Files
archie/internal/domain/taskactions/service_test.go
T
sam 36981c2024 feat(statestore): CancelExecution, and one cancel path for daemon stop and operator actions
Plan item 5 of docs/prds/execution-tree-state-machine.md (archie-core-1217).
Until now a stop was a context cancel plus a caller-side transition, and the
step rows a worker still held running could never be told: the execution moved
while the work carried on, and the old interrupt-first ordering existed only
because the record could not defend itself. The step table can: the PRD's
ordering -- the store is the record, the context cancel is the delivery --
became safe the moment a worker's next step write fails ErrStaleTransition
against a cancelled step.

- storecontract.ExecutionCanceller and the CancelExecution RPC: one
  transaction moves every non-terminal StepExecution of the execution's
  current attempt to cancelled (one event row each, in the same
  transaction), and the execution itself to the status the operator action
  names, under the shared transition table with its audit row. Staleness is
  decided before legality, the way guardTransition decides it. Earlier
  attempts are never rewritten. Deliberately administrative: the
  deny-by-default grant rule covers it, pinned by a test.
- taskactions.Service routes stop, cancel, abandon and reject through it and
  the old Transition/CancelTask combinations are deleted; the daemon's
  in-memory cancel is demoted to the delivery hook the PRD names, called
  after the store record.
- The gateway's chat controller records first and delivers second, the same
  order, and records even when no runtime is wired here (a standalone
  gateway cannot reach the daemon's goroutines): the record is what unsticks
  a lost-cancel task, and the delivery is best effort. StopRunning and
  TaskRuntime.CancelRunningTasks are deleted outright -- no production
  caller remained once the task_action tool owned stopping, and the daemon's
  CancelRunningTasks/runningTasks.stopAll go with them.
- The gateway's chat adapter writes through CancelExecution instead of a
  bare Transition, so chat /cancel and the dashboard's action service land
  on the same path.

PRD decisions recorded here:
- Each cancelled step's event row is stage_finish with {cancelled: true} and
  the reason as its detail -- the exact-one-event-per-step-transition rule
  covers the pending -> cancelled transition too, even for a step that never
  ran.
- A parked cancel carries the needs_human class, matching the Transition
  path this replaces; the transient class stays with parkRunningTask's
  unwind parks, which now meet ErrStaleTransition for cancelled runs and are
  ignored there.
- Cancelling a wait:true callee is deferred to #1216's agent-call children:
  the callee's own tree and the wait flag are what make
  "cancelling a caller cancels each callee it waits on" decidable, and the
  callee writes it cannot fake are this issue's half of the contract.
- RecoverStale's interrupted sweep is plan item 6, untouched here.
2026-09-27 06:34:01 +10:00

245 lines
8.7 KiB
Go

package taskactions
import (
"context"
"testing"
"github.com/samcharles93/archie-core/internal/domain/identity"
"github.com/samcharles93/archie-core/internal/events"
"github.com/samcharles93/archie-core/internal/taskstate"
)
type fakeStore struct {
task *Task
events []events.Event
}
func (f *fakeStore) TaskByID(context.Context, int64) (*Task, error) { return f.task, nil }
func (f *fakeStore) Transition(context.Context, int64, string, string, string) error {
return nil
}
func (f *fakeStore) CancelExecution(context.Context, int64, string, string) ([]int64, error) {
return nil, nil
}
func (f *fakeStore) Requeue(context.Context, int64, string, string) error { return nil }
func (f *fakeStore) RetryTask(context.Context, int64, string, string) error {
return nil
}
func (f *fakeStore) ArchiveTask(_ context.Context, _ int64, _ string, e events.Event) (int64, error) {
f.events = append(f.events, e)
return int64(len(f.events)), nil
}
func (f *fakeStore) InsertEvent(_ context.Context, e events.Event) (int64, error) {
f.events = append(f.events, e)
return int64(len(f.events)), nil
}
func (f *fakeStore) last() events.Event {
if len(f.events) == 0 {
return events.Event{}
}
return f.events[len(f.events)-1]
}
const (
humanID = identity.IdentityID("20000000-0000-5000-8000-000000000001")
agentID = identity.IdentityID("20000000-0000-5000-8000-000000000002")
)
func humanActor() Actor {
return ActorFor(identity.Identity{ID: humanID, Kind: identity.KindUser, Lifecycle: identity.LifecycleActive})
}
func agentActor() Actor {
return ActorFor(identity.Identity{ID: agentID, Kind: identity.KindBot, Lifecycle: identity.LifecycleActive})
}
// TestEventKindDescribesTheActor is the rule that the bug turned on: an agent's
// approval was written down as a human's. The kind names the actor, so an
// agent's action can never be read as a person's, and an action with no verified
// actor is recorded as a task-level event that credits nobody.
func TestEventKindDescribesTheActor(t *testing.T) {
tests := []struct {
name string
action taskstate.Action
actor Actor
wantKind string
}{
{"a person approves", taskstate.ActionApprove, humanActor(), events.KindHumanApproved},
{"an agent approves", taskstate.ActionApprove, agentActor(), events.KindAgentApproved},
{"a person rejects", taskstate.ActionReject, humanActor(), events.KindHumanRejected},
{"an agent rejects", taskstate.ActionReject, agentActor(), events.KindAgentRejected},
{"an unattributed approval", taskstate.ActionApprove, Actor{}, events.KindTaskApproved},
{"an unattributed rejection", taskstate.ActionReject, Actor{}, events.KindTaskRejected},
}
for _, tc := range tests {
t.Run(tc.name, func(t *testing.T) {
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: "waiting_human"}}
service := Service{Store: store}
if err := service.Apply(context.Background(), nil, tc.actor, 7, tc.action); err != nil {
t.Fatalf("Apply() error = %v", err)
}
got := store.last()
if got.Kind != tc.wantKind {
t.Fatalf("event kind = %q, want %q", got.Kind, tc.wantKind)
}
if tc.actor.Human() && got.Kind == events.KindAgentApproved {
t.Fatal("a person's action was recorded as an agent's")
}
if tc.actor.Attributed() && !tc.actor.Human() && got.Kind == events.KindHumanApproved {
t.Fatal("an agent's action was recorded as a human approval")
}
})
}
}
// TestActorAndPrincipalAreRecordedSeparately is the two-attribution requirement:
// an agent acting under a person's standing authority records both, so "approved
// by came from me" is answerable from the record.
func TestActorAndPrincipalAreRecordedSeparately(t *testing.T) {
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: "waiting_human"}}
service := Service{Store: store}
actor := agentActor().AuthorisedBy(humanID)
if err := service.Apply(context.Background(), nil, actor, 7, taskstate.ActionApprove); err != nil {
t.Fatalf("Apply() error = %v", err)
}
got := store.last()
if got.ActorID != string(agentID) {
t.Fatalf("actor = %q, want %q", got.ActorID, agentID)
}
if got.ActorKind != string(identity.KindBot) {
t.Fatalf("actor kind = %q, want %q", got.ActorKind, identity.KindBot)
}
if got.PrincipalID != string(humanID) {
t.Fatalf("principal = %q, want %q", got.PrincipalID, humanID)
}
if got.Kind != events.KindAgentApproved {
t.Fatalf("event kind = %q, want %q", got.Kind, events.KindAgentApproved)
}
}
// TestUnattributedActionClaimsNoApprover: with no verified actor the record must
// not credit the actor's own authority, and must not name a human.
func TestUnattributedActionClaimsNoApprover(t *testing.T) {
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: "waiting_human"}}
service := Service{Store: store}
if err := service.Apply(context.Background(), nil, Actor{}, 7, taskstate.ActionApprove); err != nil {
t.Fatalf("Apply() error = %v", err)
}
got := store.last()
if got.ActorID != "" || got.ActorKind != "" || got.PrincipalID != "" {
t.Fatalf("unattributed action recorded an actor: %+v", got)
}
if got.Kind == events.KindHumanApproved || got.Kind == events.KindAgentApproved {
t.Fatalf("unattributed action claimed an actor through its kind: %q", got.Kind)
}
}
// TestEveryActionCarriesItsAttribution: no action path may record an event that
// skipped the attribution, including the ones that also close a forge issue.
func TestEveryActionCarriesItsAttribution(t *testing.T) {
actions := []struct {
action taskstate.Action
status string
}{
{taskstate.ActionApprove, "waiting_human"},
{taskstate.ActionReject, "waiting_human"},
{taskstate.ActionRetry, "parked"},
{taskstate.ActionStop, "running"},
{taskstate.ActionCancel, "queued"},
{taskstate.ActionAbandon, "parked"},
}
for _, tc := range actions {
t.Run(string(tc.action), func(t *testing.T) {
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: tc.status}}
service := Service{
Store: store,
CancelTask: func(int64) bool { return true },
}
if err := service.Apply(context.Background(), nil, agentActor(), 7, tc.action); err != nil {
t.Fatalf("Apply(%s) error = %v", tc.action, err)
}
got := store.last()
if got.ActorID != string(agentID) || got.ActorKind != string(identity.KindBot) {
t.Fatalf("%s recorded without its actor: %+v", tc.action, got)
}
})
}
}
// TestScopeStillLimitsATaskButIsNotTheActor: scope answers which task a caller
// may touch. It is not an actor, and passing one must not attribute the action
// to it.
func TestScopeStillLimitsATaskButIsNotTheActor(t *testing.T) {
other := "sam"
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: "waiting_human", Identity: "archie"}}
service := Service{Store: store}
if err := service.Apply(context.Background(), &other, agentActor(), 7, taskstate.ActionApprove); err == nil {
t.Fatal("Apply() allowed an identity to act on another identity's task")
}
if len(store.events) != 0 {
t.Fatalf("a refused action recorded %d events", len(store.events))
}
if err := service.Apply(context.Background(), &other, agentActor(), 7, taskstate.ActionApprove); err == nil {
t.Fatal("Apply() allowed a scoped caller to act on a foreign task")
}
}
func TestActorForCarriesTheKindAndLeavesTheActionUnattributedWithoutAPrincipal(t *testing.T) {
actor := agentActor()
if actor.Principal != "" {
t.Fatalf("ActorFor() set a principal: %q", actor.Principal)
}
if !actor.Attributed() {
t.Fatal("ActorFor() produced an unattributed actor for a real identity")
}
if (Actor{}).Attributed() {
t.Fatal("the zero Actor reported itself as attributed")
}
if (Actor{}).Human() {
t.Fatal("the zero Actor reported itself as a human")
}
}
// TestOnlyRejectAndCancelCloseTheIssue: abandoning is archie giving up on its
// own run, not a verdict on the issue, so the issue stays open for a human.
func TestOnlyRejectAndCancelCloseTheIssue(t *testing.T) {
for _, tc := range []struct {
action taskstate.Action
status string
closes bool
}{
{taskstate.ActionReject, "waiting_human", true},
{taskstate.ActionCancel, "queued", true},
{taskstate.ActionAbandon, "parked", false},
} {
t.Run(string(tc.action), func(t *testing.T) {
store := &fakeStore{task: &Task{ID: 7, Owner: "acme", Repo: "widgets", Status: tc.status, ForgeBacked: true}}
closed := false
service := Service{Store: store, CloseIssue: func(context.Context, string, string, int, string) error {
closed = true
return nil
}}
if err := service.Apply(context.Background(), nil, agentActor(), 7, tc.action); err != nil {
t.Fatalf("Apply(%s) error = %v", tc.action, err)
}
if closed != tc.closes {
t.Fatalf("Apply(%s) closed the issue = %v, want %v", tc.action, closed, tc.closes)
}
})
}
}