2016-11-16 00:56:29 +09:00
|
|
|
# frozen_string_literal: true
|
|
|
|
|
2016-09-26 23:42:38 +09:00
|
|
|
class FetchAtomService < BaseService
|
|
|
|
def call(url)
|
2017-02-13 01:28:15 +09:00
|
|
|
return if url.blank?
|
|
|
|
|
2017-07-15 03:41:49 +09:00
|
|
|
response = Request.new(:head, url).perform
|
2016-09-26 23:42:38 +09:00
|
|
|
|
|
|
|
Rails.logger.debug "Remote status HEAD request returned code #{response.code}"
|
2016-11-09 03:09:22 +09:00
|
|
|
|
2017-07-15 03:41:49 +09:00
|
|
|
response = Request.new(:get, url).perform if response.code == 405
|
2016-11-09 03:09:22 +09:00
|
|
|
|
|
|
|
Rails.logger.debug "Remote status GET request returned code #{response.code}"
|
|
|
|
|
2016-09-26 23:42:38 +09:00
|
|
|
return nil if response.code != 200
|
2016-11-16 00:56:29 +09:00
|
|
|
return [url, fetch(url)] if response.mime_type == 'application/atom+xml'
|
2017-06-08 20:24:28 +09:00
|
|
|
return process_headers(url, response) if response['Link'].present?
|
2016-11-16 00:56:29 +09:00
|
|
|
process_html(fetch(url))
|
2016-10-05 20:26:44 +09:00
|
|
|
rescue OpenSSL::SSL::SSLError => e
|
|
|
|
Rails.logger.debug "SSL error: #{e}"
|
2017-06-29 20:04:07 +09:00
|
|
|
nil
|
|
|
|
rescue HTTP::ConnectionError => e
|
|
|
|
Rails.logger.debug "HTTP ConnectionError: #{e}"
|
|
|
|
nil
|
2016-09-26 23:42:38 +09:00
|
|
|
end
|
|
|
|
|
|
|
|
private
|
|
|
|
|
|
|
|
def process_html(body)
|
2016-09-30 04:28:21 +09:00
|
|
|
Rails.logger.debug 'Processing HTML'
|
2016-09-26 23:42:38 +09:00
|
|
|
|
|
|
|
page = Nokogiri::HTML(body)
|
|
|
|
alternate_link = page.xpath('//link[@rel="alternate"]').find { |link| link['type'] == 'application/atom+xml' }
|
|
|
|
|
|
|
|
return nil if alternate_link.nil?
|
2016-11-16 00:56:29 +09:00
|
|
|
[alternate_link['href'], fetch(alternate_link['href'])]
|
2016-09-26 23:42:38 +09:00
|
|
|
end
|
|
|
|
|
2016-09-27 00:04:05 +09:00
|
|
|
def process_headers(url, response)
|
2016-09-30 04:28:21 +09:00
|
|
|
Rails.logger.debug 'Processing link header'
|
2016-09-26 23:42:38 +09:00
|
|
|
|
2016-09-28 06:49:12 +09:00
|
|
|
link_header = LinkHeader.parse(response['Link'].is_a?(Array) ? response['Link'].first : response['Link'])
|
2016-11-16 00:56:29 +09:00
|
|
|
alternate_link = link_header.find_link(%w(rel alternate), %w(type application/atom+xml))
|
2016-09-26 23:42:38 +09:00
|
|
|
|
2016-09-27 00:04:05 +09:00
|
|
|
return process_html(fetch(url)) if alternate_link.nil?
|
2016-11-16 00:56:29 +09:00
|
|
|
[alternate_link.href, fetch(alternate_link.href)]
|
2016-09-26 23:42:38 +09:00
|
|
|
end
|
|
|
|
|
|
|
|
def fetch(url)
|
2017-07-15 03:41:49 +09:00
|
|
|
Request.new(:get, url).perform.to_s
|
2016-09-26 23:42:38 +09:00
|
|
|
end
|
|
|
|
end
|