Refactor models so that we define attribute API permissions in policy
files instead of directly in models.
This is cleaner because a) permissions are better handled by policies
and b) which attributes are visible to the API is an API-level concern
that models shouldn't have to care about.
This fixes an issue with not being able to precompile CSS/JS assets
unless the database was up and running. This was a problem when building
Docker images because we don't have a database at build time. We needed
the database because `api_attributes` was a class-level macro in some
places, which meant it ran at boot time, but this triggered a database
call because api_attributes used database introspection to get the list
of allowed API attributes.
Partial fix for #4389.
* Fix invalid username searches returning all posts instead of no posts.
* Fix "user:A user:B" returning results for user:B instead of no results.
* Fix "approver:A approver:B" returning results for approver:B instead of no results.
* Add support for negated -commenter, -noter, -noteupdater, -upvote, -downvote metatags.
* Add support for "any" and "none" values for all username metatags,
including negated metatags that didn't support "any" or "none" before.
* Change noter:any and commenter:any to include posts with deleted notes
or comments. Note that commenter:<username> already included deleted
comments before. This is so that commenter:any has the same behavior
as commenter:<username>
Remove the ability to authenticate to the API with the `login` and
`password_hash` url parameters. This is a legacy authentication method
from Danbooru 1. How to actually generate the password_hash for this
method hasn't been fully documented for many years now. It required
taking the SHA1 hash of your password combined with an undocumented salt
value (i.e., password_hash = sha1("choujin-steiner--#{password}")).
This authentication method was also slow because it required checking
the password on every API call. Checking passwords is deliberately slow
because passwords are hashed with BCrypt. BCrypt takes about ~200ms per
request, so using this method effectively limited you to ~5 requests per
second in a single thread.
* Make authentication methods into User instance methods instead of
class methods.
* Fix API key authentication to use a secure string comparison. Fixes a
hypothetical (unlikely to be exploitable) timing attack.
* Move login logic from SessionCreator to SessionLoader.
Require users who signup using proxies to verify their email addresses
before they can perform any edits. For verification purposes, the email
must be a nondisposable address from a whitelist of trusted email
providers.
Pull the password reauthentication logic out of the user model and put
it in the password update controller where it belongs.
This fixes an issue where when a new user was created the user model had
an incorrect password error set on it by `encrypt_password_on_update`.
It was trying to verify the old password even though we don't have one
when creating a new user. This error caused the user create action to
redirect back to the signup page because `respond_with` thought that
creating the user failed.
Side effects:
* The data-current-user-is-voter <body> attribute has been removed.
* {{upvote:self}} no longer works. {{upvote:<name>}} should be used instead.
Fix discrepancy between index action and show action. The index
action allowed members to see name changes for undeleted users, but the
show action didn't.
* Move emails from users table to email_addresses table.
* Validate that addresses are formatted correctly and are unique across
users. Existing invalid emails are grandfathered in.
* Add is_verified flag (the address has been confirmed by the user).
* Add is_deliverable flag (an undeliverable address is an address that bounces).
* Normalize addresses to prevent registering multiple accounts with the
same email address (using tricks like Gmail's plus addressing).
The old password reset flow:
* User requests a password reset.
* Danbooru generates a password reset nonce.
* Danbooru emails user a password reset confirmation link.
* User follows link to password reset confirmation page.
* The link contains a nonce authenticating the user.
* User confirms password reset.
* Danbooru resets user's password to a random string.
* Danbooru emails user their new password in plaintext.
The new password reset flow:
* User requests a password reset.
* Danbooru emails user a password reset link.
* User follows link to password edit page.
* The link contains a signed_user_id param authenticating the user.
* User changes their own password.
Refactor how model visibility works in index actions:
* Call `visible` in the controller instead of in model `search`
methods. This decouples model visibility from model searching.
* Explicitly pass CurrentUser when calling `visible`. This reduces
hidden dependencies on the current user inside models.
* Standardize on calling the method `visible`. In some places it was
called `permitted` instead.
* Add a `visible` base method to ApplicationModel.
- The only string works much the same as before with its comma separation
-- Nested includes are indicated with square brackets "[ ]"
-- The nested include is the value immediately preceding the square brackets
-- The only string is the comma separated string inside those brackets
- Default includes are split between format types when necessary
-- This prevents unnecessary includes from being added on page load
- Available includes are those items which are allowed to be accessible to the user
-- Some aren't because they are sensitive, such as the creator of a flag
-- Some aren't because the number of associated items is too large
- The amount of times the same model can be included to prevent recursions
-- One exception is the root model may include the same model once
--- e.g. the user model can include the inviter which is also the user model
-- Another exception is if the include is a has_many association
--- e.g. artist urls can include the artist, and then artist urls again
* Add ability to mark dmails as unread.
* Fix users.unread_dmail_count to not count deleted dmails.
* Fix show action so that API calls don't mark dmails as read.
* Don't show the unread dmail notice on the /dmails page itself.
* Stop using users.has_mail flag.
* Add ability to report dmails.
* Enable reports for comments, forum posts, and dmails.
* Allow Members to send reports.
* Don't allow users to report the same thing twice.
Avoid doing one SQL query per topic when checking for new topics on the
forum index.
This also changes it so that forum topics aren't always marked as new
for anonymous users.
Few people used dmail filters (~900 users in 5 years) and even fewer
used them correctly. Most people used them to try to block dmail spam,
but usually they either blocked too much (by adding common words that
are present in nearly all dmails, causing all mails to them to be
filtered) or too little (blocking specific email addresses or urls,
which usually are never seen again after the spammer is banned).
Nowadays the spam detection system does a better job of filtering spam.
The belongs_to_creator macro was used to initialize the creator_id field
to the CurrentUser. This made tests complicated because it meant you had
to create and set the current user every time you wanted to create an
object, when lead to the current user being set over and over again. It
also meant you had to constantly be aware of what the CurrentUser was in
many different contexts, which was often confusing. Setting creators
explicitly simplifies everything greatly.
- Limited to Builders+
-- Moderator+ can also use as they may be too busy ATM
- Only on users, comments, and forum posts
- Multiple reports can be generated per instance
- Primarily posts to a moderator-only topic for viewability
- Secondarily has a table for searchability
-- Viewable only by moderators
