Remove spam check and dependency on nilsimsa gem (#16011)
parent
7183d9a113
commit
619fad6cf8
@ -1,198 +0,0 @@ |
||||
# frozen_string_literal: true |
||||
|
||||
class SpamCheck |
||||
include Redisable |
||||
include ActionView::Helpers::TextHelper |
||||
|
||||
# Threshold over which two Nilsimsa values are considered |
||||
# to refer to the same text |
||||
NILSIMSA_COMPARE_THRESHOLD = 95 |
||||
|
||||
# Nilsimsa doesn't work well on small inputs, so below |
||||
# this size, we check only for exact matches with MD5 |
||||
NILSIMSA_MIN_SIZE = 10 |
||||
|
||||
# How long to keep the trail of digests between updates, |
||||
# there is no reason to store it forever |
||||
EXPIRE_SET_AFTER = 1.week.seconds |
||||
|
||||
# How many digests to keep in an account's trail. If it's |
||||
# too small, spam could rotate around different message templates |
||||
MAX_TRAIL_SIZE = 10 |
||||
|
||||
# How many detected duplicates to allow through before |
||||
# considering the message as spam |
||||
THRESHOLD = 5 |
||||
|
||||
def initialize(status) |
||||
@account = status.account |
||||
@status = status |
||||
end |
||||
|
||||
def skip? |
||||
disabled? || already_flagged? || trusted? || no_unsolicited_mentions? || solicited_reply? |
||||
end |
||||
|
||||
def spam? |
||||
if insufficient_data? |
||||
false |
||||
elsif nilsimsa? |
||||
digests_over_threshold?('nilsimsa') { |_, other_digest| nilsimsa_compare_value(digest, other_digest) >= NILSIMSA_COMPARE_THRESHOLD } |
||||
else |
||||
digests_over_threshold?('md5') { |_, other_digest| other_digest == digest } |
||||
end |
||||
end |
||||
|
||||
def flag! |
||||
auto_report_status! |
||||
end |
||||
|
||||
def remember! |
||||
# The scores in sorted sets don't actually have enough bits to hold an exact |
||||
# value of our snowflake IDs, so we use it only for its ordering property. To |
||||
# get the correct status ID back, we have to save it in the string value |
||||
|
||||
redis.zadd(redis_key, @status.id, digest_with_algorithm) |
||||
redis.zremrangebyrank(redis_key, 0, -(MAX_TRAIL_SIZE + 1)) |
||||
redis.expire(redis_key, EXPIRE_SET_AFTER) |
||||
end |
||||
|
||||
def reset! |
||||
redis.del(redis_key) |
||||
end |
||||
|
||||
def hashable_text |
||||
return @hashable_text if defined?(@hashable_text) |
||||
|
||||
@hashable_text = @status.text |
||||
@hashable_text = remove_mentions(@hashable_text) |
||||
@hashable_text = strip_tags(@hashable_text) unless @status.local? |
||||
@hashable_text = normalize_unicode(@status.spoiler_text + ' ' + @hashable_text) |
||||
@hashable_text = remove_whitespace(@hashable_text) |
||||
end |
||||
|
||||
def insufficient_data? |
||||
hashable_text.blank? |
||||
end |
||||
|
||||
def digest |
||||
@digest ||= begin |
||||
if nilsimsa? |
||||
Nilsimsa.new(hashable_text).hexdigest |
||||
else |
||||
Digest::MD5.hexdigest(hashable_text) |
||||
end |
||||
end |
||||
end |
||||
|
||||
def digest_with_algorithm |
||||
if nilsimsa? |
||||
['nilsimsa', digest, @status.id].join(':') |
||||
else |
||||
['md5', digest, @status.id].join(':') |
||||
end |
||||
end |
||||
|
||||
class << self |
||||
def perform(status) |
||||
spam_check = new(status) |
||||
|
||||
return if spam_check.skip? |
||||
|
||||
if spam_check.spam? |
||||
spam_check.flag! |
||||
else |
||||
spam_check.remember! |
||||
end |
||||
end |
||||
end |
||||
|
||||
private |
||||
|
||||
def disabled? |
||||
!Setting.spam_check_enabled |
||||
end |
||||
|
||||
def remove_mentions(text) |
||||
return text.gsub(Account::MENTION_RE, '') if @status.local? |
||||
|
||||
Nokogiri::HTML.fragment(text).tap do |html| |
||||
mentions = @status.mentions.map { |mention| ActivityPub::TagManager.instance.url_for(mention.account) } |
||||
|
||||
html.traverse do |element| |
||||
element.unlink if element.name == 'a' && mentions.include?(element['href']) |
||||
end |
||||
end.to_s |
||||
end |
||||
|
||||
def normalize_unicode(text) |
||||
text.unicode_normalize(:nfkc).downcase |
||||
end |
||||
|
||||
def remove_whitespace(text) |
||||
text.gsub(/\s+/, ' ').strip |
||||
end |
||||
|
||||
def auto_report_status! |
||||
status_ids = Status.where(visibility: %i(public unlisted)).where(id: matching_status_ids).pluck(:id) + [@status.id] if @status.distributable? |
||||
ReportService.new.call(Account.representative, @account, status_ids: status_ids, comment: I18n.t('spam_check.spam_detected')) |
||||
end |
||||
|
||||
def already_flagged? |
||||
@account.silenced? || @account.targeted_reports.unresolved.where(account_id: -99).exists? |
||||
end |
||||
|
||||
def trusted? |
||||
@account.trust_level > Account::TRUST_LEVELS[:untrusted] || (@account.local? && @account.user_staff?) |
||||
end |
||||
|
||||
def no_unsolicited_mentions? |
||||
@status.mentions.all? { |mention| mention.silent? || (!@account.local? && !mention.account.local?) || mention.account.following?(@account) } |
||||
end |
||||
|
||||
def solicited_reply? |
||||
!@status.thread.nil? && @status.thread.mentions.where(account: @account).exists? |
||||
end |
||||
|
||||
def nilsimsa_compare_value(first, second) |
||||
first = [first].pack('H*') |
||||
second = [second].pack('H*') |
||||
bits = 0 |
||||
|
||||
0.upto(31) do |i| |
||||
bits += Nilsimsa::POPC[255 & (first[i].ord ^ second[i].ord)].ord |
||||
end |
||||
|
||||
128 - bits # -128 <= Nilsimsa Compare Value <= 128 |
||||
end |
||||
|
||||
def nilsimsa? |
||||
hashable_text.size > NILSIMSA_MIN_SIZE |
||||
end |
||||
|
||||
def other_digests |
||||
redis.zrange(redis_key, 0, -1) |
||||
end |
||||
|
||||
def digests_over_threshold?(filter_algorithm) |
||||
other_digests.select do |record| |
||||
algorithm, other_digest, status_id = record.split(':') |
||||
|
||||
next unless algorithm == filter_algorithm |
||||
|
||||
yield algorithm, other_digest, status_id |
||||
end.size >= THRESHOLD |
||||
end |
||||
|
||||
def matching_status_ids |
||||
if nilsimsa? |
||||
other_digests.filter_map { |record| record.split(':')[2] if record.start_with?('nilsimsa') && nilsimsa_compare_value(digest, record.split(':')[1]) >= NILSIMSA_COMPARE_THRESHOLD } |
||||
else |
||||
other_digests.filter_map { |record| record.split(':')[2] if record.start_with?('md5') && record.split(':')[1] == digest } |
||||
end |
||||
end |
||||
|
||||
def redis_key |
||||
@redis_key ||= "spam_check:#{@account.id}" |
||||
end |
||||
end |
@ -1,192 +0,0 @@ |
||||
# frozen_string_literal: true |
||||
|
||||
require 'rails_helper' |
||||
|
||||
RSpec.describe SpamCheck do |
||||
let!(:sender) { Fabricate(:account) } |
||||
let!(:alice) { Fabricate(:account, username: 'alice') } |
||||
let!(:bob) { Fabricate(:account, username: 'bob') } |
||||
|
||||
def status_with_html(text, options = {}) |
||||
status = PostStatusService.new.call(sender, { text: text }.merge(options)) |
||||
status.update_columns(text: Formatter.instance.format(status), local: false) |
||||
status |
||||
end |
||||
|
||||
describe '#hashable_text' do |
||||
it 'removes mentions from HTML for remote statuses' do |
||||
status = status_with_html('@alice Hello') |
||||
expect(described_class.new(status).hashable_text).to eq 'hello' |
||||
end |
||||
|
||||
it 'removes mentions from text for local statuses' do |
||||
status = PostStatusService.new.call(alice, text: "Hey @#{sender.username}, how are you?") |
||||
expect(described_class.new(status).hashable_text).to eq 'hey , how are you?' |
||||
end |
||||
end |
||||
|
||||
describe '#insufficient_data?' do |
||||
it 'returns true when there is no text' do |
||||
status = status_with_html('@alice') |
||||
expect(described_class.new(status).insufficient_data?).to be true |
||||
end |
||||
|
||||
it 'returns false when there is text' do |
||||
status = status_with_html('@alice h') |
||||
expect(described_class.new(status).insufficient_data?).to be false |
||||
end |
||||
end |
||||
|
||||
describe '#digest' do |
||||
it 'returns a string' do |
||||
status = status_with_html('@alice Hello world') |
||||
expect(described_class.new(status).digest).to be_a String |
||||
end |
||||
end |
||||
|
||||
describe '#spam?' do |
||||
it 'returns false for a unique status' do |
||||
status = status_with_html('@alice Hello') |
||||
expect(described_class.new(status).spam?).to be false |
||||
end |
||||
|
||||
it 'returns false for different statuses to the same recipient' do |
||||
status1 = status_with_html('@alice Hello') |
||||
described_class.new(status1).remember! |
||||
status2 = status_with_html('@alice Are you available to talk?') |
||||
expect(described_class.new(status2).spam?).to be false |
||||
end |
||||
|
||||
it 'returns false for statuses with different content warnings' do |
||||
status1 = status_with_html('@alice Are you available to talk?') |
||||
described_class.new(status1).remember! |
||||
status2 = status_with_html('@alice Are you available to talk?', spoiler_text: 'This is a completely different matter than what I was talking about previously, I swear!') |
||||
expect(described_class.new(status2).spam?).to be false |
||||
end |
||||
|
||||
it 'returns false for different statuses to different recipients' do |
||||
status1 = status_with_html('@alice How is it going?') |
||||
described_class.new(status1).remember! |
||||
status2 = status_with_html('@bob Are you okay?') |
||||
expect(described_class.new(status2).spam?).to be false |
||||
end |
||||
|
||||
it 'returns false for very short different statuses to different recipients' do |
||||
status1 = status_with_html('@alice 🙄') |
||||
described_class.new(status1).remember! |
||||
status2 = status_with_html('@bob Huh?') |
||||
expect(described_class.new(status2).spam?).to be false |
||||
end |
||||
|
||||
it 'returns false for statuses with no text' do |
||||
status1 = status_with_html('@alice') |
||||
described_class.new(status1).remember! |
||||
status2 = status_with_html('@bob') |
||||
expect(described_class.new(status2).spam?).to be false |
||||
end |
||||
|
||||
it 'returns true for duplicate statuses to the same recipient' do |
||||
described_class::THRESHOLD.times do |
||||
status1 = status_with_html('@alice Hello') |
||||
described_class.new(status1).remember! |
||||
end |
||||
|
||||
status2 = status_with_html('@alice Hello') |
||||
expect(described_class.new(status2).spam?).to be true |
||||
end |
||||
|
||||
it 'returns true for duplicate statuses to different recipients' do |
||||
described_class::THRESHOLD.times do |
||||
status1 = status_with_html('@alice Hello') |
||||
described_class.new(status1).remember! |
||||
end |
||||
|
||||
status2 = status_with_html('@bob Hello') |
||||
expect(described_class.new(status2).spam?).to be true |
||||
end |
||||
|
||||
it 'returns true for nearly identical statuses with random numbers' do |
||||
source_text = 'Sodium, atomic number 11, was first isolated by Humphry Davy in 1807. A chemical component of salt, he named it Na in honor of the saltiest region on earth, North America.' |
||||
|
||||
described_class::THRESHOLD.times do |
||||
status1 = status_with_html('@alice ' + source_text + ' 1234') |
||||
described_class.new(status1).remember! |
||||
end |
||||
|
||||
status2 = status_with_html('@bob ' + source_text + ' 9568') |
||||
expect(described_class.new(status2).spam?).to be true |
||||
end |
||||
end |
||||
|
||||
describe '#skip?' do |
||||
it 'returns true when the sender is already silenced' do |
||||
status = status_with_html('@alice Hello') |
||||
sender.silence! |
||||
expect(described_class.new(status).skip?).to be true |
||||
end |
||||
|
||||
it 'returns true when the mentioned person follows the sender' do |
||||
status = status_with_html('@alice Hello') |
||||
alice.follow!(sender) |
||||
expect(described_class.new(status).skip?).to be true |
||||
end |
||||
|
||||
it 'returns false when even one mentioned person doesn\'t follow the sender' do |
||||
status = status_with_html('@alice @bob Hello') |
||||
alice.follow!(sender) |
||||
expect(described_class.new(status).skip?).to be false |
||||
end |
||||
|
||||
it 'returns true when the sender is replying to a status that mentions the sender' do |
||||
parent = PostStatusService.new.call(alice, text: "Hey @#{sender.username}, how are you?") |
||||
status = status_with_html('@alice @bob Hello', thread: parent) |
||||
expect(described_class.new(status).skip?).to be true |
||||
end |
||||
end |
||||
|
||||
describe '#remember!' do |
||||
let(:status) { status_with_html('@alice') } |
||||
let(:spam_check) { described_class.new(status) } |
||||
let(:redis_key) { spam_check.send(:redis_key) } |
||||
|
||||
it 'remembers' do |
||||
expect(Redis.current.exists?(redis_key)).to be true |
||||
spam_check.remember! |
||||
expect(Redis.current.exists?(redis_key)).to be true |
||||
end |
||||
end |
||||
|
||||
describe '#reset!' do |
||||
let(:status) { status_with_html('@alice') } |
||||
let(:spam_check) { described_class.new(status) } |
||||
let(:redis_key) { spam_check.send(:redis_key) } |
||||
|
||||
before do |
||||
spam_check.remember! |
||||
end |
||||
|
||||
it 'resets' do |
||||
expect(Redis.current.exists?(redis_key)).to be true |
||||
spam_check.reset! |
||||
expect(Redis.current.exists?(redis_key)).to be false |
||||
end |
||||
end |
||||
|
||||
describe '#flag!' do |
||||
let!(:status1) { status_with_html('@alice General Kenobi you are a bold one') } |
||||
let!(:status2) { status_with_html('@alice @bob General Kenobi, you are a bold one') } |
||||
|
||||
before do |
||||
described_class.new(status1).remember! |
||||
described_class.new(status2).flag! |
||||
end |
||||
|
||||
it 'creates a report about the account' do |
||||
expect(sender.targeted_reports.unresolved.count).to eq 1 |
||||
end |
||||
|
||||
it 'attaches both matching statuses to the report' do |
||||
expect(sender.targeted_reports.first.status_ids).to include(status1.id, status2.id) |
||||
end |
||||
end |
||||
end |
Loading…
Reference in new issue