polars-df 0.11.0 → 0.12.0

data/lib/polars/series.rb

@@ -1017,18 +1017,18 @@ module Polars
     #   s.rle.struct.unnest
     #   # =>
     #   # shape: (6, 2)
-    #   # ┌─────────┬────────┐
-    #   # │ lengths ┆ values │
-    #   # │ ---     ┆ ---    │
-    #   # │ i32     ┆ i64    │
-    #   # ╞═════════╪════════╡
-    #   # │ 2       ┆ 1      │
-    #   # │ 1       ┆ 2      │
-    #   # │ 1       ┆ 1      │
-    #   # │ 1       ┆ null   │
-    #   # │ 1       ┆ 1      │
-    #   # │ 2       ┆ 3      │
-    #   # └─────────┴────────┘
+    #   # ┌─────┬───────┐
+    #   # │ len ┆ value │
+    #   # │ --- ┆ ---   │
+    #   # │ u32 ┆ i64   │
+    #   # ╞═════╪═══════╡
+    #   # │ 2   ┆ 1     │
+    #   # │ 1   ┆ 2     │
+    #   # │ 1   ┆ 1     │
+    #   # │ 1   ┆ null  │
+    #   # │ 1   ┆ 1     │
+    #   # │ 2   ┆ 3     │
+    #   # └─────┴───────┘
     def rle
       super
     end
@@ -1082,8 +1082,24 @@ module Polars
     #   # │ 2   ┆ 2      │
     #   # │ 3   ┆ 1      │
     #   # └─────┴────────┘
-    def value_counts(sort: false)
-      Utils.wrap_df(_s.value_counts(sort))
+    def value_counts(
+      sort: false,
+      parallel: false,
+      name: nil,
+      normalize: false
+    )
+      if name.nil?
+        if normalize
+          name = "proportion"
+        else
+          name = "count"
+        end
+      end
+      DataFrame._from_rbdf(
+        self._s.value_counts(
+          sort, parallel, name, normalize
+        )
+      )
     end
 
     # Return a count of the unique values in the order of appearance.
@@ -1362,7 +1378,7 @@ module Polars
     #   #         2
     #   # ]
     def limit(n = 10)
-      to_frame.select(Utils.col(name).limit(n)).to_series
+      to_frame.select(F.col(name).limit(n)).to_series
     end
 
     # Get a slice of this Series.
@@ -1386,7 +1402,7 @@ module Polars
     #   #         3
     #   # ]
     def slice(offset, length = nil)
-      super
+      self.class._from_rbseries(_s.slice(offset, length))
     end
 
     # Append a Series to this one.
@@ -1494,7 +1510,7 @@ module Polars
     #   #         2
     #   # ]
     def head(n = 10)
-      to_frame.select(Utils.col(name).head(n)).to_series
+      to_frame.select(F.col(name).head(n)).to_series
     end
 
     # Get the last `n` rows.
@@ -1515,7 +1531,7 @@ module Polars
     #   #         3
     #   # ]
     def tail(n = 10)
-      to_frame.select(Utils.col(name).tail(n)).to_series
+      to_frame.select(F.col(name).tail(n)).to_series
     end
 
     # Take every nth value in the Series and return as new Series.
@@ -1594,7 +1610,7 @@ module Polars
     #   #         4
     #   #         3
     #   # ]
-    def top_k(k: 5, nulls_last: false, multithreaded: true)
+    def top_k(k: 5)
       super
     end
 
@@ -1616,7 +1632,7 @@ module Polars
     #   #         2
     #   #         3
     #   # ]
-    def bottom_k(k: 5, nulls_last: false, multithreaded: true)
+    def bottom_k(k: 5)
       super
     end
 
@@ -2054,11 +2070,12 @@ module Polars
     #
     # @param other [Series]
     #   Series to compare with.
+    # @param strict [Boolean]
+    #   Require data types to match.
+    # @param check_names [Boolean]
+    #   Require names to match.
     # @param null_equal [Boolean]
     #   Consider null values as equal.
-    # @param strict [Boolean]
-    #   Don't allow different numerical dtypes, e.g. comparing `:u32` with a
-    #   `:i64` will return `false`.
     #
     # @return [Boolean]
     #
@@ -2069,8 +2086,8 @@ module Polars
     #   # => true
     #   s.equals(s2)
     #   # => false
-    def equals(other, null_equal: false, strict: false)
-      _s.equals(other._s, null_equal, strict)
+    def equals(other, strict: false, check_names: false, null_equal: false)
+      _s.equals(other._s, strict, check_names, null_equal)
     end
     alias_method :series_equal, :equals
 
@@ -3004,16 +3021,7 @@ module Polars
       min_periods: nil,
       center: false
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_min(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center
-          )
-        )
-        .to_series
+      super
     end
 
     # Apply a rolling max (moving max) over the values in this array.
@@ -3054,16 +3062,7 @@ module Polars
       min_periods: nil,
       center: false
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_max(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center
-          )
-        )
-        .to_series
+      super
     end
 
     # Apply a rolling mean (moving mean) over the values in this array.
@@ -3104,16 +3103,7 @@ module Polars
       min_periods: nil,
       center: false
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_mean(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center
-          )
-        )
-        .to_series
+      super
     end
 
     # Apply a rolling sum (moving sum) over the values in this array.
@@ -3154,16 +3144,7 @@ module Polars
       min_periods: nil,
       center: false
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_sum(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center
-          )
-        )
-        .to_series
+      super
     end
 
     # Compute a rolling std dev.
@@ -3204,21 +3185,9 @@ module Polars
       weights: nil,
       min_periods: nil,
       center: false,
-      ddof: 1,
-      warn_if_unsorted: true
+      ddof: 1
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_std(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center,
-            ddof: ddof,
-            warn_if_unsorted: warn_if_unsorted
-          )
-        )
-        .to_series
+      super
     end
 
     # Compute a rolling variance.
@@ -3259,21 +3228,9 @@ module Polars
       weights: nil,
       min_periods: nil,
       center: false,
-      ddof: 1,
-      warn_if_unsorted: true
+      ddof: 1
     )
-      to_frame
-        .select(
-          Polars.col(name).rolling_var(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center,
-            ddof: ddof,
-            warn_if_unsorted: warn_if_unsorted
-          )
-        )
-        .to_series
+      super
     end
 
     # def rolling_apply
@@ -3312,24 +3269,9 @@ module Polars
       window_size,
       weights: nil,
       min_periods: nil,
-      center: false,
-      warn_if_unsorted: true
+      center: false
     )
-      if min_periods.nil?
-        min_periods = window_size
-      end
-
-      to_frame
-        .select(
-          Polars.col(name).rolling_median(
-            window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center,
-            warn_if_unsorted: warn_if_unsorted
-          )
-        )
-        .to_series
+      super
     end
 
     # Compute a rolling quantile.
@@ -3385,26 +3327,9 @@ module Polars
       window_size: 2,
       weights: nil,
       min_periods: nil,
-      center: false,
-      warn_if_unsorted: true
+      center: false
     )
-      if min_periods.nil?
-        min_periods = window_size
-      end
-
-      to_frame
-        .select(
-          Polars.col(name).rolling_quantile(
-            quantile,
-            interpolation: interpolation,
-            window_size: window_size,
-            weights: weights,
-            min_periods: min_periods,
-            center: center,
-            warn_if_unsorted: warn_if_unsorted
-          )
-        )
-        .to_series
+      super
     end
 
     # Compute a rolling skew.
@@ -3806,7 +3731,7 @@ module Polars
     #   #         null
     #   #         10
     #   # ]
-    def clip(min_val, max_val)
+    def clip(min_val = nil, max_val = nil)
       super
     end
 
@@ -3883,33 +3808,20 @@ module Polars
     #   #         200
     #   # ]
     #
-    # @example Passing a mapping with replacements is also supported as syntactic sugar. Specify a default to set all values that were not matched.
+    # @example Passing a mapping with replacements is also supported as syntactic sugar.
     #   mapping = {2 => 100, 3 => 200}
-    #   s.replace(mapping, default: -1)
+    #   s.replace(mapping)
     #   # =>
     #   # shape: (4,)
     #   # Series: '' [i64]
     #   # [
-    #   #         -1
+    #   #         1
     #   #         100
     #   #         100
     #   #         200
     #   # ]
     #
-    # @example The default can be another Series.
-    #   default = Polars::Series.new([2.5, 5.0, 7.5, 10.0])
-    #   s.replace(2, 100, default: default)
-    #   # =>
-    #   # shape: (4,)
-    #   # Series: '' [f64]
-    #   # [
-    #   #         2.5
-    #   #         100.0
-    #   #         100.0
-    #   #         10.0
-    #   # ]
-    #
-    # @example Replacing by values of a different data type sets the return type based on a combination of the `new` data type and either the original data type or the default data type if it was set.
+    # @example The original data type is preserved when replacing by values of a different data type.
     #   s = Polars::Series.new(["x", "y", "z"])
     #   mapping = {"x" => 1, "y" => 2, "z" => 3}
     #   s.replace(mapping)
@@ -3921,28 +3833,6 @@ module Polars
     #   #         "2"
     #   #         "3"
     #   # ]
-    #
-    # @example
-    #   s.replace(mapping, default: nil)
-    #   # =>
-    #   # shape: (3,)
-    #   # Series: '' [i64]
-    #   # [
-    #   #         1
-    #   #         2
-    #   #         3
-    #   # ]
-    #
-    # @example Set the `return_dtype` parameter to control the resulting data type directly.
-    #   s.replace(mapping, return_dtype: Polars::UInt8)
-    #   # =>
-    #   # shape: (3,)
-    #   # Series: '' [u8]
-    #   # [
-    #   #         1
-    #   #         2
-    #   #         3
-    #   # ]
     def replace(old, new = Expr::NO_DEFAULT, default: Expr::NO_DEFAULT, return_dtype: nil)
       super
     end
@@ -3989,7 +3879,8 @@ module Polars
       half_life: nil,
       alpha: nil,
       adjust: true,
-      min_periods: 1
+      min_periods: 1,
+      ignore_nulls: true
     )
       super
     end
@@ -4004,7 +3895,8 @@ module Polars
       alpha: nil,
       adjust: true,
       bias: false,
-      min_periods: 1
+      min_periods: 1,
+      ignore_nulls: true
     )
       super
     end
@@ -4019,7 +3911,8 @@ module Polars
       alpha: nil,
       adjust: true,
       bias: false,
-      min_periods: 1
+      min_periods: 1,
+      ignore_nulls: true
     )
       super
     end
@@ -4218,12 +4111,12 @@ module Polars
       end
 
       if other.is_a?(::Time) && dtype.is_a?(Datetime)
-        ts = Utils._datetime_to_pl_timestamp(other, time_unit)
+        ts = Utils.datetime_to_int(other, time_unit)
         f = ffi_func("#{op}_<>", Int64, _s)
         fail if f.nil?
         return Utils.wrap_s(f.call(ts))
       elsif other.is_a?(::Date) && dtype == Date
-        d = Utils._date_to_pl_date(other)
+        d = Utils.date_to_int(other)
         f = ffi_func("#{op}_<>", Int32, _s)
         fail if f.nil?
         return Utils.wrap_s(f.call(d))
@@ -4309,10 +4202,18 @@ module Polars
           # TODO improve performance
           constructor.call(name, values.to_a, strict)
         end
-      elsif values.shape.length == 2
+      elsif values.shape.sum == 0
         raise Todo
       else
-        raise Todo
+        original_shape = values.shape
+        values = values.reshape(original_shape.inject(&:*))
+        rb_s = numo_to_rbseries(
+          name,
+          values,
+          strict: strict,
+          nan_to_null: nan_to_null
+        )
+        Utils.wrap_s(rb_s).reshape(original_shape)._s
       end
     end
 
@@ -4362,6 +4263,8 @@ module Polars
         end
 
         constructor = polars_type_to_constructor(dtype)
+        # TODO remove
+        strict = false if dtype == Decimal
         rbseries = constructor.call(name, values, strict)
 
         base_type = dtype.is_a?(DataType) ? dtype.class : dtype
@@ -4370,11 +4273,11 @@ module Polars
             rbseries = rbseries.cast(dtype, true)
           end
         end
-        return rbseries
+        rbseries
       elsif dtype == Struct
         struct_schema = dtype.is_a?(Struct) ? dtype.to_schema : nil
         empty = {}
-        return DataFrame.sequence_to_rbdf(
+        DataFrame.sequence_to_rbdf(
           values.map { |v| v.nil? ? empty : v },
           schema: struct_schema,
           orient: "row",
@@ -4399,13 +4302,13 @@ module Polars
           # TODO
           time_unit = nil
 
-          rb_series = RbSeries.new_from_anyvalues(name, values, strict)
+          rb_series = RbSeries.new_from_any_values(name, values, strict)
           if time_unit.nil?
             s = Utils.wrap_s(rb_series)
           else
             s = Utils.wrap_s(rb_series).dt.cast_time_unit(time_unit)
           end
-          return s._s
+          s._s
         elsif defined?(Numo::NArray) && value.is_a?(Numo::NArray) && value.shape.length == 1
           raise Todo
         elsif ruby_dtype == ::Array
@@ -4419,11 +4322,11 @@ module Polars
             end
             return srs
           end
-          return sequence_from_anyvalue_or_object(name, values)
+          sequence_from_anyvalue_or_object(name, values)
         elsif ruby_dtype == Series
-          return RbSeries.new_series_list(name, values.map(&:_s), strict)
+          RbSeries.new_series_list(name, values.map(&:_s), strict)
         elsif ruby_dtype == RbSeries
-          return RbSeries.new_series_list(name, values, strict)
+          RbSeries.new_series_list(name, values, strict)
         else
           constructor =
             if value.is_a?(::String)
@@ -4438,13 +4341,26 @@ module Polars
             else
               rb_type_to_constructor(value.class)
             end
-          constructor.call(name, values, strict)
+
+          construct_series_with_fallbacks(constructor, name, values, dtype, strict: strict)
+        end
+      end
+    end
+
+    def construct_series_with_fallbacks(constructor, name, values, dtype, strict:)
+      begin
+        constructor.call(name, values, strict)
+      rescue
+        if dtype.nil?
+          RbSeries.new_from_any_values(name, values, strict)
+        else
+          RbSeries.new_from_any_values_and_dtype(name, values, dtype, strict)
         end
       end
     end
 
     def sequence_from_anyvalue_or_object(name, values)
-      RbSeries.new_from_anyvalues(name, values, true)
+      RbSeries.new_from_any_values(name, values, true)
     rescue
       RbSeries.new_object(name, values, false)
     end
@@ -4461,10 +4377,10 @@ module Polars
       UInt32 => RbSeries.method(:new_opt_u32),
       UInt64 => RbSeries.method(:new_opt_u64),
       Decimal => RbSeries.method(:new_decimal),
-      Date => RbSeries.method(:new_from_anyvalues),
-      Datetime => RbSeries.method(:new_from_anyvalues),
-      Duration => RbSeries.method(:new_from_anyvalues),
-      Time => RbSeries.method(:new_from_anyvalues),
+      Date => RbSeries.method(:new_from_any_values),
+      Datetime => RbSeries.method(:new_from_any_values),
+      Duration => RbSeries.method(:new_from_any_values),
+      Time => RbSeries.method(:new_from_any_values),
       Boolean => RbSeries.method(:new_opt_bool),
       Utf8 => RbSeries.method(:new_str),
       Object => RbSeries.method(:new_object),