Geo test pre refactor

NextGen_Forcings_Engine_BMI/NextGen_Forcings_Engine/core/forcingInputMod.py

@@ -1101,33 +1101,35 @@ def init_dict(
         # temporary temperature arrays that are un-downscaled.
         if input_dict[force_key].q2dDownscaleOpt > 0:
             if config_options.grid_type == "gridded":
-                input_dict[force_key].t2dTmp = np.empty(
+                input_dict[force_key].t2dTmp = np.full(
                     [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
-                    np.float32,
+                    np.nan,
+                    dtype=np.float32,
                 )
-                input_dict[force_key].psfcTmp = np.empty(
+                input_dict[force_key].psfcTmp = np.full(
                     [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
-                    np.float32,
+                    np.nan,
+                    dtype=np.float32,
                 )
             elif config_options.grid_type == "unstructured":
-                input_dict[force_key].t2dTmp = np.empty(
-                    [geo_meta_wrf_hydro.ny_local], np.float32
+                input_dict[force_key].t2dTmp = np.full(
+                    [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
                 )
-                input_dict[force_key].psfcTmp = np.empty(
-                    [geo_meta_wrf_hydro.ny_local], np.float32
+                input_dict[force_key].psfcTmp = np.full(
+                    [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
                 )
-                input_dict[force_key].t2dTmp_elem = np.empty(
-                    [geo_meta_wrf_hydro.ny_local_elem], np.float32
+                input_dict[force_key].t2dTmp_elem = np.full(
+                    [geo_meta_wrf_hydro.ny_local_elem], np.nan, dtype=np.float32
                 )
-                input_dict[force_key].psfcTmp_elem = np.empty(
-                    [geo_meta_wrf_hydro.ny_local_elem], np.float32
+                input_dict[force_key].psfcTmp_elem = np.full(
+                    [geo_meta_wrf_hydro.ny_local_elem], np.nan, dtype=np.float32
                 )
             elif config_options.grid_type == "hydrofabric":
-                input_dict[force_key].t2dTmp = np.empty(
-                    [geo_meta_wrf_hydro.ny_local], np.float32
+                input_dict[force_key].t2dTmp = np.full(
+                    [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
                 )
-                input_dict[force_key].psfcTmp = np.empty(
-                    [geo_meta_wrf_hydro.ny_local], np.float32
+                input_dict[force_key].psfcTmp = np.full(
+                    [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
                 )
         # Initialize the local final grid of values. This is represntative
         # of the local grid for this forcing, for a specific output timesetp.
@@ -1139,56 +1141,65 @@ def init_dict(
             input_dict[force_key].grib_vars = input_dict[force_key].grib_vars[:-1]
 
         if config_options.grid_type == "gridded":
-            input_dict[force_key].final_forcings = np.empty(
+            input_dict[force_key].final_forcings = np.full(
                 [force_count, geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
-                np.float64,
+                np.nan,
+                dtype=np.float64,
             )
-            input_dict[force_key].height = np.empty(
-                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local], np.float32
+            input_dict[force_key].height = np.full(
+                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
+                np.nan,
+                dtype=np.float32,
             )
-            input_dict[force_key].regridded_mask = np.empty(
-                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local], np.float32
+            input_dict[force_key].regridded_mask = np.full(
+                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
+                np.nan,
+                dtype=np.float32,
             )
-            input_dict[force_key].regridded_mask_AORC = np.empty(
-                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local], np.float32
+            input_dict[force_key].regridded_mask_AORC = np.full(
+                [geo_meta_wrf_hydro.ny_local, geo_meta_wrf_hydro.nx_local],
+                np.nan,
+                dtype=np.float32,
             )
         elif config_options.grid_type == "unstructured":
-            input_dict[force_key].final_forcings = np.empty(
-                [force_count, geo_meta_wrf_hydro.ny_local], np.float64
+            input_dict[force_key].final_forcings = np.full(
+                [force_count, geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float64
             )
-            input_dict[force_key].height = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].height = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].regridded_mask = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask_AORC = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].regridded_mask_AORC = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
-            input_dict[force_key].final_forcings_elem = np.empty(
-                [force_count, geo_meta_wrf_hydro.ny_local_elem], np.float64
+            input_dict[force_key].final_forcings_elem = np.full(
+                [force_count, geo_meta_wrf_hydro.ny_local_elem],
+                np.nan,
+                dtype=np.float64,
             )
-            input_dict[force_key].height_elem = np.empty(
-                [geo_meta_wrf_hydro.ny_local_elem], np.float32
+            input_dict[force_key].height_elem = np.full(
+                [geo_meta_wrf_hydro.ny_local_elem], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask_elem = np.empty(
-                [geo_meta_wrf_hydro.ny_local_elem], np.float32
+            input_dict[force_key].regridded_mask_elem = np.full(
+                [geo_meta_wrf_hydro.ny_local_elem], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask_elem_AORC = np.empty(
-                [geo_meta_wrf_hydro.ny_local_elem], np.float32
+            input_dict[force_key].regridded_mask_elem_AORC = np.full(
+                [geo_meta_wrf_hydro.ny_local_elem], np.nan, dtype=np.float32
             )
         elif config_options.grid_type == "hydrofabric":
-            input_dict[force_key].final_forcings = np.empty(
-                [force_count, geo_meta_wrf_hydro.ny_local], np.float64
+            input_dict[force_key].final_forcings = np.full(
+                [force_count, geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float64
             )
-            input_dict[force_key].height = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].height = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].regridded_mask = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
-            input_dict[force_key].regridded_mask_AORC = np.empty(
-                [geo_meta_wrf_hydro.ny_local], np.float32
+            input_dict[force_key].regridded_mask_AORC = np.full(
+                [geo_meta_wrf_hydro.ny_local], np.nan, dtype=np.float32
             )
         # Obtain custom input cycle frequencies
         if force_key == 10 or force_key == 11:

NextGen_Forcings_Engine_BMI/NextGen_Forcings_Engine/general_utils.py

@@ -3,6 +3,7 @@
 import json
 import logging
 import typing
+from collections import OrderedDict
 
 import numpy as np
 
@@ -73,6 +74,7 @@ def assert_equal_with_tol(
     keys_to_check: tuple | None = None,
     absolute_tolerance: float = 1e-6,
     relative_tolerance: float = 1e-10,
+    new_keys_in_actual_ok: bool = False,
 ):
     """Assert that the key,value pairs in `expect` have matching key,value pairs in `actual`, with numerical tolerance.
     It is okay if actual has extra keys that are not present in expect.
@@ -93,21 +95,26 @@ def assert_equal_with_tol(
         errors.append(KeyError(f"Missing keys from actual: {keys_missing_from_actual}"))
 
     keys_missing_from_expected = set(keys_to_check) - set(expect)
-    if keys_missing_from_expected:
-        errors.append(
-            KeyError(f"Missing keys from expected: {keys_missing_from_expected}")
-        )
+    if not new_keys_in_actual_ok:
+        if keys_missing_from_expected:
+            errors.append(
+                KeyError(f"Missing keys from expected: {keys_missing_from_expected}")
+            )
 
     for k in keys_to_check:
         ### Check key existence
         try:
             v_expect = expect[k]
+            if isinstance(v_expect, dict):
+                v_expect = OrderedDict(sorted(list(v_expect.items())))
         except KeyError:
             errors.append(KeyError(f"Key {k} is missing from expected"))
             continue
 
         try:
             v_actual = actual[k]
+            if isinstance(v_actual, dict):
+                v_actual = OrderedDict(sorted(list(v_actual.items())))
         except KeyError:
             msg = f"Key {k} is missing from actual"
             errors.append(KeyError(msg))
@@ -155,14 +162,32 @@ def assert_equal_with_tol(
         except np.exceptions.DTypePromotionError:
             errors.append(
                 ValueError(
-                    f"Expected not equal to actual, and could not apply np.allclose. expect={expect}, actual={actual}."
+                    f"Expected not equal to actual, and could not apply np.allclose. key={k}, expect={expect}, actual={actual}."
                 )
             )
             continue
-
+        except TypeError:
+            if isinstance(v_expect, (dict, OrderedDict)) and isinstance(
+                v_actual, (dict, OrderedDict)
+            ):
+                keys_not_in_v_actual = set(dict(v_expect)) - set(dict(v_actual))
+                keys_in_v_expect_and_v_actual = set(dict(v_expect)) & set(
+                    dict(v_actual)
+                )
+                keys_with_vals_not_matching = [
+                    key
+                    for key in keys_in_v_expect_and_v_actual
+                    if v_expect[key] != v_actual[key]
+                ]
+                errors.append(
+                    ValueError(
+                        f"Expected not equal to actual for key: {k}. Keys not in actual for {k}: {keys_not_in_v_actual} | keys in actual with values not matching expected for {k}: {keys_with_vals_not_matching}"
+                    )
+                )
+                continue
         errors.append(
             ValueError(
-                f"Objects not equal, and numerical tolerances (atol={absolute_tolerance} rtol={relative_tolerance}) exceeded for at least one element. {v_expect} vs {v_actual}."
+                f"Objects not equal, and numerical tolerances (atol={absolute_tolerance} rtol={relative_tolerance}) exceeded for {k}. {v_expect} vs {v_actual}."
             )
         )
 

pytest.ini

@@ -6,3 +6,5 @@ log_cli_level = INFO
 addopts = --full-trace -vv
 testpaths =
     tests/esmf_regrid
+    tests/geomod
+    tests/input_forcing

tests/README.md

@@ -0,0 +1,155 @@
+# Tests README
+
+This directory contains tests for the NextGen Forcing BMI Engine.
+
+## Initial test data
+
+Tests data is included in the `test_data` directory and includes configs, gpkgs, esmf_meshes, expected results and actual results. While the configs, gpkgs, esmf_meshes and expectd results are included in the repo and can be used as is, the following steps can be taken to re-create these test inputs.
+
+---   
+The initial test data was generated using `nwm-rte` to create a calibration realization
+for gage 01123000, starting at time 2013-07-01 00:00:00, and running for 3 timesteps,
+using `nwm-rte's` run_suite.sh.  See RETRO_FORCING_CONFIG_FILE__AORC_CONUS.
+
+More specifically the initial expected test data was developed with these specific configurations in `config.bashrc`. 
+```
+REPO_TAG_FCST_MGR="856fc0e1201076df909e56c7cd384f58e82965a2"
+REPO_TAG_MSW_MGR="693c206a22b5e9ffcca3103166c0ca59e2b11b25"
+REPO_TAG_CAL_MGR="7e56bf01477ea77e72dfb25a166ac26ff6090ecb"
+REPO_TAG_NGEN_FORCING="LOCAL"
+NGEN_SOURCE_MODE="ghcr"
+NGEN_BASE__REMOTE_GHCR_TAG="844c5f6"
+```
+
+And these two commands in `nwm-rte's` `run_suite.sh`:
+```bash
+docker_run python "/ngen-app/bin/bin_mounted/run_calibration.py" -n 2 -fsrc "aorc" -start "2013-07-01 00:00:00" -dur 3
+
+docker_run python "/ngen-app/bin/bin_mounted/run_forecast.py" -fconfig "short_range" -dt "2025-07-10 04:00:00" -rname "fcst_run1_short_range"
+```
+## Test Structure
+
+The test suite is organized into the following modules:
+
+- **`esmf_regrid/`** - Tests for ESMF regridding functionality
+- **`geomod/`** - Tests for geomod components
+- **`input_forcing/`** - Tests for input forcing data processing
+- **`nextgen_forcings_ewts/`** - Tests for EWTS (Error, Warning, and Trapping System) forcings
+- **`test_utils.py`** - Shared test utilities and fixtures
+- **`conftest.py`** - Pytest configuration and shared fixtures
+
+## Prerequisites
+### Setup requirements:
+    1. Create the forcing config.yml files using RTE.
+    2. Enter the RTE devcontainer.
+
+### Required Dependencies
+
+The test suite requires Python 3.11 or higher. Install the package with test dependencies inside of the `dev container`:
+
+```bash
+# From the repository root directory
+pip install -e ".[develop]"
+```
+
+Or install pytest directly inside of the `dev container`:
+
+```bash
+pip install pytest
+```
+
+### Additional Requirements
+
+Ensure all main package dependencies are installed inside of the `dev container` (this typically should happen when the `dev container` is built):
+
+```bash
+pip install -e .
+```
+
+## Running Tests
+
+### Run All Tests From the Dev Container
+
+```bash
+Single processor: (cd src/ngen-forcing && pytest )
+Multiple processors: ( cd src/ngen-forcing && mpirun -n 2 pytest )
+```
+### Run Specific Test Modules From the Dev Container
+
+Run tests for a specific module:
+
+```bash
+# ESMF regridding tests
+Single processor: ( cd src/ngen-forcing && pytest tests/esmf_regrid)
+Multiple processors: ( cd src/ngen-forcing && mpirun -n 2 pytest tests/esmf_regrid)
+
+# GeoMod tests
+Single processor: ( cd src/ngen-forcing && pytest tests/geomod)
+Multiple processors: ( cd src/ngen-forcing && mpirun -n 2 pytest tests/geomod)
+
+# Input forcing tests
+Single processor: ( cd src/ngen-forcing && pytest tests/input_forcing)
+Multiple processors: ( cd src/ngen-forcing && mpirun -n 2 pytest tests/input_forcing)
+```
+
+Create new test output data (creates expected outputs for subsequent tests)
+```bash
+# ESMF regridding tests
+Single processor: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true pytest tests/esmf_regrid)
+Multiple processors: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true mpirun -n 2 pytest tests/esmf_regrid)
+
+# GeoMod tests
+Single processor: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true pytest tests/geomod)
+Multiple processors: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true mpirun -n 2 pytest tests/geomod)
+
+# Input forcing tests
+Single processor: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true pytest tests/input_forcing)
+Multiple processors: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true mpirun -n 2 pytest tests/input_forcing)
+```
+
+In the rare case where you want to create new `expected` data and run the tests using `old` variable names use the following for `Input Forcing Tests`: 
+```bash
+# Input forcing tests
+Single processor: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true pytest tests/input_forcing --map_old_to_new_var_names False)
+Multiple processors: ( cd src/ngen-forcing && FORCING_PYTEST_WRITE_TEST_EXPECTED_DATA=true mpirun -n 2 pytest tests/input_forcing --map_old_to_new_var_names False)
+```
+## Test Configuration
+
+The test suite is configured via `pytest.ini` at the repository root:
+
+- **Python path**: Set to repository root (`.`)
+- **Logging**: Enabled with INFO level (DEBUG available by uncommenting)
+- **Verbosity**: Full trace with verbose output (`-vv`)
+- **Test paths**: Pre-configured to discover tests in `esmf_regrid`, `geomod`, and `input_forcing`
+
+
+## Test Data
+
+Test data is stored in the `test_data/` directory. Tests may reference files from this location for input data and expected results validation.
+
+## Writing New Tests
+
+When adding new tests:
+
+1. Place test files in the appropriate subdirectory
+2. Name test files with the `test_*.py` prefix
+3. Name test functions with the `test_*` prefix
+4. Use fixtures from `conftest.py` for common setup
+5. Place test data files in `test_data/` with descriptive names
+
+Example test structure:
+
+```python
+import pytest
+
+def test_my_feature():
+    """Test description."""
+    # Arrange
+    input_data = ...
+
+    # Act
+    result = function_under_test(input_data)
+
+    # Assert
+    assert result == expected_output
+```