Added ISO 8601 Duration string constructor for Timedelta

Author: WillAyd

asv_bench/benchmarks/timedelta.py

@@ -1,9 +1,40 @@
+import datetime
+
 import numpy as np
 import pandas as pd
 
 from pandas import to_timedelta, Timestamp, Timedelta
 
 
+class TimedeltaConstructor(object):
+    goal_time = 0.2
+
+    def time_from_int(self):
+        Timedelta(123456789)
+
+    def time_from_unit(self):
+        Timedelta(1, unit='d')
+
+    def time_from_components(self):
+        Timedelta(days=1, hours=2, minutes=3, seconds=4, milliseconds=5,
+                  microseconds=6, nanoseconds=7)
+
+    def time_from_datetime_timedelta(self):
+        Timedelta(datetime.timedelta(days=1, seconds=1))
+
+    def time_from_np_timedelta(self):
+        Timedelta(np.timedelta64(1, 'ms'))
+
+    def time_from_string(self):
+        Timedelta('1 days')
+
+    def time_from_iso_format(self):
+        Timedelta('P4DT12H30M5S')
+
+    def time_from_missing(self):
+        Timedelta('nat')
+
+
 class ToTimedelta(object):
     goal_time = 0.2
 

doc/source/timedeltas.rst

@@ -62,6 +62,13 @@ You can construct a ``Timedelta`` scalar through various arguments:
    pd.Timedelta('nan')
    pd.Timedelta('nat')
 
+   # ISO 8601 Duration strings
+   pd.Timedelta('P0DT0H1M0S')
+   pd.Timedelta('P0DT0H0M0.000000123S')
+
+.. versionadded:: 0.23.0
+   Added constructor for `ISO 8601 Duration`_ strings
+
 :ref:`DateOffsets<timeseries.offsets>` (``Day, Hour, Minute, Second, Milli, Micro, Nano``) can also be used in construction.
 
 .. ipython:: python

doc/source/whatsnew/v0.23.0.txt

@@ -208,6 +208,7 @@ Other API Changes
 - In :func:`read_excel`, the ``comment`` argument is now exposed as a named parameter (:issue:`18735`)
 - Rearranged the order of keyword arguments in :func:`read_excel()` to align with :func:`read_csv()` (:issue:`16672`)
 - The options ``html.border`` and ``mode.use_inf_as_null`` were deprecated in prior versions, these will now show ``FutureWarning`` rather than a ``DeprecationWarning`` (:issue:`19003`)
+- The default ``Timedelta`` constructor now accepts an ``ISO 8601 Duration`` string as an argument (:issue:`19040`)
 
 .. _whatsnew_0230.deprecations:
 

pandas/_libs/tslibs/timedeltas.pyx

@@ -1,6 +1,7 @@
 # -*- coding: utf-8 -*-
 # cython: profile=False
 import collections
+import re
 
 import sys
 cdef bint PY3 = (sys.version_info[0] >= 3)
@@ -235,6 +236,25 @@ cpdef inline int64_t cast_from_unit(object ts, object unit) except? -1:
     return <int64_t> (base *m) + <int64_t> (frac *m)
 
 
+cpdef match_iso_format(object ts):
+    """
+    Match a provided string against an ISO 8601 pattern, providing a group for
+    each ``Timedelta`` component.
+    """
+    pater = re.compile(r"""P
+                       (?P<days>-?[0-9]*)DT
+                       (?P<hours>[0-9]{1,2})H
+                       (?P<minutes>[0-9]{1,2})M
+                       (?P<seconds>[0-9]{0,2})
+                       (\.
+                       (?P<milliseconds>[0-9]{0,3})
+                       (?P<microseconds>[0-9]{0,3})
+                       (?P<nanoseconds>[0-9]{0,3})
+                       )?S""", re.VERBOSE)
+
+    return re.match(pater, ts)
+
+
 cdef inline parse_timedelta_string(object ts):
     """
     Parse a regular format timedelta string. Return an int64_t (in ns)
@@ -506,6 +526,33 @@ def _binary_op_method_timedeltalike(op, name):
 # ----------------------------------------------------------------------
 # Timedelta Construction
 
+def _value_from_iso_match(match):
+    """
+    Extracts and cleanses the appropriate values from a match object with
+    groups for each component of an ISO 8601 duration
+
+    Parameters
+    ----------
+    match:
+        Regular expression with groups for each component of an ISO 8601
+        duration
+
+    Returns
+    -------
+    int
+        Precision in nanoseconds of matched ISO 8601 duration
+    """
+    match_dict = {k: v for k, v in match.groupdict().items() if v}
+    for comp in ['milliseconds', 'microseconds', 'nanoseconds']:
+        if comp in match_dict:
+            match_dict[comp] ='{:0<3}'.format(match_dict[comp])
+
+    match_dict = {k: int(v) for k, v in match_dict.items()}
+    nano = match_dict.pop('nanoseconds', 0)
+
+    return nano + convert_to_timedelta64(timedelta(**match_dict), 'ns')
+
+
 cdef _to_py_int_float(v):
     # Note: This used to be defined inside Timedelta.__new__
     # but cython will not allow `cdef` functions to be defined dynamically.
@@ -825,7 +872,11 @@ class Timedelta(_Timedelta):
         if isinstance(value, Timedelta):
             value = value.value
         elif is_string_object(value):
-            value = np.timedelta64(parse_timedelta_string(value))
+            if len(value) > 0 and value[0] == 'P':  # hackish
+                match = match_iso_format(value)
+                value = _value_from_iso_match(match)
+            else:
+                value = np.timedelta64(parse_timedelta_string(value))
         elif PyDelta_Check(value):
             value = convert_to_timedelta64(value, 'ns')
         elif is_timedelta64_object(value):

pandas/tests/scalar/test_timedelta.py

@@ -853,3 +853,18 @@ def test_isoformat(self):
         result = Timedelta(minutes=1).isoformat()
         expected = 'P0DT0H1M0S'
         assert result == expected
+
+    @pytest.mark.parametrize('fmt,exp', [
+        ('P6DT0H50M3.010010012S', Timedelta(days=6, minutes=50, seconds=3,
+                                            milliseconds=10, microseconds=10,
+                                            nanoseconds=12)),
+        ('P-6DT0H50M3.010010012S', Timedelta(days=-6, minutes=50, seconds=3,
+                                             milliseconds=10, microseconds=10,
+                                             nanoseconds=12)),
+        ('P4DT12H30M5S', Timedelta(days=4, hours=12, minutes=30, seconds=5)),
+        ('P0DT0H0M0.000000123S', Timedelta(nanoseconds=123)),
+        ('P0DT0H0M0.00001S', Timedelta(microseconds=10)),
+        ('P0DT0H0M0.001S', Timedelta(milliseconds=1)),
+        ('P0DT0H1M0S', Timedelta(minutes=1))])
+    def test_iso_constructor(self, fmt, exp):
+        assert Timedelta(fmt) == exp