infospace-bench/tests/test_archive.py

from __future__ import annotations

from pathlib import Path

import pytest
import yaml

from infospace_bench import (
    ArchiveRecord,
    InfospaceError,
    add_artifact,
    archive_infospace,
    create_infospace,
    list_archives,
)
from infospace_bench.archive import (
    ARCHIVE_INDEX_PATH,
    ARCHIVE_STORE_DIR,
    DEFAULT_RETENTION_CLASS,
    PRODUCER,
)


def _seed_infospace(workspace: Path, slug: str = "demo") -> Path:
    create_infospace(workspace, slug, name="Demo", topic_domain="Test")
    root = workspace / "infospaces" / slug
    source = workspace / "source.md"
    source.write_text("# source\n", encoding="utf-8")
    add_artifact(root, source, kind="source", title="Source One")
    (root / "reports" / "summary.md").write_text("# summary\n", encoding="utf-8")
    return root


def test_archive_infospace_writes_index_and_finalizes_package(tmp_path: Path) -> None:
    root = _seed_infospace(tmp_path)

    record = archive_infospace(root, note="first archive")

    assert isinstance(record, ArchiveRecord)
    assert record.package_id
    assert record.manifest_digest.startswith("blake3:")
    assert record.retention_class == DEFAULT_RETENTION_CLASS
    assert record.producer == PRODUCER
    assert record.subject == "demo"
    assert record.note == "first archive"
    assert record.file_count >= 4  # infospace.yaml, index.yaml, source.md, summary.md

    index_path = root / ARCHIVE_INDEX_PATH
    assert index_path.is_file()
    data = yaml.safe_load(index_path.read_text(encoding="utf-8"))
    assert isinstance(data, dict)
    assert len(data["archives"]) == 1
    assert data["archives"][0]["package_id"] == record.package_id

    store_root = root / ARCHIVE_STORE_DIR
    assert (store_root / "registry.sqlite").is_file()
    assert (store_root / "storage").is_dir()


def test_list_archives_returns_recorded_entries(tmp_path: Path) -> None:
    root = _seed_infospace(tmp_path)

    assert list_archives(root) == []
    first = archive_infospace(root, note="alpha")
    second = archive_infospace(root, note="beta")

    archives = list_archives(root)
    assert [a.package_id for a in archives] == [first.package_id, second.package_id]
    assert [a.note for a in archives] == ["alpha", "beta"]


def test_archive_excludes_store_dir_to_avoid_recursive_capture(tmp_path: Path) -> None:
    root = _seed_infospace(tmp_path)

    first = archive_infospace(root)
    second = archive_infospace(root)

    # The store dir grows on the first call; the second call must not pick up
    # any of its bytes (otherwise file_count would balloon).
    assert second.file_count == first.file_count
    second_record = list_archives(root)[1]
    assert all(
        not path.startswith(ARCHIVE_STORE_DIR)
        for path in second_record.included_paths
    )


def test_archive_respects_caller_supplied_include_set(tmp_path: Path) -> None:
    root = _seed_infospace(tmp_path)

    record = archive_infospace(root, include=["infospace.yaml"])
    assert record.included_paths == ["infospace.yaml"]
    assert record.file_count == 1


def test_archive_rejects_empty_include(tmp_path: Path) -> None:
    root = _seed_infospace(tmp_path)

    with pytest.raises(InfospaceError) as excinfo:
        archive_infospace(root, include=["does-not-exist"])
    assert excinfo.value.code == "empty_archive"