tests/sync/test_delta.py

from sync.delta import compute_content_hash, compute_delta, build_state


def _paper(id_, title="T", abstract="A"):
    return {"id": id_, "title": title, "abstract": abstract, "authors": "X", "venue": "ACL", "year": 2023, "url": "http://x"}


def test_compute_content_hash_is_stable():
    p = _paper("p1")
    assert compute_content_hash(p) == compute_content_hash(_paper("p1"))


def test_compute_content_hash_changes_with_content():
    assert compute_content_hash(_paper("p1", title="A")) != compute_content_hash(_paper("p1", title="B"))


def test_compute_delta_detects_new_paper():
    current = [_paper("p1")]
    result = compute_delta(current, last_state={})
    assert [p["id"] for p in result.new_or_changed] == ["p1"]
    assert result.removed_ids == []


def test_compute_delta_detects_changed_paper():
    old_hash = compute_content_hash(_paper("p1", title="Old"))
    current = [_paper("p1", title="New")]
    result = compute_delta(current, last_state={"p1": old_hash})
    assert [p["id"] for p in result.new_or_changed] == ["p1"]


def test_compute_delta_skips_unchanged_paper():
    p = _paper("p1")
    current = [p]
    result = compute_delta(current, last_state={"p1": compute_content_hash(p)})
    assert result.new_or_changed == []


def test_compute_delta_detects_removed_paper():
    current = []
    result = compute_delta(current, last_state={"p1": "somehash"})
    assert result.removed_ids == ["p1"]


def test_build_state_maps_id_to_hash():
    p = _paper("p1")
    state = build_state([p])
    assert state == {"p1": compute_content_hash(p)}