stack/tests/pfs/test_files.py

"""Tests for pfs.files — Format/Sieve definitions and registry integration.

Covers every line of pfs/files/__init__.py:
- All 6 column map dicts (_RVU_COLUMNS through _CARRIER_COLUMNS)
- All 4 sieve definitions (_rvu_sieve, _gpci_sieve, _pe_input_sieve, _carrier_sieve)
- All 6 format definitions in the ``formats`` dict
- Registry integration (``registry.update(formats)``)
"""

from __future__ import annotations

import re

import pytest

from pfs.files import formats
from rex.format import Format, registry
from rex.sieve import Sieve

# ── Format dict structure ────────────────────────────────────────


class TestFormatsDict:
    """The ``formats`` dict has exactly 6 entries."""

    EXPECTED_KEYS = [
        "pfs_rvu",
        "pfs_gpci",
        "pfs_labor",
        "pfs_supply",
        "pfs_equipment",
        "pfs_carrier",
    ]

    def test_has_six_keys(self):
        assert len(formats) == 6

    def test_key_names(self):
        assert sorted(formats.keys()) == sorted(self.EXPECTED_KEYS)

    @pytest.mark.parametrize("key", EXPECTED_KEYS)
    def test_values_are_format_instances(self, key):
        assert isinstance(formats[key], Format)


# ── Registry integration ─────────────────────────────────────────


class TestRegistryIntegration:
    """All PFS formats are registered in rex.format.registry at import time."""

    @pytest.mark.parametrize(
        "key",
        [
            "pfs_rvu",
            "pfs_gpci",
            "pfs_labor",
            "pfs_supply",
            "pfs_equipment",
            "pfs_carrier",
        ],
    )
    def test_format_in_registry(self, key):
        assert key in registry
        assert registry[key] is formats[key]


# ── Per-format details ───────────────────────────────────────────

_FORMAT_SPECS = [
    ("pfs_rvu", "pfs.rvu", "pfs_rvu"),
    ("pfs_gpci", "pfs.gpci", "pfs_gpci"),
    ("pfs_labor", "pfs.clinical_labor", "pfs_pe_input"),
    ("pfs_supply", "pfs.medical_supply", "pfs_pe_input"),
    ("pfs_equipment", "pfs.medical_equipment", "pfs_pe_input"),
    ("pfs_carrier", "pfs.carrier_locality", "pfs_carrier"),
]


@pytest.mark.parametrize(
    "fmt_key, expected_target, expected_sieve_name",
    _FORMAT_SPECS,
    ids=[s[0] for s in _FORMAT_SPECS],
)
class TestFormatDetails:
    """Each format has the correct name, target, sieve, and column_map."""

    def test_name_matches_key(self, fmt_key, expected_target, expected_sieve_name):
        fmt = formats[fmt_key]
        assert fmt.name == fmt_key

    def test_target(self, fmt_key, expected_target, expected_sieve_name):
        fmt = formats[fmt_key]
        assert fmt.target == expected_target

    def test_sieve_is_sieve(self, fmt_key, expected_target, expected_sieve_name):
        fmt = formats[fmt_key]
        assert isinstance(fmt.sieve, Sieve)

    def test_sieve_name(self, fmt_key, expected_target, expected_sieve_name):
        fmt = formats[fmt_key]
        assert fmt.sieve.name == expected_sieve_name

    def test_column_map_is_nonempty_dict(
        self, fmt_key, expected_target, expected_sieve_name
    ):
        fmt = formats[fmt_key]
        assert isinstance(fmt.column_map, dict)
        assert len(fmt.column_map) > 0


# ── Sieve regex patterns ────────────────────────────────────────


class TestSievePatterns:
    """Verify that each sieve has compiled regex patterns for data/head/skip."""

    @pytest.mark.parametrize("fmt_key", list(formats.keys()))
    def test_data_pattern(self, fmt_key):
        sieve = formats[fmt_key].sieve
        assert sieve.data is not None
        assert isinstance(sieve.data, re.Pattern)

    @pytest.mark.parametrize("fmt_key", list(formats.keys()))
    def test_head_pattern(self, fmt_key):
        sieve = formats[fmt_key].sieve
        assert sieve.head is not None
        assert isinstance(sieve.head, re.Pattern)

    @pytest.mark.parametrize("fmt_key", list(formats.keys()))
    def test_skip_pattern(self, fmt_key):
        sieve = formats[fmt_key].sieve
        assert sieve.skip is not None
        assert isinstance(sieve.skip, re.Pattern)


# ── Column map spot checks ──────────────────────────────────────


class TestColumnMaps:
    """Spot-check that key raw CMS headers map to expected field names."""

    def test_rvu_columns_hcpcs(self):
        assert formats["pfs_rvu"].column_map["HCPCS"] == "hcpcs"

    def test_rvu_columns_work_rvu(self):
        assert formats["pfs_rvu"].column_map["WORK RVU"] == "work_rvu"

    def test_rvu_columns_non_fac_pe_variant(self):
        cm = formats["pfs_rvu"].column_map
        assert cm["NON-FACILITY PE RVU"] == "non_fac_pe_rvu"
        assert cm["NON-FAC PE RVU"] == "non_fac_pe_rvu"

    def test_gpci_columns_carrier_alias(self):
        cm = formats["pfs_gpci"].column_map
        assert cm["MAC"] == "mac"
        assert cm["CARRIER"] == "mac"

    def test_labor_columns_nf_minutes_variant(self):
        cm = formats["pfs_labor"].column_map
        assert cm["NF MINUTES"] == "nf_minutes"
        assert cm["NON-FACILITY MINUTES"] == "nf_minutes"

    def test_supply_columns_unit_price_variant(self):
        cm = formats["pfs_supply"].column_map
        assert cm["UNIT PRICE"] == "unit_price"
        assert cm["PRICE"] == "unit_price"

    def test_equipment_columns_useful_life(self):
        assert formats["pfs_equipment"].column_map["USEFUL LIFE"] == "useful_life"

    def test_carrier_columns_fac_fee(self):
        cm = formats["pfs_carrier"].column_map
        assert cm["FACILITY FEE SCHEDULE AMOUNT"] == "fac_fee"
        assert cm["FACILITY LIMITING CHARGE"] == "fac_limiting_charge"