2016-09-26 15:42:38 +01:00
|
|
|
class FetchAtomService < BaseService
|
|
|
|
def call(url)
|
|
|
|
response = http_client.head(url)
|
|
|
|
|
|
|
|
Rails.logger.debug "Remote status HEAD request returned code #{response.code}"
|
|
|
|
return nil if response.code != 200
|
|
|
|
|
|
|
|
if response.mime_type == 'application/atom+xml'
|
|
|
|
return [url, fetch(url)]
|
|
|
|
elsif !response['Link'].blank?
|
2016-09-26 16:04:05 +01:00
|
|
|
return process_headers(url, response)
|
2016-09-26 15:42:38 +01:00
|
|
|
else
|
|
|
|
return process_html(fetch(url))
|
|
|
|
end
|
|
|
|
end
|
|
|
|
|
|
|
|
private
|
|
|
|
|
|
|
|
def process_html(body)
|
2016-09-29 20:28:21 +01:00
|
|
|
Rails.logger.debug 'Processing HTML'
|
2016-09-26 15:42:38 +01:00
|
|
|
|
|
|
|
page = Nokogiri::HTML(body)
|
|
|
|
alternate_link = page.xpath('//link[@rel="alternate"]').find { |link| link['type'] == 'application/atom+xml' }
|
|
|
|
|
|
|
|
return nil if alternate_link.nil?
|
|
|
|
return [alternate_link['href'], fetch(alternate_link['href'])]
|
|
|
|
end
|
|
|
|
|
2016-09-26 16:04:05 +01:00
|
|
|
def process_headers(url, response)
|
2016-09-29 20:28:21 +01:00
|
|
|
Rails.logger.debug 'Processing link header'
|
2016-09-26 15:42:38 +01:00
|
|
|
|
2016-09-27 22:49:12 +01:00
|
|
|
link_header = LinkHeader.parse(response['Link'].is_a?(Array) ? response['Link'].first : response['Link'])
|
2016-09-26 15:42:38 +01:00
|
|
|
alternate_link = link_header.find_link(['rel', 'alternate'], ['type', 'application/atom+xml'])
|
|
|
|
|
2016-09-26 16:04:05 +01:00
|
|
|
return process_html(fetch(url)) if alternate_link.nil?
|
2016-09-26 15:42:38 +01:00
|
|
|
return [alternate_link.href, fetch(alternate_link.href)]
|
|
|
|
end
|
|
|
|
|
|
|
|
def fetch(url)
|
|
|
|
http_client.get(url).to_s
|
|
|
|
end
|
|
|
|
|
|
|
|
def http_client
|
2016-09-26 16:04:05 +01:00
|
|
|
HTTP.timeout(:per_operation, write: 20, connect: 20, read: 50).follow
|
2016-09-26 15:42:38 +01:00
|
|
|
end
|
|
|
|
end
|