Optimized reprojection function.

climax/core/utils.py

@@ -3,24 +3,61 @@
 # !/usr/bin/env python
 # -*- coding: utf-8 -*-
 
+# builtins
+import pathlib
+import logging
+
 # externals
 import pandas as pd
+import cdo
 
 # locals
 from pysegcnn.core.utils import search_files
-from climax.core.constants import CORDEX_PARAMETERS
+from climax.core.constants import CORDEX_PARAMETERS, CDO_RESAMPLING_MODES
+
+# module level logger
+LOGGER = logging.getLogger(__name__)
 
 
-def get_inventory(path):
+def get_inventory(path, pattern='(.*).nc$', return_df=False):
 
     # find all netcdf files in path
-    inventory = search_files(path, '(.*).nc$')
+    inventory = search_files(path, pattern)
 
     # create dictionary: (filename: [simulation parameters])
     inventory = {k: k.stem.split('_') for k in inventory}
 
     # create a DataFrame
-    df = pd.DataFrame(data=inventory.values(), index=inventory.keys(),
-                      columns=CORDEX_PARAMETERS)
+    if return_df:
+        inventory = pd.DataFrame(data=inventory.values(),
+                                 index=inventory.keys(),
+                                 columns=CORDEX_PARAMETERS)
+
+    return inventory
+
+
+def reproject_cdo(grid, src_ds, trg_ds, mode='bilinear', overwrite=False):
+
+    # instanciate the cdo
+    operator = cdo.Cdo()
+
+    # check if target dataset exists
+    if pathlib.Path(trg_ds).exists() and not overwrite:
+        LOGGER.info('{} already exists. Aborting ...'.format(trg_ds))
+        return trg_ds
+
+    # check if mode is supported
+    if mode not in CDO_RESAMPLING_MODES:
+        raise ValueError('Resampling mode "{}" not supported.'.format(mode))
+    else:
+        # check which resampling mode to use
+        LOGGER.info('Reproject: {}'.format(trg_ds))
+        if mode == 'bilinear':
+            operator.remapbil(str(grid),
+                              infile=str(src_ds), outfile=str(trg_ds))
+
+        if mode == 'con':
+            operator.remapcon(str(grid),
+                              infile=str(src_ds), outfile=str(trg_ds))
 
-    return df
+    return trg_ds