TypeORM promises to support composite primary keys,
but that does not work in reality.
This replaces the composite key used in the permission entities with
a single generated primary key and
a unique index on the relation columns.
See https://github.com/typeorm/typeorm/issues/8513
Signed-off-by: David Mehren <git@herrmehren.de>
TypeORM promises to support composite primary keys,
but that does not work in reality.
This replaces the composite key used in the permission entities with
a single generated primary key and
a unique index on the relation columns.
See https://github.com/typeorm/typeorm/issues/8513
Signed-off-by: David Mehren <git@herrmehren.de>
To keep the createdAt date stable, even when the revisions are dropped,
this adds a separate column
to store this data separately from revisions.
Signed-off-by: David Mehren <git@herrmehren.de>
The primary address is never null.
If no alias is set, the id is returned.
To still easily get the primary alias, the complete Alias DTOs
are now included in the metadata.
Signed-off-by: David Mehren <git@herrmehren.de>
MariaDB seems to order the returned media objects in a different way,
making our tests fail.
This refactors the tests to be independent of the order of returned data.
Signed-off-by: David Mehren <git@herrmehren.de>
The email address should only be available
in /me routes.
This commit splits the email address into a new FullUserInfoDto.
Signed-off-by: David Mehren <git@herrmehren.de>
In the e2e tests the global filter must be added via the special provider 'APP_FILTER' and not with useGlobalFilters, because if not the filter breaks, because of the way supertest handles the http-connection.
See: https://github.com/nestjs/nest/issues/1160#issuecomment-468698640
Signed-off-by: Philip Molares <philip.molares@udo.edu>
This commit separates the app config object from a new note config object. This was done to separate different concerns in different config files. Especially if the number of settings that are about notes increase, it is a good idea to keep them separate from the app config.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
The previous pattern used async methods for the builder pattern,
which were hideous to use when chained multiple times.
This extracts the builder into a separate class
and uses normal functions in the builder.
These queue async functions in arrays, which are executed
at the correct time, when the new async `build` function is called.
Signed-off-by: David Mehren <git@herrmehren.de>
This is part of an effort to consistently not embed User objects
in API responses. Usernames are returned instead.
Signed-off-by: David Mehren <git@herrmehren.de>
This is part of an effort to consistently not embed User objects
in API responses. Usernames are returned instead.
Signed-off-by: David Mehren <git@herrmehren.de>
Because the rejection now happens automatically in the permissions guard it does not get to the controller method and does not report the Content-Type to text/markdown
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Because the rejection now happens automatically in the permissions guard it now returns a 403 instead of 401
Signed-off-by: Philip Molares <philip.molares@udo.edu>
To make the create method easier to use in conjunction
with the authentication framework, this commit changes the type of
the `owner` parameter from `User | undefined` to `User | null`.
Signed-off-by: David Mehren <git@herrmehren.de>
Including both PublicApiModule and PrivateApiModule in the test setup
lead to the API routes overwriting each other.
This adds a router to separate the APIs as they are in the normal app.
Signed-off-by: David Mehren <git@herrmehren.de>
This should help to make clear why code is executed when the TokenAuthGuard is encountered by a request. Currently, one has to connect both files via the string 'token', which is a bit cryptic
Signed-off-by: Philip Molares <philip.molares@udo.edu>
The function now expects a `Note` object instead of a noteId
and a `User` instead of a username to
make it more consistent with other functions.
Signed-off-by: David Mehren <git@herrmehren.de>
The function now expects a `Note` object instead of a noteId to
make it more consistent with other functions.
Signed-off-by: David Mehren <git@herrmehren.de>
This renames `createOrUpdateHistoryEntry` to `updateHistoryEntryTimestamp`,
which reduces confusion with the similarly named
`updateHistoryEntry` function.
Signed-off-by: David Mehren <git@herrmehren.de>
