auth: update user type constants
Those need to be kept in sync with the database/GraphQL values. The ones
not in sync are not currently used, though.
server: let AUTH_INTERNAL access @anoninternal
Both methods enforce the internal aspect, but @anoninternal really just
means the resolver does not require an "authenticated user" context,
which means it's still perfectly safe if there is one.
With this in place, any resolver that may have to be called from an
anonymous context can be switched from @internal to @anoninternal
without breaking existing users. Of course it can only be switched if it
really does not require a user context.
webhooks: do not limit webhooks by token_hash
When a user re-authenticates with an OAuth2 app, the app would not be
able to query or delete webhooks it created with a previous session.
This was a problem for a tool I am developing and removing this
constraint fixes the issue.
server: allow setting ports for metrics, pprof
With an eye towards Kubernetes, it is desirable to avoid having the API
pick random ports for anything. This commit introduces additional flags
for the API server to specify the metrics and pprof ports. This change
is backwards-compatible insofar as the default behavior is preserved if
the flags are not used.
The log output is changed to include the full address that the
respective listeners are listening on.
Note that using "0.0.0.0:1234" might not quite work as expected, because
Go [1]. But that does not change the default behavior, so I guess it's
fine.
[1] https://github.com/golang/go/issues/48723
Run golangci-lint in CI
See https://golangci-lint.run
This has already helped uncover some issues, and we should make more use
of it. Currently we have to disable one of the default lints, as we have
a lot of issues it complains about. I plan to address those one by one
and enable it. Until then, put this in place to make sure we don't
re-introduce any of the issues we've eliminated so far.
Don't uppercase error messages
Fix a bunch of linter complaints
auth/middleware: return semantic errors on auth failure
This time with a much simpler approach than the last one that caused
some authentication issues for chat.sr.ht.
server: add error presenter to make context cancellation semantic
auth: ensure semantic errors are bubbled up to user properly
auth: add error code on unauthorized request response
auth: handle edge case when fetching users from meta
feature: add submodule + middleware for feature flags
model: introduce ID type
Based on UUIDv7 but represented to users in a more compact base32 form,
though all of that are internal implementation details.
database: add WithReadOnlyTx
Refactor config loading and server initialization
As is, LoadConfig() does some things that are not strictly related to
the configuration, such as parsing command line arguments. This has led
to a proliferation of different ways to load the config based on various
needs and also prevents tools that need a config but are not services to
use custom command line arguments.
This commit aims to decouple config loading from everything else and
do nothing but loading the config files.
On a high level, this commit:
- renames server.NewServer() to server.New()
- moves config.Debug and config.Addr into the server package
- moves crypto.InitCrypto() call into server.New()
- moves command line parsing into server.New(), using passed-in values
rather than os.Args
The only changes required for services would be changing
cfg := config.LoadConfig(":5100")
server := server.NewServer("meta.sr.ht", cfg)
to
cfg := config.LoadConfig()
server := server.New("meta.sr.ht", ":5100", cfg, os.Args)
All other tools will be switched to just LoadConfig() and, optionally, a
call to crypto.InitCrypto(). I managed to completely remove some global
state (addr) and at least make the rest private, so that users are
forced to use the designated functions.
The config module gained support for custom FS implementation, mainly
for testing.