PERF: Speed up Period construction (#50149)

* PERF: Speed up Period construction * Try to fix CI? * Avoid hackiness * debug CI * Modify condition * revert whitespace * fix tests
pandas-dev · Jan 18, 2023 · d2d1797 · d2d1797
1 parent 3ec3ac2
commit d2d1797
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 13 deletions.
diff --git a/doc/source/whatsnew/v2.0.0.rst b/doc/source/whatsnew/v2.0.0.rst
@@ -868,6 +868,7 @@ Performance improvements
 - Performance improvement in :func:`merge` when not merging on the index - the new index will now be :class:`RangeIndex` instead of :class:`Int64Index` (:issue:`49478`)
 - Performance improvement in :meth:`DataFrame.to_dict` and :meth:`Series.to_dict` when using any non-object dtypes (:issue:`46470`)
 - Performance improvement in :func:`read_html` when there are multiple tables (:issue:`49929`)
+- Performance improvement in :class:`Period` constructor when constructing from a string or integer (:issue:`38312`)
 - Performance improvement in :func:`to_datetime` when using ``'%Y%m%d'`` format (:issue:`17410`)
 - Performance improvement in :func:`to_datetime` when format is given or can be inferred (:issue:`50465`)
 - Performance improvement in :func:`read_csv` when passing :func:`to_datetime` lambda-function to ``date_parser`` and inputs have mixed timezone offsetes (:issue:`35296`)

diff --git a/pandas/_libs/tslibs/parsing.pyx b/pandas/_libs/tslibs/parsing.pyx
@@ -377,7 +377,11 @@ def parse_datetime_string_with_reso(
         &out_tzoffset, False
     )
     if not string_to_dts_failed:
-        if out_bestunit == NPY_DATETIMEUNIT.NPY_FR_ns or out_local:
+        timestamp_units = {NPY_DATETIMEUNIT.NPY_FR_ns,
+                           NPY_DATETIMEUNIT.NPY_FR_ps,
+                           NPY_DATETIMEUNIT.NPY_FR_fs,
+                           NPY_DATETIMEUNIT.NPY_FR_as}
+        if out_bestunit in timestamp_units or out_local:
             # TODO: the not-out_local case we could do without Timestamp;
             #  avoid circular import
             from pandas import Timestamp
@@ -389,9 +393,7 @@ def parse_datetime_string_with_reso(
         # Match Timestamp and drop picoseconds, femtoseconds, attoseconds
         # The new resolution will just be nano
         # GH 50417
-        if out_bestunit in {NPY_DATETIMEUNIT.NPY_FR_ps,
-                            NPY_DATETIMEUNIT.NPY_FR_fs,
-                            NPY_DATETIMEUNIT.NPY_FR_as}:
+        if out_bestunit in timestamp_units:
             out_bestunit = NPY_DATETIMEUNIT.NPY_FR_ns
         reso = {
             NPY_DATETIMEUNIT.NPY_FR_Y: "year",

diff --git a/pandas/_libs/tslibs/period.pyx b/pandas/_libs/tslibs/period.pyx
@@ -2592,18 +2592,13 @@ class Period(_Period):
 
             freqstr = freq.rule_code if freq is not None else None
             dt, reso = parse_datetime_string_with_reso(value, freqstr)
-            try:
-                ts = Timestamp(value)
-            except ValueError:
-                nanosecond = 0
-            else:
-                nanosecond = ts.nanosecond
-                if nanosecond != 0:
-                    reso = "nanosecond"
+            if reso == "nanosecond":
+                nanosecond = dt.nanosecond
             if dt is NaT:
                 ordinal = NPY_NAT
 
-            if freq is None:
+            if freq is None and ordinal != NPY_NAT:
+                # Skip NaT, since it doesn't have a resolution
                 try:
                     freq = attrname_to_abbrevs[reso]
                 except KeyError: