Merge pull request #6153 from jreback/index_perf

PERF: perf regression index construction from series (GH6150)

Author: jreback

doc/source/release.rst

@@ -104,6 +104,7 @@ Improvements to existing features
   - add ability to recognize '%p' format code (am/pm) to date parsers when the specific format
     is supplied (:issue:`5361`)
   - Fix performance regression in JSON IO (:issue:`5765`)
+  - performance regression in Index construction from Series (:issue:`6150`)
 
 .. _release.bug_fixes-0.13.1:
 

pandas/core/index.py

@@ -15,7 +15,7 @@
 from pandas.util.decorators import cache_readonly, deprecate
 from pandas.core.common import isnull
 import pandas.core.common as com
-from pandas.core.common import _values_from_object, is_float, is_integer
+from pandas.core.common import _values_from_object, is_float, is_integer, ABCSeries
 from pandas.core.config import get_option
 
 # simplify
@@ -105,7 +105,7 @@ def __new__(cls, data, dtype=None, copy=False, name=None, fastpath=False,
             return subarr
 
         from pandas.tseries.period import PeriodIndex
-        if isinstance(data, np.ndarray):
+        if isinstance(data, (np.ndarray, ABCSeries)):
             if issubclass(data.dtype.type, np.datetime64):
                 from pandas.tseries.index import DatetimeIndex
                 result = DatetimeIndex(data, copy=copy, name=name, **kwargs)
@@ -212,7 +212,7 @@ def _coerce_to_ndarray(cls, data):
                 cls._scalar_data_error(data)
 
             # other iterable of some kind
-            if not isinstance(data, (list, tuple)):
+            if not isinstance(data, (ABCSeries, list, tuple)):
                 data = list(data)
             data = np.asarray(data)
         return data
@@ -767,7 +767,7 @@ def asof(self, label):
         For a sorted index, return the most recent label up to and including
         the passed label. Return NaN if not found
         """
-        if isinstance(label, (Index, np.ndarray)):
+        if isinstance(label, (Index, ABCSeries, np.ndarray)):
             raise TypeError('%s' % type(label))
 
         if label not in self:
@@ -1535,7 +1535,7 @@ def slice_locs(self, start=None, end=None):
 
                     # get_loc will return a boolean array for non_uniques
                     # if we are not monotonic
-                    if isinstance(start_slice, np.ndarray):
+                    if isinstance(start_slice, (ABCSeries, np.ndarray)):
                         raise KeyError("cannot peform a slice operation "
                                        "on a non-unique non-monotonic index")
 

pandas/tseries/index.py

@@ -8,7 +8,7 @@
 
 from pandas.core.common import (isnull, _NS_DTYPE, _INT64_DTYPE,
                                 is_list_like,_values_from_object, _maybe_box,
-                                notnull)
+                                notnull, ABCSeries)
 from pandas.core.index import Index, Int64Index, _Identity
 import pandas.compat as compat
 from pandas.compat import u
@@ -52,9 +52,9 @@ def f(self):
 def _join_i8_wrapper(joinf, with_indexers=True):
     @staticmethod
     def wrapper(left, right):
-        if isinstance(left, np.ndarray):
+        if isinstance(left, (np.ndarray, ABCSeries)):
             left = left.view('i8', type=np.ndarray)
-        if isinstance(right, np.ndarray):
+        if isinstance(right, (np.ndarray, ABCSeries)):
             right = right.view('i8', type=np.ndarray)
         results = joinf(left, right)
         if with_indexers:
@@ -77,7 +77,7 @@ def wrapper(self, other):
             other = DatetimeIndex(other)
         elif isinstance(other, compat.string_types):
             other = _to_m8(other, tz=self.tz)
-        elif not isinstance(other, np.ndarray):
+        elif not isinstance(other, (np.ndarray, ABCSeries)):
             other = _ensure_datetime64(other)
         result = func(other)
 
@@ -195,7 +195,7 @@ def __new__(cls, data=None,
                                  tz=tz, normalize=normalize, closed=closed,
                                  infer_dst=infer_dst)
 
-        if not isinstance(data, np.ndarray):
+        if not isinstance(data, (np.ndarray, ABCSeries)):
             if np.isscalar(data):
                 raise ValueError('DatetimeIndex() must be called with a '
                                  'collection of some kind, %s was passed'
@@ -228,6 +228,8 @@ def __new__(cls, data=None,
                                       yearfirst=yearfirst)
 
         if issubclass(data.dtype.type, np.datetime64):
+            if isinstance(data, ABCSeries):
+                data = data.values
             if isinstance(data, DatetimeIndex):
                 if tz is None:
                     tz = data.tz
@@ -1400,7 +1402,7 @@ def freqstr(self):
     nanosecond = _field_accessor('nanosecond', 'ns')
     weekofyear = _field_accessor('weekofyear', 'woy')
     week = weekofyear
-    dayofweek = _field_accessor('dayofweek', 'dow', 
+    dayofweek = _field_accessor('dayofweek', 'dow',
                                 "The day of the week with Monday=0, Sunday=6")
     weekday = dayofweek
     dayofyear = _field_accessor('dayofyear', 'doy')

vb_suite/ctors.py

@@ -29,3 +29,11 @@
 """
 
 ctor_index_array_string = Benchmark('Index(data)', setup=setup)
+
+# index constructors
+setup = common_setup + """
+s = Series([Timestamp('20110101'),Timestamp('20120101'),Timestamp('20130101')]*1000)
+"""
+index_from_series_ctor = Benchmark('Index(s)', setup=setup)
+
+dtindex_from_series_ctor = Benchmark('DatetimeIndex(s)', setup=setup)