school_compare/backend/tests/test_school_groups.py

"""Parent-facing groups over GIAS establishment types and religious characters.

Every GIAS code must be accounted for, so a new DfE code fails here instead of
silently vanishing from the filter.
"""

from pathlib import Path

import numpy as np
import pytest
import yaml

from backend.gias_codes import RELIGIOUS_CHARACTER, SCHOOL_TYPE
from backend.school_groups import (
    FAITH_GROUPS,
    TYPE_GROUPS,
    UNOFFERED_TYPE_CODES,
    faith_groups_for,
    type_group_for,
    type_group_key,
)

DBT_PROJECT = Path(__file__).resolve().parents[2] / "pipeline" / "transform" / "dbt_project.yml"


def _non_england_codes() -> set[int]:
    return set(yaml.safe_load(DBT_PROJECT.read_text())["vars"]["non_england_school_type_codes"])


def test_every_type_code_is_in_exactly_one_place():
    places = [codes for _, _, codes in TYPE_GROUPS] + [UNOFFERED_TYPE_CODES, _non_england_codes()]
    for code in SCHOOL_TYPE:
        homes = sum(code in p for p in places)
        assert homes == 1, f"type code {code} ({SCHOOL_TYPE[code]}) is in {homes} places"


def test_every_religion_code_has_a_faith():
    for code, name in RELIGIOUS_CHARACTER.items():
        assert any(code in codes for _, _, codes in FAITH_GROUPS), f"{code} {name!r}"


def test_type_groups_in_display_order():
    assert [k for k, _, _ in TYPE_GROUPS] == [
        "state", "independent", "special", "post16", "alternative"]


def test_faiths_in_display_order():
    assert [k for k, _, _ in FAITH_GROUPS] == [
        "none", "church_of_england", "roman_catholic", "other_christian",
        "jewish", "muslim", "other_faith"]


@pytest.mark.parametrize("name, group", [
    ("Academy converter", "state"),
    ("University technical college", "state"),
    ("Voluntary aided school", "state"),
    ("Local authority nursery school", "state"),
    ("Other independent school", "independent"),
    ("Other independent special school", "special"),
    ("Special post 16 institution", "special"),
    ("Further education", "post16"),
    ("Pupil referral unit", "alternative"),
    ("academy CONVERTER", "state"),
])
def test_type_group_by_name(name, group):
    assert type_group_for(name) == group


@pytest.mark.parametrize("value, key", [
    ("state", "state"),
    ("Special", "special"),
    # The two state groups that preceded "state", kept so their links still work.
    ("academy", "state"),
    ("Council", "state"),
    ("Community school", None),
    ("", None),
])
def test_type_group_key_resolves_keys_and_old_keys(value, key):
    assert type_group_key(value) == key


@pytest.mark.parametrize("name", [
    "Higher education institutions", "Miscellaneous", "Unknown (9999)", "Academy", "", None, np.nan,
])
def test_unoffered_or_unknown_types_have_no_group(name):
    assert type_group_for(name) is None


@pytest.mark.parametrize("name, faiths", [
    ("Roman Catholic/Church of England", ("church_of_england", "roman_catholic")),
    ("Roman Catholic/Anglican", ("church_of_england", "roman_catholic")),
    ("Church of England/Methodist", ("church_of_england", "other_christian")),
    ("Church of England/Christian", ("church_of_england",)),
    ("Catholic", ("roman_catholic",)),
    ("Inter- / non- denominational", ("other_christian",)),
    ("Orthodox Jewish", ("jewish",)),
    ("Sunni Deobandi", ("muslim",)),
    ("Hindu", ("other_faith",)),
    ("Does not apply", ("none",)),
    ("None", ("none",)),
])
def test_faiths_by_name(name, faiths):
    assert faith_groups_for(name) == faiths


@pytest.mark.parametrize("missing", [None, np.nan, "", "  "])
def test_a_missing_religion_is_no_religious_character(missing):
    assert faith_groups_for(missing) == ("none",)


def test_an_unknown_religion_has_no_faith():
    assert faith_groups_for("Unknown (77)") == ()