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

2026-09-05 06:37:35 +05:30
parent 48402f7d61
commit 9c1a301c1a
10 changed files with 477 additions and 92 deletions
@@ -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()