Skip to main content

Access Levels

The AccessLevel enum defines the access control hierarchy used throughout the permission engine.

Enum Definition​

class AccessLevel(str, Enum):
NONE = "none"
READ = "read"
WRITE = "write"
ACTIVE = "active"

def grants(self, required: "AccessLevel") -> bool:
order = {
AccessLevel.NONE: 0,
AccessLevel.READ: 1,
AccessLevel.WRITE: 2,
AccessLevel.ACTIVE: 1,
}
return order[self] >= order[required]

Hierarchy​

WRITE (2) ─── implies READ
│
READ (1) ──── implies ACTIVE (only in command/tool context)
│
NONE (0) ──── explicit deny, overrides everything

For Paths​

AccessPermissions
noneCannot access the path at all
readCan list directories, read file contents
writeCan create, update, delete files (implies read)

For Commands​

AccessPermissions
noneCommand cannot be executed
activeCommand can be executed

For Tools (Proxy Servers)​

AccessPermissions
noneTool cannot be used
activeTool can be used

The grants() Method​

The grants() method checks if one access level is sufficient to satisfy a required level:

AccessLevel.WRITE.grants(AccessLevel.READ) # True (write implies read)
AccessLevel.READ.grants(AccessLevel.WRITE) # False (read does not imply write)
AccessLevel.READ.grants(AccessLevel.NONE) # True (anything grants none)
AccessLevel.NONE.grants(AccessLevel.READ) # False (none grants nothing)

This is used in the check() method:

def check(self, required_access, path, tool=""):
required = AccessLevel(required_access)
granted = self._path_resolver.resolve(path)

if not granted.grants(required):
raise ForbiddenError(...)

How ACTIVE Relates​

ACTIVE is at the same numeric level as READ (1) in the hierarchy. It is used for binary on/off access for commands and tools where there is no concept of "read vs write":

AccessLevel.ACTIVE.grants(AccessLevel.ACTIVE) # True
AccessLevel.ACTIVE.grants(AccessLevel.NONE) # True
AccessLevel.NONE.grants(AccessLevel.ACTIVE) # False

Commands and tools always require ACTIVE access, which means any rule with access: active satisfies the requirement.