red-arrow 16.0.0 → 17.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 021e0bb1714a4dbaf54622b66dcea24682f78d2045497a8109148620f6090e83
-  data.tar.gz: 8c61eedc5f5f94011b3c7a37aed40583c8b94f36e2e91f86a1d8c76fbe768a5c
+  metadata.gz: 219ff5131490e19062c8ec0813d2478182b0e4804fd3b574cfae56c684fe14a2
+  data.tar.gz: e3a579210e075a7bf2bae488ab59d4c764efdddf6a684d2ec3bf85d121f0aca3
 SHA512:
-  metadata.gz: 73e30cc705260b29ab70be6f98ad30075a5d6c07878c4af25b1650d48d9d7b5df6a7884eaf0db98f4e20cfa8ca0e3a25c49c625fe64f5c8aa4bd529dd10c9b8c
-  data.tar.gz: 30c65035dcf46283e7e64533604c16dcba2dd1b54e00bb8b7dfd10e9ecb2c8440f32a5cfc6adf24bcd7bee5a159aeba110d180f1bdd2d5746f28f66cabb696a8
+  metadata.gz: ed735f057c60826172842e0f9a24e23ac950e8a972bb5e3b9d0ab0b5dc16ee87e4996a5f0552cb97e70f1e134c882372110c488945ad8d12e548c55234e40d5b
+  data.tar.gz: ae524564569d6889e2cde7616f5e937490c30ae55a4c3e362af4af0ef4444f2062fb3910164c55ba29cb4d34c3f158abb478d9d8332024bf85a68beff8f069e8

data/lib/arrow/field-containable.rb

@@ -29,7 +29,7 @@ module Arrow
         return nil if index < 0 or index >= n_fields
         get_field(index)
       else
-        message = "field name or index must be String, Symbol or Integer"
+        message = +"field name or index must be String, Symbol or Integer"
         message << ": <#{name_or_index.inspect}>"
         raise ArgumentError, message
       end

data/lib/arrow/table-formatter.rb

@@ -24,7 +24,8 @@ module Arrow
       attr_reader :head_values
       attr_reader :tail_values
       attr_reader :sample_values
-      def initialize(column, head_values, tail_values)
+      def initialize(table_formatter, column, head_values, tail_values)
+        @table_formatter = table_formatter
         @column = column
         @head_values = head_values
         @tail_values = tail_values
@@ -36,6 +37,15 @@ module Arrow
         @data_type ||= @column.data_type
       end
 
+      def formatted_data_type_name
+        @formatted_data_type_name ||= "(#{data_type.name})"
+      end
+
+      def aligned_data_type_name
+        @aligned_data_type_name ||=
+          "%*s" % [aligned_name.size, formatted_data_type_name]
+      end
+
       def name
         @name ||= @column.name
       end
@@ -63,7 +73,7 @@ module Arrow
             formatted_value = format_value(value[field_name], field_value_width)
             "#{formatted_name}: #{formatted_value}"
           end
-          formatted = "{"
+          formatted = +"{"
           formatted << formatted_values.join(", ")
           formatted << "}"
           "%-*s" % [width, formatted]
@@ -90,9 +100,16 @@ module Arrow
       end
 
       def format_aligned_name(name, data_type, sample_values)
+        if @table_formatter.show_column_type?
+          min_width = formatted_data_type_name.size
+        else
+          min_width = 0
+        end
         case data_type
         when TimestampDataType
-          "%*s" % [::Time.now.iso8601.size, name]
+          width = ::Time.now.iso8601.size
+          width = min_width if width < min_width
+          "%*s" % [width, name]
         when IntegerDataType
           have_null = false
           have_negative = false
@@ -118,9 +135,12 @@ module Arrow
           end
           width += 1 if have_negative # Need "-"
           width = [width, FORMATTED_NULL.size].max if have_null
+          width = min_width if width < min_width
           "%*s" % [width, name]
         when FloatDataType, DoubleDataType
-          "%*s" % [FLOAT_N_DIGITS, name]
+          width = FLOAT_N_DIGITS
+          width = min_width if width < min_width
+          "%*s" % [width, name]
         when StructDataType
           field_widths = data_type.fields.collect do |field|
             field_value_width = compute_field_value_width(field, sample_values)
@@ -130,9 +150,11 @@ module Arrow
           if field_widths.size > 0
             width += (", ".size * (field_widths.size - 1))
           end
+          width = min_width if width < min_width
           "%*s" % [width, name]
         else
-          name
+          width = min_width
+          "%*s" % [width, name]
         end
       end
     end
@@ -143,7 +165,7 @@ module Arrow
     end
 
     def format
-      text = ""
+      text = +""
       n_rows = @table.n_rows
       border = @options[:border] || 10
 
@@ -159,7 +181,7 @@ module Arrow
         else
           tail_values = []
         end
-        ColumnFormatter.new(column, head_values, tail_values)
+        ColumnFormatter.new(self, column, head_values, tail_values)
       end
 
       format_header(text, column_formatters)
@@ -186,5 +208,9 @@ module Arrow
 
       text
     end
+
+    def show_column_type?
+      @options.fetch(:show_column_type, true)
+    end
   end
 end

data/lib/arrow/table-list-formatter.rb

@@ -27,9 +27,9 @@ module Arrow
         text << ("=" * 20 + " #{start_offset + nth_row} " + "=" * 20 + "\n")
         row.each_with_index do |column_value, nth_column|
           column_formatter = column_formatters[nth_column]
-          formatted_name = column_formatter.name
-          formatted_value = column_formatter.format_value(column_value)
-          text << "#{formatted_name}: #{formatted_value}\n"
+          text << column_formatter.name
+          text << "(#{column_formatter.data_type.name})" if show_column_type?
+          text << ": #{column_formatter.format_value(column_value)}\n"
         end
       end
     end

data/lib/arrow/table-table-formatter.rb

@@ -26,6 +26,13 @@ module Arrow
         text << "\t"
         text << column_formatter.aligned_name
       end
+      if show_column_type?
+        text << "\n"
+        column_formatters.each do |column_formatter|
+          text << "\t"
+          text << column_formatter.aligned_data_type_name
+        end
+      end
       text << "\n"
     end
 

data/lib/arrow/version.rb

@@ -16,7 +16,7 @@
 # under the License.
 
 module Arrow
-  VERSION = "16.0.0"
+  VERSION = "17.0.0"
 
   module Version
     numbers, TAG = VERSION.split("-")

data/test/test-csv-loader.rb

@@ -27,80 +27,88 @@ class CSVLoaderTest < Test::Unit::TestCase
     test("String: data: with header") do
       data = fixture_path("with-header-float.csv").read
       assert_equal(<<-TABLE, load_csv(data).to_s)
-	name	     score
-0	alice	 10.100000
-1	bob 	 29.200000
-2	chris	 -1.300000
+	  name	     score
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("String: data: without header") do
       data = fixture_path("without-header-float.csv").read
       assert_equal(<<-TABLE, load_csv(data).to_s)
-	0	         1
-0	alice	 10.100000
-1	bob	 29.200000
-2	chris	 -1.300000
+	     0	         1
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("String: path: with header") do
       path = fixture_path("with-header-float.csv").to_s
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	name	     score
-0	alice	 10.100000
-1	bob 	 29.200000
-2	chris	 -1.300000
+	  name	     score
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("String: path: without header") do
       path = fixture_path("without-header-float.csv").to_s
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	0	         1
-0	alice	 10.100000
-1	bob	 29.200000
-2	chris	 -1.300000
+	     0	         1
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("Pathname: with header") do
       path = fixture_path("with-header-float.csv")
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	name	     score
-0	alice	 10.100000
-1	bob 	 29.200000
-2	chris	 -1.300000
+	  name	     score
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("Pathname: without header") do
       path = fixture_path("without-header-float.csv")
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	0	         1
-0	alice	 10.100000
-1	bob	 29.200000
-2	chris	 -1.300000
+	     0	         1
+	(utf8)	  (double)
+0	alice 	 10.100000
+1	bob   	 29.200000
+2	chris 	 -1.300000
       TABLE
     end
 
     test("null: with double quote") do
       path = fixture_path("null-with-double-quote.csv").to_s
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	name	 score
-0	alice	    10
-1	bob 	(null)
-2	chris	    -1
+	  name	 score
+	(utf8)	(int8)
+0	alice 	    10
+1	bob   	(null)
+2	chris 	    -1
       TABLE
     end
 
     test("null: without double quote") do
       path = fixture_path("null-without-double-quote.csv").to_s
       assert_equal(<<-TABLE, load_csv(path).to_s)
-	name	 score
-0	alice	    10
-1	bob 	(null)
-2	chris	    -1
+	  name	 score
+	(utf8)	(int8)
+0	alice 	    10
+1	bob   	(null)
+2	chris 	    -1
       TABLE
     end
 

data/test/test-group.rb

@@ -43,6 +43,7 @@ class GroupTest < Test::Unit::TestCase
       table = Arrow::Table.new(raw_table)
       assert_equal(<<-TABLE, table.group(:time).count.to_s)
 	                     time	count(int)
+	              (timestamp)	   (int64)
 0	#{time_values[0].iso8601}	         1
 1	#{time_values[1].iso8601}	         1
       TABLE
@@ -53,6 +54,7 @@ class GroupTest < Test::Unit::TestCase
     test("single") do
       assert_equal(<<-TABLE, @table.group(:group_key1).count.to_s)
 	group_key1	count(group_key2)	count(int)	count(uint)	count(float)	count(string)
+	   (uint8)	          (int64)	   (int64)	    (int64)	     (int64)	      (int64)
 0	         1	                2	         2	          1	           1	            2
 1	         2	                1	         0	          1	           1	            1
 2	         3	                3	         3	          3	           3	            2
@@ -62,6 +64,7 @@ class GroupTest < Test::Unit::TestCase
     test("multiple") do
       assert_equal(<<-TABLE, @table.group(:group_key1, :group_key2).count.to_s)
 	group_key1	group_key2	count(int)	count(uint)	count(float)	count(string)
+	   (uint8)	   (uint8)	   (int64)	    (int64)	     (int64)	      (int64)
 0	         1	         1	         2	          1	           1	            2
 1	         2	         1	         0	          1	           1	            1
 2	         3	         1	         1	          1	           1	            0
@@ -73,6 +76,7 @@ class GroupTest < Test::Unit::TestCase
       group = @table.group(:group_key1, :group_key2)
       assert_equal(<<-TABLE, group.count(:int, :uint).to_s)
 	group_key1	group_key2	count(int)	count(uint)
+	   (uint8)	   (uint8)	   (int64)	    (int64)
 0	         1	         1	         2	          1
 1	         2	         1	         0	          1
 2	         3	         1	         1	          1
@@ -85,6 +89,7 @@ class GroupTest < Test::Unit::TestCase
     test("single") do
       assert_equal(<<-TABLE, @table.group(:group_key1).sum.to_s)
 	group_key1	sum(group_key2)	sum(int)	sum(uint)	sum(float)
+	   (uint8)	       (uint64)	 (int64)	 (uint64)	  (double)
 0	         1	              2	      -3	        1	  2.200000
 1	         2	              1	  (null)	        3	  3.300000
 2	         3	              5	     -15	       15	 16.500000
@@ -94,6 +99,7 @@ class GroupTest < Test::Unit::TestCase
     test("multiple") do
       assert_equal(<<-TABLE, @table.group(:group_key1, :group_key2).sum.to_s)
 	group_key1	group_key2	sum(int)	sum(uint)	sum(float)
+	   (uint8)	   (uint8)	 (int64)	 (uint64)	  (double)
 0	         1	         1	      -3	        1	  2.200000
 1	         2	         1	  (null)	        3	  3.300000
 2	         3	         1	      -4	        4	  4.400000
@@ -106,6 +112,7 @@ class GroupTest < Test::Unit::TestCase
     test("single") do
       assert_equal(<<-TABLE, @table.group(:group_key1).mean.to_s)
 	group_key1	mean(group_key2)	 mean(int)	mean(uint)	mean(float)
+	   (uint8)	        (double)	  (double)	  (double)	   (double)
 0	         1	        1.000000	 -1.500000	  1.000000	   2.200000
 1	         2	        1.000000	    (null)	  3.000000	   3.300000
 2	         3	        1.666667	 -5.000000	  5.000000	   5.500000
@@ -115,6 +122,7 @@ class GroupTest < Test::Unit::TestCase
     test("multiple") do
       assert_equal(<<-TABLE, @table.group(:group_key1, :group_key2).mean.to_s)
 	group_key1	group_key2	 mean(int)	mean(uint)	mean(float)
+	   (uint8)	   (uint8)	  (double)	  (double)	   (double)
 0	         1	         1	 -1.500000	  1.000000	   2.200000
 1	         2	         1	    (null)	  3.000000	   3.300000
 2	         3	         1	 -4.000000	  4.000000	   4.400000
@@ -127,6 +135,7 @@ class GroupTest < Test::Unit::TestCase
     test("single") do
       assert_equal(<<-TABLE, @table.group(:group_key1).min.to_s)
 	group_key1	min(group_key2)	min(int)	min(uint)	min(float)
+	   (uint8)	        (uint8)	 (int32)	 (uint32)	   (float)
 0	         1	              1	      -2	        1	  2.200000
 1	         2	              1	  (null)	        3	  3.300000
 2	         3	              1	      -6	        4	  4.400000
@@ -136,6 +145,7 @@ class GroupTest < Test::Unit::TestCase
     test("multiple") do
       assert_equal(<<-TABLE, @table.group(:group_key1, :group_key2).min.to_s)
 	group_key1	group_key2	min(int)	min(uint)	min(float)
+	   (uint8)	   (uint8)	 (int32)	 (uint32)	   (float)
 0	         1	         1	      -2	        1	  2.200000
 1	         2	         1	  (null)	        3	  3.300000
 2	         3	         1	      -4	        4	  4.400000
@@ -148,6 +158,7 @@ class GroupTest < Test::Unit::TestCase
     test("single") do
       assert_equal(<<-TABLE, @table.group(:group_key1).max.to_s)
 	group_key1	max(group_key2)	max(int)	max(uint)	max(float)
+	   (uint8)	        (uint8)	 (int32)	 (uint32)	   (float)
 0	         1	              1	      -1	        1	  2.200000
 1	         2	              1	  (null)	        3	  3.300000
 2	         3	              2	      -4	        6	  6.600000
@@ -157,6 +168,7 @@ class GroupTest < Test::Unit::TestCase
     test("multiple") do
       assert_equal(<<-TABLE, @table.group(:group_key1, :group_key2).max.to_s)
 	group_key1	group_key2	max(int)	max(uint)	max(float)
+	   (uint8)	   (uint8)	 (int32)	 (uint32)	   (float)
 0	         1	         1	      -1	        1	  2.200000
 1	         2	         1	  (null)	        3	  3.300000
 2	         3	         1	      -4	        4	  4.400000
@@ -170,6 +182,7 @@ class GroupTest < Test::Unit::TestCase
       group = @table.group(:group_key1, :group_key2)
       assert_equal(<<-TABLE, group.aggregate("count(int)", "sum(uint)").to_s)
 	group_key1	group_key2	count(int)	sum(uint)
+	   (uint8)	   (uint8)	   (int64)	 (uint64)
 0	         1	         1	         2	        1
 1	         2	         1	         0	        3
 2	         3	         1	         1	        4

data/test/test-schema.rb

@@ -95,7 +95,7 @@ class SchemaTest < Test::Unit::TestCase
 
       test("[invalid]") do
         invalid = []
-        message = "field name or index must be String, Symbol or Integer"
+        message = +"field name or index must be String, Symbol or Integer"
         message << ": <#{invalid.inspect}>"
         assert_raise(ArgumentError.new(message)) do
           @schema[invalid]