updated domain table fetch (#140)

* updated domain table fetch

* test updates

* Update whats_new.rst

* added deprecation warning

cordex/domain.py

@@ -13,6 +13,19 @@
 from .utils import get_tempfile
 
 
+def _locate_domain_id(domain_id, tables):
+ """Locate domain_id in domain table trying different indexes."""
+ indexes = ["short_name", "domain_id", "CORDEX_domain"]
+
+ for i in indexes:
+ if domain_id in tables.reset_index()[i].values:
+ return (
+ tables.reset_index().replace(np.nan, None).set_index(i).loc[domain_id]
+ )
+
+ return tables.replace(np.nan, None).loc[domain_id]
+
+
 def domain_names(table_name=None):
  """Returns a list of short names of all availabe Cordex domains
 
@@ -112,11 +125,12 @@ def cordex_domain(
  tables = domains.table
  if isinstance(tables, list):
  tables = pd.concat(tables)
- config = tables.replace(np.nan, None).loc[domain_id]
+
+ config = _locate_domain_id(domain_id, tables)
 
  return create_dataset(
  **config,
- domain_id=domain_id,
+ # domain_id=domain_id,
  dummy=dummy,
  add_vertices=False,
  attrs=attrs,
@@ -214,11 +228,15 @@ def create_dataset(
  attrs = cv["default_global_attrs"]
  elif attrs is None:
  attrs = {}
+
  if name:
  attrs[cv["domain_id"]] = name
  # remove inconsistencies in keyword names
  if domain_id:
  attrs[cv["domain_id"]] = domain_id
+ if cv["domain_id"] in kwargs:
+ attrs[cv["domain_id"]] = kwargs[cv["domain_id"]]
+
  if pollon is None or pollat is None:
  rotated = False
  try:
@@ -424,20 +442,6 @@ def _bounds(coord, include="left"):
  return xr.merge([left, right])
 
 
-def bounds_coordinates(ds, coords):
- """Adds coordinate bounds as coordinates to the dataset."""
- if isinstance(coords, str):
- coords = (coords,)
- bounds = []
- for coord in coords:
- lr = _bounds(ds.coords[coord])
- name = ds.coords[coord].name + "_b"
- bounds.append(
- xr.DataArray(np.append(lr.left, lr.right[-1]), dims=name, name=name)
- )
- return ds.assign_coords({b.name: b for b in bounds})
-
-
 def bounds(coords):
  if isinstance(coords, xr.DataArray):
  coords = (coords,)

cordex/preprocessing/preprocessing.py

@@ -5,6 +5,8 @@
 """
 
 
+from warnings import warn
+
 import numpy as np
 import xarray as xr
 
@@ -447,6 +449,12 @@ def get_grid_mapping(ds):
  Dataarray containing the grid mapping meta data.
 
  """
+ message = (
+ "get_grid_mapping is deprecated, please use cf_xarray "
+ 'accessor ds.cf["grid_mapping"] instead.'
+ )
+ warn(message, DeprecationWarning, stacklevel=2)
+
  return ds[get_grid_mapping_varname(ds)]
 
 
@@ -553,35 +561,6 @@ def member_id_to_dset_id(ds_dict):
  return ds_split
 
 
-# def dset_ids_to_coord(ds_dict):
-# """Creates a DataArray from dataset ids"""
-# dset_ids = list(ds_dict.keys())
-# dim = xr.DataArray(
-# dset_ids, dims="dset_id", name="dset_id", coords={"dset_id": dset_ids}
-# )
-# return dim
-
-
-# def align_time_axis(ds_dict):
-# from datetime import datetime as dt
-
-# for ds in ds_dict.values():
-# # ds = ds.copy()
-# ds.coords["time"] = [dt(date.year, date.month, 15) for date in ds.time.values]
-# return ds_dict
-
-
-# def concat_along_dset_id(ds_dict, coords="minimal", compat="override", **kwargs):
-# dset_coord = dset_ids_to_coord(ds_dict)
-# ds_dict = align_time_axis(ds_dict)
-# ds_list = []
-# for ds in ds_dict.values():
-# ds = replace_rlon_rlat(ds)
-# ds = replace_lon_lat(ds)
-# ds_list.append(ds)
-# return xr.concat(ds_list, dim=dset_coord, coords=coords, compat=compat, **kwargs)
-
-
 def sort_ds_dict_by_attr(ds_dict, attr):
  """Sorts the dataset dict by a certain attribute.
 

cordex/tables/__init__.py

@@ -1,22 +1,19 @@
-from warnings import warn
-
 import pandas as pd
 
-from ._resources import (
+from ._resources import ( # fetch_cmip6_cmor_table,
  cmor_tables_inpath,
  ecmwf_tables,
- fetch_cmip6_cmor_table,
  fetch_cordex_cmor_table,
- read_cordex_domain_tables,
+ read_domain_table,
 )
 
 # __cmor_table_version__ = cmor_table_version
 __all__ = [
  "cmor_tables_inpath",
  "ecmwf_tables",
- "fetch_cmpi6_cmor_table",
+ # "fetch_cmpi6_cmor_table",
  "fetch_cordex_cmor_table",
- "read_cordex_domain_tables",
+ "read_cordex_domain_table",
  "cmor_tables_inpath",
 ]
 
@@ -31,13 +28,15 @@ def tables(self):
 
  @property
  def table(self):
- return pd.concat(self.tables.values())
+ if isinstance(self.tables, dict):
+ return pd.concat(self.tables.values())
+ return self.tables
 
  # def __getattr__(self, table):
  # return self.tables[table]
 
 
-domains = read_cls(read_cordex_domain_tables)
+domains = read_cls(read_domain_table)
 
 ecmwf = read_cls(ecmwf_tables)
 
@@ -69,26 +68,26 @@ def cordex_cmor_table(table, table_dir=None):
  return fetch_cordex_cmor_table(table)
 
 
-def cmip6_cmor_table(table):
- """fetch a cmip6 cmor table
-
- If required, the table will be download from github.
- The tables are experimental right now and only used
- for development purposes.
-
- Parameters
- ----------
- table: str
- Name of the cmip6 table.
-
- Returns
- -------
- filename : str
- Filepath to the cmip6 cmor table.
- """
- warn(
- "CMIP6 cmor table fetching is deprecated and will be removed in the future. Please use cordex_cmor_table instead.",
- DeprecationWarning,
- stacklevel=2,
- )
- return fetch_cmip6_cmor_table(table)
+# def cmip6_cmor_table(table):
+# """fetch a cmip6 cmor table
+#
+# If required, the table will be download from github.
+# The tables are experimental right now and only used
+# for development purposes.
+#
+# Parameters
+# ----------
+# table: str
+# Name of the cmip6 table.
+#
+# Returns
+# -------
+# filename : str
+# Filepath to the cmip6 cmor table.
+# """
+# warn(
+# "CMIP6 cmor table fetching is deprecated and will be removed in the future. Please use cordex_cmor_table instead.",
+# DeprecationWarning,
+# stacklevel=2,
+# )
+# return fetch_cmip6_cmor_table(table)

cordex/tables/_resources.py

@@ -26,13 +26,9 @@ def _construct_cache_dir(path):
  # Use the default cache folder for the OS
  path=cache_url, # pooch.os_cache("cordex"),
  # The remote data is on Github
- base_url=base_url + "domains/",
+ base_url="https://raw.githubusercontent.com/WCRP-CORDEX/domain-tables/main/",
  registry={
- "cordex.csv": None,
- "cordex-high-res.csv": None,
- "cordex-fps.csv": None,
- "cordex-core.csv": None,
- "cordex-regular.csv": None,
+ "rotated-latitude-longitude.csv": None,
  },
 )
 
@@ -70,10 +66,10 @@ def fetch_cordex_cmor_table(table):
  )
 
 
-def fetch_cmip6_cmor_table(table):
- return retrieve_cmor_table(
- table, url="https://github.com/PCMDI/cmip6-cmor-tables/raw/master/Tables"
- )
+# def fetch_cmip6_cmor_table(table):
+# return retrieve_cmor_table(
+# table, url="https://github.com/PCMDI/cmip6-cmor-tables/raw/master/Tables"
+# )
 
 
 def retrieve_cmor_table(table, url):
@@ -106,12 +102,11 @@ def read_region_table(name):
  return read_remote_table(name, resource=REGION_RESOURCE, index_col="area")
 
 
-def read_cordex_domain_tables():
+def read_domain_table():
  resource = DOMAIN_RESOURCE
- return {
- table.split(".")[0]: read_remote_table(table, resource, index_col="short_name")
- for table in resource.registry.keys()
- }
+ return read_remote_table(
+ "rotated-latitude-longitude.csv", resource, index_col="short_name"
+ )
 
 
 def region_tables():

docs/whats_new.rst

@@ -11,11 +11,11 @@ What's New
 v0.6.0 (Unreleased)
 -------------------
 
-New Features
-~~~~~~~~~~~~
+Internal Changes
+~~~~~~~~~~~~~~~~
 
-- Support for upcoming CORDEX-CMIP6 vocabulary. This includes the new keyword `mip_era` in :py:meth:`cmor.cordex_domain` and
-:py:meth:`cmor.create_dataset` (:pull:`129`).
+- Support for upcoming CORDEX-CMIP6 vocabulary. This includes the new keyword `mip_era` in :py:meth:`cmor.cordex_domain` and :py:meth:`cmor.create_dataset` (:pull:`129`).
+- Updated domain table fetching. There is now only one table that contains all domain definitions which is now located in the `WCRP CORDEX github table repository <https://github.com/WCRP-CORDEX/domain-tables/blob/main/rotated-latitude-longitude.csv>`_ (:pull:`140`). This table allows for selection by different naming conventions for the domain identifier, e.g., ``EUR-12`` is equivilant to ``EUR-11``, etc.
 
 Breaking Changes
 ~~~~~~~~~~~~~~~~

tests/test_tables.py

@@ -2,21 +2,20 @@
 
 import cordex as cx
 
-
-@pytest.mark.parametrize(
- "table",
- [
- "CMIP6_Amon",
- "CMIP6_day",
- "CMIP6_E1hr",
- "CMIP6_E3hr",
- "CMIP6_coordinate",
- "CMIP6_fx",
- "CMIP6_grids",
- ],
-)
-def test_download_cmip6_cmor_tables(table):
- cx.tables.cmip6_cmor_table(table)
+# @pytest.mark.parametrize(
+# "table",
+# [
+# "CMIP6_Amon",
+# "CMIP6_day",
+# "CMIP6_E1hr",
+# "CMIP6_E3hr",
+# "CMIP6_coordinate",
+# "CMIP6_fx",
+# "CMIP6_grids",
+# ],
+# )
+# def test_download_cmip6_cmor_tables(table):
+# cx.tables.cmip6_cmor_table(table)
 
 
 @pytest.mark.parametrize("table", ["CORDEX_CV", "CORDEX_remo_example"])