Feature: add toy data

.gitignore

@@ -130,7 +130,9 @@ dmypy.json
 # Pyre type checker
 .pyre/
 
-/data/
+# ignore data except for .msk files
+data/*
+!data/*.msk
 tests/testdata/tmp
 /tmp/
 

docs/_quarto.yml

@@ -125,6 +125,8 @@ website:
           href: api/settings.qmd
         - text: "Quantity"
           href: api/Quantity.qmd
+        - text: "Data"
+          href: api/data.qmd
 
 
 filters:
@@ -277,6 +279,7 @@ quartodoc:
         - settings
         - TimeSeries
         - Quantity
+        - data
 
 #metadata-files:
 #  - api/_sidebar.yml

modelskill/__init__.py

@@ -38,6 +38,7 @@
 from .configuration import from_config
 from .settings import options, get_option, set_option, reset_option, load_style
 from . import plotting
+from . import data
 from .comparison import ComparerCollection, Comparer
 from .skill import SkillTable
 from .timeseries import TimeSeries
@@ -89,4 +90,5 @@ def load(filename: Union[str, Path]) -> ComparerCollection:
     "load_style",
     "plotting",
     "from_config",
+    "data",
 ]

modelskill/data/__init__.py

@@ -0,0 +1,53 @@
+"""
+Toy datasets for testing and demonstration purposes
+
+Examples
+--------
+```{python}
+>>> import modelskill as ms
+>>> cc = ms.data.vistula()
+>>> cc
+```
+```{python}
+>>> cc = ms.data.oresund()
+>>> cc
+```
+"""
+
+from importlib.resources import files
+
+import modelskill as ms
+from ..comparison import ComparerCollection
+
+
+def vistula() -> ComparerCollection:
+    """5-year daily discharge data for Vistula catchment, Poland
+
+    Contains discharge data for 8 stations along the Vistula river
+    compared with two hydrological models "sim1" and "sim2".
+
+    The dataset additionally contains precipitation data as aux data
+    and metadata about the river and the catchment area in the attrs dictionary.
+
+    Returns
+    -------
+    ComparerCollection
+    """
+    fn = str(files("modelskill.data") / "vistula.msk")
+    return ms.load(fn)
+
+
+def oresund() -> ComparerCollection:
+    """Oresund water level data for Jan-June 2022 compared with MIKE21 model
+
+    Contains water level data for 7 stations along the Oresund strait with
+    metadata about the country in the attrs dictionary.
+
+    The dataset contains additional ERA5 wind-components U10 and V10 aux data.
+
+    Returns
+    -------
+    ComparerCollection
+    """
+    fn = str(files("modelskill.data") / "oresund.msk")
+    return ms.load(fn)

pyproject.toml

@@ -2,8 +2,10 @@
 requires = ["hatchling"]
 build-backend = "hatchling.build"
 
-[tool.hatch.build.targets.sdist]
-exclude = ["notebooks", "tests"]
+[tool.hatch.build]
+include = [
+    "modelskill/**/*",
+]
 
 [tool.hatch.metadata]
 allow-direct-references = true

tests/test_data.py

@@ -0,0 +1,52 @@
+import os
+
+import pytest
+import modelskill as ms
+
+
+@pytest.fixture
+def change_test_directory(tmp_path):
+    original_directory = os.getcwd()
+    os.chdir(tmp_path)
+    yield tmp_path
+    os.chdir(original_directory)
+
+
+def test_load_vistula(change_test_directory):
+    cc = ms.data.vistula()
+    assert isinstance(cc, ms.ComparerCollection)
+    assert len(cc) == 8
+    assert cc.mod_names == ["sim1", "sim2"]
+    assert cc[0].name == "Tczew"
+    assert cc[-1].n_points == 1827
+    assert cc[-1].y == 52.94889
+
+    assert cc[0].quantity.name == "Discharge"
+    assert cc[0].quantity.unit == "m3/s"
+
+    assert cc.aux_names == ["Precipitation"]
+    assert float(cc[1].data.Precipitation[0]) == pytest.approx(1.18)
+    assert cc[0].attrs["River"] == "Vistula"
+    assert cc[0].attrs["Area"] == 193922.9
+
+    assert cc[1].raw_mod_data["sim2"].n_points == 1827
+    assert isinstance(cc[0].raw_mod_data["sim1"], ms.PointModelResult)
+
+
+def test_load_oresund(change_test_directory):
+    cc = ms.data.oresund()
+    assert isinstance(cc, ms.ComparerCollection)
+    assert len(cc) == 7
+    assert cc.mod_names == ["MIKE21"]
+    assert cc[-1].name == "Vedbaek"
+    assert cc[0].n_points == 8422
+    assert cc[0].x == pytest.approx(12.7117)
+
+    assert cc[0].quantity.name == "Surface Elevation"
+    assert cc[0].quantity.unit == "meter"
+
+    assert cc.aux_names == ["U10", "V10"]
+    assert cc[-1].attrs["Country"] == "DK"
+
+    assert cc[1].raw_mod_data["MIKE21"].n_points == 4344
+    assert isinstance(cc[0].raw_mod_data["MIKE21"], ms.PointModelResult)