app/services/fetch_atom_service.rb

   1 # frozen_string_literal: true
   2
   3 class FetchAtomService < BaseService
   4   include JsonLdHelper
   5
   6   def call(url)
   7     return if url.blank?
   8
   9     result = process(url)
  10
  11     # retry without ActivityPub
  12     result ||= process(url) if @unsupported_activity
  13
  14     result
  15   rescue OpenSSL::SSL::SSLError => e
  16     Rails.logger.debug "SSL error: #{e}"
  17     nil
  18   rescue HTTP::ConnectionError => e
  19     Rails.logger.debug "HTTP ConnectionError: #{e}"
  20     nil
  21   end
  22
  23   private
  24
  25   def process(url, terminal = false)
  26     @url = url
  27     perform_request { |response| process_response(response, terminal) }
  28   end
  29
  30   def perform_request(&block)
  31     accept = 'text/html'
  32     accept = 'application/activity+json, application/ld+json; profile="https://www.w3.org/ns/activitystreams", application/atom+xml, ' + accept unless @unsupported_activity
  33
  34     Request.new(:get, @url).add_headers('Accept' => accept).perform(&block)
  35   end
  36
  37   def process_response(response, terminal = false)
  38     return nil if response.code != 200
  39
  40     if response.mime_type == 'application/atom+xml'
  41       [@url, { prefetched_body: response.body_with_limit }, :ostatus]
  42     elsif ['application/activity+json', 'application/ld+json'].include?(response.mime_type)
  43       body = response.body_with_limit
  44       json = body_to_json(body)
  45       if supported_context?(json) && equals_or_includes_any?(json['type'], ActivityPub::FetchRemoteAccountService::SUPPORTED_TYPES) && json['inbox'].present?
  46         [json['id'], { prefetched_body: body, id: true }, :activitypub]
  47       elsif supported_context?(json) && expected_type?(json)
  48         [json['id'], { prefetched_body: body, id: true }, :activitypub]
  49       else
  50         @unsupported_activity = true
  51         nil
  52       end
  53     elsif !terminal
  54       link_header = response['Link'] && parse_link_header(response)
  55
  56       if link_header&.find_link(%w(rel alternate))
  57         process_link_headers(link_header)
  58       elsif response.mime_type == 'text/html'
  59         process_html(response)
  60       end
  61     end
  62   end
  63
  64   def expected_type?(json)
  65     equals_or_includes_any?(json['type'], ActivityPub::Activity::Create::SUPPORTED_TYPES + ActivityPub::Activity::Create::CONVERTED_TYPES)
  66   end
  67
  68   def process_html(response)
  69     page = Nokogiri::HTML(response.body_with_limit)
  70
  71     json_link = page.xpath('//link[@rel="alternate"]').find { |link| ['application/activity+json', 'application/ld+json; profile="https://www.w3.org/ns/activitystreams"'].include?(link['type']) }
  72     atom_link = page.xpath('//link[@rel="alternate"]').find { |link| link['type'] == 'application/atom+xml' }
  73
  74     result ||= process(json_link['href'], terminal: true) unless json_link.nil? || @unsupported_activity
  75     result ||= process(atom_link['href'], terminal: true) unless atom_link.nil?
  76
  77     result
  78   end
  79
  80   def process_link_headers(link_header)
  81     json_link = link_header.find_link(%w(rel alternate), %w(type application/activity+json)) || link_header.find_link(%w(rel alternate), ['type', 'application/ld+json; profile="https://www.w3.org/ns/activitystreams"'])
  82     atom_link = link_header.find_link(%w(rel alternate), %w(type application/atom+xml))
  83
  84     result ||= process(json_link.href, terminal: true) unless json_link.nil? || @unsupported_activity
  85     result ||= process(atom_link.href, terminal: true) unless atom_link.nil?
  86
  87     result
  88   end
  89
  90   def parse_link_header(response)
  91     LinkHeader.parse(response['Link'].is_a?(Array) ? response['Link'].first : response['Link'])
  92   end
  93 end