feat: add ordered agent runtime

backend/apps/agents/tests/test_runtime.py

@@ -0,0 +1,114 @@
+from datetime import timedelta
+
+from django.utils import timezone
+
+from apps.agents.models import AgentRun
+from apps.agents.services import build_config_snapshot, execute_agent_run
+from apps.agents.tasks import cleanup_expired_agent_runs
+from apps.modelhub.services import CompletionResult
+
+from .helpers import create_agent
+
+
+def create_run(version):
+    return AgentRun.objects.create(
+        definition=version.definition,
+        agent_version=version,
+        sandbox=version.status == "draft",
+        input_payload={"problem": "无法登录"},
+        config_snapshot=build_config_snapshot(version),
+        total_steps=version.steps.count(),
+    )
+
+
+def test_ordered_runtime_records_steps_usage_and_final_output(db, monkeypatch):
+    _, version = create_agent()
+    outputs = iter(
+        [
+            CompletionResult('{"summary":"账号异常"}', 10, 5),
+            CompletionResult('{"reply":"请重置密码"}', 12, 6),
+        ]
+    )
+    monkeypatch.setattr(
+        "apps.modelhub.services._request_completion",
+        lambda *args, **kwargs: next(outputs),
+    )
+    run = create_run(version)
+
+    execute_agent_run(run.id)
+
+    run.refresh_from_db()
+    assert run.status == "succeeded"
+    assert run.final_output == {"reply": "请重置密码"}
+    assert run.input_tokens == 22
+    assert run.output_tokens == 11
+    assert list(run.steps.values_list("code", flat=True)) == ["structure", "reply"]
+
+
+def test_invalid_structure_is_repaired_once(db, monkeypatch):
+    _, version = create_agent(two_steps=False)
+    outputs = iter(
+        [
+            CompletionResult("not-json", 4, 2),
+            CompletionResult('{"summary":"已修复"}', 5, 3),
+        ]
+    )
+    monkeypatch.setattr(
+        "apps.modelhub.services._request_completion",
+        lambda *args, **kwargs: next(outputs),
+    )
+    run = create_run(version)
+
+    execute_agent_run(run.id)
+
+    run.refresh_from_db()
+    step = run.steps.get()
+    assert run.status == "succeeded"
+    assert step.attempts.count() == 2
+    assert list(step.attempts.values_list("attempt_type", flat=True)) == ["initial", "repair"]
+
+
+def test_second_invalid_structure_fails_and_stops(db, monkeypatch):
+    _, version = create_agent()
+    outputs = iter(
+        [
+            CompletionResult("not-json", 4, 2),
+            CompletionResult("still-not-json", 5, 3),
+        ]
+    )
+    monkeypatch.setattr(
+        "apps.modelhub.services._request_completion",
+        lambda *args, **kwargs: next(outputs),
+    )
+    run = create_run(version)
+
+    execute_agent_run(run.id)
+
+    run.refresh_from_db()
+    assert run.status == "failed"
+    assert run.steps.count() == 1
+    assert run.steps.get().status == "failed"
+
+
+def test_cancellation_before_first_step_stops_run(db):
+    _, version = create_agent()
+    run = create_run(version)
+    run.cancellation_requested_at = timezone.now()
+    run.save(update_fields=["cancellation_requested_at"])
+
+    execute_agent_run(run.id)
+
+    run.refresh_from_db()
+    assert run.status == "cancelled"
+    assert run.steps.count() == 0
+
+
+def test_cleanup_removes_expired_runs_but_keeps_audit(db, settings):
+    _, version = create_agent()
+    run = create_run(version)
+    AgentRun.objects.filter(id=run.id).update(created_at=timezone.now() - timedelta(days=91))
+
+    deleted = cleanup_expired_agent_runs()
+
+    assert deleted == 1
+    assert not AgentRun.objects.filter(id=run.id).exists()