Performance improvement in Series.to_sql and DataFrame.to_sql (SQLiteTable): processing time arrays can be up to 65% faster ! (related to pandas-dev#44764)

Author: Sylvain MARIE

doc/source/whatsnew/v1.5.0.rst

@@ -387,7 +387,9 @@ Performance improvements
 - Performance improvement in :class:`DataFrame` and :class:`Series` constructors for extension dtype scalars (:issue:`45854`)
 - Performance improvement in :class:`BusinessHour`, ``str`` and ``repr`` are now 4 times faster ! (related to :issue:`44764`)
 - Performance improvement in :class:`DatetimeArray` and :class:`DatetimeIndex`: string formatting is now up to 80% faster (as fast as default) when one of the default strftime formats ``"%Y-%m-%d %H:%M:%S"`` or ``"%Y-%m-%d %H:%M:%S.%f"`` is used. (:issue:`44764`)
-- Performance improvement in :meth:`Series.to_excel` and :meth:`DataFrame.to_excel` (:class:`ExcelFormatter`): processing dates can be up to 4% faster. (related to :issue:`44764`)
+- Performance improvement in :meth:`Series.to_excel` and :meth:`DataFrame.to_excel` (:class:`ExcelFormatter`): processing date cells can be up to 4% faster. (related to
+:issue:`44764`)
+- Performance improvement in :meth:`Series.to_sql` and :meth:`DataFrame.to_sql` (:class:`SQLiteTable`): processing time arrays can be up to 65% faster ! (related to :issue:`44764`)
 
 .. ---------------------------------------------------------------------------
 .. _whatsnew_150.bug_fixes:

pandas/io/sql.py

@@ -1836,7 +1836,20 @@ def __init__(self, *args, **kwargs) -> None:
 
         # this will transform time(12,34,56,789) into '12:34:56.000789'
         # (this is what sqlalchemy does)
-        sqlite3.register_adapter(time, lambda _: _.strftime("%H:%M:%S.%f"))
+        def _adapt(dt):
+            if dt.tzinfo is None:
+                # This is faster than strftime
+                return "%02d:%02d:%02d.%06d" % (
+                    dt.hour,
+                    dt.minute,
+                    dt.second,
+                    dt.microsecond,
+                )
+            else:
+                # Slow TODO we can probably find a way to use string formatting too
+                return dt.strftime("%H:%M:%S.%f")
+
+        sqlite3.register_adapter(time, _adapt)
         super().__init__(*args, **kwargs)
 
     def sql_schema(self):