ge-aerospace/shopdb-flask
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add prebuilt user scopes and guard invalid Reporter grants

Browse Source 
A newly created user held no roles and no permissions, so the sidebar filtered every nav item away and the account appeared broken. Adds three prebuilt scopes (Reporter, Operator, Technician) derived from the live permission catalog by verb rather than hand-listed, so a new plugin permission lands in the right scope automatically instead of silently going ungranted. users.view is granted to all three since a user who cannot see the directory cannot collaborate; users.create/edit/delete stay admin-only. Role.seedscopes() is additive and idempotent, seeded via flask seed permissions, and preserves custom grants on rerun.

Reporter guard: the auth model is additive across admin, direct grants and roles, so Reporter cannot be expressed as merely fewer grants - a direct endpoint-action grant would silently re-arm a read-only account. Rather than change the global permission predicate and alter every route's semantics, _check_scope_grants() rejects the invalid combination at write time on both create and update. The update path validates the effective post-update state, so demoting a Technician to Reporter is refused while a stale endpoint grant remains rather than quietly leaving a read-only user with write authority.

Validation runs before the User object is constructed. Assigning user.roles first put the relationship into the session autoflush path while the User was still unattached; nothing broke because the refusal returns before db.session.add(), but it left a trap where any query added below that point would autoflush a half-built user and surface later as an orphan row. Verified with -W error::sqlalchemy.exc.SAWarning: 25 passed.
This commit is contained in:
benjamin.baker
2026-09-07 12:39:24 -04:00
parent 52c9d2b009
commit 0e43ac6b8d
9 changed files with 1107 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
shopdb/cli/__init__.py
View File

@@ -810,7 +810,7 @@ def fix_controls_direction():
@seed_cli.command('permissions')
@with_appcontext
def seed_permissions():
"""Seed predefined permissions, and the guest role that can hold them."""
"""Seed predefined permissions, the guest role, and the prebuilt scopes."""
from shopdb.extensions import db
from shopdb.core.models import Permission, Role
@@ -819,11 +819,23 @@ def seed_permissions():
# on the upgrade it already runs. It holds nothing until somebody grants
# it something, so seeding it changes no behaviour.
guest = Role.seedguest()
# AFTER Permission.seed(), which is load-bearing: seedscopes only links
# permissions that exist as rows, so seeding the catalog first is what
# lets a newly installed plugin's permissions reach the scopes in the
# same command rather than on the next run.
scopes = Role.seedscopes()
db.session.commit()
click.echo(click.style(f"{created} permissions created.", fg='green'))
if guest:
click.echo(click.style(
"guest role created (holds no permissions).", fg='green'))
for rolename in scopes['created']:
click.echo(click.style(f"{rolename} role created.", fg='green'))
for rolename, count in sorted(scopes['granted'].items()):
click.echo(click.style(
f"{rolename}: {count} permission(s) granted.", fg='green'))
if not scopes['created'] and not scopes['granted']:
click.echo("Prebuilt scopes already up to date.")
@seed_cli.command('settings')

158
shopdb/core/api/users.py
View File

@@ -7,13 +7,82 @@ from werkzeug.security import generate_password_hash
from shopdb.extensions import db
from shopdb.core.models import (
User, Role, Permission, AuditLog, full_permission_catalog,
guest_grantable_permissions)
guest_grantable_permissions, SCOPE_ENDPOINT_ACTIONS,
SCOPES_DENIED_ENDPOINT_ACTIONS)
from shopdb.utils.responses import success_response, error_response, ErrorCodes
from shopdb.utils.authz import require_role
users_bp = Blueprint('users', __name__)
def _check_scope_grants(rolenames, permissionnames):
"""Refuse direct grants a prebuilt scope is defined as never holding.
Returns an error_response, or None when the combination is allowed.
This is the server-side half of the scope dropdown, and it is the half that
is actually load-bearing. The UI greys out the endpoint action checkboxes
for a reporter, but a greyed-out checkbox is a suggestion: the same account
can be created with a restart grant by anyone willing to POST JSON. Reporter
was specified as holding no action capability EVEN IF expressly permitted,
and only a check on the write path can make that true.
Enforced here rather than in User.haspermission() on purpose. A subtractive
ceiling in the read predicate would sit in the hot path of every
permission-gated route in the app and change the meaning of a grant
everywhere, to fix a combination that should simply never have been stored.
Refusing the write is the narrow version: bad state cannot get in, and the
thousands of existing reads keep their current semantics.
Note this deliberately does NOT restrict the admin role, which bypasses
permission checks by rule, nor any site-created custom role. It constrains
only the scopes that were seeded with a documented promise about what they
hold.
"""
denied_scopes = SCOPES_DENIED_ENDPOINT_ACTIONS & set(rolenames)
if not denied_scopes:
return None
conflicts = sorted(set(permissionnames) & set(SCOPE_ENDPOINT_ACTIONS))
if not conflicts:
return None
scope = sorted(denied_scopes)[0]
return error_response(
ErrorCodes.VALIDATION_ERROR,
f"The {scope} scope cannot hold endpoint actions "
f"({', '.join(conflicts)}). Assign the technician or operator scope "
f"if this user needs to act on endpoints.")
def _resolve_directpermissions(names):
"""Resolve permission names to rows, rejecting any that do not exist.
Returns (permissions, error_response). A None permissions value with no
error means the caller did not mention direct grants at all, which is not
the same as sending an empty list - that clears them.
Unknown names are refused rather than dropped. Silently ignoring a typo
would report success while granting nothing, and the admin would believe
the capability was handed out.
"""
if names is None:
return None, None
if not isinstance(names, list):
return None, error_response(
ErrorCodes.VALIDATION_ERROR,
'directpermissions must be a list of permission names')
if not names:
return [], None
wanted = {str(n) for n in names}
permissions = Permission.query.filter(Permission.name.in_(wanted)).all()
missing = sorted(wanted - {p.name for p in permissions})
if missing:
return None, error_response(
ErrorCodes.VALIDATION_ERROR,
'Unknown permissions: ' + ', '.join(missing))
return permissions, None
@users_bp.route('', methods=['GET'])
@jwt_required()
@require_role('admin')
@@ -61,6 +130,28 @@ def create_user():
if User.query.filter_by(email=data['email']).first():
return error_response(ErrorCodes.CONFLICT, 'Email already exists', http_code=409)
# Resolve and VALIDATE roles and grants before touching the User object.
#
# Ordering is deliberate. Assigning user.roles first put the relationship
# into the session's autoflush path while the User itself was still
# unattached, which SQLAlchemy warned about ("Object of type <User> not in
# session, add operation along 'Role.users' won't proceed"). Nothing broke,
# because the refusal returns before db.session.add(). But it left a live
# trap: move or add a query below this point and the autoflush becomes a
# real write of a half-built user, surfacing later as an orphan row nobody
# can explain. Validate everything, then build.
role_ids = data.get('roles', [])
roles = Role.query.filter(Role.roleid.in_(role_ids)).all() if role_ids else []
permissions, err = _resolve_directpermissions(data.get('directpermissions'))
if err:
return err
err = _check_scope_grants(
[r.rolename for r in roles],
[p.name for p in (permissions or [])])
if err:
return err
# Admin-created accounts are forced to change the password on first login
# unless the admin explicitly opts out.
mustchange = data.get('mustchangepassword', True)
@@ -75,12 +166,13 @@ def create_user():
mustchangepassword=bool(mustchange)
)
# Assign roles
role_ids = data.get('roles', [])
if role_ids:
roles = Role.query.filter(Role.roleid.in_(role_ids)).all()
if roles:
user.roles = roles
# Direct per-user permission grants, held alongside (not instead of) roles.
if permissions is not None:
user.directpermissions = permissions
db.session.add(user)
# Audit log
@@ -181,6 +273,39 @@ def update_user(userid: int):
changes['roles'] = {'old': old_roles, 'new': new_roles}
user.roles = roles
# Direct grants are audited by NAME rather than id: the point of the
# audit row is to be readable years later, when a permission id means
# nothing to the person reading it.
if 'directpermissions' in data:
permissions, err = _resolve_directpermissions(
data['directpermissions'])
if err:
return err
old_direct = user.getdirectpermissionnames()
new_direct = sorted(p.name for p in (permissions or []))
if old_direct != new_direct:
changes['directpermissions'] = {
'old': old_direct, 'new': new_direct}
user.directpermissions = permissions or []
# Validate the COMBINATION, after both assignments and against the
# user's effective state rather than only what this request mentioned.
# A payload may change roles without mentioning grants, or grants
# without mentioning roles; checking only the supplied half would let a
# request that demotes an operator to reporter keep the restart grant
# they already held, which is exactly the state the scope promises is
# impossible. Reading back off `user` covers all three shapes.
#
# Safe to check after mutating because nothing has committed yet - the
# early return leaves the session to be rolled down by the request
# teardown, so a refused combination is never persisted.
err = _check_scope_grants(
[r.rolename for r in user.roles],
user.getdirectpermissionnames())
if err:
db.session.rollback()
return err
# Unlock user
if data.get('unlock'):
user.lockeduntil = None
@@ -237,6 +362,21 @@ def delete_user(userid: int):
# revoke their API tokens outright, and DETACH their audit history
# (userid -> NULL) - the log rows themselves are kept, entityname and
# details still tell the story.
#
# What a departing user leaves behind is deliberately NOT undone. Policies
# carry no FK to users, so configuration and business policies they authored
# keep working; the audit rows keep username, which is denormalized on
# auditlogs for exactly this reason. Leaving is not a reason to unpick
# someone's work.
#
# TODO (future): offer to EXPORT this user's audit history at the moment of
# removal, and on a significant scope reduction too. Detaching costs the
# userid link, so after this runs their trail is only reachable by matching
# the username string - which is fragile if the name is ever reused. An
# export taken at the point of change preserves the joined-up record while
# it still exists. Deactivating (isactive=False) instead of deleting keeps
# the userid link intact and is the better choice when the account may ever
# need to be audited in depth.
from shopdb.core.models import ApiToken
ApiToken.query.filter_by(userid=userid).delete(synchronize_session=False)
AuditLog.query.filter_by(userid=userid).update(
@@ -452,6 +592,14 @@ def user_to_dict(user: User) -> dict:
'lastlogindate': user.lastlogindate.isoformat() + 'Z' if user.lastlogindate else None,
'failedlogins': user.failedlogins,
'roles': [{'roleid': r.roleid, 'rolename': r.rolename, 'color': r.color} for r in user.roles],
# Split deliberately. directpermissions is what is STORED on the user and
# is what the editor's checkboxes bind to; permissions is the EFFECTIVE
# set including everything the roles supply, which is what a reader needs
# to answer "can this person do X". A checkbox bound to the effective set
# would appear ticked for a capability the user holds through a role, and
# unticking it would silently do nothing.
'directpermissions': user.getdirectpermissionnames(),
'permissions': sorted(user.getpermissions()),
'createddate': user.createddate.isoformat() + 'Z' if user.createddate else None,
'modifieddate': user.modifieddate.isoformat() + 'Z' if user.modifieddate else None
}

9
shopdb/core/models/__init__.py
View File

@@ -14,7 +14,9 @@ from .operatingsystem import OperatingSystem
from .relationship import AssetRelationship, RelationshipType, RelationshipTypePropagation
from .communication import Communication, CommunicationType
from .user import (User, Role, Permission, full_permission_catalog,
guest_grantable_permissions)
guest_grantable_permissions, scope_permission_names,
PREBUILT_SCOPES, SCOPE_ENDPOINT_ACTIONS,
SCOPE_NEVER_GRANT, SCOPES_DENIED_ENDPOINT_ACTIONS)
from .application import Application, AppVersion
from .script_library import (Script, RUNCONTEXTS, RUNCONTEXT_SYSTEM,
RUNCONTEXT_USER)
@@ -63,6 +65,11 @@ __all__ = [
'Permission',
'full_permission_catalog',
'guest_grantable_permissions',
'scope_permission_names',
'PREBUILT_SCOPES',
'SCOPE_ENDPOINT_ACTIONS',
'SCOPE_NEVER_GRANT',
'SCOPES_DENIED_ENDPOINT_ACTIONS',
# Applications
'Application',
'AppVersion',

259
shopdb/core/models/user.py
View File

@@ -19,6 +19,25 @@ rolepermissions = db.Table(
db.Column('permissionid', db.Integer, db.ForeignKey('permissions.permissionid'), primary_key=True)
)
# Association table for DIRECT per-user permissions (many-to-many).
#
# A grant here is an ordinary permission, identical in power to one held through
# a role - it counts toward the API token scope ceiling and everything else. It
# exists because a site of 5-7 operators should be able to give one person one
# capability without minting a role for them, which would otherwise turn the
# role list into a mixture of job titles and single-capability flags.
#
# Anything that answers "what may this user do" MUST fold this in. There are two
# such surfaces and they are computed separately: haspermission() (the predicate
# authorization uses) and getpermissions() (the list baked into JWT claims and
# read by the frontend authStore). Teaching only one of them makes the server and
# the UI disagree silently.
userpermissions = db.Table(
'userpermissions',
db.Column('userid', db.Integer, db.ForeignKey('users.userid'), primary_key=True),
db.Column('permissionid', db.Integer, db.ForeignKey('permissions.permissionid'), primary_key=True)
)
class Permission(db.Model):
"""
@@ -179,6 +198,155 @@ CORE_GUEST_GRANTABLE = {
}
# ---------------------------------------------------------------------------
# Prebuilt user scopes
# ---------------------------------------------------------------------------
#
# Why these exist: before them the only roles were `admin` and `guest`, so a new
# user was given nothing, held no permissions, and the sidebar correctly hid
# every link. There was no rung between "runs the site" and "sees nothing".
#
# The four scopes, as specified:
# admin - unchanged. Bypasses every check in haspermission(); NOT seeded
# here and NOT described by a grant list, because it holds
# everything by rule rather than by grant.
# operator - full record access except deletion, which stays admin-only.
# technician - read-only on records. Cannot create or edit.
# reporter - reads everything and exports reports. No action capability.
#
# DERIVED, not hardcoded, and this is the opposite choice to
# CORE_GUEST_GRANTABLE above. That set is declared name-by-name because it
# governs what an ANONYMOUS visitor may reach, where the bar is "could this be
# public forever" and a wrong answer leaks to the internet. These scopes govern
# authenticated internal staff, and the catalog is plugin-driven: roughly sixty
# permissions across fifteen plugins, growing whenever a plugin is added. A
# hardcoded list would silently omit every future plugin, so a Technician would
# lose access to new modules by default and the failure would look like a
# permissions bug rather than a stale list.
#
# So scopes are computed from the catalog by verb, minus an explicit exclusion
# set. Adding a plugin extends the scopes automatically; keeping something OUT
# of them is the thing that requires a deliberate code change.
# Verbs that mutate records. Split by which scope may hold them.
_EDIT_VERBS = ('create', 'edit', 'restock', 'manage', 'publish')
_DESTRUCTIVE_VERBS = ('delete', 'purge')
# Never granted by ANY prebuilt scope, regardless of verb. Each is either an
# administrative capability, a machine-to-machine credential, or a payload that
# a scope-holder should not get merely for being staff.
SCOPE_NEVER_GRANT = {
# Administrative surface. A scope is not a way to become an admin.
'settings.view', 'settings.edit',
# users.view is deliberately ABSENT: every scope may see who exists, so an
# operator can tell whether an account is already there before asking an
# admin to make one. Creating, editing, and deleting accounts stay
# admin-only, so seeing the directory grants no power over it.
'users.create', 'users.edit', 'users.delete',
'audit.view', 'audit.purge',
'apitokens.create',
# Service-token permissions: these authorize machines, not people. Granting
# one to a human scope would let a person impersonate the collector or the
# GE-Enforce client.
'collector.ingest',
'geenforce.fetch',
'geenforce.report',
# Bytes, not metadata: SYSTEM-context PowerShell bodies and vendor
# installers. Deliberately separate from applications.view for exactly this
# reason (see the CORE_PERMISSIONS comment), so scopes must not undo that.
'applications.download',
# The big four endpoint actions. These are per-user direct grants chosen by
# checkbox in the user editor, NOT role grants - that is the whole point of
# the Phase A2 split. A scope that granted them would take the choice away
# from the person creating the account.
'endpoints.restart',
'endpoints.runscript',
'endpoints.updateapp',
'endpoints.corptools',
# Fleet-wide and agent-distribution capabilities, admin-only by nature.
'endpoints.fleet',
'endpoints.agentpackage',
'endpoints.agentupdate',
}
# Endpoint actions offered as per-user checkboxes in the user editor. Held here
# rather than in the endpoints plugin because the USER EDITOR is what needs the
# list, and core cannot import a plugin. Kept in sync by
# test_scope_endpoint_actions_match_the_plugin.
SCOPE_ENDPOINT_ACTIONS = (
'endpoints.restart',
'endpoints.runscript',
'endpoints.updateapp',
'endpoints.corptools',
)
def _permission_verb(name):
"""The verb half of a dotted permission name ('assets.edit' -> 'edit')."""
return name.rsplit('.', 1)[-1] if '.' in name else ''
def scope_permission_names(scope, catalog=None):
"""Permission names a prebuilt scope grants, derived from the catalog.
Returns a sorted list so seeding and tests are deterministic.
reporter - every non-excluded permission whose verb is neither a mutation
nor a destruction, plus reports.export. The export is a read
that produces a file, not an action against a machine, and the
user asked for it by name.
technician - the same reads. Identical to reporter at ROLE level: the two
differ by what the user editor offers alongside the role, where
a technician may be given endpoint action checkboxes and a
reporter may not.
operator - the reads plus the editing verbs, still never the destructive
ones. Deletion stays admin-only, per the user's instruction.
"""
if catalog is None:
catalog = full_permission_catalog()
names = {entry[0] for entry in normalize_permission_entries(catalog)}
names -= SCOPE_NEVER_GRANT
readable = {
name for name in names
if _permission_verb(name) not in _EDIT_VERBS
and _permission_verb(name) not in _DESTRUCTIVE_VERBS
}
if scope in ('reporter', 'technician'):
return sorted(readable)
if scope == 'operator':
return sorted(
readable | {
name for name in names
if _permission_verb(name) in _EDIT_VERBS
})
raise ValueError(f'Unknown scope: {scope!r}')
# The seeded scopes, in the order the user editor should offer them. `admin` is
# absent on purpose: it already exists, is created by `seed admin`, and holds
# everything by bypass rather than by grant.
PREBUILT_SCOPES = (
('operator',
'Full access to records except deletion, which stays with admins. '
'Endpoint actions are granted per user by checkbox.',
'#0D6EFD'),
('technician',
'Read-only on records: cannot create, edit, or delete, and cannot author '
'policy. Endpoint actions are granted per user by checkbox.',
'#0DCAF0'),
('reporter',
'Views everything and exports reports. Holds no action capability and '
'cannot be granted one.',
'#6F42C1'),
)
# Scopes that may NOT hold the per-user endpoint action grants, enforced
# server-side in the user API. Reporter's defining property is that it holds no
# action capability EVEN when someone tries to grant one directly.
SCOPES_DENIED_ENDPOINT_ACTIONS = frozenset({'reporter'})
def guest_grantable_permissions():
"""Which permissions the guest role may be granted.
@@ -297,6 +465,60 @@ class Role(BaseModel):
))
return True
@classmethod
def seedscopes(cls):
"""Create the prebuilt scope roles and sync their grants.
Returns {'created': [rolename...], 'granted': {rolename: n}}.
Idempotent, and ADDITIVE on re-run. A second run creates nothing and
adds only grants the role is missing; it never REMOVES a permission from
an existing role.
That asymmetry is deliberate. Re-running has to add, because the grant
lists are derived from a plugin-driven catalog: install a plugin, re-run
the seed, and Technicians should gain read access to the new module
without anyone editing a list. But it must not remove, because the role
grid is an administrative surface and a site may have deliberately given
Operator something extra. A revoking seed would silently undo that
decision on the next deploy, which is a worse failure than a role
holding one grant more than the default - the first is invisible, the
second is on screen in the role editor.
Consequence worth stating plainly: this cannot be used to TIGHTEN a
scope after the fact. If a permission is ever added to
SCOPE_NEVER_GRANT, roles already holding it keep it, and revoking is a
separate deliberate migration.
"""
catalog = full_permission_catalog()
result = {'created': [], 'granted': {}}
for rolename, description, color in PREBUILT_SCOPES:
role = cls.query.filter_by(rolename=rolename).first()
if role is None:
role = cls(rolename=rolename, description=description,
color=color)
db.session.add(role)
db.session.flush()
result['created'].append(rolename)
wanted = set(scope_permission_names(rolename, catalog=catalog))
held = {p.name for p in role.permissions}
missing = wanted - held
if not missing:
continue
# Only link permissions that actually exist as rows. A name in the
# catalog with no Permission row means `seed permissions` has not
# run for that plugin yet; the next run picks it up.
rows = Permission.query.filter(Permission.name.in_(missing)).all()
for permission in rows:
role.permissions.append(permission)
if rows:
result['granted'][rolename] = len(rows)
return result
def getpermissionnames(self) -> list:
"""Get list of permission names."""
if self.rolename == 'admin':
@@ -332,6 +554,14 @@ class User(BaseModel):
backref=db.backref('users', lazy='dynamic')
)
# Permissions granted to this user directly, outside any role. See the
# userpermissions table comment for why these exist and what must fold them.
directpermissions = db.relationship(
'Permission',
secondary=userpermissions,
lazy='joined'
)
def __repr__(self):
return f"<User {self.username}>"
@@ -347,18 +577,43 @@ class User(BaseModel):
return any(r.rolename == rolename for r in self.roles)
def haspermission(self, permission_name: str) -> bool:
"""Check if user has a specific permission through any role."""
"""Check if user has a permission, through any role or directly.
A direct grant is not a lesser kind of permission: it is honoured here
exactly like a role grant, which is what keeps the API token scope
ceiling correct without any change to the ceiling itself. Both ceiling
checks (mint time in core/api/apitokens.py and per-request in
utils/authz.py) call this method.
"""
# Admin role has all permissions
if self.hasrole('admin'):
return True
if any(p.name == permission_name for p in self.directpermissions):
return True
return any(r.haspermission(permission_name) for r in self.roles)
def getdirectpermissionnames(self) -> list:
"""Permission names granted to this user directly, ignoring roles.
The user editor needs the DIRECT grants alone so a checkbox reflects
what is stored on the user rather than what a role happens to supply;
getpermissions() deliberately cannot answer that because it folds both.
"""
return sorted(p.name for p in self.directpermissions)
def getpermissions(self) -> list:
"""Get list of all permission names from all roles."""
"""Get list of all permission names, from roles and direct grants.
This is the LIST surface: it feeds JWT claims and the frontend
authStore. It must agree with haspermission(), the predicate surface, or
the UI hides controls the server would allow (or worse, shows ones it
would refuse).
"""
if self.hasrole('admin'):
return [p[0] for p in full_permission_catalog()]
perms = set()
for role in self.roles:
perms.update(role.getpermissionnames())
perms.update(p.name for p in self.directpermissions)
return list(perms)