        end
      end

      unless layout == 'page' then
        path = "_#{name}/#{date}-#{slug}.#{extension}"
        unless path == document.relative_path then
          raise "date/filename mismatch:\ndate+slug: #{path}\nfilename:  #{document.relative_path}"
        end

        if lang == 'en' then
          unless ref == slug then
            raise "ref/slug mismatch:\nref:  #{ref}\nslug: #{slug}"
          end
        end
      end

      if name == 'podcasts' then
        flac = "resources/podcasts/#{date}-#{slug}.flac"
        unless File.exist? flac then
          raise "Missing FLAC file '#{flac}'"
        end

      end

      if name == 'screencasts' then
        mkv = "resources/screencasts/#{date}-#{slug}.mkv"
        unless File.exist? mkv then
          raise "Missing MKV file '#{mkv}'"
        end
      end
    end

    def assert_frontmatter(site)
      all_documents(site) do |collection_name, document|
        assert_frontmatter_fields collection_name, document
      end
    end

    def assert_git_annex(site)
      url = site.config['url']
      stdout = `git annex find --not --in web`
      if stdout != '' or not $?.success? then
        puts 'Files in Git Annex not published to "web" remote:'
        puts stdout
        puts 'Add them with:'
        stdout.strip.split('\n').each do |file|
          puts "  git annex addurl --file #{file} #{url}/#{file}"
        end
        raise 'Build again after files above are added'
      end
    end

    def generate(site)
      assert_unique_ids(site)
      assert_frontmatter(site)
      assert_git_annex(site)
    end

    def all_documents(site)
      site.collections.each do |name, collection|
        collection.docs.each do |document|
          unless document.data['generated']
            yield name, document
          end
        end
      end

      site.pages.each do |page|
        unless IGNORED_PAGES.include? page.path
          unless page.data['generated']
            yield 'page', page
          end
        end
      end
    end
  end
end