implement cancellation handling in CaptureService and Database, update job states accordingly

tests/test_database.py

@@ -97,13 +97,73 @@ def test_worker_cannot_cancel_without_a_user_cancellation_request(tmp_path) -> N
     assert running is not None
     assert running["state"] == "running"
 
-    database.request_cancel(job["id"])
-    assert database.mark_cancelled(job["id"], "worker-a", "Cancelled by user")
+    cancellation = database.request_cancel(job["id"])
+    assert cancellation is not None
+    assert cancellation["state"] == "cancelled"
+    assert not database.mark_cancelled(job["id"], "worker-a", "Cancelled by user")
     cancelled = database.get_job(job["id"])
     assert cancelled is not None
     assert cancelled["state"] == "cancelled"
 
 
+def test_worker_stop_check_detects_a_cancellation_request(tmp_path) -> None:
+    database = Database(tmp_path / "siteharbor.sqlite3")
+    database.initialize()
+    job = database.create_job(
+        "https://example.com",
+        "https://example.com/",
+        capture_config(),
+    )
+    assert database.claim_next("worker-a", lease_seconds=30)
+
+    assert not database.worker_should_stop(job["id"], "worker-a")
+    assert database.request_cancel(job["id"])
+    assert database.worker_should_stop(job["id"], "worker-a")
+
+
+def test_cancel_terminalizes_an_existing_cancellation_request(tmp_path) -> None:
+    database = Database(tmp_path / "siteharbor.sqlite3")
+    database.initialize()
+    job = database.create_job(
+        "https://example.com",
+        "https://example.com/",
+        capture_config(),
+    )
+    assert database.claim_next("worker-a", lease_seconds=30)
+    with database.connection() as connection:
+        connection.execute(
+            "UPDATE jobs SET cancel_requested = 1 WHERE id = ?",
+            (job["id"],),
+        )
+
+    cancelled = database.request_cancel(job["id"])
+
+    assert cancelled is not None
+    assert cancelled["state"] == "cancelled"
+
+
+def test_initialize_recovers_a_legacy_cancellation_request(tmp_path) -> None:
+    database = Database(tmp_path / "siteharbor.sqlite3")
+    database.initialize()
+    job = database.create_job(
+        "https://example.com",
+        "https://example.com/",
+        capture_config(),
+    )
+    assert database.claim_next("worker-a", lease_seconds=30)
+    with database.connection() as connection:
+        connection.execute(
+            "UPDATE jobs SET cancel_requested = 1 WHERE id = ?",
+            (job["id"],),
+        )
+
+    database.initialize()
+
+    recovered = database.get_job(job["id"])
+    assert recovered is not None
+    assert recovered["state"] == "cancelled"
+
+
 def test_expired_cancel_requested_job_is_terminalized_before_reclaim(tmp_path) -> None:
     database = Database(tmp_path / "siteharbor.sqlite3")
     database.initialize()