RepoPulse-Public/backend/tests/test_store.py

import json
from datetime import UTC, datetime, timedelta
from pathlib import Path

import app.store as store_module
from app.scanner import scan_repository
from app.store import ScanStore


def test_scan_history_comparison_and_trend(tmp_path: Path) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")
    (repository / ".gitignore").write_text("*.db\n", encoding="utf-8")

    store = ScanStore(str(tmp_path / "repopulse.db"))
    first = store.save(scan_repository(str(repository)))

    (repository / "app.py").write_text("print('hello')\n", encoding="utf-8")
    (repository / "tests").mkdir()
    (repository / "tests" / "test_app.py").write_text("def test_ok():\n    assert True\n", encoding="utf-8")
    second = store.save(scan_repository(str(repository)))

    assert store.get(second.id) == second
    assert store.recent(1) == [second]
    assert store.previous_for_repository(second.repository_path, second.id) == first

    comparison = store.compare(second)
    assert comparison.previous_scan_id == first.id
    assert comparison.health_score_delta == second.health_score - first.health_score
    assert comparison.metric_deltas["source_files"] == 2

    trend = store.trend(second.repository_path)
    assert [item.scan_id for item in trend] == [first.id, second.id]
    assert trend[-1].health_score == second.health_score


def test_comparison_uses_the_scan_before_the_selected_scan(tmp_path: Path) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")
    (repository / ".gitignore").write_text("*.db\n", encoding="utf-8")

    store = ScanStore(str(tmp_path / "repopulse.db"))
    first = store.save(scan_repository(str(repository)))
    second = store.save(scan_repository(str(repository)))
    third = store.save(scan_repository(str(repository)))

    assert store.previous_for_repository(second.repository_path, second.id) == first
    assert store.previous_for_repository(third.repository_path, third.id) == second


def test_repository_summary_exposes_compact_remediation_context(tmp_path: Path) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")
    (repository / ".gitignore").write_text(".env\n", encoding="utf-8")
    (repository / ".env").write_text("TOKEN=actual-secret-value-123456\n", encoding="utf-8")

    store = ScanStore(str(tmp_path / "repopulse.db"))
    store.save(scan_repository(str(repository)))

    (repository / ".env").unlink()
    (repository / "app.py").write_text("api_key = 'actual-secret-value-123456'\n", encoding="utf-8")
    store.save(scan_repository(str(repository)))

    summary = store.list_repositories()[0]

    assert summary.last_scan_blockers == 1
    assert summary.last_scan_primary_signal is not None
    assert summary.last_scan_primary_signal.code == "security.secret-assignment"
    assert summary.last_scan_primary_signal.count == 1
    assert summary.last_scan_added_findings >= 1
    assert summary.last_scan_resolved_findings == 1


def test_corrupt_database_is_quarantined_and_recreated(tmp_path: Path) -> None:
    database = tmp_path / "repopulse.db"
    database.write_bytes(b"not a sqlite database")

    store = ScanStore(str(database))

    assert store.recent() == []
    assert list(tmp_path.glob("repopulse.db.corrupt-*"))
    assert store.database_recovery_status()["status"] == "degraded_recovered"


def test_schema_migration_records_the_durable_backup_artifact(tmp_path: Path) -> None:
    database = tmp_path / "repopulse.db"
    initial_store = ScanStore(str(database))
    with initial_store._connection() as db:
        db.execute("PRAGMA user_version = 3")

    migrated_store = ScanStore(str(database))

    recorded = migrated_store.get_state("last_schema_migration_backup")
    assert recorded is not None
    metadata = json.loads(recorded)
    assert metadata["from_version"] == 3
    assert metadata["to_version"] == store_module.SCHEMA_VERSION
    assert (tmp_path / metadata["backup_file"]).is_file()
    assert migrated_store.storage_stats()["last_schema_migration_backup"] == recorded


def test_history_is_bounded_per_repository(tmp_path: Path, monkeypatch) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")

    monkeypatch.setattr(store_module, "MAX_HISTORY_PER_REPOSITORY", 2)
    store = ScanStore(str(tmp_path / "repopulse.db"))
    first = store.save(scan_repository(str(repository)))
    second = store.save(scan_repository(str(repository)))
    third = store.save(scan_repository(str(repository)))

    assert [item.id for item in store.recent_for_repository(third.repository_path)] == [third.id, second.id]
    assert store.get(first.id) is None


def test_retention_compacts_old_detailed_payloads_and_expires_old_snapshots(tmp_path: Path, monkeypatch) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")
    (repository / ".gitignore").write_text("*.db\n", encoding="utf-8")
    monkeypatch.setenv("REPOPULSE_FULL_REPORTS_PER_REPOSITORY", "1")
    monkeypatch.setenv("REPOPULSE_REPORT_RETENTION_DAYS", "30")
    store = ScanStore(str(tmp_path / "repopulse.db"))

    first = store.save(scan_repository(str(repository)))
    second = store.save(scan_repository(str(repository)))
    compacted = store.get(first.id)

    assert compacted is not None and compacted.compacted is True
    assert compacted.findings == []
    assert store.get(second.id) is not None

    expired = second.model_copy(
        update={"id": "expired-scan", "scanned_at": datetime.now(UTC) - timedelta(days=31)}
    )
    store.save(expired)
    assert store.get("expired-scan") is None


def test_catalog_summary_uses_denormalized_fields_without_report_json_parsing(tmp_path: Path, monkeypatch) -> None:
    repository = tmp_path / "repo"
    repository.mkdir()
    (repository / "README.md").write_text("# Demo\n", encoding="utf-8")
    (repository / ".gitignore").write_text("*.db\n", encoding="utf-8")
    store = ScanStore(str(tmp_path / "repopulse.db"))
    store.save(scan_repository(str(repository)))

    def fail_if_called(_row):
        raise AssertionError("catalog list must not parse report_json")

    monkeypatch.setattr(store, "_report_from_row", fail_if_called)
    summary = store.list_repositories()[0]
    assert summary.last_scan_id
    storage = store.storage_stats()
    assert storage["scan_count"] == 1
    assert storage["mirror_count"] >= 0
    assert storage["worktree_count"] >= 0