mastodon/app/services/process_feed_service.rb

238 lines
6.8 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
2016-02-24 12:57:29 +01:00
class ProcessFeedService < BaseService
ACTIVITY_NS = 'http://activitystrea.ms/spec/1.0/'
THREAD_NS = 'http://purl.org/syndication/thread/1.0'
2016-10-10 18:05:52 +02:00
2016-02-20 22:53:20 +01:00
def call(body, account)
xml = Nokogiri::XML(body)
xml.encoding = 'utf-8'
2016-11-08 01:32:34 +01:00
update_author(xml, account)
process_entries(xml, account)
end
2016-02-20 22:53:20 +01:00
private
2016-11-08 01:32:34 +01:00
def update_author(xml, account)
return if xml.at_xpath('/xmlns:feed').nil?
UpdateRemoteProfileService.new.call(xml.at_xpath('/xmlns:feed'), account, true)
2016-11-08 01:32:34 +01:00
end
2016-11-08 01:32:34 +01:00
def process_entries(xml, account)
xml.xpath('//xmlns:entry').reverse_each.map { |entry| ProcessEntry.new.call(entry, account) }.compact
end
2016-03-16 10:46:15 +01:00
2016-11-08 01:32:34 +01:00
class ProcessEntry
def call(xml, account)
@account = account
@xml = xml
2016-11-08 01:32:34 +01:00
return if skip_unsupported_type?
2016-11-08 01:32:34 +01:00
case verb
when :post, :share
return create_status
when :delete
return delete_status
end
rescue ActiveRecord::RecordInvalid => e
Rails.logger.debug "Nothing was saved for #{id} because: #{e}"
nil
end
2016-02-25 00:17:01 +01:00
2016-11-08 01:32:34 +01:00
private
2016-11-08 01:32:34 +01:00
def create_status
Rails.logger.debug "Creating remote status #{id}"
status = status_from_xml(@xml)
2016-11-05 15:20:05 +01:00
2016-11-08 01:32:34 +01:00
if verb == :share
2016-11-08 01:48:17 +01:00
original_status = status_from_xml(@xml.at_xpath('.//activity:object', activity: ACTIVITY_NS))
2016-11-08 01:32:34 +01:00
status.reblog = original_status
2016-11-08 19:37:08 +01:00
if original_status.nil?
status.destroy
return nil
elsif original_status.reblog?
status.reblog = original_status.reblog
2016-11-08 19:37:08 +01:00
end
2016-11-08 01:32:34 +01:00
end
2016-11-08 01:32:34 +01:00
status.save!
NotifyService.new.call(status.reblog.account, status) if status.reblog? && status.reblog.account.local?
2016-11-08 01:32:34 +01:00
Rails.logger.debug "Queuing remote status #{status.id} (#{id}) for distribution"
DistributionWorker.perform_async(status.id)
2016-11-08 01:32:34 +01:00
status
end
2016-11-08 01:32:34 +01:00
def delete_status
Rails.logger.debug "Deleting remote status #{id}"
status = Status.find_by(uri: id)
RemoveStatusService.new.call(status) unless status.nil?
nil
end
2016-11-08 01:32:34 +01:00
def skip_unsupported_type?
!([:post, :share, :delete].include?(verb) && [:activity, :note, :comment].include?(type))
end
2016-11-08 01:32:34 +01:00
def status_from_xml(entry)
# Return early if status already exists in db
status = find_status(id(entry))
return status unless status.nil?
# If status embeds an author, find that author
# If that author cannot be found, don't record the status (do not misattribute)
if account?(entry)
begin
account = find_or_resolve_account(acct(entry))
return nil if account.nil?
rescue Goldfinger::Error
return nil
end
else
account = @account
2016-11-08 19:37:08 +01:00
end
status = Status.create!(
2016-11-08 01:32:34 +01:00
uri: id(entry),
url: url(entry),
2016-11-08 19:37:08 +01:00
account: account,
2016-11-08 01:32:34 +01:00
text: content(entry),
created_at: published(entry)
)
2016-11-08 01:32:34 +01:00
if thread?(entry)
2016-11-08 19:37:08 +01:00
Rails.logger.debug "Trying to attach #{status.id} (#{id(entry)}) to #{thread(entry).first}"
2016-11-08 01:32:34 +01:00
status.thread = find_or_resolve_status(status, *thread(entry))
end
2016-11-08 01:32:34 +01:00
mentions_from_xml(status, entry)
hashtags_from_xml(status, entry)
media_from_xml(status, entry)
2016-09-29 21:28:21 +02:00
2016-11-08 01:32:34 +01:00
status
end
2016-11-08 01:32:34 +01:00
def find_or_resolve_account(acct)
FollowRemoteAccountService.new.call(acct)
end
2016-11-08 01:32:34 +01:00
def find_or_resolve_status(parent, uri, url)
status = find_status(uri)
ThreadResolveWorker.perform_async(parent.id, url) if status.nil?
2016-11-08 01:32:34 +01:00
status
end
2016-11-08 01:32:34 +01:00
def find_status(uri)
if TagManager.instance.local_id?(uri)
local_id = TagManager.instance.unique_tag_to_local_id(uri, 'Status')
return Status.find(local_id)
end
2016-11-08 01:32:34 +01:00
Status.find_by(uri: uri)
end
2016-11-08 01:32:34 +01:00
def mentions_from_xml(parent, xml)
processed_account_ids = []
2016-11-08 01:32:34 +01:00
xml.xpath('./xmlns:link[@rel="mentioned"]').each do |link|
next if link['href'] == 'http://activityschema.org/collection/public'
2016-11-08 01:32:34 +01:00
url = Addressable::URI.parse(link['href'])
2016-11-08 01:32:34 +01:00
mentioned_account = if TagManager.instance.local_domain?(url.host)
Account.find_local(url.path.gsub('/users/', ''))
else
Account.find_by(url: link['href']) || FetchRemoteAccountService.new.call(link['href'])
end
2016-09-29 21:28:21 +02:00
2016-11-08 01:32:34 +01:00
next if mentioned_account.nil? || processed_account_ids.include?(mentioned_account.id)
mention = mentioned_account.mentions.where(status: parent).first_or_create(status: parent)
# Notify local user
NotifyService.new.call(mentioned_account, mention) if mentioned_account.local?
2016-02-24 17:23:59 +01:00
2016-11-08 01:32:34 +01:00
# So we can skip duplicate mentions
processed_account_ids << mentioned_account.id
end
end
2016-11-08 01:32:34 +01:00
def hashtags_from_xml(parent, xml)
2016-11-08 19:37:08 +01:00
tags = xml.xpath('./xmlns:category').map { |category| category['term'] }.select { |t| !t.blank? }
2016-11-08 01:32:34 +01:00
ProcessHashtagsService.new.call(parent, tags)
end
2016-02-20 22:53:20 +01:00
2016-11-08 01:32:34 +01:00
def media_from_xml(parent, xml)
xml.xpath('./xmlns:link[@rel="enclosure"]').each do |link|
next unless link['href']
2016-03-16 10:46:15 +01:00
2016-11-08 01:32:34 +01:00
media = MediaAttachment.where(status: parent, remote_url: link['href']).first_or_initialize(account: parent.account, status: parent, remote_url: link['href'])
2016-02-20 22:53:20 +01:00
2016-11-08 01:32:34 +01:00
begin
media.file_remote_url = link['href']
media.save
rescue OpenURI::HTTPError, Paperclip::Errors::NotIdentifiedByImageMagickError
2016-11-08 01:32:34 +01:00
next
end
end
2016-02-20 22:53:20 +01:00
end
2016-02-24 17:23:59 +01:00
2016-11-08 01:32:34 +01:00
def id(xml = @xml)
xml.at_xpath('./xmlns:id').content
2016-02-24 17:23:59 +01:00
end
2016-11-08 01:32:34 +01:00
def verb(xml = @xml)
raw = xml.at_xpath('./activity:verb', activity: ACTIVITY_NS).content
raw.gsub('http://activitystrea.ms/schema/1.0/', '').gsub('http://ostatus.org/schema/1.0/', '').to_sym
rescue
:post
end
2016-11-08 01:32:34 +01:00
def type(xml = @xml)
raw = xml.at_xpath('./activity:object-type', activity: ACTIVITY_NS).content
raw.gsub('http://activitystrea.ms/schema/1.0/', '').gsub('http://ostatus.org/schema/1.0/', '').to_sym
rescue
:activity
end
2016-11-08 01:32:34 +01:00
def url(xml = @xml)
link = xml.at_xpath('./xmlns:link[@rel="alternate"]')
link.nil? ? nil : link['href']
2016-11-08 01:32:34 +01:00
end
2016-11-08 01:32:34 +01:00
def content(xml = @xml)
xml.at_xpath('./xmlns:content').content
end
2016-02-24 17:23:59 +01:00
2016-11-08 01:32:34 +01:00
def published(xml = @xml)
xml.at_xpath('./xmlns:published').content
end
2016-02-24 17:23:59 +01:00
2016-11-08 01:32:34 +01:00
def thread?(xml = @xml)
!xml.at_xpath('./thr:in-reply-to', thr: THREAD_NS).nil?
end
2016-11-08 01:32:34 +01:00
def thread(xml = @xml)
thr = xml.at_xpath('./thr:in-reply-to', thr: THREAD_NS)
[thr['ref'], thr['href']]
end
2016-11-08 01:32:34 +01:00
def account?(xml = @xml)
!xml.at_xpath('./xmlns:author').nil?
end
2016-02-25 00:17:01 +01:00
2016-11-08 01:32:34 +01:00
def acct(xml = @xml)
username = xml.at_xpath('./xmlns:author/xmlns:name').content
url = xml.at_xpath('./xmlns:author/xmlns:uri').content
domain = Addressable::URI.parse(url).host
2016-11-08 01:32:34 +01:00
"#{username}@#{domain}"
end
end
2016-02-20 22:53:20 +01:00
end