PlantAI-dev/tests/unit/test_string_species_registry.py

from pathlib import Path
import json

import pytest

from scripts.prepare_string_species_snapshot import prepare_snapshot


PROJECT_ROOT = Path(__file__).resolve().parents[2]
FIXTURE_SOURCE = PROJECT_ROOT / "tests" / "fixtures" / "string_species" / "v12.0"
FULL_SNAPSHOT = PROJECT_ROOT / "data" / "string" / "v12.0"


@pytest.fixture()
def snapshot(tmp_path: Path):
    from gfa_core.string_species_registry import load_string_snapshot

    destination = tmp_path / "snapshot"
    prepare_snapshot(destination, source_directory=FIXTURE_SOURCE)
    return load_string_snapshot(destination)


def test_exact_string_uses_explicit_target_taxon(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    result = resolve_ppi_string_species(
        snapshot,
        "Os",
        ppi_target_tree_taxon_id="39947",
    )

    assert result["selection_mode"] == "exact_string"
    assert result["target_tree_taxon_id"] == "39947"
    assert result["selected_string_taxon_id"] == "39947"
    assert result["requires_user_confirmation"] is False


def test_os_alias_uses_string_japonica_taxon(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    result = resolve_ppi_string_species(snapshot, "Os")

    assert result["target_tree_taxon_id"] == "39947"
    assert result["selected_string_taxon_id"] == "39947"
    assert result["resolution_source"] == "project_alias"


def test_explicit_taxon_overrides_project_alias(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    result = resolve_ppi_string_species(
        snapshot,
        "Os",
        ppi_target_tree_taxon_id="39946",
    )

    assert result["target_tree_taxon_id"] == "39946"
    assert result["selected_string_taxon_id"] == "39946"
    assert result["resolution_source"] == "explicit_tree_taxon_id"


def test_registry_species_outside_legacy_map_is_exact_string(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    result = resolve_ppi_string_species(
        snapshot,
        "",
        target_scientific_name="Brassica napus",
    )

    assert result["selection_mode"] == "exact_string"
    assert result["selected_string_taxon_id"] == "3708"


def test_ambiguous_name_is_unresolved_not_at(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    result = resolve_ppi_string_species(
        snapshot,
        "At",
        target_scientific_name="Ambiguous plant",
    )

    assert result["selection_mode"] == "unresolved"
    assert result["selected_string_taxon_id"] == ""
    assert result["issue_code"] == "target_string_tree_unresolved"


def test_tree_nearest_is_stable_top_three(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    first = resolve_ppi_string_species(snapshot, "Ft")
    second = resolve_ppi_string_species(snapshot, "Ft")

    assert first["selection_mode"] == "tree_nearest"
    assert [row["selected_string_taxon_id"] for row in first["candidates"]] == [
        "3562",
        "3702",
        "3708",
    ]
    assert first["candidates"] == second["candidates"]
    assert first["requires_user_confirmation"] is True


def test_unresolved_requires_explicit_model_fallback_choice(snapshot) -> None:
    from gfa_core.string_species_registry import resolve_ppi_string_species

    unresolved = resolve_ppi_string_species(snapshot, "Unknown")
    fallback = resolve_ppi_string_species(
        snapshot,
        "Unknown",
        allow_model_fallback=True,
    )

    assert unresolved["selection_mode"] == "unresolved"
    assert unresolved["selected_string_taxon_id"] == ""
    assert fallback["selection_mode"] == "model_fallback"
    assert fallback["selected_string_taxon_id"] == "3702"
    assert fallback["requires_user_confirmation"] is True


def test_corrupt_snapshot_is_resource_error(tmp_path: Path) -> None:
    from gfa_core.string_species_registry import StringSnapshotError, load_string_snapshot

    with pytest.raises(StringSnapshotError) as error:
        load_string_snapshot(tmp_path)

    assert error.value.code in {
        "string_registry_unavailable",
        "string_registry_integrity_failed",
    }


def test_malformed_manifest_file_metadata_is_integrity_error(tmp_path: Path) -> None:
    from gfa_core.string_species_registry import StringSnapshotError, load_string_snapshot

    (tmp_path / "source_manifest.json").write_text(
        json.dumps(
            {
                "snapshot_id": "invalid",
                "string_version": "12.0",
                "files": {
                    "species.v12.0.txt": "not-an-object",
                    "species.tree.v12.0.txt": {},
                },
            }
        ),
        encoding="utf-8",
    )

    with pytest.raises(StringSnapshotError) as error:
        load_string_snapshot(tmp_path)

    assert error.value.code == "string_registry_integrity_failed"


def test_full_snapshot_supports_registry_only_species_and_second_non_ft_target() -> None:
    from gfa_core.string_species_registry import (
        load_string_snapshot,
        resolve_ppi_string_species,
    )

    full = load_string_snapshot(FULL_SNAPSHOT)
    brassica = resolve_ppi_string_species(
        full,
        "",
        target_scientific_name="Brassica napus",
    )
    camellia = resolve_ppi_string_species(
        full,
        "",
        ppi_target_tree_taxon_id="4442",
    )

    assert brassica["selection_mode"] == "exact_string"
    assert brassica["selected_string_taxon_id"] == "3708"
    assert camellia["selection_mode"] == "tree_nearest"
    assert len(camellia["candidates"]) == 3
    assert camellia["candidate_source"] == "string_species_tree"