BUG: Don't error in pd.to_timedelta when errors=ignore

Title is self-explanatory.  Closes pandas-dev#13613.

Author: gfyoung <gfyoung17@gmail.com>

Closes pandas-dev#13832 from gfyoung/to-timedelta-error-bug and squashes the following commits:

dc39205 [gfyoung] BUG: Don't error in pd.to_timedelta when errors=ignore

asv_bench/benchmarks/timedelta.py

@@ -31,4 +31,19 @@ def setup(self):
         self.arr = ['00:00:{0:02d}'.format(i) for i in self.arr]
 
     def time_timedelta_convert_string_seconds(self):
-        to_timedelta(self.arr)
+        to_timedelta(self.arr)
+
+
+class timedelta_convert_bad_parse(object):
+    goal_time = 0.2
+
+    def setup(self):
+        self.arr = np.random.randint(0, 1000, size=10000)
+        self.arr = ['{0} days'.format(i) for i in self.arr]
+        self.arr[-1] = 'apple'
+
+    def time_timedelta_convert_coerce(self):
+        to_timedelta(self.arr, errors='coerce')
+
+    def time_timedelta_convert_ignore(self):
+        to_timedelta(self.arr, errors='ignore')

doc/source/whatsnew/v0.19.0.txt

@@ -858,6 +858,7 @@ Bug Fixes
 - Bug in ``groupby().cumsum()`` calculating ``cumprod`` when ``axis=1``. (:issue:`13994`)
 - Bug in ``pd.read_csv()``, which may cause a segfault or corruption when iterating in large chunks over a stream/file under rare circumstances (:issue:`13703`)
 - Bug in ``pd.read_csv()``, which caused BOM files to be incorrectly parsed by not ignoring the BOM (:issue:`4793`)
+- Bug in ``pd.to_timedelta()`` in which the ``errors`` parameter was not being respected (:issue:`13613`)
 - Bug in ``io.json.json_normalize()``, where non-ascii keys raised an exception (:issue:`13213`)
 - Bug when passing a not-default-indexed ``Series`` as ``xerr`` or ``yerr`` in ``.plot()`` (:issue:`11858`)
 - Bug in area plot draws legend incorrectly if subplot is enabled or legend is moved after plot (matplotlib 1.5.0 is required to draw area plot legend properly) (issue:`9161`, :issue:`13544`)

pandas/src/inference.pyx

@@ -780,7 +780,7 @@ def maybe_convert_objects(ndarray[object] objects, bint try_float=0,
                 break
         elif is_timedelta(val):
             if convert_timedelta:
-                itimedeltas[i] = convert_to_timedelta64(val, 'ns', False)
+                itimedeltas[i] = convert_to_timedelta64(val, 'ns')
                 seen_timedelta = 1
             else:
                 seen_object = 1

pandas/tseries/tests/test_timedeltas.py

@@ -845,6 +845,11 @@ def testit(unit, transform):
 
     def test_to_timedelta_invalid(self):
 
+        # bad value for errors parameter
+        msg = "errors must be one of"
+        tm.assertRaisesRegexp(ValueError, msg, to_timedelta,
+                              ['foo'], errors='never')
+
         # these will error
         self.assertRaises(ValueError, lambda: to_timedelta([1, 2], unit='foo'))
         self.assertRaises(ValueError, lambda: to_timedelta(1, unit='foo'))
@@ -862,6 +867,24 @@ def test_to_timedelta_invalid(self):
                               to_timedelta(['1 day', 'bar', '1 min'],
                                            errors='coerce'))
 
+        # gh-13613: these should not error because errors='ignore'
+        invalid_data = 'apple'
+        self.assertEqual(invalid_data, to_timedelta(
+            invalid_data, errors='ignore'))
+
+        invalid_data = ['apple', '1 days']
+        tm.assert_numpy_array_equal(
+            np.array(invalid_data, dtype=object),
+            to_timedelta(invalid_data, errors='ignore'))
+
+        invalid_data = pd.Index(['apple', '1 days'])
+        tm.assert_index_equal(invalid_data, to_timedelta(
+            invalid_data, errors='ignore'))
+
+        invalid_data = Series(['apple', '1 days'])
+        tm.assert_series_equal(invalid_data, to_timedelta(
+            invalid_data, errors='ignore'))
+
     def test_to_timedelta_via_apply(self):
         # GH 5458
         expected = Series([np.timedelta64(1, 's')])

pandas/tseries/timedeltas.py

@@ -3,7 +3,9 @@
 """
 
 import numpy as np
+import pandas as pd
 import pandas.tslib as tslib
+
 from pandas.types.common import (_ensure_object,
                                  is_integer_dtype,
                                  is_timedelta64_dtype,
@@ -64,37 +66,22 @@ def to_timedelta(arg, unit='ns', box=True, errors='raise', coerce=None):
     """
     unit = _validate_timedelta_unit(unit)
 
-    def _convert_listlike(arg, box, unit, name=None):
-
-        if isinstance(arg, (list, tuple)) or not hasattr(arg, 'dtype'):
-            arg = np.array(list(arg), dtype='O')
-
-        # these are shortcutable
-        if is_timedelta64_dtype(arg):
-            value = arg.astype('timedelta64[ns]')
-        elif is_integer_dtype(arg):
-            value = arg.astype('timedelta64[{0}]'.format(
-                unit)).astype('timedelta64[ns]', copy=False)
-        else:
-            value = tslib.array_to_timedelta64(_ensure_object(arg),
-                                               unit=unit, errors=errors)
-            value = value.astype('timedelta64[ns]', copy=False)
-
-        if box:
-            from pandas import TimedeltaIndex
-            value = TimedeltaIndex(value, unit='ns', name=name)
-        return value
+    if errors not in ('ignore', 'raise', 'coerce'):
+        raise ValueError("errors must be one of 'ignore', "
+                         "'raise', or 'coerce'}")
 
     if arg is None:
         return arg
     elif isinstance(arg, ABCSeries):
         from pandas import Series
-        values = _convert_listlike(arg._values, box=False, unit=unit)
-        return Series(values, index=arg.index, name=arg.name, dtype='m8[ns]')
+        values = _convert_listlike(arg._values, unit=unit,
+                                   box=False, errors=errors)
+        return Series(values, index=arg.index, name=arg.name)
     elif isinstance(arg, ABCIndexClass):
-        return _convert_listlike(arg, box=box, unit=unit, name=arg.name)
+        return _convert_listlike(arg, unit=unit, box=box,
+                                 errors=errors, name=arg.name)
     elif is_list_like(arg) and getattr(arg, 'ndim', 1) == 1:
-        return _convert_listlike(arg, box=box, unit=unit)
+        return _convert_listlike(arg, unit=unit, box=box, errors=errors)
     elif getattr(arg, 'ndim', 1) > 1:
         raise TypeError('arg must be a string, timedelta, list, tuple, '
                         '1-d array, or Series')
@@ -142,13 +129,55 @@ def _validate_timedelta_unit(arg):
 
 
 def _coerce_scalar_to_timedelta_type(r, unit='ns', box=True, errors='raise'):
-    """
-    convert strings to timedelta; coerce to Timedelta (if box), else
-    np.timedelta64
-    """
+    """Convert string 'r' to a timedelta object."""
+
+    try:
+        result = tslib.convert_to_timedelta64(r, unit)
+    except ValueError:
+        if errors == 'raise':
+            raise
+        elif errors == 'ignore':
+            return r
+
+        # coerce
+        result = pd.NaT
 
-    result = tslib.convert_to_timedelta(r, unit, errors)
     if box:
         result = tslib.Timedelta(result)
-
     return result
+
+
+def _convert_listlike(arg, unit='ns', box=True, errors='raise', name=None):
+    """Convert a list of objects to a timedelta index object."""
+
+    if isinstance(arg, (list, tuple)) or not hasattr(arg, 'dtype'):
+        arg = np.array(list(arg), dtype='O')
+
+    # these are shortcut-able
+    if is_timedelta64_dtype(arg):
+        value = arg.astype('timedelta64[ns]')
+    elif is_integer_dtype(arg):
+        value = arg.astype('timedelta64[{0}]'.format(
+            unit)).astype('timedelta64[ns]', copy=False)
+    else:
+        try:
+            value = tslib.array_to_timedelta64(_ensure_object(arg),
+                                               unit=unit, errors=errors)
+            value = value.astype('timedelta64[ns]', copy=False)
+        except ValueError:
+            if errors == 'ignore':
+                return arg
+            else:
+                # This else-block accounts for the cases when errors='raise'
+                # and errors='coerce'. If errors == 'raise', these errors
+                # should be raised. If errors == 'coerce', we shouldn't
+                # expect any errors to be raised, since all parsing errors
+                # cause coercion to pd.NaT. However, if an error / bug is
+                # introduced that causes an Exception to be raised, we would
+                # like to surface it.
+                raise
+
+    if box:
+        from pandas import TimedeltaIndex
+        value = TimedeltaIndex(value, unit='ns', name=name)
+    return value

pandas/tslib.pxd

@@ -1,7 +1,7 @@
 from numpy cimport ndarray, int64_t
 
 cdef convert_to_tsobject(object, object, object, bint, bint)
-cdef convert_to_timedelta64(object, object, object)
+cpdef convert_to_timedelta64(object, object)
 cpdef object maybe_get_tz(object)
 cdef bint _is_utc(object)
 cdef bint _is_tzlocal(object)