GoSCIM is the LEGO blocks for building a custom SCIM v2 implementation. It already implemented most of, if not all, the functionalities specified in the IETF document.

Functionalities like validation, read-only attributes assignment, query resolution, persistence and JSON serialization are included. In addition, a sample server is included with the project just to demonstrate how easy it is to put together these components into a functioning SCIM API Server.

GoSCIM does not compile to a runnable artifact. Instead, it enables building one. If you want to quickly have a preview of an example server that has been already built:

cd $GOPATH/src/github.com/davidiamyou/go-scim/example
go run server.go

This section explains some of the design decisions. Knowing these may save you some time in figuring out about your own implementations.

The SCIM protocol defines resources based on schemas. Naturally GoSCIM uses these schema attribute definitions to perform validation, serialize JSON and so on. However, GoSCIM extended the SCIM defined schema attribute to include some more useful information called Assists. Some key assists include:

• JSON name of the field (i.e. displayName)
• Path of the field (i.e. emails.value, active, name.familyName)
• Full URN name of the field (i.e. id, urn:ietf:params:scim:schemas:core:2.0:User:name.familyName)
• Array key field (i.e. the field that could be used to uniquely identify a complex array entry, for instance, value)

GoSCIM uses the extended schema internally and renders SCIM defined schema. It is important to know, although one extends another, they are separate entities inside GoSCIM.

The following table relates SCIM type to Go type:

The entry point of serializing an object in GoSCIM is MarshalJSON in shared/json.go. Underneath, it tries to utilize go's native JSON capabilities whenever possible. However, when serializing resources, it does not rely on tags, rather it seeks advice from SCIM schema.

GoSCIM tries to parse the query text into an abstract syntax tree first. The tree then can be flattened and transformed to whichever query language the database understands.

GoSCIM supports MongoDB. The mongo directory contains an example of how the AST can be flattened to MongoDB query. It should work similarly at least with other document based databases.

GoSCIM supports MongoDB, but it does not restrict adopters to it. It provides a Repository interface in shared/persistence.go which other database choices can implement. The MongoDB implementation is contained in the mongo folder.

• WebRequest: an abstraction of HTTP request. Useful when delegating mock requests, for instance, during bulk operation.
• WebResponse: similar to WebRequest in intentions.
• PropertySource: abstraction of a property provider. The example server uses a map to implement this. Actual implementations can be projects like viper
• Logger: abstraction of a logger. The example server implementations just prints to console. Actual logger can be used in real implementations.
• ReadOnlyAssignment: logic to assign value to read only fields. GoSCIM already provides id, meta and group assignment, plus copying any read only value from existing resource reference during update. User needs to implement this interface per custom readonly field.