This ensures stack traces are helpful at the cost of a slightly
lower performance (one more tick in the event loop).
Fixes#838
Signed-off-by: David Mehren <git@herrmehren.de>
This is necessary as typescript is unable to determine that `year` in this object is of type "numeric" and not string.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
As the NotesController has the note already, because it checked with it if the user has the permission to perform the action, it's not necessary to get the note from the DB again, instead we should just provide the note to the functions directly.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Handle the AlreadyInDB and PermissionsUpdateInconsistent errors and correctly show them to the api user as BadRequest errors.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Fixed toUserDto method of UsersService: If a user has no email an empty string should be returned (like with the photoUrl) instead of null.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
This service is necessary as we plan to have functions to create and manipulate groups in the future.
The GroupInfoDto was moved from the file note-permissions.dto.ts to mimic the UserInfoDto.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
As discussed in #835 we don't want to have default constructors and prefer .create methods.
Because the created NoteGroupPermission and NoteUserPermission are not saved to the DB by themselves, but are saved via a change to the Note using a Pick<Class, attributes>-style return type is not helpful here as every single time the .create functions are called a full object is required.
The mock calls in the PermissionService test are not needed and break the .create calls so they got removed.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
The permission check also tries to get the note and a non existing note needs to be handled there too.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
The unit test now uses per test mocking of the necessary functions instead of one mock in the beforeEach call.
Also some tests got expanded to cover more error cases.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Using nest-router for this purpose as it is a rather easy addition to our structure. As we don't add the Router to any e2e tests we don't need to change them.
fixes#523
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Our custom logger supports providing the name of the function that
calls the logger, this commit adds this context string where it
was previously missing.
Signed-off-by: David Mehren <git@herrmehren.de>
For a better structure of the autogenerated apidoc website tags are used. Each Controller get it's own tag and will be put in a separate section.
See https://docs.nestjs.com/openapi/operations#tags
Signed-off-by: Philip Molares <philip.molares@udo.edu>
This makes it possible to create the user before the mock auth guard does it's magic. This is necessary for some test, where we need the user object before the api is called.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
Add history service to allow for CRUD operations.
Use history service in controllers to:
1. Allow manipulating of history entries
2. Guaranty the correct existence of history entries
Signed-off-by: Philip Molares <philip.molares@udo.edu>
With this the backend now can hold a history entry.
Also included in this commit are some minor changes to tests and services so they can still work.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
DTO should only be used for sending information to and from user.
Services now have methods which return normal internal objects and
methods which convert them to DTOs. This conversion is done in the
controlers
Signed-off-by: Yannick Bungers <git@innay.de>
Move conversion of Errors from AuthService to TokenStrategy.
This is necessary to correctly test the validateToken method.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
The former length of 64 bytes (512-bit) is transformed into base64url (a 6-bit code) ~86 characters long. This is too long for bcrypt as it ignores any characters beyond the 72th.
This fix therefore reduces the amount of generated bytes to 54 (as 72*6/8 = 54) characters. This ensures that removing one character from the token the hash won't be the same anymore.
Signed-off-by: Philip Molares <philip.molares@udo.edu>
