start on extended metadata (ACDD) validators

Closes #40

cchdo/hydro/metadata.py

@@ -0,0 +1,48 @@
+from collections import defaultdict
+from itertools import groupby, islice
+from logging import getLogger
+
+import xarray as xr
+
+log = getLogger(__name__)
+
+
+def all_equal(iterable, key=None):
+    "Returns True if all the elements are equal to each other."
+    # see https://docs.python.org/3/library/itertools.html#itertools-recipes
+    return len(list(islice(groupby(iterable, key), 2))) <= 1
+
+
+SAME_KEYS = (
+    "processing_level",
+    "comment",
+    "creator_name",
+)
+ALLOWED_DIFFER = (
+    "date_modified",
+    "date_metadata_modified",
+)
+
+
+def validate(ds: xr.Dataset):
+    exceptions = []
+
+    # cannot use filter_by_attrs since we want to filter on the coordinates too
+    projects = defaultdict(list)
+    for name, da in ds.variables.items():
+        if (project := da.attrs.get("project")) is not None:
+            projects[project].append(name)
+
+    for project, vars in projects.items():
+        log.debug(f"Checking project '{project}' which includes {vars}")
+        for key in SAME_KEYS:
+            values = {var: ds[var].attrs.get(key) for var in vars}
+            valid = all_equal(values.values())
+            if not valid:
+                exception = ValueError(
+                    f"Project '{project}' key '{key}' is not the same: {values}"
+                )
+                log.debug(exception)
+                exceptions.append(exception)
+    if exceptions:
+        raise ExceptionGroup("Metadata has failed to validate", exceptions)

cchdo/hydro/tests/test_metadata.py

@@ -0,0 +1,24 @@
+from io import BytesIO
+
+import pytest
+
+from cchdo.hydro import read_csv
+from cchdo.hydro.metadata import validate
+
+
+def test_invalid_createor_name():
+    test_data = BytesIO(
+        b"""EXPOCODE,STNNBR,CASTNO,SAMPNO,LATITUDE,LONGITUDE,DATE,TIME,CTDPRS [DBAR],SILCAT [UMOL/KG],SILCAT [UMOL/KG]_FLAG_W,PHSPHT [UMOL/KG],PHSPHT [UMOL/KG]_FLAG_W
+TEST,1,1,1,0,0,20220101,0000,0,0,2,0,3"""
+    )
+    ds = read_csv(test_data)
+
+    ds.silicate.attrs["project"] = "nutrients"
+    ds.phosphate.attrs["project"] = "nutrients"
+
+    ds.silicate.attrs["creator_name"] = "Susan Becker"
+
+    with pytest.raises(ExceptionGroup) as excinfo:
+        validate(ds)
+
+        assert excinfo.group_contains(ValueError)