ENH: adapt stdlib _strptime module to Cython for 2.5x speedup

Author: wesm

pandas/tseries/tests/test_timeseries.py

@@ -675,6 +675,18 @@ def test_to_datetime_default(self):
         xp = datetime(2001, 1, 1)
         self.assert_(rs, xp)
 
+    def test_to_datetime_format(self):
+        values = ['1/1/2000', '1/2/2000', '1/3/2000']
+
+        def _parse_format(fmt, values):
+            return to_datetime([datetime.strptime(x, fmt)
+                                for x in values])
+
+        for fmt in ['%d/%m/%Y', '%m/%d/%Y']:
+            result = to_datetime(values, format=fmt)
+            expected = _parse_format(fmt, values)
+            self.assert_(result.equals(expected))
+
     def test_to_datetime_on_datetime64_series(self):
         # #2699
         s = Series(date_range('1/1/2000', periods=10))
@@ -2445,7 +2457,7 @@ def test_intercept_astype_object(self):
         # Work around NumPy 1.6 bugs
         #result = self.series.astype(object)
         #result2 = self.series.astype('O')
- 
+
         expected = Series(self.series, dtype=object)
 
         #assert_series_equal(result, expected)

pandas/tseries/tools.py

@@ -48,7 +48,8 @@ def _maybe_get_tz(tz):
     return tz
 
 
-def to_datetime(arg, errors='ignore', dayfirst=False, utc=None, box=True):
+def to_datetime(arg, errors='ignore', dayfirst=False, utc=None, box=True,
+                format=None):
     """
     Convert argument to datetime
 
@@ -72,8 +73,11 @@ def _convert_f(arg):
         arg = com._ensure_object(arg)
 
         try:
-            result = tslib.array_to_datetime(arg, raise_=errors == 'raise',
-                                             utc=utc, dayfirst=dayfirst)
+            if format is not None:
+                result = tslib.array_strptime(arg, format)
+            else:
+                result = tslib.array_to_datetime(arg, raise_=errors == 'raise',
+                                                 utc=utc, dayfirst=dayfirst)
             if com.is_datetime64_dtype(result) and box:
                 result = DatetimeIndex(result, tz='utc' if utc else None)
             return result