Merge branch 'master' into dev/yigoshev-value_counts

docs/supported_apis/series_supported.rst

@@ -130,7 +130,7 @@ the related section on `Defaulting to pandas`_.
 +-----------------------------+---------------------------------+
 | ``dropna``                  | Y                               |
 +-----------------------------+---------------------------------+
-| ``dt``                      | D                               |
+| ``dt``                      | Y                               |
 +-----------------------------+---------------------------------+
 | ``dtype``                   | Y                               |
 +-----------------------------+---------------------------------+

docs/supported_apis/utilities_supported.rst

@@ -19,7 +19,7 @@ default to pandas.
 +---------------------------+---------------------------------+----------------------------------------------------+
 | `pd.eval`_                | Y                               |                                                    |
 +---------------------------+---------------------------------+----------------------------------------------------+
-| `pd.unique`_              | D                               |                                                    |
+| `pd.unique`_              | Y                               |                                                    |
 +---------------------------+---------------------------------+----------------------------------------------------+
 | ``pd.value_counts``       | Y                               |                                                    |
 +---------------------------+---------------------------------+----------------------------------------------------+

modin/backends/pandas/query_compiler.py

@@ -60,6 +60,61 @@ def str_op_builder(df, *args, **kwargs):
     return str_op_builder
 
 
+def _dt_prop_map(property_name):
+    """
+    Create a function that call property of property `dt` of the series.
+
+    Parameters
+    ----------
+    property_name
+        The property of `dt`, which will be applied.
+
+    Returns
+    -------
+        A callable function to be applied in the partitions
+
+    Notes
+    -----
+    This applies non-callable properties of `Series.dt`.
+    """
+
+    def dt_op_builder(df, *args, **kwargs):
+        prop_val = getattr(df.squeeze().dt, property_name)
+        if isinstance(prop_val, pandas.Series):
+            return prop_val.to_frame()
+        else:
+            return pandas.DataFrame([prop_val])
+
+    return dt_op_builder
+
+
+def _dt_func_map(func_name):
+    """
+    Create a function that call method of property `dt` of the series.
+
+    Parameters
+    ----------
+    func_name
+        The method of `dt`, which will be applied.
+
+    Returns
+    -------
+        A callable function to be applied in the partitions
+
+    Notes
+    -----
+    This applies callable methods of `Series.dt`.
+    """
+
+    def dt_op_builder(df, *args, **kwargs):
+        dt_s = df.squeeze().dt
+        return pandas.DataFrame(
+            getattr(pandas.Series.dt, func_name)(dt_s, *args, **kwargs)
+        )
+
+    return dt_op_builder
+
+
 def copy_df_for_func(func):
     """Create a function that copies the dataframe, likely because `func` is inplace.
 
@@ -506,6 +561,50 @@ def unique(self):
         )
         return self.__constructor__(new_modin_frame)
 
+    # Dt map partitions operations
+
+    dt_date = MapFunction.register(_dt_prop_map("date"))
+    dt_time = MapFunction.register(_dt_prop_map("time"))
+    dt_timetz = MapFunction.register(_dt_prop_map("timetz"))
+    dt_year = MapFunction.register(_dt_prop_map("year"))
+    dt_month = MapFunction.register(_dt_prop_map("month"))
+    dt_day = MapFunction.register(_dt_prop_map("day"))
+    dt_hour = MapFunction.register(_dt_prop_map("hour"))
+    dt_minute = MapFunction.register(_dt_prop_map("minute"))
+    dt_second = MapFunction.register(_dt_prop_map("second"))
+    dt_microsecond = MapFunction.register(_dt_prop_map("microsecond"))
+    dt_nanosecond = MapFunction.register(_dt_prop_map("nanosecond"))
+    dt_week = MapFunction.register(_dt_prop_map("week"))
+    dt_weekofyear = MapFunction.register(_dt_prop_map("weekofyear"))
+    dt_dayofweek = MapFunction.register(_dt_prop_map("dayofweek"))
+    dt_weekday = MapFunction.register(_dt_prop_map("weekday"))
+    dt_dayofyear = MapFunction.register(_dt_prop_map("dayofyear"))
+    dt_quarter = MapFunction.register(_dt_prop_map("quarter"))
+    dt_is_month_start = MapFunction.register(_dt_prop_map("is_month_start"))
+    dt_is_month_end = MapFunction.register(_dt_prop_map("is_month_end"))
+    dt_is_quarter_start = MapFunction.register(_dt_prop_map("is_quarter_start"))
+    dt_is_quarter_end = MapFunction.register(_dt_prop_map("is_quarter_end"))
+    dt_is_year_start = MapFunction.register(_dt_prop_map("is_year_start"))
+    dt_is_year_end = MapFunction.register(_dt_prop_map("is_year_end"))
+    dt_is_leap_year = MapFunction.register(_dt_prop_map("is_leap_year"))
+    dt_daysinmonth = MapFunction.register(_dt_prop_map("daysinmonth"))
+    dt_days_in_month = MapFunction.register(_dt_prop_map("days_in_month"))
+    dt_tz = MapFunction.register(_dt_prop_map("tz"))
+    dt_freq = MapFunction.register(_dt_prop_map("freq"))
+    dt_to_period = MapFunction.register(_dt_func_map("to_period"))
+    dt_to_pydatetime = MapFunction.register(_dt_func_map("to_pydatetime"))
+    dt_tz_localize = MapFunction.register(_dt_func_map("tz_localize"))
+    dt_tz_convert = MapFunction.register(_dt_func_map("tz_convert"))
+    dt_normalize = MapFunction.register(_dt_func_map("normalize"))
+    dt_strftime = MapFunction.register(_dt_func_map("strftime"))
+    dt_round = MapFunction.register(_dt_func_map("round"))
+    dt_floor = MapFunction.register(_dt_func_map("floor"))
+    dt_ceil = MapFunction.register(_dt_func_map("ceil"))
+    dt_month_name = MapFunction.register(_dt_func_map("month_name"))
+    dt_day_name = MapFunction.register(_dt_func_map("day_name"))
+
+    # END Dt map partitions operations
+
     def astype(self, col_dtypes, **kwargs):
         """Converts columns dtypes to given dtypes.
 

modin/pandas/__init__.py

@@ -27,7 +27,6 @@
 
 from pandas import (
     eval,
-    unique,
     cut,
     to_numeric,
     factorize,
@@ -131,6 +130,7 @@
     notnull,
     notna,
     pivot,
+    unique,
     value_counts,
 )
 from .plotting import Plotting as plotting
@@ -283,7 +283,6 @@ def import_pandas(*args):
     "json_normalize",
     "concat",
     "eval",
-    "unique",
     "cut",
     "to_numeric",
     "factorize",
@@ -362,6 +361,7 @@ def import_pandas(*args):
     "notnull",
     "notna",
     "pivot",
+    "unique",
     "value_counts",
     "datetime",
     "NamedAgg",

modin/pandas/general.py

@@ -220,6 +220,21 @@ def pivot(data, index=None, columns=None, values=None):
     return data.pivot(index=index, columns=columns, values=values)
 
 
+def unique(values):
+    """
+    Return unique values of input data.
+
+    Uniques are returned in order of appearance. Hash table-based unique,
+    therefore does NOT sort.
+
+    Returns
+    -------
+    ndarray
+        The unique values returned as a NumPy array.
+    """
+    return Series(values).unique()
+
+
 def value_counts(
     values, sort=True, ascending=False, normalize=False, bins=None, dropna=True,
 ):

modin/pandas/series.py

@@ -1338,7 +1338,7 @@ def cat(self):
 
     @property
     def dt(self):
-        return self._default_to_pandas(pandas.Series.dt)
+        return DatetimeProperties(self)
 
     @property
     def dtype(self):
@@ -1426,6 +1426,163 @@ def _to_pandas(self):
         return series
 
 
+class DatetimeProperties(object):
+    def __init__(self, series):
+        self._series = series
+        self._query_compiler = series._query_compiler
+
+    @property
+    def date(self):
+        return Series(query_compiler=self._query_compiler.dt_date())
+
+    @property
+    def time(self):
+        return Series(query_compiler=self._query_compiler.dt_time())
+
+    @property
+    def timetz(self):
+        return Series(query_compiler=self._query_compiler.dt_timetz())
+
+    @property
+    def year(self):
+        return Series(query_compiler=self._query_compiler.dt_year())
+
+    @property
+    def month(self):
+        return Series(query_compiler=self._query_compiler.dt_month())
+
+    @property
+    def day(self):
+        return Series(query_compiler=self._query_compiler.dt_day())
+
+    @property
+    def hour(self):
+        return Series(query_compiler=self._query_compiler.dt_hour())
+
+    @property
+    def minute(self):
+        return Series(query_compiler=self._query_compiler.dt_minute())
+
+    @property
+    def second(self):
+        return Series(query_compiler=self._query_compiler.dt_second())
+
+    @property
+    def microsecond(self):
+        return Series(query_compiler=self._query_compiler.dt_microsecond())
+
+    @property
+    def nanosecond(self):
+        return Series(query_compiler=self._query_compiler.dt_nanosecond())
+
+    @property
+    def week(self):
+        return Series(query_compiler=self._query_compiler.dt_week())
+
+    @property
+    def weekofyear(self):
+        return Series(query_compiler=self._query_compiler.dt_weekofyear())
+
+    @property
+    def dayofweek(self):
+        return Series(query_compiler=self._query_compiler.dt_dayofweek())
+
+    @property
+    def weekday(self):
+        return Series(query_compiler=self._query_compiler.dt_weekday())
+
+    @property
+    def dayofyear(self):
+        return Series(query_compiler=self._query_compiler.dt_dayofyear())
+
+    @property
+    def quarter(self):
+        return Series(query_compiler=self._query_compiler.dt_quarter())
+
+    @property
+    def is_month_start(self):
+        return Series(query_compiler=self._query_compiler.dt_is_month_start())
+
+    @property
+    def is_month_end(self):
+        return Series(query_compiler=self._query_compiler.dt_is_month_end())
+
+    @property
+    def is_quarter_start(self):
+        return Series(query_compiler=self._query_compiler.dt_is_quarter_start())
+
+    @property
+    def is_quarter_end(self):
+        return Series(query_compiler=self._query_compiler.dt_is_quarter_end())
+
+    @property
+    def is_year_start(self):
+        return Series(query_compiler=self._query_compiler.dt_is_year_start())
+
+    @property
+    def is_year_end(self):
+        return Series(query_compiler=self._query_compiler.dt_is_year_end())
+
+    @property
+    def is_leap_year(self):
+        return Series(query_compiler=self._query_compiler.dt_is_leap_year())
+
+    @property
+    def daysinmonth(self):
+        return Series(query_compiler=self._query_compiler.dt_daysinmonth())
+
+    @property
+    def days_in_month(self):
+        return Series(query_compiler=self._query_compiler.dt_days_in_month())
+
+    @property
+    def tz(self):
+        return self._query_compiler.dt_tz().to_pandas().squeeze()
+
+    @property
+    def freq(self):
+        return self._query_compiler.dt_freq().to_pandas().squeeze()
+
+    def to_period(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_to_period(*args, **kwargs))
+
+    def to_pydatetime(self):
+        return Series(query_compiler=self._query_compiler.dt_to_pydatetime()).to_numpy()
+
+    def tz_localize(self, *args, **kwargs):
+        return Series(
+            query_compiler=self._query_compiler.dt_tz_localize(*args, **kwargs)
+        )
+
+    def tz_convert(self, *args, **kwargs):
+        return Series(
+            query_compiler=self._query_compiler.dt_tz_convert(*args, **kwargs)
+        )
+
+    def normalize(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_normalize(*args, **kwargs))
+
+    def strftime(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_strftime(*args, **kwargs))
+
+    def round(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_round(*args, **kwargs))
+
+    def floor(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_floor(*args, **kwargs))
+
+    def ceil(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_ceil(*args, **kwargs))
+
+    def month_name(self, *args, **kwargs):
+        return Series(
+            query_compiler=self._query_compiler.dt_month_name(*args, **kwargs)
+        )
+
+    def day_name(self, *args, **kwargs):
+        return Series(query_compiler=self._query_compiler.dt_day_name(*args, **kwargs))
+
+
 class StringMethods(object):
     def __init__(self, series):
         # Check if dtypes is objects