Revert some of the changes in commit 703420a to attain a backward com…

…patibility with older python/pandas versions.

adda/adda.py

@@ -172,8 +172,8 @@ def main(args):
     data_adc,meta_adc=rpl.fetch_station_product(urls, return_sample_min=args.return_sample_min, fort63_style=fort63_style )
 
     # Revert Harvester filling of nans to -99999 back to nans
-    data_adc = data_adc.replace('-99999',np.nan).infer_objects(copy=False)
-    meta_adc = meta_adc.replace('-99999',np.nan).infer_objects(copy=False)
+    data_adc = data_adc.replace('-99999',np.nan)
+    meta_adc = meta_adc.replace('-99999',np.nan)
 
     # Get the grid coordinates for the url 
     adc_coords = get_adcirc_stations.extract_adcirc_grid_coords( urls )
@@ -242,8 +242,8 @@ def main(args):
                 knockout_dict=None, station_list_file=station_file)
     # Get data at highest resolution
     data_obs,meta_obs=obs.fetch_station_product((obs_starttime,obs_endtime), return_sample_min=0)
-    data_obs = data_obs.replace('-99999',np.nan).infer_objects(copy=False)
-    meta_obs = meta_obs.replace('-99999',np.nan).infer_objects(copy=False)
+    data_obs = data_obs.replace('-99999',np.nan)
+    meta_obs = meta_obs.replace('-99999',np.nan)
     temp=io_utilities.write_csv(data_obs, rootdir=rootdir,subdir=iosubdir,fileroot='data_obs')
 
     # Remove stations with too many nans 

harvester/fetch_station_data.py

@@ -94,7 +94,7 @@ def replace_and_fill(df):
     """
     Replace all Nans with 'None" values with GLOBAL_FILL_VALUE
     """
-    df=df.fillna(GLOBAL_FILL_VALUE).infer_objects(copy=False)
+    df=df.fillna(GLOBAL_FILL_VALUE)
     return df
 
 def stations_resample(df, sample_mins=15)->pd.DataFrame:
@@ -191,7 +191,7 @@ def interpolate_and_resample(self, dx, n_pad=0, sample_mins=15, int_limit=3)->pd
                 timeout = dt.datetime.strptime(max(dx.index+np.timedelta64(n_pad,'h')).strftime(dformat), dformat)
                 # Generate the NEW augmented time range
                 actualRange = dx.index
-                normalRange = pd.date_range(str(timein), str(timeout), freq=f'{sample_mins*60.0}s') # This gets us the stepping we want
+                normalRange = pd.date_range(str(timein), str(timeout), freq=f'{sample_mins*60.0}S') # This gets us the stepping we want
                 datanormal=[x for x in normalRange if x not in actualRange] 
                 # Assemble the union of values for the final data set. Exclude entries that already exist in the real data
                 dappend = pd.concat([dx,pd.DataFrame(index=datanormal)],axis=0)

processing/compute_error_field.py

@@ -44,7 +44,7 @@ def interpolate_and_sample( diurnal_range, df_in )-> pd.DataFrame:
     df_out.sort_index(inplace=True) # this is sorted with intervening nans that need to be imputed
     #df_out.to_csv('check_pre.csv',float_format='%.3f')
     #df_out_int = df_out.interpolate(method='linear')
-    df_out_int = df_out.infer_objects(copy=False).interpolate(method='values')
+    df_out_int = df_out.interpolate(method='values')
     #df_out_int.to_csv('check_po1.csv',float_format='%.3f')
     df_out_int = df_out_int.loc[diurnal_range]
     df_out_int.index.name='TIME' 
@@ -217,13 +217,13 @@ def _tidal_transform_data(self):
         n_range = self.adc.index.tolist()
         n_range.sort()
         timein, timeout = n_range[0], n_range[-1]
-        normalRange = pd.date_range(str(timein), str(timeout), freq='3600s') 
+        normalRange = pd.date_range(str(timein), str(timeout), freq='3600S') 
         n_hours_per_period = self.n_hours_per_period
         n_hours_per_tide = self.n_hours_per_tide
         n_pad = self.n_pad # This is used to push inteprlation end-nans to outside the time bounds
 
         time_step =  int(3600*n_hours_per_tide/n_hours_per_period) # Always scale to an hour (3600s)
-        diurnal_range = pd.date_range(timein, timeout+np.timedelta64(n_pad,'h'), freq=str(time_step)+'s').to_list()
+        diurnal_range = pd.date_range(timein, timeout+np.timedelta64(n_pad,'h'), freq=str(time_step)+'S').to_list()
 
         #self.adc.to_csv('check_adc_po3a.csv',float_format='%.3f')
         self.adc = interpolate_and_sample( diurnal_range, self.adc )