fix #803 :

- included key associated with Axis and Group objects when exporting Session objects to CSV or Excel files
- removed useless kwargs['name'] = key in PandasHDFHandler._read_item()
fix #804 : wildcard axes are handled correctly when reading/exporting from/to CSV or Excel files

Author: alixdamman

doc/source/changes/version_0_32.rst.inc

@@ -55,4 +55,5 @@ Miscellaneous improvements
 Fixes
 ^^^^^
 
-* fixed something (closes :issue:`1`).
+* fixed reading/exporting sessions containing two or more axes/groups
+  with the same name (or anonymous) from/to CSV, Excel and HDF files (closes :issue:`803`).

larray/inout/csv.py

@@ -329,15 +329,15 @@ def list_items(self):
             pass
         try:
             fnames.remove('__axes__')
-            items = [(name, 'Axis') for name in sorted(self.axes.keys())]
+            items = [(key, 'Axis') for key in sorted(self.axes.keys())]
         except:
             pass
         try:
             fnames.remove('__groups__')
-            items += [(name, 'Group') for name in sorted(self.groups.keys())]
+            items += [(key, 'Group') for key in sorted(self.groups.keys())]
         except:
             pass
-        items += [(name, 'Array') for name in fnames]
+        items += [(key, 'Array') for key in fnames]
         return items
 
     def _read_item(self, key, type, *args, **kwargs):
@@ -375,10 +375,10 @@ def _dump_metadata(self, metadata):
 
     def save(self):
         if len(self.axes) > 0:
-            df = _axes_to_df(self.axes.values())
+            df = _axes_to_df(self.axes)
             df.to_csv(self._to_filepath('__axes__'), sep=self.sep, index=False)
         if len(self.groups) > 0:
-            df = _groups_to_df(self.groups.values())
+            df = _groups_to_df(self.groups)
             df.to_csv(self._to_filepath('__groups__'), sep=self.sep, index=False)
 
     def close(self):

larray/inout/excel.py

@@ -268,15 +268,15 @@ def list_items(self):
             pass
         try:
             sheet_names.remove('__axes__')
-            items = [(name, 'Axis') for name in sorted(self.axes.keys())]
+            items = [(key, 'Axis') for key in sorted(self.axes.keys())]
         except:
             pass
         try:
             sheet_names.remove('__groups__')
-            items += [(name, 'Group') for name in sorted(self.groups.keys())]
+            items += [(key, 'Group') for key in sorted(self.groups.keys())]
         except:
             pass
-        items += [(name, 'Array') for name in sheet_names]
+        items += [(key, 'Array') for key in sheet_names]
         return items
 
     def _read_item(self, key, type, *args, **kwargs):
@@ -316,10 +316,10 @@ def _dump_metadata(self, metadata):
 
     def save(self):
         if len(self.axes) > 0:
-            df = _axes_to_df(self.axes.values())
+            df = _axes_to_df(self.axes)
             df.to_excel(self.handle, '__axes__', index=False, engine='xlsxwriter')
         if len(self.groups) > 0:
-            df = _groups_to_df(self.groups.values())
+            df = _groups_to_df(self.groups)
             df.to_excel(self.handle, '__groups__', index=False, engine='xlsxwriter')
 
     def close(self):
@@ -373,15 +373,15 @@ def list_items(self):
             pass
         try:
             sheet_names.remove('__axes__')
-            items = [(name, 'Axis') for name in sorted(self.axes.keys())]
+            items = [(key, 'Axis') for key in sorted(self.axes.keys())]
         except:
             pass
         try:
             sheet_names.remove('__groups__')
-            items += [(name, 'Group') for name in sorted(self.groups.keys())]
+            items += [(key, 'Group') for key in sorted(self.groups.keys())]
         except:
             pass
-        items += [(name, 'Array') for name in sheet_names]
+        items += [(key, 'Array') for key in sheet_names]
         return items
 
     def _read_item(self, key, type, *args, **kwargs):
@@ -419,11 +419,11 @@ def _dump_metadata(self, metadata):
 
     def save(self):
         if len(self.axes) > 0:
-            df = _axes_to_df(self.axes.values())
+            df = _axes_to_df(self.axes)
             self.handle['__axes__'] = ''
             self.handle['__axes__'][:].options(pd.DataFrame, index=False).value = df
         if len(self.groups) > 0:
-            df = _groups_to_df(self.groups.values())
+            df = _groups_to_df(self.groups)
             self.handle['__groups__'] = ''
             self.handle['__groups__'][:].options(pd.DataFrame, index=False).value = df
         self.handle.save()

larray/inout/hdf.py

@@ -132,10 +132,8 @@ def _read_item(self, key, type, *args, **kwargs):
             hdf_key = '/' + key
         elif type == 'Axis':
             hdf_key = '__axes__/' + key
-            kwargs['name'] = key
         elif type == 'Group':
             hdf_key = '__groups__/' + key
-            kwargs['name'] = key
         else:
             raise TypeError()
         return read_hdf(self.handle, hdf_key, *args, **kwargs)

larray/inout/pandas.py

@@ -338,51 +338,61 @@ def df_aslarray(df, sort_rows=False, sort_columns=False, raw=False, parse_header
 #    SERIES <--> AXIS, GROUP, META     #
 # #################################### #
 
-def _axis_to_series(axis, dtype=None):
-    return pd.Series(data=axis.labels, name=str(axis), dtype=dtype)
+def _axis_to_series(key, axis, dtype=None):
+    name = '{}:{}'.format(key, str(axis))
+    labels = len(axis) if axis.iswildcard else axis.labels
+    return pd.Series(data=labels, name=name, dtype=dtype)
 
 
 def _series_to_axis(series):
-    return Axis(labels=series.values, name=series.name)
+    name = str(series.name)
+    if ':' in name:
+        key, axis_name = name.split(':')
+        labels = int(series.values[0]) if '*' in axis_name else series.values
+        if '{?}' in axis_name:
+            axis_name = None
+    else:
+        # for backward compatibility
+        key = axis_name = name
+        labels = series.values
+    return key, Axis(labels=labels, name=axis_name)
 
 
-def _group_to_series(group, dtype=None):
-    name = group.name if group.name is not None else '{?}'
+def _group_to_series(key, group, dtype=None):
     if group.axis.name is None:
         raise ValueError("Cannot save a group with an anonymous associated axis")
-    name += '@{}'.format(group.axis.name)
+    name = '{}:{}@{}'.format(key, group.name, group.axis.name)
     return pd.Series(data=group.eval(), name=name, dtype=dtype)
 
 
-def _series_to_group(series, axis):
-    name = series.name.split('@')[0]
-    return LGroup(key=series.values, name=name, axis=axis)
+def _series_to_group(series, axes):
+    key, name = str(series.name).split(':')
+    group_name, axis_name = name.split('@')
+    if group_name == 'None':
+        group_name = None
+    axis = axes[axis_name]
+    return key, LGroup(key=series.values, name=group_name, axis=axis)
 
 
 # ######################################## #
 #    DATAFRAME <--> AXES, GROUPS, META     #
 # ######################################## #
 
 def _df_to_axes(df):
-    return OrderedDict([(col_name, _series_to_axis(df[col_name])) for col_name in df.columns.values])
+    return OrderedDict([_series_to_axis(df[col_name]) for col_name in df.columns.values])
 
 
 def _axes_to_df(axes):
     # set dtype to np.object otherwise pd.concat below may convert an int row/column as float
     # if trailing NaN need to be added
-    return pd.concat([_axis_to_series(axis, dtype=np.object) for axis in axes], axis=1)
+    return pd.concat([_axis_to_series(key, axis, dtype=np.object) for key, axis in axes.items()], axis=1)
 
 
 def _df_to_groups(df, axes):
-    groups = OrderedDict()
-    for name, values in df.iteritems():
-        group_name, axis_name = name.split('@')
-        axis = axes[axis_name]
-        groups[group_name] = _series_to_group(values, axis)
-    return groups
+    return OrderedDict([_series_to_group(df[col_name], axes) for col_name in df.columns.values])
 
 
 def _groups_to_df(groups):
     # set dtype to np.object otherwise pd.concat below may convert an int row/column as float
     # if trailing NaN need to be added
-    return pd.concat([_group_to_series(group, dtype=np.object) for group in groups], axis=1)
+    return pd.concat([_group_to_series(key, group, dtype=np.object) for key, group in groups.items()], axis=1)