feat(growth): automate sales source reconciliation

tests/test_pchome_sales_acquisition_service.py

@@ -177,12 +177,19 @@ def test_no_candidate_with_missing_freshness_is_blocked_and_receipted(monkeypatc
     _disable_remote_sources(monkeypatch)
     monkeypatch.delenv("PCHOME_SALES_LOCAL_DROP_DIR", raising=False)
     monkeypatch.setattr(import_service, "drive_service", EmptyDrive())
+    monkeypatch.setattr(acquisition, "drive_service", ReadyDrive())
 
     result = acquisition.PChomeSalesAcquisitionService().run(trigger="test")
 
     assert result["success"] is False
     assert result["status"] == "blocked_with_safe_next_action"
-    assert result["safe_next_action"] == "scheduler_retry_authorized_sources"
+    assert result["safe_next_action"] == "scheduled_retry_and_enabled_fallback_probe"
+    assert result["next_retry_at"]
+    assert result["source_reconciliation"]["contract"] == "pchome_sales_source_reconciliation_v1"
+    assert result["source_reconciliation"]["manual_review_required"] is False
+    assert result["providers"][1]["status"] == "disabled"
+    assert result["providers"][2]["status"] == "disabled"
+    assert result["providers"][3]["status"] == "disabled"
     assert result["receipt_persisted"] is True
     with import_service.engine.connect() as conn:
         receipt = conn.execute(text(
@@ -303,6 +310,7 @@ def test_auto_import_api_uses_governed_acquisition_service(monkeypatch):
 def test_scheduler_continues_when_legacy_hitl_pause_is_set(monkeypatch):
     import scheduler
     import services.agent_actions as agent_actions
+    import services.notification_manager as notification_manager
     import services.pchome_sales_acquisition_service as acquisition
 
     calls = []
@@ -322,8 +330,19 @@ def test_scheduler_continues_when_legacy_hitl_pause_is_set(monkeypatch):
             "data_lag_days": 1,
             "decision_ready": True,
             "safe_next_action": "scheduler_retry_authorized_sources",
+            "next_retry_at": "2026-07-14T19:00:00+08:00",
+            "notification_policy": {
+                "send": False,
+                "reason": "duplicate_within_cooldown",
+                "fingerprint": "same-state",
+            },
         },
     )
+    monkeypatch.setattr(
+        notification_manager,
+        "NotificationManager",
+        lambda: (_ for _ in ()).throw(AssertionError("suppressed notification must not send")),
+    )
     monkeypatch.setattr(scheduler, "_save_stats", lambda name, stats: saved.update({"name": name, "stats": stats}))
 
     scheduler.run_auto_import_task()
@@ -332,3 +351,5 @@ def test_scheduler_continues_when_legacy_hitl_pause_is_set(monkeypatch):
     assert saved["name"] == "auto_import_task"
     assert saved["stats"]["status"] == "Degraded"
     assert saved["stats"]["receipt_persisted"] is True
+    assert saved["stats"]["next_retry_at"] == "2026-07-14T19:00:00+08:00"
+    assert saved["stats"]["notification_policy"]["send"] is False

tests/test_pchome_sales_source_reconciliation.py

@@ -0,0 +1,126 @@
+from datetime import datetime, timedelta, timezone
+
+from services.pchome_sales_source_reconciliation import build_source_reconciliation
+
+
+NOW = datetime(2026, 7, 14, 18, 30, tzinfo=timezone(timedelta(hours=8)))
+
+
+def _source(source_id, *, enabled=True, ready=True, status="ready"):
+    return {
+        "id": source_id,
+        "enabled": enabled,
+        "ready": ready,
+        "status": status,
+    }
+
+
+def _receipt(provider, *, completed_at, reconciliation=None):
+    stages = {"providers": [provider]}
+    if reconciliation:
+        stages["source_reconciliation"] = reconciliation
+    return {
+        "status": "blocked_with_safe_next_action",
+        "completed_at": completed_at,
+        "stages": stages,
+    }
+
+
+def test_stale_single_source_schedules_retry_and_reports_fallback_gap():
+    payload = build_source_reconciliation(
+        sources=[
+            _source("google_drive"),
+            _source("authorized_https", enabled=False, ready=False, status="disabled"),
+        ],
+        provider_attempts=[
+            {"source": "google_drive", "status": "upstream_stale", "file_count": 0},
+            {"source": "authorized_https", "status": "disabled"},
+        ],
+        prior_receipts=[],
+        freshness={"latest_sales_date": "2026-06-24", "data_lag_days": 20, "decision_ready": False},
+        now=NOW,
+    )
+
+    assert payload["contract"] == "pchome_sales_source_reconciliation_v1"
+    assert payload["status"] == "degraded"
+    assert payload["fallback_chain"] == ["google_drive"]
+    assert payload["single_source_dependency"] is True
+    assert payload["next_retry_at"] == "2026-07-14T19:00:00+08:00"
+    assert payload["safe_next_action"] == "scheduled_retry_and_enabled_fallback_probe"
+    assert payload["notification_policy"]["severity"] == "critical"
+    assert payload["notification_policy"]["send"] is True
+    assert payload["sources"][0]["consecutive_empty_runs"] == 1
+    assert payload["sources"][1]["runtime_state"] == "disabled"
+
+
+def test_same_state_within_cooldown_suppresses_duplicate_notification():
+    first = build_source_reconciliation(
+        sources=[_source("google_drive")],
+        provider_attempts=[{"source": "google_drive", "status": "upstream_stale"}],
+        prior_receipts=[],
+        freshness={"data_lag_days": 20, "decision_ready": False},
+        now=NOW - timedelta(minutes=10),
+    )
+    history = [_receipt(
+        {"source": "google_drive", "status": "upstream_stale"},
+        completed_at=(NOW - timedelta(minutes=10)).isoformat(),
+        reconciliation=first,
+    )]
+
+    second = build_source_reconciliation(
+        sources=[_source("google_drive")],
+        provider_attempts=[{"source": "google_drive", "status": "upstream_stale"}],
+        prior_receipts=history,
+        freshness={"data_lag_days": 20, "decision_ready": False},
+        now=NOW,
+    )
+
+    assert second["notification_policy"]["fingerprint"] == first["notification_policy"]["fingerprint"]
+    assert second["notification_policy"]["send"] is False
+    assert second["notification_policy"]["reason"] == "duplicate_within_cooldown"
+    assert second["sources"][0]["consecutive_empty_runs"] == 2
+
+
+def test_provider_failures_use_bounded_exponential_backoff():
+    failed = {"source": "authorized_https", "status": "failed", "error_kind": "http_fetch_failed"}
+    history = [
+        _receipt(failed, completed_at=(NOW - timedelta(minutes=30)).isoformat()),
+        _receipt(failed, completed_at=(NOW - timedelta(minutes=60)).isoformat()),
+    ]
+
+    payload = build_source_reconciliation(
+        sources=[_source("authorized_https")],
+        provider_attempts=[failed],
+        prior_receipts=history,
+        freshness={"data_lag_days": 5, "decision_ready": False},
+        now=NOW,
+        retry_interval_minutes=30,
+        max_failure_backoff_minutes=240,
+    )
+
+    source = payload["sources"][0]
+    assert source["runtime_state"] == "failed"
+    assert source["consecutive_failure_runs"] == 3
+    assert source["retry_minutes"] == 120
+    assert source["next_retry_at"] == "2026-07-14T20:30:00+08:00"
+    assert source["action"] == "retry_with_bounded_backoff_and_probe_fallback"
+
+
+def test_duplicate_source_remains_on_scheduled_retry_without_writing():
+    payload = build_source_reconciliation(
+        sources=[_source("google_drive")],
+        provider_attempts=[{
+            "source": "google_drive",
+            "status": "duplicate",
+            "duplicate_count": 1,
+        }],
+        prior_receipts=[],
+        freshness={"data_lag_days": 10, "decision_ready": False},
+        now=NOW,
+    )
+
+    source = payload["sources"][0]
+    assert source["runtime_state"] == "duplicate"
+    assert source["retry_minutes"] == 30
+    assert source["next_retry_at"] == "2026-07-14T19:00:00+08:00"
+    assert payload["safe_next_action"] == "scheduled_retry_and_enabled_fallback_probe"