Merge pull request #17 from mikapfl/multiple-sources

More than one source in a dataset

CHANGELOG.rst

@@ -6,6 +6,7 @@ unreleased
 ----------
 * Add the ``da.pr.set()`` and ``ds.pr.set()`` functions for overwriting / infilling /
   extending data.
+* Allow for more than one source in a single Dataset / DataArray.
 
 0.3.1
 -----

docs/api.rst

@@ -67,7 +67,6 @@ Attributes
     Dataset.pr.comment
     Dataset.pr.institution
     Dataset.pr.history
-    Dataset.pr.source
 
 .. _dsmeth:
 

docs/data_format_details.rst

@@ -50,7 +50,6 @@ category-set (for example, ``animal (FAOSTAT)``).
 
 Additional rules:
 
-* The size of the ``source`` dimension must be exactly 1.
 * The valid values for the ``provenance`` are ``measured``, ``projected``, and
   ``derived``.
 

docs/data_format_examples.ipynb

@@ -201,7 +201,7 @@
     "    \"scenario (FAOSTAT)\": np.array([\"highpop\", \"lowpop\"]),\n",
     "    \"provenance\": np.array([\"projected\"]),\n",
     "    \"model\": np.array([\"FANCYFAO\"]),\n",
-    "    \"source\": np.array([\"RAND2020\"]),\n",
+    "    \"source\": np.array([\"RAND2020\", \"RAND2021\"]),\n",
     "}\n",
     "\n",
     "opulent = xr.Dataset(\n",

primap2/_data_format.py

@@ -385,10 +385,6 @@ def ensure_valid_dimensions(ds: xr.Dataset):
             f" sec_cats."
         )
 
-    if ds.dims["source"] != 1:
-        logger.error("Exactly one source required per data set.")
-        raise ValueError("Exactly one source required")
-
     for dim in required_indirect_dims.union(optional_indirect_dims):
         if dim in ds.attrs:
             split_dim_name(ds.attrs[dim])

primap2/_metadata.py

@@ -72,12 +72,3 @@ def history(self) -> str:
     @history.setter
     def history(self, value: str):
         self._ds.attrs["history"] = value
-
-    @property
-    def source(self) -> str:
-        """a short identifier for the source of the data"""
-        return self._ds["source"].item()
-
-    @source.setter
-    def source(self, value: str):
-        self._ds["source"] = [value]

primap2/tests/examples.py

@@ -48,7 +48,7 @@ def opulent_ds():
         "scenario (FAOSTAT)": np.array(["highpop", "lowpop"]),
         "provenance": np.array(["projected"]),
         "model": np.array(["FANCYFAO"]),
-        "source": np.array(["RAND2020"]),
+        "source": np.array(["RAND2020", "RAND2021"]),
     }
 
     # seed the rng with a constant to achieve predictable "randomness"

primap2/tests/test_data_format.py

@@ -20,10 +20,9 @@ def test_something_else_entirely(caplog):
     assert "object is not an xarray Dataset." in caplog.text
 
 
-def test_valid_ds_pass(minimal_ds, opulent_ds, caplog):
+def test_valid_ds_pass(any_ds, caplog):
     caplog.set_level(logging.INFO)
-    minimal_ds.pr.ensure_valid()
-    opulent_ds.pr.ensure_valid()
+    any_ds.pr.ensure_valid()
     assert not caplog.records
 
 
@@ -109,22 +108,6 @@ def test_wrong_dimension_key(minimal_ds, caplog):
     assert "'asdf' not in the format 'dim (category_set)'." in caplog.text
 
 
-def test_not_one_source(minimal_ds: xr.Dataset, caplog):
-    ds = minimal_ds.loc[{"source": "RAND2020"}]
-    ds["source"] = []
-    with pytest.raises(ValueError, match=r"Exactly one source required"):
-        ds.pr.ensure_valid()
-    assert "ERROR" in caplog.text
-    assert "Exactly one source required per data set." in caplog.text
-
-    caplog.clear()
-    ds["source"] = ["a", "b"]
-    with pytest.raises(ValueError, match=r"Exactly one source required"):
-        ds.pr.ensure_valid()
-    assert "ERROR" in caplog.text
-    assert "Exactly one source required per data set." in caplog.text
-
-
 def test_missing_sec_cat(minimal_ds, caplog):
     minimal_ds.attrs["sec_cats"] = ["missing"]
     with pytest.raises(ValueError, match="Secondary category 'missing' not in dims"):

primap2/tests/test_metadata.py

@@ -12,7 +12,6 @@ def test_metadata_properties(opulent_ds):
     assert ds.pr.history == (
         "2021-01-14 14:50 data invented\n" "2021-01-14 14:51 additional processing step"
     )
-    assert ds.pr.source == "RAND2020"
 
     ds.pr.references = "references"
     assert ds.pr.references == "references"
@@ -28,5 +27,3 @@ def test_metadata_properties(opulent_ds):
     assert ds.pr.institution == "institution"
     ds.pr.history = "history"
     assert ds.pr.history == "history"
-    ds.pr.source = "source"
-    assert ds.pr.source == "source"