charty 0.2.10 → 0.2.11

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: d885d476eadfadd3f76f138707ffaf6166eec006b4528fb19c3acce0b97b1c85
-  data.tar.gz: 0630f23f19b65177f3b6dc238717a81a019f9dfa40a40c143d71dcc5c5f760f0
+  metadata.gz: eac796ba17aeb9a82a312d7099c898a2350abaf34c3fad80ea1afd1483de1797
+  data.tar.gz: 8001339bd86377ab3693cc2eeb0a826c04c4ea4e08d58e3a10e99849a273cd58
 SHA512:
-  metadata.gz: 535a69b8d794ccbd30b4328f3188c8a8b50c850cfd84285f79b31162296a9d981313e561ea1089a68be3368b7eb1c83ccc3cb2ff511b8f79aef77e544e261bbe
-  data.tar.gz: 804de2468433db38b5381bcbccab4c28e84c7592bce5c5a98a539ce99c589449759e65ffe7f2e808bd3c5cb5e002ee2219219111561f66f590dbb0ac2c0757bc
+  metadata.gz: 3679334bb56834f7fb1067878c60e7601375e05d6adf63e566eb2cd3847a7ce1423bb27d0104f666854c8343840706a8e14dcc12bc8420ae062b8144d3f6ebc2
+  data.tar.gz: f4bbec96ae815cd758b67a993355fe5e42a10b91d7ef351e63570ec84fa618e65e4c2df4006f2d3431738c8b82b6a5dcf74d79a9b4ce005f89f18fd2ae88d31d

data/lib/charty/backends/unicode_plot.rb

@@ -20,12 +20,12 @@ module Charty
         @figure = {
           type: :bar,
           bar_pos: bar_pos,
-          values: values
+          values: values,
         }
       end
 
-      def box_plot(plot_data, positions, **kwargs)
-        @figure = { type: :box, data: plot_data }
+      def box_plot(plot_data, positions, orient:, **kwargs)
+        @figure = { type: :box, data: plot_data, orient: orient }
       end
 
       def set_xlabel(label)
@@ -57,7 +57,12 @@ module Charty
                 when :bar
                   ::UnicodePlot.barplot(@layout[:xtick_labels], @figure[:values], xlabel: @layout[:xlabel])
                 when :box
-                  ::UnicodePlot.boxplot(@layout[:xtick_labels], @figure[:data], xlabel: @layout[:xlabel])
+                  xlabel = if @figure[:orient] == :v
+                             @layout[:ylabel]
+                           else
+                             @layout[:xlabel]
+                           end
+                  ::UnicodePlot.boxplot(@layout[:xtick_labels], @figure[:data], xlabel: xlabel)
                 end
         sio = StringIO.new
         class << sio

data/lib/charty/index.rb

@@ -199,6 +199,15 @@ module Charty
       end
     end
 
+    def loc(key)
+      case values
+      when Pandas::Index
+        values.get_loc(key)
+      else
+        super
+      end
+    end
+
     def union(other)
       other = PandasIndex.try_convert(other)
       # NOTE: Using `sort=False` in pandas.Index#union does not produce pandas.RangeIndex.

data/lib/charty/statistics.rb

@@ -109,9 +109,9 @@ module Charty
       a = a.sort
       n = a.size
       q.map do |x|
-        x = n * (x / 100.0)
+        x = (n-1) * (x / 100.0)
         i = x.floor
-        if i == n-1
+        if i == x
           a[i]
         else
           t = x - i

data/lib/charty/table_adapters/arrow_adapter.rb

@@ -0,0 +1,53 @@
+module Charty
+  module TableAdapters
+    class ArrowAdapter < BaseAdapter
+      TableAdapters.register(:arrow, self)
+
+      def self.supported?(data)
+        defined?(Arrow::Table) && data.is_a?(Arrow::Table)
+      end
+
+      def initialize(data)
+        @data = data
+        @column_names = @data.columns.map(&:name)
+        self.columns = Index.new(@column_names)
+        self.index = RangeIndex.new(0 ... length)
+      end
+
+      attr_reader :data
+
+      def length
+        @data.n_rows
+      end
+
+      def column_length
+        @column_names.length
+      end
+
+      def compare_data_equality(other)
+        case other
+        when ArrowAdapter
+          data == other.data
+        else
+          super
+        end
+      end
+
+      def [](row, column)
+        if row
+          @data[column][row]
+        else
+          case column
+          when Array
+            Table.new(@data.select_columns(*column))
+          else
+            column_data = @data[column]
+            Vector.new(column_data.data.combine,
+                       index: index,
+                       name: column_data.name)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/charty/table_adapters/hash_adapter.rb

@@ -75,6 +75,8 @@ module Charty
                 arrays[i] << record[key]
               end
             end
+          when Vector
+            arrays = data
           when self.class.method(:array?)
             unsupported_data_format unless data.all?(&self.class.method(:array?))
             arrays = data.map(&:to_a).transpose
@@ -121,20 +123,15 @@ module Charty
         index = union_indexes(*indexes)
 
         arrays = arrays.map do |array|
-          case array
-          when Charty::Vector
-            array.data
-          when Hash
-            raise NotImplementedError
-          when self.class.method(:array?)
-            array
-          else
-            Array.try_convert(array)
-          end
+          Vector.try_convert(array)
         end
 
         columns = generate_column_names(arrays.length, columns)
 
+        arrays.zip(columns) do |array, column|
+          array.name = column.to_sym if array.name.to_s != column
+        end
+
         return arrays, columns, index
       end
 
@@ -199,9 +196,7 @@ module Charty
                         else
                           @data[str_key]
                         end
-          # FIXME: Here column_data need to be dupped to
-          # prevent to overwrite the name of Pandas::Series
-          Vector.new(column_data.dup, index: index, name: column)
+          Vector.new(column_data, index: index, name: column)
         end
       end
 
@@ -216,15 +211,10 @@ module Charty
         end
 
         orig_values = values
-        case values
-        when Charty::Vector
-          values = values.data
-        else
-          values = Array.try_convert(values)
-        end
+        values = Vector.try_convert(values)
         if values.nil?
           raise ArgumentError,
-                "`values` must be convertible to Array"
+                "`values` must be convertible to Charty::Vector"
         end
 
         if values.length != self.length

data/lib/charty/table_adapters.rb

@@ -23,3 +23,4 @@ require_relative 'table_adapters/daru_adapter'
 require_relative 'table_adapters/active_record_adapter'
 require_relative 'table_adapters/nmatrix_adapter'
 require_relative 'table_adapters/pandas_adapter'
+require_relative 'table_adapters/arrow_adapter'

data/lib/charty/vector.rb

@@ -49,7 +49,7 @@ module Charty
 
     alias completecases notnull
 
-    def_delegators :adapter, :mean, :stdev
+    def_delegators :adapter, :mean, :stdev, :percentile
 
     def_delegators :adapter, :scale, :scale_inverse
 

data/lib/charty/vector_adapters/array_adapter.rb

@@ -30,7 +30,7 @@ module Charty
       include NameSupport
       include IndexSupport
 
-      def_delegators :data, :values_at
+      def_delegators :data, :values_at, :to_a
 
       def where(mask)
         masked_data, masked_index = where_in_array(mask)

data/lib/charty/vector_adapters/arrow_adapter.rb

@@ -0,0 +1,156 @@
+module Charty
+  module VectorAdapters
+    class ArrowAdapter < BaseAdapter
+      VectorAdapters.register(:arrow, self)
+
+      include Enumerable
+      include NameSupport
+      include IndexSupport
+
+      def self.supported?(data)
+        (defined?(Arrow::Array) && data.is_a?(Arrow::Array)) ||
+          (defined?(Arrow::ChunkedArray) && data.is_a?(Arrow::ChunkedArray))
+      end
+
+      def initialize(data)
+        @data = check_data(data)
+        self.index = index || RangeIndex.new(0 ... length)
+      end
+
+      def size
+        @data.length
+      end
+
+      def empty?
+        @data.length.zero?
+      end
+
+      def where(mask)
+        mask = check_mask_vector(mask)
+        mask_data = mask.data
+        unless mask_data.is_a?(Arrow::BooleanArray)
+          mask_data = mask.to_a
+          mask_data = mask_data.map(&:nonzero?) if mask_data[0].is_a?(Integer)
+          mask_data = Arrow::BooleanArray.new(mask_data)
+        end
+        masked_data = @data.filter(mask_data)
+        masked_index = []
+        mask_data.to_a.each_with_index do |boolean, i|
+          masked_index << index[i] if boolean
+        end
+        Vector.new(masked_data, index: masked_index, name: name)
+      end
+
+      def boolean?
+        case @data
+        when Arrow::BooleanArray
+          true
+        when Arrow::ChunkedArray
+          @data.value_data_type.is_a?(Arrow::BooleanDataType)
+        else
+          false
+        end
+      end
+
+      def numeric?
+        case @data
+        when Arrow::NumericArray
+          true
+        when Arrow::ChunkedArray
+          @data.value_data_type.is_a?(Arrow::NumericDataType)
+        else
+          false
+        end
+      end
+
+      def categorical?
+        case @data
+        when Arrow::StringArray, Arrow::DictionaryArray
+          true
+        when Arrow::ChunkedArray
+          case @data.value_data_type
+          when Arrow::StringArray, Arrow::DictionaryDataType
+            true
+          else
+            false
+          end
+        else
+          false
+        end
+      end
+
+      def categories
+        if @data.respond_to?(:dictionary)
+          dictionary = @data.dictionary
+        else
+          dictionary = @data.dictionary_encode.dictionary
+        end
+        dictionary.to_a
+      end
+
+      def unique_values
+        @data.unique.to_a
+      end
+
+      def group_by(grouper)
+        grouper = Vector.new(grouper) unless grouper.is_a?(Vector)
+        group_keys = grouper.unique_values
+        grouper_data = grouper.data
+        unless grouper_data.is_a?(Arrow::Array)
+          grouper_data = Arrow::Array.new(grouper.to_a)
+        end
+        equal = Arrow::Function.find("equal")
+        group_keys.map { |key|
+          if key.nil?
+            target_vector = Vector.new([nil] * @data.n_nulls)
+          else
+            mask = equal.execute([grouper_data, key]).value
+            target_vector = Vector.new(@data.filter(mask))
+          end
+          [key, target_vector]
+        }.to_h
+      end
+
+      def drop_na
+        if @data.n_nulls.zero?
+          Vector.new(@data, index: index, name: name)
+        else
+          data_without_null =
+            Arrow::Function.find("drop_null").execute([@data]).value
+          Vector.new(data_without_null)
+        end
+      end
+
+      def eq(val)
+        mask = Arrow::Function.find("equal").execute([@data, val]).value
+        Vector.new(mask, index: index, name: name)
+      end
+
+      def notnull
+        if @data.n_nulls.zero?
+          mask = Arrow::BooleanArray.new([true] * @data.length)
+        else
+          mask = Arrow::BooleanArray.new(@data.length,
+                                         @data.null_bitmap,
+                                         nil,
+                                         0)
+        end
+        Vector.new(mask, index: index, name: name)
+      end
+
+      def mean
+        @data.mean
+      end
+
+      def stdev(population: false)
+        options = Arrow::VarianceOptions.new
+        if population
+          options.ddof = 0
+        else
+          options.ddof = 1
+        end
+        Arrow::Function.find("stddev").execute([@data], options).value.value
+      end
+    end
+  end
+end

data/lib/charty/vector_adapters/daru_adapter.rb

@@ -34,12 +34,8 @@ module Charty
         case other
         when DaruVectorAdapter
           data == other.data
-        when ArrayAdapter
-          data.to_a == other.data
-        when NArrayAdapter, NMatrixAdapter, NumpyAdapter, PandasSeriesAdapter
-          other.compare_data_equality(self)
         else
-          data == other.data.to_a
+          to_a == other.to_a
         end
       end
 
@@ -156,7 +152,8 @@ module Charty
       end
 
       def percentile(q)
-        data.linear_percentile(q)
+        a = data.reject_values(*Daru::MISSING_VALUES).to_a
+        Statistics.percentile(a, q)
       end
     end
   end

data/lib/charty/vector_adapters/narray_adapter.rb

@@ -22,13 +22,22 @@ module Charty
         when NumpyAdapter, PandasSeriesAdapter
           other.compare_data_equality(self)
         else
-          data == other.data.to_a
+          data == other.to_a
         end
       end
 
       include NameSupport
       include IndexSupport
 
+      def to_a
+        case data
+        when Numo::Bit
+          map {|bit| bit == 1 }
+        else
+          super
+        end
+      end
+
       def where(mask)
         mask = check_mask_vector(mask)
         case mask.data

data/lib/charty/vector_adapters/nmatrix_adapter.rb

@@ -19,12 +19,8 @@ module Charty
         case other
         when NMatrixAdapter
           data == other.data
-        when ArrayAdapter, DaruVectorAdapter
-          data.to_a == other.data.to_a
-        when NArrayAdapter, NumpyAdapter, PandasSeriesAdapter
-          other.compare_data_equality(self)
         else
-          data == other.data.to_a
+          data.to_a == other.data.to_a
         end
       end
 

data/lib/charty/vector_adapters/numpy_adapter.rb

@@ -163,6 +163,10 @@ module Charty
       def stdev(population: false)
         Numpy.std(data, ddof: population ? 0 : 1)
       end
+
+      def percentile(q)
+        Numpy.nanpercentile(data, q)
+      end
     end
   end
 end

data/lib/charty/vector_adapters/vector_adapter.rb

@@ -0,0 +1,62 @@
+module Charty
+  module VectorAdapters
+    class VectorAdapter < BaseAdapter
+      VectorAdapters.register(:vector, self)
+
+      extend Forwardable
+      include Enumerable
+
+      def self.supported?(data)
+        data.is_a?(Vector)
+      end
+
+      def initialize(data, index: nil)
+        data = check_data(data)
+        @data = reduce_nested_vector(data)
+        self.index = index || RangeIndex.new(0 ... length)
+      end
+
+      include NameSupport
+      include IndexSupport
+
+      def_delegators :data,
+                     :boolean?,
+                     :categorical?,
+                     :categories,
+                     :drop_na,
+                     :each,
+                     :eq,
+                     :first_nonnil,
+                     :group_by,
+                     :notnull,
+                     :numeric?,
+                     :to_a,
+                     :uniq,
+                     :unique_values,
+                     :values_at,
+                     :where
+
+      def compare_data_equality(other)
+        if other.is_a?(self.class)
+          other = reduce_nested_vector(other.data).adapter
+        end
+        if other.is_a?(self.class)
+          @data.adapter.data == other.data
+        elsif @data.adapter.respond_to?(:compare_data_equality)
+          @data.adapter.compare_data_equality(other)
+        elsif other.respond_to?(:compare_data_equality)
+          other.compare_data_equality(@data.adapter)
+        else
+          @data.adapter.to_a == other.to_a
+        end
+      end
+
+      private def reduce_nested_vector(vector)
+        while vector.adapter.is_a?(self.class)
+          vector = vector.adapter.data
+        end
+        vector
+      end
+    end
+  end
+end

data/lib/charty/vector_adapters.rb

@@ -100,6 +100,10 @@ module Charty
         Statistics.stdev(data, population: population)
       end
 
+      def percentile(q)
+        Statistics.percentile(data, q)
+      end
+
       def log_scale(method)
         Charty::Vector.new(
           self.map {|x| Math.log10(x) },
@@ -196,8 +200,10 @@ module Charty
 end
 
 require_relative "vector_adapters/array_adapter"
+require_relative "vector_adapters/arrow_adapter"
 require_relative "vector_adapters/daru_adapter"
 require_relative "vector_adapters/narray_adapter"
 require_relative "vector_adapters/nmatrix_adapter"
 require_relative "vector_adapters/numpy_adapter"
 require_relative "vector_adapters/pandas_adapter"
+require_relative "vector_adapters/vector_adapter"

data/lib/charty/version.rb

@@ -1,5 +1,5 @@
 module Charty
-  VERSION = "0.2.10"
+  VERSION = "0.2.11"
 
   module Version
     numbers, TAG = VERSION.split("-")

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: charty
 version: !ruby/object:Gem::Version
-  version: 0.2.10
+  version: 0.2.11
 platform: ruby
 authors:
 - youchan
@@ -10,7 +10,7 @@ authors:
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2021-08-18 00:00:00.000000000 Z
+date: 2021-09-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: red-colors
@@ -336,6 +336,7 @@ files:
 - lib/charty/table.rb
 - lib/charty/table_adapters.rb
 - lib/charty/table_adapters/active_record_adapter.rb
+- lib/charty/table_adapters/arrow_adapter.rb
 - lib/charty/table_adapters/base_adapter.rb
 - lib/charty/table_adapters/daru_adapter.rb
 - lib/charty/table_adapters/datasets_adapter.rb
@@ -347,11 +348,13 @@ files:
 - lib/charty/vector.rb
 - lib/charty/vector_adapters.rb
 - lib/charty/vector_adapters/array_adapter.rb
+- lib/charty/vector_adapters/arrow_adapter.rb
 - lib/charty/vector_adapters/daru_adapter.rb
 - lib/charty/vector_adapters/narray_adapter.rb
 - lib/charty/vector_adapters/nmatrix_adapter.rb
 - lib/charty/vector_adapters/numpy_adapter.rb
 - lib/charty/vector_adapters/pandas_adapter.rb
+- lib/charty/vector_adapters/vector_adapter.rb
 - lib/charty/version.rb
 homepage: https://github.com/red-data-tools/charty
 licenses: