* Rate limit based on remote address IP, not on potential reverse proxy

* Limit rate of unauthenticated API requests further

* Rate-limit paging requests to one every 3 seconds

* Bump rack-attack from 5.4.2 to 6.0.0

Bumps [rack-attack](https://github.com/kickstarter/rack-attack) from 5.4.2 to 6.0.0.
- [Release notes](https://github.com/kickstarter/rack-attack/releases)
- [Changelog](https://github.com/kickstarter/rack-attack/blob/master/CHANGELOG.md)
- [Commits](https://github.com/kickstarter/rack-attack/compare/v5.4.2...v6.0.0)

Signed-off-by: dependabot[bot] <support@dependabot.com>

* fix payload[:request]

30 per 30 minutes, like media uploads

Fix #8078

Otherwise, no matter what is given for CAS_VALIDATE_URL the default /serviceValidate path would be used.

Deletions take a lot of resources to execute and cause a lot of
federation traffic, so it makes sense to decrease the number
someone can queue up through the API.

30 per 30 minutes

Fix #9993
Fix #5654

I also added "public" here, as I can't think of a good reason not to add it. Perhaps it has some marginal benefit in that ISPs (or other proxies) can cache it for all users. The assets are certainly publicly available and the same for all users.

* Add REST API for creating an account

The method is available to apps with a token obtained via the client
credentials grant. It creates a user and account records, as well as
an access token for the app that initiated the request. The user is
unconfirmed, and an e-mail is sent as usual.

The method returns the access token, which the app should save for
later. The REST API is not available to users with unconfirmed
accounts, so the app must be smart to wait for the user to click a
link in their e-mail inbox.

The method is rate-limited by IP to 5 requests per 30 minutes.

* Redirect users back to app from confirmation if they were created with an app

* Add tests

* Return 403 on the method if registrations are not open

* Require agreement param to be true in the API when creating an account

Fix #8666

Fixes #8189

rack-cors being called before the application router, it does not follow
the redirection, and we need a separate rule for /users/:username.

Fixes #9411

Right now, this includes three endpoints: host-meta, webfinger, and change-password.

host-meta and webfinger are publicly available and do not use any authentication. Nothing bad can be done by accessing them in a user's browser.

change-password being CORS-enabled will only reveal the URL it redirects to (which is /auth/edit) but not anything about the actual /auth/edit page, because it does not have CORS enabled.

The documentation for hosting an instance on a different domain should also be updated to point out that Access-Control-Allow-Origin: * should be set at a minimum for the /.well-known/host-meta redirect to allow browser-based non-proxied instance discovery.

Regression from #8957

* Set CSP rules in RoR's configuration

* Override CSP setting in the embed controller to allow frames

* add ffmpeg initializer

* use different expression to check for environment var

* cleanup pass

* undo mistakes

* fixed.

* revert

* Code quality pass

* Typofix

* Update applications_controller_spec.rb

* Update applications_controller_spec.rb

* Verify link ownership with rel="me"

* Add explanation about verification to UI

* Perform link verifications

* Add click-to-copy widget for verification HTML

* Redesign edit profile page

* Redesign forms

* Improve responsive design of settings pages

* Restore landing page sign-up form

* Fix typo

* Support <link> tags, add spec

* Fix links not being verified on first discovery and passive updates

Found via `codespell -q 3 --skip="./app/javascript/mastodon/locales,./config/locales"`

CSFR-prevention is already implemented but adding this doesn't hurt.

A brief introduction to Same-Site cookies (and the difference between strict and
lax) can be found at
https://blog.mozilla.org/security/2018/04/24/same-site-cookies-in-firefox-60/

TLDR: We use lax since we want the cookies to be sent when the user navigates
safely from an external site.

Still check for S3_CLOUDFRONT_HOST for existing installs.

* Fix uncaching worker

* Revert to using Paperclip's filesystem backend instead of fog-local

fog-local has lots of concurrency issues, causing failure to delete files,
dangling file records, and spurious errors UncacheMediaWorker

Adopted from GitLab CE. Generate new migration with:

    rails g post_deployment_migration name_of_migration_here

By default they are run together with db:migrate. To not run them,
the env variable SKIP_POST_DEPLOYMENT_MIGRATIONS must be set

Code by Yorick Peterse <yorickpeterse@gmail.com>, see also:

83c8241160

* Add more granular OAuth scopes

* Add human-readable descriptions of the new scopes

* Ensure new scopes look good on the app UI

* Add tests

* Group scopes in screen and color-code dangerous ones

* Fix wrong extra scope

If Mastodon accesses to the hidden service via transparent proxy, it's needed to avoid checking whether it's a private address, since `.onion` is resolved to a private address.
I was previously using the `HIDDEN_SERVICE_VIA_TRANSPARENT_PROXY` to provide that function. However, I realized that using `HIDDEN_SERVICE_VIA_TRANSPARENT_PROXY` is redundant, since this specification is always used with `ALLOW_ACCESS_TO_HIDDEN_SERVICE`. Therefore, I decided to integrate the setting of `HIDDEN_SERVICE_VIA_TRANSPARENT_PROXY` into` ALLOW_ACCESS_TO_HIDDEN_SERVICE`.

* Add dat:// and gopher:// to URL extractor

Fix #6072

* Fix comment indent

* Add dweb, ipfs, ipns, ssb

Timeout considered harmful due to leaving the app in a broken
state, including unreaped database connections

Especially in production it's just noise and doesn't mix well with the log format

* User agent for WebFinger

* local_domain → web_domain

* 'http' is away accidentally...

- POST /api/v1/push/subscription
- PUT /api/v1/push/subscription
- DELETE /api/v1/push/subscription
- New OAuth scope: "push" (required for the above methods)

* Update paperclip.rb

* Update .env.production.sample

* Update paperclip.rb

The previous rate limit allowed to post media so fast that it is possible
to fill up the disk space even before an administrator notices. The new
rate limit is configured so that it takes 24 hours to eat 10 gigabytes:
10 * 1024 / 8 / (24 * 60 / 30) = 27 (which rounded to 30)

The period is set long so that it does not prevent from attaching several
media to one post, which would happen in a short period. For example,
if the period is 5 minutes, the rate limit would be:
10 * 1024 / 8 / (24 * 60 / 5) = 4

This long period allows to lift the limit up.

* No need to re-require sidekiq plugins, they are required via Gemfile

* Add derailed_benchmarks tool, no need to require TTY gems in Gemfile

* Replace ruby-oembed with FetchOEmbedService

Reduce startup by 45382 allocated objects

* Remove preloaded JSON-LD in favour of caching HTTP responses

Reduce boot RAM by about 6 MiB

* Fix tests

* Fix test suite by stubbing out JSON-LD contexts

* Add support for HTTP client proxy

* Add access control for darknet

Supress error when access to darknet via transparent proxy

* Fix the codes pointed out

* Lint

* Fix an omission + lint

* any? -> include?

* Change detection method to regexp to avoid test fail

Fix #3565 (oops)

Fix #7095