Shortened display of NumPy arrays in DataArray.__repr__ (#1207)

* Shortened display of NumPy arrays in DataArray.__repr__

This shortens the array portion of DataArray repr to reasonable lengths for high
dimensional arrays:

Array                         | Old length | New length
------------------------------+------------+-----------
np.random.randn(500)          | 167        | 1
np.random.randn(20, 20)       | 150        | 7
np.random.randn(5, 10, 15)    | 254        | 24
np.random.randn(5, 10, 15, 3) | 248        | 19
air_temperature['air']        | 41         | 24
rasm['Tair']                  | 37         | 24

where the last two entries refer to the corresponding datasets loaded from
`xarray.tutorial.load_dataset`.

Importantly, this change does not preclude future improvements, such as those
suggested in GH1044.

* fix spelling

* a little more explanation in what's new

Author: shoyer

doc/whats-new.rst

@@ -84,6 +84,10 @@ Breaking changes
 - By default ``to_netcdf()`` add a ``_FillValue = NaN`` attributes to float types.
   By `Frederic Laliberte <https://github.com/laliberte>`_.
 
+- ``repr`` on ``DataArray`` objects uses an shortened display for NumPy array
+  data that is less likely to overflow onto multiple pages (:issue:`1207`).
+  By `Stephan Hoyer <https://github.com/shoyer>`_.
+
 - xarray no longer supports python 3.3, versions of dask prior to v0.9.0,
   or versions of bottleneck prior to v1.0.
 

xarray/core/formatting.py

@@ -7,6 +7,7 @@
 from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
+import contextlib
 from datetime import datetime, timedelta
 import functools
 
@@ -336,6 +337,34 @@ def dim_summary(obj):
     return u', '.join(elements)
 
 
+@contextlib.contextmanager
+def set_numpy_options(*args, **kwargs):
+    original = np.get_printoptions()
+    np.set_printoptions(*args, **kwargs)
+    yield
+    np.set_printoptions(**original)
+
+
+def short_array_repr(array):
+    array = np.asarray(array)
+    # default to lower precision so a full (abbreviated) line can fit on
+    # one line with the default display_width
+    options = {
+        'precision': 6,
+        'linewidth': OPTIONS['display_width'],
+        'threshold': 200,
+    }
+    if array.ndim < 3:
+        edgeitems = 3
+    elif array.ndim == 3:
+        edgeitems = 2
+    else:
+        edgeitems = 1
+    options['edgeitems'] = edgeitems
+    with set_numpy_options(**options):
+        return repr(array)
+
+
 def array_repr(arr):
     # used for DataArray, Variable and IndexVariable
     if hasattr(arr, 'name') and arr.name is not None:
@@ -349,7 +378,7 @@ def array_repr(arr):
     if isinstance(getattr(arr, 'variable', arr)._data, dask_array_type):
         summary.append(repr(arr.data))
     elif arr._in_memory or arr.size < 1e5:
-        summary.append(repr(arr.values))
+        summary.append(short_array_repr(arr.values))
     else:
         summary.append(u'[%s values with dtype=%s]' % (arr.size, arr.dtype))
 

xarray/test/test_formatting.py

@@ -130,3 +130,26 @@ def test_format_timestamp_out_of_bounds(self):
         expected = '2300-12-01'
         result = formatting.format_timestamp(date)
         self.assertEqual(result, expected)
+
+
+def test_set_numpy_options():
+    original_options = np.get_printoptions()
+    with formatting.set_numpy_options(threshold=10):
+        assert len(repr(np.arange(500))) < 200
+    # original options are restored
+    assert np.get_printoptions() == original_options
+
+
+def test_short_array_repr():
+    cases = [
+        np.random.randn(500),
+        np.random.randn(20, 20),
+        np.random.randn(5, 10, 15),
+        np.random.randn(5, 10, 15, 3),
+    ]
+    # number of lines:
+    # for default numpy repr: 167, 140, 254, 248
+    # for short_array_repr: 1, 7, 24, 19
+    for array in cases:
+        num_lines = formatting.short_array_repr(array).count('\n') + 1
+        assert num_lines < 30