discourse/lib/email/message_id_service.rb

# frozen_string_literal: true

module Email
  ##
  # Email Message-IDs are used in both our outbound and inbound email
  # flow. For the outbound flow via Email::Sender, we assign a unique
  # Message-ID for any emails sent out from the application.
  # If we are sending an email related to a post, such as through the
  # PostAlerter class, then the Message-ID will contain references to
  # the post ID. The host must also be included on the Message-IDs.
  # The format looks like this:
  #
  # discourse/post/POST_ID@HOST
  #
  # We previously had the following formats, but support for these
  # will be removed in 2023:
  #
  # topic/TOPIC_ID/POST_ID@HOST
  # topic/TOPIC_ID@HOST
  #
  # For the inbound email flow via Email::Receiver, we use Message-IDs
  # to discern which topic and post the inbound email reply should be
  # in response to. In this case, the Message-ID is extracted from the
  # References and/or In-Reply-To headers, and compared with either
  # the IncomingEmail table, the Post table, or the IncomingEmail to
  # determine where to send the reply.
  #
  # See https://datatracker.ietf.org/doc/html/rfc2822#section-3.6.4 for
  # more specific information around Message-IDs in email.
  #
  # See https://tools.ietf.org/html/rfc850#section-2.1.7 for the
  # Message-ID format specification.
  class MessageIdService
    class << self
      def generate_default
        "<#{SecureRandom.uuid}@#{host}>"
      end

      ##
      # The outbound_message_id may be present because either:
      #
      # * The post was created via incoming email and Email::Receiver, and
      #   references a Message-ID generated by an external email client or service.
      # * At least one email has been sent because of the post being created
      #   to inform interested parties via email.
      #
      # If it is blank then we should assume Discourse was the originator
      # of the post, and generate a Message-ID to be used from now on using
      # our discourse/post/POST_ID@HOST format.
      def generate_or_use_existing(post_ids)
        post_ids = Array.wrap(post_ids)
        return [] if post_ids.empty?

        DB.exec(<<~SQL, host: host)
          UPDATE posts
          SET outbound_message_id = 'discourse/post/' || posts.id || '@' || :host
          WHERE outbound_message_id IS NULL AND posts.id IN (#{post_ids.join(",")});
        SQL

        DB.query_single(<<~SQL)
          SELECT '<' || posts.outbound_message_id || '>'
          FROM posts
          WHERE posts.id IN (#{post_ids.join(",")})
          ORDER BY posts.created_at ASC;
        SQL
      end

      ##
      # Uses extracted Message-IDs from both the In-Reply-To and References
      # headers from an incoming email.
      def find_post_from_message_ids(message_ids)
        message_ids = message_ids.map { |message_id| message_id_clean(message_id) }

        # TODO (martin) 2023-04-01 We should remove these backwards-compatible
        # formats for the Message-ID and solely use the discourse/post/999@host
        # format.
        topic_ids =
          message_ids
            .map { |message_id| message_id[message_id_topic_id_regexp, 1] }
            .compact
            .map(&:to_i)
        post_ids =
          message_ids
            .map { |message_id| message_id[message_id_post_id_regexp, 1] }
            .compact
            .map(&:to_i)

        post_ids << message_ids
          .map { |message_id| message_id[message_id_discourse_regexp, 1] }
          .compact
          .map(&:to_i)

        post_ids << Post
          .where(outbound_message_id: message_ids)
          .or(Post.where(topic_id: topic_ids, post_number: 1))
          .pluck(:id)
        post_ids << EmailLog.where(message_id: message_ids).pluck(:post_id)
        post_ids << IncomingEmail.where(message_id: message_ids).pluck(:post_id)

        post_ids.flatten!
        post_ids.compact!
        post_ids.uniq!

        return if post_ids.empty?

        Post.where(id: post_ids).order(:created_at).last
      end

      # TODO (martin) 2023-04-01 We should remove these backwards-compatible
      # formats for the Message-ID and solely use the discourse/post/999@host
      # format.
      def discourse_generated_message_id?(message_id)
        !!(message_id =~ message_id_post_id_regexp) ||
          !!(message_id =~ message_id_topic_id_regexp) ||
          !!(message_id =~ message_id_discourse_regexp)
      end

      # TODO (martin) 2023-04-01 We should remove these backwards-compatible
      # formats for the Message-ID and solely use the discourse/post/999@host
      # format.
      def message_id_post_id_regexp
        Regexp.new "topic/\\d+/(\\d+|\\d+\.\\w+)@#{Regexp.escape(host)}"
      end

      def message_id_topic_id_regexp
        Regexp.new "topic/(\\d+|\\d+\.\\w+)@#{Regexp.escape(host)}"
      end

      def message_id_discourse_regexp
        Regexp.new "discourse/post/(\\d+)@#{Regexp.escape(host)}"
      end

      def message_id_rfc_format(message_id)
        message_id.present? && !is_message_id_rfc?(message_id) ? "<#{message_id}>" : message_id
      end

      def message_id_clean(message_id)
        if message_id.present? && is_message_id_rfc?(message_id)
          message_id.gsub(/^<|>$/, "")
        else
          message_id
        end
      end

      def is_message_id_rfc?(message_id)
        message_id.start_with?("<") && message_id.include?("@") && message_id.end_with?(">")
      end

      def host
        Email::Sender.host_for(Discourse.base_url)
      end
    end
  end
end