discourse/lib/middleware/anonymous_cache.rb

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

378 lines
11 KiB
Ruby
Raw Normal View History

# frozen_string_literal: true
require "mobile_detection"
require "crawler_detection"
require "guardian"
require "http_language_parser"
module Middleware
class AnonymousCache
def self.cache_key_segments
@@cache_key_segments ||= {
m: "key_is_mobile?",
c: "key_is_crawler?",
o: "key_is_old_browser?",
d: "key_is_modern_mobile_device?",
b: "key_has_brotli?",
t: "key_cache_theme_ids",
ca: "key_compress_anon",
l: "key_locale",
}
end
# Compile a string builder method that will be called to create
# an anonymous cache key
def self.compile_key_builder
method = +"def self.__compiled_key_builder(h)\n \""
cache_key_segments.each do |k, v|
raise "Invalid key name" unless k =~ /\A[a-z]+\z/
raise "Invalid method name" unless v =~ /\Akey_[a-z_\?]+\z/
method << "|#{k}=#\{h.#{v}}"
end
method << "\"\nend"
2021-10-27 04:39:28 -04:00
eval(method) # rubocop:disable Security/Eval
@@compiled = true
end
def self.build_cache_key(helper)
compile_key_builder unless defined?(@@compiled)
__compiled_key_builder(helper)
end
def self.anon_cache(env, duration)
env["ANON_CACHE_DURATION"] = duration
end
# This gives us an API to insert anonymous cache segments
class Helper
RACK_SESSION = "rack.session"
USER_AGENT = "HTTP_USER_AGENT"
ACCEPT_ENCODING = "HTTP_ACCEPT_ENCODING"
DISCOURSE_RENDER = "HTTP_DISCOURSE_RENDER"
REDIS_STORE_SCRIPT = DiscourseRedis::EvalHelper.new <<~LUA
local current = redis.call("incr", KEYS[1])
redis.call("expire",KEYS[1],ARGV[1])
return current
LUA
FEATURE: Apply rate limits per user instead of IP for trusted users (#14706) Currently, Discourse rate limits all incoming requests by the IP address they originate from regardless of the user making the request. This can be frustrating if there are multiple users using Discourse simultaneously while sharing the same IP address (e.g. employees in an office). This commit implements a new feature to make Discourse apply rate limits by user id rather than IP address for users at or higher than the configured trust level (1 is the default). For example, let's say a Discourse instance is configured to allow 200 requests per minute per IP address, and we have 10 users at trust level 4 using Discourse simultaneously from the same IP address. Before this feature, the 10 users could only make a total of 200 requests per minute before they got rate limited. But with the new feature, each user is allowed to make 200 requests per minute because the rate limits are applied on user id rather than the IP address. The minimum trust level for applying user-id-based rate limits can be configured by the `skip_per_ip_rate_limit_trust_level` global setting. The default is 1, but it can be changed by either adding the `DISCOURSE_SKIP_PER_IP_RATE_LIMIT_TRUST_LEVEL` environment variable with the desired value to your `app.yml`, or changing the setting's value in the `discourse.conf` file. Requests made with API keys are still rate limited by IP address and the relevant global settings that control API keys rate limits. Before this commit, Discourse's auth cookie (`_t`) was simply a 32 characters string that Discourse used to lookup the current user from the database and the cookie contained no additional information about the user. However, we had to change the cookie content in this commit so we could identify the user from the cookie without making a database query before the rate limits logic and avoid introducing a bottleneck on busy sites. Besides the 32 characters auth token, the cookie now includes the user id, trust level and the cookie's generation date, and we encrypt/sign the cookie to prevent tampering. Internal ticket number: t54739.
2021-11-17 15:27:30 -05:00
def initialize(env, request = nil)
@env = env
FEATURE: Apply rate limits per user instead of IP for trusted users (#14706) Currently, Discourse rate limits all incoming requests by the IP address they originate from regardless of the user making the request. This can be frustrating if there are multiple users using Discourse simultaneously while sharing the same IP address (e.g. employees in an office). This commit implements a new feature to make Discourse apply rate limits by user id rather than IP address for users at or higher than the configured trust level (1 is the default). For example, let's say a Discourse instance is configured to allow 200 requests per minute per IP address, and we have 10 users at trust level 4 using Discourse simultaneously from the same IP address. Before this feature, the 10 users could only make a total of 200 requests per minute before they got rate limited. But with the new feature, each user is allowed to make 200 requests per minute because the rate limits are applied on user id rather than the IP address. The minimum trust level for applying user-id-based rate limits can be configured by the `skip_per_ip_rate_limit_trust_level` global setting. The default is 1, but it can be changed by either adding the `DISCOURSE_SKIP_PER_IP_RATE_LIMIT_TRUST_LEVEL` environment variable with the desired value to your `app.yml`, or changing the setting's value in the `discourse.conf` file. Requests made with API keys are still rate limited by IP address and the relevant global settings that control API keys rate limits. Before this commit, Discourse's auth cookie (`_t`) was simply a 32 characters string that Discourse used to lookup the current user from the database and the cookie contained no additional information about the user. However, we had to change the cookie content in this commit so we could identify the user from the cookie without making a database query before the rate limits logic and avoid introducing a bottleneck on busy sites. Besides the 32 characters auth token, the cookie now includes the user id, trust level and the cookie's generation date, and we encrypt/sign the cookie to prevent tampering. Internal ticket number: t54739.
2021-11-17 15:27:30 -05:00
@request = request || Rack::Request.new(@env)
end
def blocked_crawler?
@request.get? && !@request.xhr? && !@request.path.ends_with?("robots.txt") &&
!@request.path.ends_with?("srv/status") &&
@request[Auth::DefaultCurrentUserProvider::API_KEY].nil? &&
@env[Auth::DefaultCurrentUserProvider::USER_API_KEY].nil? &&
@env[Auth::DefaultCurrentUserProvider::HEADER_API_KEY].nil? &&
CrawlerDetection.is_blocked_crawler?(@env[USER_AGENT])
end
2014-01-08 23:11:04 -05:00
def is_mobile=(val)
@is_mobile = val ? :true : :false
end
2019-12-05 15:07:22 -05:00
def is_mobile?
@is_mobile ||=
begin
session = @env[RACK_SESSION]
# don't initialize params until later
# otherwise you get a broken params on the request
params = {}
MobileDetection.resolve_mobile_view!(@env[USER_AGENT], params, session) ? :true : :false
end
@is_mobile == :true
end
2019-12-05 15:07:22 -05:00
alias_method :key_is_mobile?, :is_mobile?
def key_has_brotli?
@has_brotli ||=
begin
@env[ACCEPT_ENCODING].to_s =~ /br/ ? :true : :false
end
@has_brotli == :true
end
def key_locale
if locale = Discourse.anonymous_locale(@request)
locale
else
"" # No need to key, it is the same for all anon users
end
end
2019-12-05 15:07:22 -05:00
def is_crawler?
@is_crawler ||=
begin
user_agent = @env[USER_AGENT]
if @env[DISCOURSE_RENDER] == "crawler" ||
CrawlerDetection.crawler?(user_agent, @env["HTTP_VIA"])
:true
else
if user_agent.downcase.include?("discourse") &&
!user_agent.downcase.include?("mobile")
:true
else
:false
end
end
end
@is_crawler == :true
end
2019-12-05 15:07:22 -05:00
alias_method :key_is_crawler?, :is_crawler?
def key_is_modern_mobile_device?
MobileDetection.modern_mobile_device?(@env[USER_AGENT]) if @env[USER_AGENT]
end
def key_is_old_browser?
CrawlerDetection.show_browser_update?(@env[USER_AGENT]) if @env[USER_AGENT]
end
def cache_key
return @cache_key if defined?(@cache_key)
@cache_key =
+"ANON_CACHE_#{@env["HTTP_ACCEPT"]}_#{@env[Rack::RACK_URL_SCHEME]}_#{@env["HTTP_HOST"]}#{@env["REQUEST_URI"]}"
@cache_key << AnonymousCache.build_cache_key(self)
@cache_key
end
def key_cache_theme_ids
theme_ids.join(",")
end
def key_compress_anon
GlobalSetting.compress_anon_cache
end
def theme_ids
ids, _ = @request.cookies["theme_ids"]&.split("|")
id = ids&.split(",")&.map(&:to_i)&.first
if id && Guardian.new.allow_themes?([id])
Theme.transform_ids(id)
else
[]
end
end
def cache_key_count
@cache_key_count ||= "#{cache_key}_count"
end
def cache_key_body
@cache_key_body ||= "#{cache_key}_body"
end
def cache_key_other
@cache_key_other || "#{cache_key}_other"
end
def get?
@env["REQUEST_METHOD"] == "GET"
end
def has_auth_cookie?
CurrentUser.has_auth_cookie?(@env)
end
def no_cache_bypass
request = Rack::Request.new(@env)
request.cookies["_bypass_cache"].nil? && (request.path != "/srv/status") &&
request[Auth::DefaultCurrentUserProvider::API_KEY].nil? &&
@env[Auth::DefaultCurrentUserProvider::HEADER_API_KEY].nil? &&
@env[Auth::DefaultCurrentUserProvider::USER_API_KEY].nil?
end
def force_anonymous!
@env[Auth::DefaultCurrentUserProvider::USER_API_KEY] = nil
@env[Auth::DefaultCurrentUserProvider::HEADER_API_KEY] = nil
@env["HTTP_COOKIE"] = nil
@env["HTTP_DISCOURSE_LOGGED_IN"] = nil
@env["rack.request.cookie.hash"] = {}
@env["rack.request.cookie.string"] = ""
@env["_bypass_cache"] = nil
request = Rack::Request.new(@env)
request.delete_param("api_username")
request.delete_param("api_key")
end
def logged_in_anon_limiter
@logged_in_anon_limiter ||=
RateLimiter.new(
nil,
"logged_in_anon_cache_#{@env["HTTP_HOST"]}/#{@env["REQUEST_URI"]}",
GlobalSetting.force_anonymous_min_per_10_seconds,
10,
)
end
def check_logged_in_rate_limit!
!logged_in_anon_limiter.performed!(raise_error: false)
end
MIN_TIME_TO_CHECK = 0.05
ADP = "action_dispatch.request.parameters"
def should_force_anonymous?
if (queue_time = @env["REQUEST_QUEUE_SECONDS"]) && get?
if queue_time > GlobalSetting.force_anonymous_min_queue_seconds
return check_logged_in_rate_limit!
elsif queue_time >= MIN_TIME_TO_CHECK
return check_logged_in_rate_limit! if !logged_in_anon_limiter.can_perform?
end
end
false
end
def cacheable?
!!(!has_auth_cookie? && get? && no_cache_bypass)
end
def compress(val)
if val && GlobalSetting.compress_anon_cache
require "lz4-ruby" if !defined?(LZ4)
LZ4.compress(val)
else
val
end
end
def decompress(val)
if val && GlobalSetting.compress_anon_cache
require "lz4-ruby" if !defined?(LZ4)
LZ4.uncompress(val)
else
val
end
end
def cached(env = {})
if body = decompress(Discourse.redis.get(cache_key_body))
if other = Discourse.redis.get(cache_key_other)
other = JSON.parse(other)
if req_params = other[1].delete(ADP)
env[ADP] = req_params
end
[other[0], other[1], [body]]
end
end
end
def cache_duration
@env["ANON_CACHE_DURATION"]
end
# NOTE in an ideal world cache still serves out cached content except for one magic worker
# that fills it up, this avoids a herd killing you, we can probably do this using a job or redis tricks
# but coordinating this is tricky
def cache(result, env = {})
return result if GlobalSetting.anon_cache_store_threshold == 0
status, headers, response = result
if status == 200 && cache_duration
if GlobalSetting.anon_cache_store_threshold > 1
count = REDIS_STORE_SCRIPT.eval(Discourse.redis, [cache_key_count], [cache_duration])
# technically lua will cast for us, but might as well be
# prudent here, hence the to_i
if count.to_i < GlobalSetting.anon_cache_store_threshold
headers["X-Discourse-Cached"] = "skip"
return status, headers, response
end
end
headers_stripped =
headers.dup.delete_if { |k, _| %w[Set-Cookie X-MiniProfiler-Ids].include? k }
headers_stripped["X-Discourse-Cached"] = "true"
parts = []
response.each { |part| parts << part }
if req_params = env[ADP]
headers_stripped[ADP] = {
"action" => req_params["action"],
"controller" => req_params["controller"],
}
end
Discourse.redis.setex(cache_key_body, cache_duration, compress(parts.join))
Discourse.redis.setex(cache_key_other, cache_duration, [status, headers_stripped].to_json)
headers["X-Discourse-Cached"] = "store"
else
parts = response
end
[status, headers, parts]
end
def clear_cache
Discourse.redis.del(cache_key_body)
Discourse.redis.del(cache_key_other)
end
end
def initialize(app, settings = {})
@app = app
end
PAYLOAD_INVALID_REQUEST_METHODS = %w[GET HEAD]
def call(env)
if PAYLOAD_INVALID_REQUEST_METHODS.include?(env[Rack::REQUEST_METHOD]) &&
env[Rack::RACK_INPUT].size > 0
return 413, { "Cache-Control" => "private, max-age=0, must-revalidate" }, []
end
helper = Helper.new(env)
force_anon = false
if helper.blocked_crawler?
env["discourse.request_tracker.skip"] = true
return 403, {}, ["Crawler is not allowed!"]
end
if helper.should_force_anonymous?
force_anon = env["DISCOURSE_FORCE_ANON"] = true
helper.force_anonymous!
end
if (env["HTTP_DISCOURSE_BACKGROUND"] == "true") && (queue_time = env["REQUEST_QUEUE_SECONDS"])
max_time = GlobalSetting.background_requests_max_queue_length.to_f
if max_time > 0 && queue_time.to_f > max_time
return [
429,
{ "content-type" => "application/json; charset=utf-8" },
[
{
errors: I18n.t("rate_limiter.slow_down"),
extras: {
wait_seconds: 5 + (5 * rand).round(2),
},
}.to_json,
]
]
end
end
result =
if helper.cacheable?
helper.cached(env) || helper.cache(@app.call(env), env)
else
@app.call(env)
end
2018-04-23 23:24:26 -04:00
result[1]["Set-Cookie"] = "dosp=1; Path=/" if force_anon
result
end
end
end