tests/core/test_spec.py

from __future__ import annotations
import yaml


def test_assemble_spec_yaml_creates_valid_yaml(tmp_path):
    from core.spec import assemble_spec_yaml

    route = {"sectors": ["WASH", "Food Security"], "lens": [], "type": []}
    bind_results = {
        "rcsi": {
            "indicator_id": "rcsi",
            "variables": ["cope_q1"],
            "measurable": "PROXY",
            "reasons": "Proves: ordinal coping. Cannot prove: full rCSI score.",
            "result_ids": ["rcsi_proxy"],
        }
    }
    indicator_defs = {
        "rcsi": {
            "label": "Reduced Coping Strategies Index",
            "definition": "Measures food insecurity via coping behaviour frequency.",
            "cluster": "Food Security",
        }
    }

    yaml_path = assemble_spec_yaml(tmp_path, "vasyr_2023", route, bind_results, indicator_defs)

    assert yaml_path.exists()
    assert yaml_path.name == "analysis_spec_vasyr_2023.yaml"

    data = yaml.safe_load(yaml_path.read_text())
    assert data["dataset"] == "vasyr_2023"
    assert data["unit_of_analysis"] == "key_informant"
    assert "rcsi" in data["indicators"]
    ind = data["indicators"]["rcsi"]
    assert ind["measurable"] == "PROXY"
    assert ind["definition"] == "Measures food insecurity via coping behaviour frequency."
    assert ind["variables"] == ["cope_q1"]
    assert ind["reasons"] == "Proves: ordinal coping. Cannot prove: full rCSI score."
    assert ind["label"] == "Reduced Coping Strategies Index"


def test_assemble_spec_yaml_result_ids_is_catalog_slug(tmp_path):
    """result_ids must be the catalog slug (the dict key), never the bind call's value.

    The bind call leaks dataset variable names into result_ids; code overrides it so the
    'Indicator name in the analysis' column always shows the stable catalog slug.
    """
    from core.spec import assemble_spec_yaml

    route = {"sectors": ["WASH"], "lens": [], "type": []}
    bind_results = {
        "jmp_water_safely_managed": {
            "indicator_id": "jmp_water_safely_managed",
            "variables": ["main_water_source_now"],
            "measurable": "NOT_MEASURABLE",
            "reasons": "Requires on-premises + availability + E. coli — none collected.",
            # Model garbage: a dataset variable name leaked into result_ids.
            "result_ids": ["main_water_source_now"],
        }
    }
    indicator_defs = {
        "jmp_water_safely_managed": {
            "label": "JMP Drinking Water — Safely Managed",
            "definition": "Improved source on premises, available, free of contamination.",
            "cluster": "wash",
        }
    }

    yaml_path = assemble_spec_yaml(tmp_path, "vasyr_2023", route, bind_results, indicator_defs)
    data = yaml.safe_load(yaml_path.read_text())
    ind = data["indicators"]["jmp_water_safely_managed"]
    assert ind["result_ids"] == ["jmp_water_safely_managed"]


def test_assemble_spec_yaml_dimension_uses_cluster(tmp_path):
    from core.spec import assemble_spec_yaml

    route = {"sectors": ["WASH"], "lens": [], "type": []}
    bind_results = {
        "jmp_water_basic": {
            "indicator_id": "jmp_water_basic",
            "variables": ["water_src"],
            "measurable": "MEASURABLE",
            "reasons": "Proves: JMP basic access. Cannot prove: safely managed.",
            "result_ids": ["jmp_basic"],
        }
    }
    indicator_defs = {
        "jmp_water_basic": {
            "label": "JMP Drinking Water — Basic",
            "definition": "Access to an improved source within 30 minutes.",
            "cluster": "WASH",
        }
    }

    yaml_path = assemble_spec_yaml(tmp_path, "test_slug", route, bind_results, indicator_defs)
    data = yaml.safe_load(yaml_path.read_text())

    assert data["indicators"]["jmp_water_basic"]["dimension"] == "sector::WASH"
    assert data["step1_framework_route"]["sector"] == "WASH"


def test_render_spec_md_calls_render_script(monkeypatch, tmp_path):
    from core import spec as spec_mod

    calls = []
    monkeypatch.setattr(spec_mod, "run_skill_script_raw",
                        lambda name, args: (calls.append((name, args)) or "# rendered md"))

    spec_yaml = tmp_path / "analysis_spec_test.yaml"
    spec_yaml.write_text("dataset: test\nindicators: {}")

    md_path = spec_mod.render_spec_md(spec_yaml, tmp_path)

    assert calls[0][0] == "render_spec.py"
    assert str(spec_yaml) in calls[0][1]
    assert md_path.exists()
    assert md_path.read_text() == "# rendered md"
    assert md_path.name == "analysis_spec_test.md"