small code cleanups in various functions

incidentally it also makes .reshape support string axes collection...
not that it matters much since it should be private method these days anyway

Author: gdementen

larray/core/array.py

@@ -1333,10 +1333,10 @@ def rename(self, renames=None, to=None, inplace=False, **kwargs):
             items = []
         items += kwargs.items()
         renames = {self.axes[k]: v for k, v in items}
-        axes = [a.rename(renames[a]) if a in renames else a
-                for a in self.axes]
+        axes = AxisCollection([a.rename(renames[a]) if a in renames else a
+                               for a in self.axes])
         if inplace:
-            self.axes = AxisCollection(axes)
+            self.axes = axes
             return self
         else:
             return LArray(self.data, axes)
@@ -1870,8 +1870,7 @@ def sort_key(axis):
                 key = key[::-1]
             return axis.i[key]
 
-        res = self[tuple(sort_key(axis) for axis in axes)]
-        return res
+        return self[tuple(sort_key(axis) for axis in axes)]
 
     sort_axis = renamed_to(sort_axes, 'sort_axis')
 
@@ -2113,6 +2112,7 @@ def set(self, value, **kwargs):
         """
         self.__setitem__(kwargs, value)
 
+    # TODO: this should be a private method
     def reshape(self, target_axes):
         """
         Given a list of new axes, changes the shape of the array.
@@ -2157,9 +2157,12 @@ def reshape(self, target_axes):
         #            -> 3, 8 WRONG (non adjacent dimensions)
         #            -> 8, 3 WRONG
         #    4, 3, 2 -> 2, 2, 3, 2 is potentially ok (splitting dim)
-        data = np.asarray(self).reshape([len(axis) for axis in target_axes])
+        if not isinstance(target_axes, AxisCollection):
+            target_axes = AxisCollection(target_axes)
+        data = np.asarray(self).reshape(target_axes.shape)
         return LArray(data, target_axes)
 
+    # TODO: this should be a private method
     def reshape_like(self, target):
         """
         Same as reshape but with an array as input.
@@ -6372,15 +6375,12 @@ def to_clipboard(self, *args, **kwargs):
         self.to_frame().to_clipboard(*args, **kwargs)
 
     # XXX: sep argument does not seem very useful
-    # def to_excel(self, filename, sep=None):
+    # def to_excel(self, filename, sep='_'):
     #     # Why xlsxwriter? Because it is faster than openpyxl and xlwt
     #     # currently does not .xlsx (only .xls).
     #     # PyExcelerate seem like a decent alternative too
     #     import xlsxwriter as xl
     #
-    #     if sep is None:
-    #         sep = '_'
-    #         #sep = self.sep
     #     workbook = xl.Workbook(filename)
     #     if self.ndim > 2:
     #         for key in product(*[axis.labels for axis in self.axes[:-2]]):
@@ -8529,7 +8529,7 @@ def raw_broadcastable(values, min_axes=None):
     """
     same as make_numpy_broadcastable but returns numpy arrays
     """
-    arrays, res_axes = make_numpy_broadcastable(values, min_axes)
+    arrays, res_axes = make_numpy_broadcastable(values, min_axes=min_axes)
     raw = [a.data if isinstance(a, LArray) else a
            for a in arrays]
     return raw, res_axes

larray/core/group.py

@@ -33,7 +33,6 @@ def _slice_to_str(key, repr_func=str):
     >>> _slice_to_str(slice(None, 5, 2))
     ':5:2'
     """
-    # examples of result: ":24" "25:" ":" ":5:2"
     start = repr_func(key.start) if key.start is not None else ''
     stop = repr_func(key.stop) if key.stop is not None else ''
     step = (":" + repr_func(key.step)) if key.step is not None else ''

larray/inout/pandas.py

@@ -7,7 +7,7 @@
 import pandas as pd
 
 from larray.core.array import LArray
-from larray.core.axis import Axis
+from larray.core.axis import Axis, AxisCollection
 from larray.core.group import LGroup
 from larray.core.constants import nan
 from larray.util.misc import basestring, decode, unique
@@ -67,7 +67,7 @@ def cartesian_product_df(df, sort_rows=False, sort_columns=False, fill_value=nan
     columns = sorted(df.columns) if sort_columns else list(df.columns)
     # the prodlen test is meant to avoid the more expensive array_equal test
     prodlen = np.prod([len(axis_labels) for axis_labels in labels])
-    if prodlen == len(df) and columns == list(df.columns) and np.array_equal(df.index.values, new_index.values):
+    if prodlen == len(df) and columns == list(df.columns) and np.array_equal(idx.values, new_index.values):
         return df, labels
     return df.reindex(index=new_index, columns=columns, fill_value=fill_value, **kwargs), labels
 
@@ -247,8 +247,8 @@ def from_frame(df, sort_rows=False, sort_columns=False, parse_header=False, unfo
     axes_names = [str(name) if name is not None else name
                   for name in axes_names]
 
-    axes = [Axis(labels, name) for labels, name in zip(axes_labels, axes_names)]
-    data = df.values.reshape([len(axis) for axis in axes])
+    axes = AxisCollection([Axis(labels, name) for labels, name in zip(axes_labels, axes_names)])
+    data = df.values.reshape(axes.shape)
     return LArray(data, axes, meta=meta)
 
 

larray/random.py

@@ -26,7 +26,7 @@
 import numpy as np
 
 from larray.core.axis import Axis, AxisCollection
-from larray.core.array import LArray, aslarray, stack, ndtest
+from larray.core.array import LArray, aslarray
 from larray.core.array import raw_broadcastable
 import larray as la
 
@@ -35,7 +35,7 @@
 
 
 def generic_random(np_func, args, min_axes, meta):
-    args, res_axes = raw_broadcastable(args, min_axes)
+    args, res_axes = raw_broadcastable(args, min_axes=min_axes)
     res_data = np_func(*args, size=res_axes.shape)
     return LArray(res_data, res_axes, meta=meta)
 

larray/util/misc.py

@@ -63,7 +63,7 @@ def is_interactive_interpreter():
 
 def csv_open(filename, mode='r'):
     assert 'b' not in mode and 't' not in mode
-    if sys.version < '3':
+    if PY2:
         return open(filename, mode + 'b')
     else:
         return open(filename, mode, newline='', encoding='utf8')