1
0
Fork 0
forked from fedi/mastodon
mastodon/app/services/process_feed_service.rb

305 lines
8.7 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
2016-02-24 11:57:29 +00:00
class ProcessFeedService < BaseService
2016-02-20 21:53:20 +00:00
def call(body, account)
xml = Nokogiri::XML(body)
xml.encoding = 'utf-8'
2016-11-08 00:32:34 +00:00
update_author(body, account)
2016-11-08 00:32:34 +00:00
process_entries(xml, account)
end
2016-02-20 21:53:20 +00:00
private
def update_author(body, account)
RemoteProfileUpdateWorker.perform_async(account.id, body.force_encoding('UTF-8'), true)
2016-11-08 00:32:34 +00:00
end
2016-11-08 00:32:34 +00:00
def process_entries(xml, account)
xml.xpath('//xmlns:entry', xmlns: TagManager::XMLNS).reverse_each.map { |entry| ProcessEntry.new.call(entry, account) }.compact
2016-11-08 00:32:34 +00:00
end
2016-03-16 09:46:15 +00:00
2016-11-08 00:32:34 +00:00
class ProcessEntry
include AuthorExtractor
2016-11-08 00:32:34 +00:00
def call(xml, account)
@account = account
@xml = xml
2016-11-08 00:32:34 +00:00
return if skip_unsupported_type?
2016-11-08 00:32:34 +00:00
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
2016-02-24 23:17:01 +00:00
2016-11-08 00:32:34 +00:00
private
2016-11-08 00:32:34 +00:00
def create_status
if redis.exists("delete_upon_arrival:#{id}")
Rails.logger.debug "Delete for status #{id} was queued, ignoring"
return
end
status, just_created = nil
2016-11-08 00:32:34 +00:00
Rails.logger.debug "Creating remote status #{id}"
2016-11-05 14:20:05 +00:00
if verb == :share
original_status = shared_status_from_xml(@xml.at_xpath('.//activity:object', activity: TagManager::AS_XMLNS))
return nil if original_status.nil?
end
ApplicationRecord.transaction do
status, just_created = status_from_xml(@xml)
return if status.nil?
return status unless just_created
if verb == :share
status.reblog = original_status.reblog? ? original_status.reblog : original_status
2016-11-08 18:37:08 +00:00
end
status.save!
end
if thread?(@xml) && status.thread.nil?
Rails.logger.debug "Trying to attach #{status.id} (#{id(@xml)}) to #{thread(@xml).first}"
ThreadResolveWorker.perform_async(status.id, thread(@xml).second)
end
notify_about_mentions!(status) unless status.reblog?
notify_about_reblog!(status) if status.reblog? && status.reblog.account.local?
2016-11-08 00:32:34 +00:00
Rails.logger.debug "Queuing remote status #{status.id} (#{id}) for distribution"
LinkCrawlWorker.perform_async(status.id) unless status.spoiler_text?
DistributionWorker.perform_async(status.id)
2016-11-08 00:32:34 +00:00
status
end
def notify_about_mentions!(status)
status.mentions.includes(:account).each do |mention|
mentioned_account = mention.account
next unless mentioned_account.local?
NotifyService.new.call(mentioned_account, mention)
end
end
def notify_about_reblog!(status)
NotifyService.new.call(status.reblog.account, status)
end
2016-11-08 00:32:34 +00:00
def delete_status
Rails.logger.debug "Deleting remote status #{id}"
status = Status.find_by(uri: id)
if status.nil?
redis.setex("delete_upon_arrival:#{id}", 6 * 3_600, id)
else
RemoveStatusService.new.call(status)
end
2016-11-08 00:32:34 +00:00
nil
end
2016-11-08 00:32:34 +00:00
def skip_unsupported_type?
!([:post, :share, :delete].include?(verb) && [:activity, :note, :comment].include?(type))
end
def shared_status_from_xml(entry)
status = find_status(id(entry))
return status unless status.nil?
FetchRemoteStatusService.new.call(url(entry))
end
2016-11-08 00:32:34 +00:00
def status_from_xml(entry)
# Return early if status already exists in db
status = find_status(id(entry))
return [status, false] unless status.nil?
2016-11-08 00:32:34 +00:00
# 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 = author_from_xml(entry)
return [nil, false] if account.nil?
rescue Goldfinger::Error
return [nil, false]
end
else
account = @account
2016-11-08 18:37:08 +00:00
end
return [nil, false] if account.suspended?
2016-12-05 21:59:30 +00:00
status = Status.create!(
2016-11-08 00:32:34 +00:00
uri: id(entry),
url: url(entry),
2016-11-08 18:37:08 +00:00
account: account,
2016-11-08 00:32:34 +00:00
text: content(entry),
spoiler_text: content_warning(entry),
created_at: published(entry),
reply: thread?(entry),
language: content_language(entry),
visibility: visibility_scope(entry),
conversation: find_or_create_conversation(entry),
thread: thread?(entry) ? find_status(thread(entry).first) : nil
)
2016-11-08 00:32:34 +00:00
mentions_from_xml(status, entry)
hashtags_from_xml(status, entry)
media_from_xml(status, entry)
2016-09-29 19:28:21 +00:00
[status, true]
2016-11-08 00:32:34 +00:00
end
def find_or_create_conversation(xml)
uri = xml.at_xpath('./ostatus:conversation', ostatus: TagManager::OS_XMLNS)&.attribute('ref')&.content
return if uri.nil?
if TagManager.instance.local_id?(uri)
local_id = TagManager.instance.unique_tag_to_local_id(uri, 'Conversation')
return Conversation.find_by(id: local_id)
end
Conversation.find_by(uri: uri) || Conversation.create!(uri: uri)
end
2016-11-08 00:32:34 +00:00
def find_status(uri)
if TagManager.instance.local_id?(uri)
local_id = TagManager.instance.unique_tag_to_local_id(uri, 'Status')
return Status.find_by(id: local_id)
end
2016-11-08 00:32:34 +00:00
Status.find_by(uri: uri)
end
2016-11-08 00:32:34 +00:00
def mentions_from_xml(parent, xml)
processed_account_ids = []
xml.xpath('./xmlns:link[@rel="mentioned"]', xmlns: TagManager::XMLNS).each do |link|
next if [TagManager::TYPES[:group], TagManager::TYPES[:collection]].include? link['ostatus:object-type']
mentioned_account = account_from_href(link['href'])
2016-09-29 19:28:21 +00:00
2016-11-08 00:32:34 +00:00
next if mentioned_account.nil? || processed_account_ids.include?(mentioned_account.id)
mentioned_account.mentions.where(status: parent).first_or_create(status: parent)
2016-02-24 16:23:59 +00:00
2016-11-08 00:32:34 +00:00
# So we can skip duplicate mentions
processed_account_ids << mentioned_account.id
end
end
def account_from_href(href)
url = Addressable::URI.parse(href).normalize
if TagManager.instance.web_domain?(url.host)
Account.find_local(url.path.gsub('/users/', ''))
else
Account.where(uri: href).or(Account.where(url: href)).first || FetchRemoteAccountService.new.call(href)
end
end
2016-11-08 00:32:34 +00:00
def hashtags_from_xml(parent, xml)
tags = xml.xpath('./xmlns:category', xmlns: TagManager::XMLNS).map { |category| category['term'] }.select(&:present?)
2016-11-08 00:32:34 +00:00
ProcessHashtagsService.new.call(parent, tags)
end
2016-02-20 21:53:20 +00:00
2016-11-08 00:32:34 +00:00
def media_from_xml(parent, xml)
do_not_download = DomainBlock.find_by(domain: parent.account.domain)&.reject_media?
xml.xpath('./xmlns:link[@rel="enclosure"]', xmlns: TagManager::XMLNS).each do |link|
2016-11-08 00:32:34 +00:00
next unless link['href']
2016-03-16 09:46:15 +00:00
2016-11-08 00:32:34 +00:00
media = MediaAttachment.where(status: parent, remote_url: link['href']).first_or_initialize(account: parent.account, status: parent, remote_url: link['href'])
parsed_url = Addressable::URI.parse(link['href']).normalize
next if !%w(http https).include?(parsed_url.scheme) || parsed_url.host.empty?
media.save
next if do_not_download
2016-02-20 21:53:20 +00:00
2016-11-08 00:32:34 +00:00
begin
media.file_remote_url = link['href']
media.save!
rescue ActiveRecord::RecordInvalid
2016-11-08 00:32:34 +00:00
next
end
end
2016-02-20 21:53:20 +00:00
end
2016-02-24 16:23:59 +00:00
2016-11-08 00:32:34 +00:00
def id(xml = @xml)
xml.at_xpath('./xmlns:id', xmlns: TagManager::XMLNS).content
2016-02-24 16:23:59 +00:00
end
2016-11-08 00:32:34 +00:00
def verb(xml = @xml)
raw = xml.at_xpath('./activity:verb', activity: TagManager::AS_XMLNS).content
TagManager::VERBS.key(raw)
2016-11-08 00:32:34 +00:00
rescue
:post
end
2016-11-08 00:32:34 +00:00
def type(xml = @xml)
raw = xml.at_xpath('./activity:object-type', activity: TagManager::AS_XMLNS).content
TagManager::TYPES.key(raw)
2016-11-08 00:32:34 +00:00
rescue
:activity
end
2016-11-08 00:32:34 +00:00
def url(xml = @xml)
link = xml.at_xpath('./xmlns:link[@rel="alternate"]', xmlns: TagManager::XMLNS)
link.nil? ? nil : link['href']
2016-11-08 00:32:34 +00:00
end
2016-11-08 00:32:34 +00:00
def content(xml = @xml)
xml.at_xpath('./xmlns:content', xmlns: TagManager::XMLNS).content
2016-11-08 00:32:34 +00:00
end
2016-02-24 16:23:59 +00:00
def content_language(xml = @xml)
xml.at_xpath('./xmlns:content', xmlns: TagManager::XMLNS)['xml:lang']&.presence || 'en'
end
def content_warning(xml = @xml)
xml.at_xpath('./xmlns:summary', xmlns: TagManager::XMLNS)&.content || ''
end
def visibility_scope(xml = @xml)
xml.at_xpath('./mastodon:scope', mastodon: TagManager::MTDN_XMLNS)&.content&.to_sym || :public
end
2016-11-08 00:32:34 +00:00
def published(xml = @xml)
xml.at_xpath('./xmlns:published', xmlns: TagManager::XMLNS).content
2016-11-08 00:32:34 +00:00
end
2016-02-24 16:23:59 +00:00
2016-11-08 00:32:34 +00:00
def thread?(xml = @xml)
!xml.at_xpath('./thr:in-reply-to', thr: TagManager::THR_XMLNS).nil?
2016-11-08 00:32:34 +00:00
end
2016-11-08 00:32:34 +00:00
def thread(xml = @xml)
thr = xml.at_xpath('./thr:in-reply-to', thr: TagManager::THR_XMLNS)
2016-11-08 00:32:34 +00:00
[thr['ref'], thr['href']]
end
2016-11-08 00:32:34 +00:00
def account?(xml = @xml)
!xml.at_xpath('./xmlns:author', xmlns: TagManager::XMLNS).nil?
2016-11-08 00:32:34 +00:00
end
def redis
Redis.current
end
end
2016-02-20 21:53:20 +00:00
end