update retention policy to a maximum of 2 days for completed archives and adjust related configurations

tests/test_database.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+from datetime import datetime, timedelta
+
 from app.db import Database
 
 
@@ -62,12 +64,15 @@ def test_sqlite_lease_claims_one_queued_job(tmp_path) -> None:
             "files_written": 1,
         },
         warnings=[],
-        retention_days=1,
+        retention_days=30,
     )
     ready = database.get_job(job["id"])
     assert ready is not None
     assert ready["state"] == "ready"
     assert ready["expires_at"] > ready["completed_at"]
+    expires_at = datetime.fromisoformat(ready["expires_at"])
+    completed_at = datetime.fromisoformat(ready["completed_at"])
+    assert expires_at - completed_at <= timedelta(days=2)
 
 
 def test_worker_cannot_cancel_without_a_user_cancellation_request(tmp_path) -> None:

tests/test_main.py

@@ -19,6 +19,8 @@ def test_capture_routes_are_scoped_to_the_browser_session(tmp_path, monkeypatch)
         artifacts_dir=tmp_path / "artifacts",
         reports_dir=tmp_path / "reports",
         worker_enabled=False,
+        default_retention_days=2,
+        max_retention_days=2,
     )
     test_settings.ensure_directories()
     test_database = Database(test_settings.database_path)
@@ -31,7 +33,8 @@ def test_capture_routes_are_scoped_to_the_browser_session(tmp_path, monkeypatch)
     with TestClient(main.app) as client:
         homepage = client.get("/")
         assert homepage.status_code == 200
-        assert "Copy the web." in homepage.text
+        assert "Archive the web." in homepage.text
+        assert "Website Downloader & Offline Archive Tool" in homepage.text
         assert client.get("/api/captures").status_code == 401
 
         created = client.post(
@@ -50,6 +53,18 @@ def test_capture_routes_are_scoped_to_the_browser_session(tmp_path, monkeypatch)
         assert created_job["options"]["parallel_connections"] == 2
         assert created_job["options"]["download_asset_types"] == ["styles", "images"]
         assert created_job["options"]["skip_file_types"] == ["pdf", "zip"]
+        assert created_job["options"]["retention_days"] == 2
+
+        capabilities = client.get("/api/capabilities")
+        assert capabilities.status_code == 200
+        assert capabilities.json()["limits"]["max_retention_days"] == 2
+
+        rejected_retention = client.post(
+            "/api/captures",
+            headers={"X-SiteHarbor-Session": owner_a},
+            json={"url": "https://example.com", "retention_days": 3},
+        )
+        assert rejected_retention.status_code == 422
 
         assert client.get(
             f"/api/captures/{job_id}", headers={"X-SiteHarbor-Session": owner_b}

tests/test_scheduler.py

@@ -0,0 +1,69 @@
+from __future__ import annotations
+
+from dataclasses import replace
+
+from app.config import settings
+from app.db import Database
+from app.scheduler import JobRunner
+
+
+def test_maintenance_removes_archives_after_two_days(tmp_path) -> None:
+    test_settings = replace(
+        settings,
+        data_dir=tmp_path,
+        database_path=tmp_path / "siteharbor.sqlite3",
+        work_dir=tmp_path / "work",
+        artifacts_dir=tmp_path / "artifacts",
+        reports_dir=tmp_path / "reports",
+        worker_enabled=False,
+    )
+    test_settings.ensure_directories()
+    database = Database(test_settings.database_path)
+    database.initialize()
+    job = database.create_job(
+        "https://example.com",
+        "https://example.com/",
+        {
+            "include_external_assets": True,
+            "max_pages": 20,
+            "max_depth": 3,
+            "max_bytes": 1_000_000,
+            "max_duration_seconds": 60,
+            "retention_days": 2,
+        },
+    )
+    assert database.claim_next("worker-a", lease_seconds=30)
+
+    archive_path = test_settings.artifacts_dir / "capture.zip"
+    report_path = test_settings.reports_dir / "capture.json"
+    archive_path.write_bytes(b"zip contents")
+    report_path.write_text("{}", encoding="utf-8")
+    assert database.mark_ready(
+        job["id"],
+        "worker-a",
+        artifact_path=str(archive_path),
+        report_path=str(report_path),
+        archive_size=archive_path.stat().st_size,
+        entry_point="example.com/index.html",
+        stats={},
+        warnings=[],
+        retention_days=2,
+    )
+    with database.connection() as connection:
+        connection.execute(
+            """
+            UPDATE jobs
+            SET completed_at = '2000-01-01T00:00:00+00:00',
+                expires_at = '2100-01-01T00:00:00+00:00'
+            WHERE id = ?
+            """,
+            (job["id"],),
+        )
+
+    JobRunner(database, test_settings)._expire_artifacts_if_due()
+
+    expired = database.get_job(job["id"])
+    assert expired is not None
+    assert expired["state"] == "expired"
+    assert not archive_path.exists()
+    assert not report_path.exists()