Completely nuke the lib.py class with refactoring

- Rename `infiles` to `vars_to_filepaths` which is a dict with the key being a var and value being list of strings

e3sm_to_cmip/cmor_handlers/handler.py

@@ -1,5 +1,7 @@
+from __future__ import annotations
+
 import abc
-from typing import Any, Dict, List, Literal, Optional, TypedDict, Union
+from typing import Any, Dict, List, Literal, Optional, TypedDict
 
 import cmor
 import numpy as np
@@ -14,13 +16,6 @@
 # Used by areacella.py
 RADIUS = 6.37122e6
 
-# The names for hybrid levels.
-LEVEL_NAMES = [
-    "atmosphere_sigma_coordinate",
-    "standard_hybrid_sigma",
-    "standard_hybrid_sigma_half",
-]
-
 
 class Levels(TypedDict):
     name: str
@@ -162,7 +157,7 @@ def to_dict(self) -> Dict[str, Any]:
 
     def cmorize(
         self,
-        infiles: List[str],  # command-line arg (--input-path)
+        vars_to_filepaths: Dict[str, List[str]],  # command-line arg (--input-path)
         tables: str,  # command-line arg (--tables-path)
         user_input_path: str,  # command-line arg (--user-metadata)
         **kwargs,
@@ -176,7 +171,7 @@ def cmorize(
         # outpath=None,  # command-line arg (--output-path)
         if self.unit_conversion is None:
             return handle_variables(
-                infiles=infiles,
+                vars_to_filepaths=vars_to_filepaths,
                 raw_variables=self.raw_variables,
                 write_data=self._write_data,
                 outvar_name=self.name,
@@ -187,14 +182,11 @@ def cmorize(
                 serial=kwargs.get("serial"),
                 positive=self.positive,
                 levels=self.levels,
-                axis=kwargs.get("axis"),
                 logdir=kwargs.get("logdir"),
-                simple=kwargs.get("simple"),
-                outpath=kwargs.get("outpath"),
             )
 
         return default_handler(
-            infiles,
+            vars_to_filepaths,
             tables=tables,
             user_input_path=user_input_path,
             raw_variables=self.raw_variables,
@@ -212,43 +204,35 @@ def cmorize(
     def _write_data(
         self,
         varid: int,
-        data: Dict[str, Union[xr.DataArray, np.ndarray]],
+        data: Dict[str, xr.DataArray],
         timeval: float,
         timebnds: List[List[float]],
         index: int,
         **kwargs,
     ):
         # TODO: Replace **kwargs with explicit method parameters.
-
-        # Convert all variable arrays to np.ndarray to ensure compatibility with
-        # the CMOR library. For handlers that use `self.levels`, the array data
-        # structures are already `np.ndarray` due to additional processing.
-        for var in self.raw_variables:
-            if isinstance(data[var], xr.DataArray):
-                data[var] = data[var].values  # type: ignore
-
         if self.formula is not None:
             outdata = self.formula_method(data, index)
-        else:
-            outdata = data[self.raw_variables[0]][index, :]
-
-        if self._has_hybrid_sigma_levels():
+        elif self._has_hybrid_sigma_levels():
             outdata = self._hybrid_to_pressure(
-                data["ps"], data["hyai"], data["hybi"], p0=100000  # type: ignore
+                data["PS"][index, :], data["hyai"], data["hybi"], p0=100000  # type: ignore
             )
+        else:
+            outdata = data[self.raw_variables[0]][index, :]
 
-        # Replace `np.nan` with static FILL_VALUE
+        # Replace `np.nan` with static FILL_VALUE.
+        outdata = outdata.values
         outdata[np.isnan(outdata)] = FILL_VALUE
 
         if kwargs["simple"]:
             return outdata
 
         cmor.write(varid, outdata, time_vals=timeval, time_bnds=timebnds)
 
-        if self.levels is not None and self.levels.get("name") in LEVEL_NAMES:
+        if "PS" in data.keys() and "ips" in data.keys():
             cmor.write(
                 data["ips"],
-                data["ps"],
+                data["PS"].values,
                 time_vals=timeval,
                 time_bnds=timebnds,
                 store_with=varid,

e3sm_to_cmip/cmor_handlers/handlers.yaml

@@ -452,30 +452,30 @@
   levels: null
 - name: pfull
   units: Pa
-  raw_variables: ["PS", "hyai", "hybi", "hyam", "hybm"]
+  raw_variables: ["hyai", "hybi", "hyam", "hybm", "PS"]
   table: CMIP6_Amon.json
   unit_conversion: null
   formula: null
   positive: null
   levels:
     {
       name: "standard_hybrid_sigma",
-      unit": "1",
+      units: "1",
       e3sm_axis_name: "lev",
       e3sm_axis_bnds: "ilev",
       time_name: "time2",
     }
 - name: phalf
   units: Pa
-  raw_variables: ["PS", "hyai", "hybi", "hyam", "hybm"]
+  raw_variables: ["hyai", "hybi", "hyam", "hybm", "PS"]
   table: CMIP6_Amon.json
   unit_conversion: null
   formula: null
   positive: null
   levels:
     {
       name: "atmosphere_sigma_coordinate",
-      unit": "1",
+      units: "1",
       e3sm_axis_name: "lev",
       e3sm_axis_bnds: "ilev",
       time_name: "time2",

e3sm_to_cmip/cmor_handlers/vars/clcalipso.py

@@ -38,7 +38,7 @@ def write_data(varid, data, timeval, timebnds, index, **kwargs):
 # ------------------------------------------------------------------
 
 
-def handle(infiles, tables, user_input_path, **kwargs):
+def handle(vars_to_filepaths, tables, user_input_path, **kwargs):
     """
     Parameters
     ----------
@@ -51,19 +51,17 @@ def handle(infiles, tables, user_input_path, **kwargs):
     """
 
     return handle_variables(
+        vars_to_filepaths=vars_to_filepaths,
         metadata_path=user_input_path,
         tables=tables,
         table=kwargs.get("table", TABLE),
-        infiles=infiles,
         raw_variables=RAW_VARIABLES,
         write_data=write_data,
         outvar_name=VAR_NAME,
         outvar_units=VAR_UNITS,
         serial=kwargs.get("serial"),
         levels=LEVELS,
         logdir=kwargs.get("logdir"),
-        simple=kwargs.get("simple"),
-        outpath=kwargs.get("outpath"),
     )
 
 

e3sm_to_cmip/default.py

@@ -6,13 +6,11 @@
 from e3sm_to_cmip.lib import handle_variables
 
 
-def default_handler(infiles, tables, user_input_path, **kwargs):
-
+def default_handler(vars_to_filepaths, tables, user_input_path, **kwargs):  # noqa: C901
     RAW_VARIABLES = kwargs["raw_variables"]
     unit_conversion = kwargs.get("unit_conversion")
 
     def write_data(varid, data, timeval=None, timebnds=None, index=None, **kwargs):
-
         if timeval is not None:
             if unit_conversion is not None:
                 if unit_conversion == "g-to-kg":
@@ -62,16 +60,14 @@ def write_data(varid, data, timeval=None, timebnds=None, index=None, **kwargs):
         metadata_path=user_input_path,
         tables=tables,
         table=kwargs["table"],
-        infiles=infiles,
+        vars_to_filepaths=vars_to_filepaths,
         raw_variables=RAW_VARIABLES,
         write_data=write_data,
         outvar_name=kwargs["name"],
         outvar_units=kwargs["units"],
         serial=kwargs.get("serial"),
         positive=kwargs.get("positive"),
         logdir=kwargs.get("logdir"),
-        simple=kwargs.get("simple"),
-        outpath=kwargs.get("outpath"),
     )