Assign ww3 data to wave realm in access-om3 builder (#162)

* assign ww3 data to wave realm in access-om3 builder

* add specific tests for builder parsers

Author: dougiesquire

src/access_nri_intake/source/builders.py

@@ -221,12 +221,8 @@ def __init__(self, path):
     @staticmethod
     def parser(file):
         try:
-            match_groups = re.match(
-                r".*/([^/]*)/([^/]*)/output\d+/([^/]*)/.*\.nc", file
-            ).groups()
-            # configuration = match_groups[0]
-            # exp_id = match_groups[1]
-            realm = match_groups[2]
+            match_groups = re.match(r".*/output\d+/([^/]*)/.*\.nc", file).groups()
+            realm = match_groups[0]
 
             if realm == "ice":
                 realm = "seaIce"
@@ -323,8 +319,10 @@ def parser(file):
                 variable_units_list,
             ) = parse_access_ncfile(file)
 
-            if ("mom6" in filename) or ("ww3" in filename):
+            if "mom6" in filename:
                 realm = "ocean"
+            elif "ww3" in filename:
+                realm = "wave"
             elif "cice" in filename:
                 realm = "seaIce"
             else:

tests/test_builders.py

@@ -62,6 +62,83 @@ def test_builder_build(
     assert len(cat) == num_datasets
 
 
+@pytest.mark.parametrize(
+    "filename, builder, realm, member, file_id",
+    [
+        (
+            "access-om2/output000/ocean/ocean.nc",
+            "AccessOm2Builder",
+            "ocean",
+            None,
+            "ocean",
+        ),
+        (
+            "access-om2/output000/ice/OUTPUT/iceh.1900-01.nc",
+            "AccessOm2Builder",
+            "seaIce",
+            None,
+            "iceh_XXXX_XX",
+        ),
+        (
+            "access-cm2/by578/history/atm/netCDF/by578a.pd201501_dai.nc",
+            "AccessCm2Builder",
+            "atmos",
+            "by578",
+            "a_pdXXXXXX_dai",
+        ),
+        (
+            "access-cm2/by578a/history/atm/netCDF/by578aa.pd201501_dai.nc",
+            "AccessCm2Builder",
+            "atmos",
+            "by578a",
+            "a_pdXXXXXX_dai",
+        ),
+        (
+            "access-cm2/by578/history/ice/iceh_d.2015-01.nc",
+            "AccessCm2Builder",
+            "seaIce",
+            "by578",
+            "iceh_d_XXXX_XX",
+        ),
+        (
+            "access-cm2/by578/history/ocn/ocean_daily.nc-20150630",
+            "AccessCm2Builder",
+            "ocean",
+            "by578",
+            "ocean_daily",
+        ),
+        (
+            "access-om3/output000/GMOM_JRA_WD.mom6.h.sfc_1900_01_02.nc",
+            "AccessOm3Builder",
+            "ocean",
+            None,
+            "GMOM_JRA_WD_mom6_h_sfc_XXXX_XX_XX",
+        ),
+        (
+            "access-om3/output000/GMOM_JRA_WD.cice.h.1900-01-01.nc",
+            "AccessOm3Builder",
+            "seaIce",
+            None,
+            "GMOM_JRA_WD_cice_h_XXXX_XX_XX",
+        ),
+        (
+            "access-om3/output000/GMOM_JRA_WD.ww3.hi.1900-01-02-00000.nc",
+            "AccessOm3Builder",
+            "wave",
+            None,
+            "GMOM_JRA_WD_ww3_hi_XXXX_XX_XX_XXXXX",
+        ),
+    ],
+)
+def test_builder_parser(test_data, filename, builder, realm, member, file_id):
+    Builder = getattr(builders, builder)
+    info = Builder.parser(str(test_data / filename))
+    assert info["realm"] == realm
+    if member:
+        assert info["member"] == member
+    assert info["file_id"] == file_id
+
+
 def test_builder_columns_with_iterables(test_data):
     builder = builders.AccessOm2Builder(str(test_data / "access-om2"))
     assert not builder.columns_with_iterables