markdown_filter.rb

# frozen_string_literal: true

require "redcarpet"
require "rouge/plugins/redcarpet"
require "nokogiri"

module Homeland
  class Pipeline
    class MarkdownFilter < HTML::Pipeline::TextFilter
      class Render < Redcarpet::Render::HTML
        include Rouge::Plugins::Redcarpet

        def domain
          Setting.domain
        end

        def block_code(code, lang)
          lang = lang.downcase if lang.is_a?(String)
          code = code.strip_heredoc
          super(code, lang)
        end

        def table(header, body)
          %(<div class="table-responsive"><table class="table table-bordered table-striped">#{header}#{body}</table></div>)
        end

        def header(text, header_level)
          l = (header_level <= 2) ? 2 : header_level
          raw_text = Nokogiri::HTML(text).xpath("//text()")
          %(<h#{l} id="#{raw_text}">#{text}</h#{l}>)
        end

        # Extend to support img width
        # ![](foo.jpg =300x)
        # ![](foo.jpg =300x200)
        # Example: https://gist.github.com/uupaa/f77d2bcf4dc7a294d109
        def image(link, title, alt_text)
          link ||= ""
          links = link.split(" ")
          link = links[0]
          if links.count > 1
            # 原本 Markdown 的 title 部分是需要引号的 ![](foo.jpg "Title")
            # ![](foo.jpg =300x)
            title = links[1]
          end

          if title =~ /=(\d+)x(\d+)/
            %(<img src="#{link}" width="#{Regexp.last_match(1)}px" height="#{Regexp.last_match(2)}px" alt="#{alt_text}">)
          elsif title =~ /=(\d+)x/
            %(<img src="#{link}" width="#{Regexp.last_match(1)}px" alt="#{alt_text}">)
          elsif title =~ /=x(\d+)/
            %(<img src="#{link}" height="#{Regexp.last_match(1)}px" alt="#{alt_text}">)
          else
            %(<img src="#{link}" title="#{title}" alt="#{alt_text}">)
          end
        end

        def link(link, title, content)
          external = false
          safe_link = link&.split("?")&.first
          uri = begin
            URI.parse(safe_link)
          rescue
            nil
          end
          if uri&.host && uri&.host&.downcase != domain
            external = true
          end

          if external
            %(<a href="#{link}" rel="nofollow" target="_blank" title="#{title}">#{content}</a>)
          else
            %(<a href="#{link}" title="#{title}">#{content}</a>)
          end
        end

        # Fix Chinese neer the URL
        def autolink(link, link_type)
          # return link
          if link_type.to_s == "email"
            link
          else
            begin
              # 防止 C 的 autolink 出来的内容有编码错误，万一有就直接跳过转换
              # 比如这句: 此版本并非线上的http://yavaeye.com的源码.
              link.match(/.+?/)
            rescue
              return link
            end
            # Fix Chinese neer the URL
            bad_text = link.match(%r{[^\w:@/\-~,$!.=?&#+|%]+}im).to_s
            link.gsub!(bad_text, "")
            %(<a href="#{link}" rel="nofollow" target="_blank">#{link}</a>#{bad_text})
          end
        end
      end

      RENDERDER = Redcarpet::Markdown.new(Render, {
        no_styles: true,
        hard_wrap: true,
        autolink: true,
        fenced_code_blocks: true,
        strikethrough: true,
        underline: true,
        superscript: false,
        footnotes: false,
        highlight: false,
        tables: true,
        lax_spacing: true,
        space_after_headers: true,
        disable_indented_code_blocks: true,
        no_intra_emphasis: true
      })

      def call
        RENDERDER.render(@text)
      end
    end
  end
end