Add .tests.report.legacy.test_pp_utils

Copied from iiasa/message_data branch `dev`.

Author: khaeru

message_ix_models/tests/report/legacy/test_pp_utils.py

@@ -0,0 +1,109 @@
+from itertools import product
+
+import pandas as pd
+import pytest
+from message_ix_models.model.structure import get_codes
+from message_ix_models.util import nodes_ex_world
+
+from message_data.tools.post_processing import pp_utils
+from message_data.tools.post_processing.pp_utils import fil
+
+#: These format strings appear in .default_tables.retr_hfc() in calls to :func:`.fil`.
+FACTOR_EXPR = "refAC foam Fire ANMDI mVac Solv".split()
+
+#: These values appear in default_run_config.yaml. The commented lines are those for
+#: which :func:`.fil` is *not* called.
+FACTOR_VAL_UNIT = (
+    # (0, "kt HFC134a-equiv/yr"),  # Total  # Not called because hfc: "Total"
+    (125, "kt HFC125/yr"),  # HFC125
+    (134, "kt HFC134a/yr"),  # HFC134a
+    (143, "kt HFC143a/yr"),  # HFC143a
+    (227, "kt HFC227ea/yr"),  # HFC227ea
+    # (0, "kt HFC23/yr"),  # HFC23  # Not called because run: "empty"
+    (245, "kt HFC245fa/yr"),  # HFC245fa
+    (32, "kt HFC32/yr"),  # HFC32
+    (431, "kt HFC43-10/yr"),  # HFC430
+    (365, "kt HFC365mfc/yr"),  # HFC365mfc  # Not called because run: False
+    (152, "kt HFC152a/yr"),  # HFC152a
+    (236, "kt HFC236fa/yr"),  # HFC236fa  # Not called because run: False
+)
+
+
+#: Input data expected by :func:`.fil`.
+#:
+#: - Axis named "Region" with short labels, e.g. "ABC" rather than "R##_ABC"
+#: - Some columns with period labels, as integers.
+#: - Float or NaN values.
+FIL_DF = pd.DataFrame(
+    [[1.0, None, 1.0]], columns=[2000, 2050, 2100], index=["AFR"]
+).rename_axis("Region")
+
+
+def pp_utils_globals(monkeypatch, regions):
+    """Set :mod:`.pp_utils` global module variables.
+
+    This utility function uses pytest's `monkeypatch` fixture to set the variables; the
+    pre-existing values are restored after the end of each test.
+
+    This only sets the globals that are used by :func:`.fil()`.
+    """
+    # region_id: the ID of the node code list; i.e. Context.regions
+    monkeypatch.setattr(pp_utils, "region_id", regions)
+
+    # regions: a mapping from model's native "R##_ABC" to "ABC", *plus* e.g. "R##_GLB"
+    # mapped to # "World". This normally set by .iamc_report_hackathon.report().
+    nodes = get_codes(f"node/{regions}")
+    nodes = nodes[nodes.index("World")].child
+    regions_map = {n.id: n.id.split("_")[1] for n in nodes_ex_world(nodes)}
+    regions_map[f"{regions}_GLB"] = "World"
+    monkeypatch.setattr(pp_utils, "regions", regions_map)
+
+    # unit_conversion: a mapping of mappings. fil() directly sets "???" on its data and
+    # then looks up mappings from "???" to `units_out`
+    monkeypatch.setattr(
+        pp_utils,
+        "unit_conversion",
+        {"???": {units: 1.0 for _, units in FACTOR_VAL_UNIT}},
+    )
+
+    # Expected index of data frame returned by fil()
+    return set(regions_map.values())
+
+
+@pytest.mark.parametrize("regions", ["R11", "R12"])
+@pytest.mark.parametrize(
+    "fil_arg, min_filled_values",
+    [
+        ("HFC_fac", 1),  # The only supported value
+        ("foo", 0),  # Other values are essentially a no-op
+    ],
+)
+def test_fil(monkeypatch, regions, fil_arg, min_filled_values):
+    """Test :func:`.pp_utils.fil`.
+
+    The test is parametrized for `regions` and each argument to :func:`.fil`.
+
+    `min_filled_values` checks that at least N values in the resulting data frame are
+    populated from the file(s).
+    """
+    # Monkey-patch the global variables in the pp_utils module used by fil() and
+    # retrieve the expected index of the result
+    expected_regions = pp_utils_globals(monkeypatch, regions)
+
+    # Iterate over all values for the `factor` and `units` arguments that fil() is
+    # expected to support
+    for factor_expr, (factor_val, units) in product(FACTOR_EXPR, FACTOR_VAL_UNIT):
+        # Mirror how values for the `factor` argument are constructed in
+        # .default_tables.retr_hfc(), e.g. "Fire134"
+        factor = f"{factor_expr}{factor_val}"
+
+        # Function executes without error
+        result = fil(FIL_DF, fil_arg, factor, units)
+
+        # Returned data are a pd.DataFrame with the expected index
+        assert expected_regions == set(result.index)
+
+        # Number of NaN values in results is as expected
+        assert min_filled_values <= (
+            result.notnull().sum().sum() - FIL_DF.notnull().sum().sum()
+        )