parqueteur 1.1.1 → 1.3.2

data/lib/parqueteur/converter.rb

@@ -2,9 +2,7 @@
 
 module Parqueteur
   class Converter
-    attr_reader :schema
-
-    DEFAULT_BATCH_SIZE = 25
+    DEFAULT_BATCH_SIZE = 10
 
     def self.inline(&block)
       Class.new(self, &block)
@@ -26,106 +24,137 @@ module Parqueteur
       transforms << (method_name || block)
     end
 
-    def self.convert(input, output: nil, batch_size: DEFAULT_BATCH_SIZE)
-      converter = new(input, bactch_size: batch_size)
-      if !output.nil?
-        converter.write(output)
-      else
-        converter.to_blob
+    def self.convert(input, **kwargs)
+      new(input, **kwargs).to_io
+    end
+
+    def self.convert_to(input, output_path, **kwargs)
+      converter = new(input, **kwargs)
+      converter.write(output_path)
+    end
+
+    # @param [Enumerable] An enumerable object
+    # @option [Symbol] compression - :gzip
+    def initialize(input, **kwargs)
+      @input = Parqueteur::Input.from(input)
+      @batch_size = kwargs.fetch(:batch_size, DEFAULT_BATCH_SIZE)
+      @compression = kwargs.fetch(:compression, nil)&.to_sym
+    end
+
+    def split(size, batch_size: nil, compression: nil)
+      Enumerator.new do |arr|
+        options = {
+          batch_size: batch_size || @batch_size,
+          compression: compression || @compression
+        }
+        @input.each_slice(size) do |records|
+          local_converter = self.class.new(records, **options)
+          file = local_converter.to_tmpfile
+          arr << file
+          file.close
+          file.unlink
+        end
       end
     end
 
-    def initialize(input, options = {})
-      @input = Parqueteur::Input.from(input, options)
-      @batch_size = options.fetch(:batch_size, DEFAULT_BATCH_SIZE)
-    end
-
-    def write(output)
-      case output
-      when :io
-        to_io
-      when String
-        to_arrow_table.save(output)
-      when StringIO, IO
-        buffer = Arrow::ResizableBuffer.new(0)
-        to_arrow_table.save(buffer, format: :parquet)
-        output.write(buffer.data.to_s)
-        output.rewind
-        output
-      else
-        raise ArgumentError, "unsupported output: #{output.class}, accepted: String (filename), IO, StringIO"
+    def split_by_io(size, batch_size: nil, compression: nil)
+      Enumerator.new do |arr|
+        options = {
+          batch_size: batch_size || @batch_size,
+          compression: compression || @compression
+        }
+        @input.each_slice(size) do |records|
+          local_converter = self.class.new(records, **options)
+          arr << local_converter.to_io
+        end
+      end
+    end
+
+    def write(path, batch_size: nil, compression: nil)
+      compression = @compression if compression.nil?
+      batch_size = @batch_size if batch_size.nil?
+      arrow_schema = self.class.columns.arrow_schema
+      writer_properties = Parquet::WriterProperties.new
+      if !compression.nil? && compression != false
+        writer_properties.set_compression(compression)
+      end
+
+      Arrow::FileOutputStream.open(path, false) do |output|
+        Parquet::ArrowFileWriter.open(arrow_schema, output, writer_properties) do |writer|
+          @input.each_slice(batch_size) do |records|
+            arrow_table = build_arrow_table(records)
+            writer.write_table(arrow_table, 1024)
+          end
+        end
       end
+
+      true
+    end
+
+    def to_tmpfile(options = {})
+      tempfile = Tempfile.new
+      tempfile.binmode
+      write(tempfile.path, **options)
+      tempfile.rewind
+      tempfile
     end
 
-    def to_s
-      inspect
+    def to_io(options = {})
+      tmpfile = to_tmpfile(options)
+      strio = StringIO.new(tmpfile.read)
+      tmpfile.close
+      tmpfile.unlink
+      strio
     end
 
-    def to_io
-      write(StringIO.new)
+    def to_arrow_table(options = {})
+      file = to_tmpfile(options)
+      table = Arrow::Table.load(file.path, format: :parquet)
+      file.close
+      file.unlink
+      table
     end
 
-    def to_blob
-      write(StringIO.new).read
+    def to_blob(options = {})
+      to_tmpfile(options).read
     end
 
-    def to_arrow_table
+    private
+
+    def build_arrow_table(records)
       transforms = self.class.transforms
 
-      chunks = self.class.columns.each_with_object({}) do |column, hash|
+      values = self.class.columns.each_with_object({}) do |column, hash|
         hash[column.name] = []
       end
 
-      items_count = 0
-      @input.each_slice(@batch_size) do |items|
-        values = self.class.columns.each_with_object({}) do |column, hash|
-          hash[column.name] = []
-        end
-
-        items.each do |item|
-          if transforms.length > 0
-            transforms.each do |transform|
-              item = \
-                if transform.is_a?(Symbol)
-                  __send__(transform, item)
-                else
-                  transform.call(item)
-                end
-            end
+      records.each do |item|
+        if transforms.length > 0
+          transforms.each do |transform|
+            item = \
+              if transform.is_a?(Symbol)
+                __send__(transform, item)
+              else
+                transform.call(item)
+              end
           end
+        end
 
-          values.each_key do |value_key|
-            if item.key?(value_key)
-              values[value_key] << item[value_key]
-            else
-              values[value_key] << nil
-            end
+        values.each_key do |value_key|
+          if item.key?(value_key)
+            values[value_key] << item[value_key]
+          else
+            values[value_key] << nil
           end
         end
+      end
 
-        values.each_with_object(chunks) do |item, hash|
+      Arrow::Table.new(
+        values.each_with_object({}) do |item, hash|
           column = self.class.columns.find(item[0])
-          hash[item[0]].push(
-            column.type.build_value_array(item[1])
-          )
+          hash[item[0]] = column.type.build_value_array(item[1])
         end
-
-        items_count += items.length
-      end
-
-      if items_count > 0
-        Arrow::Table.new(
-          chunks.transform_values! do |value|
-            Arrow::ChunkedArray.new(value)
-          end
-        )
-      else
-        Arrow::Table.new(
-          self.class.columns.each_with_object({}) do |column, hash|
-            hash[column.name] = column.type.build_value_array([])
-          end
-        )
-      end
+      )
     end
   end
 end

data/lib/parqueteur/input.rb

@@ -4,40 +4,25 @@ module Parqueteur
   class Input
     include Enumerable
 
-    def self.from(arg, options = {})
-      new(
-        case arg
-        when String
-          if File.exist?(arg)
-            File.new(arg, 'r')
-          else
-            arg.split("\n")
-          end
-        when Enumerable
-          arg
-        end,
-        options
-      )
+    def self.from(arg)
+      return arg if arg.is_a?(self)
+
+      new(arg)
     end
 
-    def initialize(source, options = {})
+    def initialize(source)
+      unless source.is_a?(Enumerable)
+        raise ArgumentError, 'Enumerable object expected'
+      end
+
       @source = source
-      @options = options
     end
 
     def each(&block)
-      case @source
-      when File
-        if @options.fetch(:json_newlines, true) == true
-          @source.each_line do |line|
-            yield(JSON.parse(line.strip))
-          end
-        else
-          JSON.parse(@source.read).each(&block)
-        end
-        @source.rewind
-      when Enumerable
+      if block_given?
         @source.each(&block)
+      else
+        @source.to_enum(:each)
       end
     end
   end

data/lib/parqueteur/type_resolver.rb

@@ -7,14 +7,24 @@ module Parqueteur
     def self.registered_types
       @registered_types ||= {
         array: Parqueteur::Types::ArrayType,
+        bigdecimal: Parqueteur::Types::Decimal256Type,
         bigint: Parqueteur::Types::Int64Type,
         boolean: Parqueteur::Types::BooleanType,
+        date: Parqueteur::Types::Date32Type,
+        date32: Parqueteur::Types::Date32Type,
+        date64: Parqueteur::Types::Date64Type,
+        decimal: Parqueteur::Types::Decimal128Type,
+        decimal128: Parqueteur::Types::Decimal128Type,
+        decimal256: Parqueteur::Types::Decimal256Type,
         int32: Parqueteur::Types::Int32Type,
         int64: Parqueteur::Types::Int64Type,
         integer: Parqueteur::Types::Int32Type,
         map: Parqueteur::Types::MapType,
         string: Parqueteur::Types::StringType,
         struct: Parqueteur::Types::StructType,
+        time: Parqueteur::Types::Time32Type,
+        time32: Parqueteur::Types::Time32Type,
+        time64: Parqueteur::Types::Time64Type,
         timestamp: Parqueteur::Types::TimestampType
       }
     end

data/lib/parqueteur/types/date32_type.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Date32Type < Parqueteur::Type
+      def build_value_array(values)
+        Arrow::Date32ArrayBuilder.build(values)
+      end
+
+      def arrow_type_builder
+        Arrow::Date32DataType.new
+      end
+    end
+  end
+end

data/lib/parqueteur/types/date64_type.rb

@@ -0,0 +1,15 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Date64Type < Parqueteur::Type
+      def build_value_array(values)
+        Arrow::Date64ArrayBuilder.build([values])
+      end
+
+      def arrow_type_builder
+        Arrow::Date64DataType.new
+      end
+    end
+  end
+end

data/lib/parqueteur/types/decimal128_type.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Decimal128Type < Parqueteur::Type
+      def initialize(options = {}, &block)
+        @scale = options.fetch(:scale)
+        @precision = options.fetch(:precision)
+        @format_str = "%.#{@scale}f"
+        super(options, &block)
+      end
+
+      def build_value_array(values)
+        Arrow::Decimal128ArrayBuilder.build(
+          @arrow_type,
+          values.map do |value|
+            Arrow::Decimal128.new(format(@format_str, BigDecimal(value)))
+          end
+        )
+      end
+
+      def arrow_type_builder
+        Arrow::Decimal128DataType.new(
+          @precision, @scale
+        )
+      end
+    end
+  end
+end

data/lib/parqueteur/types/decimal256_type.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Decimal256Type < Parqueteur::Type
+      def initialize(options = {}, &block)
+        @scale = options.fetch(:scale)
+        @precision = options.fetch(:precision)
+        @format_str = "%.#{@scale}f"
+        super(options, &block)
+      end
+
+      def build_value_array(values)
+        Arrow::Decimal256ArrayBuilder.build(
+          @arrow_type,
+          values.map do |value|
+            Arrow::Decimal256.new(format(@format_str, BigDecimal(value)))
+          end
+        )
+      end
+
+      def arrow_type_builder
+        Arrow::Decimal256DataType.new(
+          @precision, @scale
+        )
+      end
+    end
+  end
+end

data/lib/parqueteur/types/int32_type.rb

@@ -21,5 +21,3 @@ module Parqueteur
     end
   end
 end
-
-# when :integer

data/lib/parqueteur/types/int64_type.rb

@@ -21,5 +21,3 @@ module Parqueteur
     end
   end
 end
-
-# when :integer

data/lib/parqueteur/types/time32_type.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Time32Type < Parqueteur::Type
+      def build_value_array(values)
+        Arrow::Time32Array.new(
+          @options.fetch(:precision, :second), values
+        )
+      end
+
+      def arrow_type_builder
+        Arrow::Time32DataType.new(
+          options.fetch(:unit, :second)
+        )
+      end
+    end
+  end
+end

data/lib/parqueteur/types/time64_type.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module Parqueteur
+  module Types
+    class Time64Type < Parqueteur::Type
+      def build_value_array(values)
+        Arrow::Time64Array.new(
+          @options.fetch(:precision, :second), values
+        )
+      end
+
+      def arrow_type_builder
+        Arrow::Time64DataType.new(
+          options.fetch(:unit, :second)
+        )
+      end
+    end
+  end
+end

data/lib/parqueteur/types/timestamp_type.rb

@@ -9,7 +9,9 @@ module Parqueteur
   module Types
     class TimestampType < Parqueteur::Type
       def build_value_array(values)
-        Arrow::TimestampArray.new(values)
+        Arrow::TimestampArray.new(
+          options.fetch(:unit, :second), values
+        )
       end
 
       def arrow_type_builder

data/lib/parqueteur/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Parqueteur
-  VERSION = '1.1.1'
+  VERSION = '1.3.2'
 end

data/lib/parqueteur.rb

@@ -2,9 +2,10 @@
 
 require 'json'
 require 'singleton'
+require 'tempfile'
+require 'parquet'
 
-require_relative "parqueteur/version"
-require 'parqueteur/chunked_converter'
+require_relative 'parqueteur/version'
 require 'parqueteur/column'
 require 'parqueteur/column_collection'
 require 'parqueteur/converter'
@@ -14,16 +15,20 @@ require 'parqueteur/type'
 require 'parqueteur/type_resolver'
 require 'parqueteur/types/array_type'
 require 'parqueteur/types/boolean_type'
+require 'parqueteur/types/date32_type'
+require 'parqueteur/types/date64_type'
+require 'parqueteur/types/decimal128_type'
+require 'parqueteur/types/decimal256_type'
 require 'parqueteur/types/int32_type'
 require 'parqueteur/types/int64_type'
 require 'parqueteur/types/map_type'
 require 'parqueteur/types/string_type'
 require 'parqueteur/types/struct_type'
+require 'parqueteur/types/time32_type'
+require 'parqueteur/types/time64_type'
 require 'parqueteur/types/timestamp_type'
-require 'parquet'
 
 module Parqueteur
   class Error < StandardError; end
   class TypeNotFound < Error; end
-  # Your code goes here...
 end

data/parqueteur.gemspec

@@ -8,8 +8,8 @@ Gem::Specification.new do |spec|
   spec.authors       = ["Julien D."]
   spec.email         = ["julien@pocketsizesun.com"]
   spec.license       = 'Apache-2.0'
-  spec.summary       = 'Parqueteur - A Ruby gem that convert JSON to Parquet'
-  spec.description   = 'Convert JSON to Parquet'
+  spec.summary       = 'Parqueteur - A Ruby gem that convert data to Parquet'
+  spec.description   = 'Convert data to Parquet'
   spec.homepage      = 'https://github.com/pocketsizesun/parqueteur-ruby'
   spec.required_ruby_version = Gem::Requirement.new(">= 2.3.0")
 
@@ -30,7 +30,7 @@ Gem::Specification.new do |spec|
 
   # Uncomment to register a new dependency of your gem
   # spec.add_dependency "example-gem", "~> 1.0"
-  spec.add_dependency 'red-parquet', '~> 5.0'
+  spec.add_dependency 'red-parquet', '~> 6.0'
 
   # For more information and examples about making a new gem, checkout our
   # guide at: https://bundler.io/guides/creating_gem.html

data/scripts/apache-arrow-ubuntu-install.sh

@@ -0,0 +1,18 @@
+#!/bin/sh
+
+if [ $(dpkg-query -W -f='${Status}' apache-arrow-apt-source 2>/dev/null | grep -c "ok installed") -eq 1 ]
+then
+  exit 0
+fi
+
+LSB_RELEASE_CODENAME_SHORT=$(lsb_release --codename --short)
+
+apt-get update
+apt-get install -y -V ca-certificates lsb-release wget
+wget https://apache.jfrog.io/artifactory/arrow/$(lsb_release --id --short | tr 'A-Z' 'a-z')/apache-arrow-apt-source-latest-$(lsb_release --codename --short).deb
+apt-get install -y -V ./apache-arrow-apt-source-latest-${LSB_RELEASE_CODENAME_SHORT}.deb
+rm ./apache-arrow-apt-source-latest-${LSB_RELEASE_CODENAME_SHORT}.deb
+apt-get update
+apt-get install -y libgirepository1.0-dev libarrow-dev libarrow-glib-dev libparquet-dev libparquet-glib-dev
+
+exit 0

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: parqueteur
 version: !ruby/object:Gem::Version
-  version: 1.1.1
+  version: 1.3.2
 platform: ruby
 authors:
 - Julien D.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2021-10-02 00:00:00.000000000 Z
+date: 2021-11-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: red-parquet
@@ -16,15 +16,15 @@ dependencies:
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '5.0'
+        version: '6.0'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - "~>"
       - !ruby/object:Gem::Version
-        version: '5.0'
-description: Convert JSON to Parquet
+        version: '6.0'
+description: Convert data to Parquet
 email:
 - julien@pocketsizesun.com
 executables: []
@@ -38,9 +38,15 @@ files:
 - Rakefile
 - bin/console
 - bin/setup
-- example.rb
+- examples/cars.rb
+- examples/convert-and-compression.rb
+- examples/convert-methods.rb
+- examples/convert-to-io.rb
+- examples/convert-with-chunks.rb
+- examples/convert-without-compression.rb
+- examples/hello-world.rb
+- examples/readme-example.rb
 - lib/parqueteur.rb
-- lib/parqueteur/chunked_converter.rb
 - lib/parqueteur/column.rb
 - lib/parqueteur/column_collection.rb
 - lib/parqueteur/converter.rb
@@ -50,15 +56,21 @@ files:
 - lib/parqueteur/type_resolver.rb
 - lib/parqueteur/types/array_type.rb
 - lib/parqueteur/types/boolean_type.rb
+- lib/parqueteur/types/date32_type.rb
+- lib/parqueteur/types/date64_type.rb
+- lib/parqueteur/types/decimal128_type.rb
+- lib/parqueteur/types/decimal256_type.rb
 - lib/parqueteur/types/int32_type.rb
 - lib/parqueteur/types/int64_type.rb
 - lib/parqueteur/types/map_type.rb
 - lib/parqueteur/types/string_type.rb
 - lib/parqueteur/types/struct_type.rb
+- lib/parqueteur/types/time32_type.rb
+- lib/parqueteur/types/time64_type.rb
 - lib/parqueteur/types/timestamp_type.rb
 - lib/parqueteur/version.rb
 - parqueteur.gemspec
-- test.json
+- scripts/apache-arrow-ubuntu-install.sh
 homepage: https://github.com/pocketsizesun/parqueteur-ruby
 licenses:
 - Apache-2.0
@@ -82,5 +94,5 @@ requirements: []
 rubygems_version: 3.2.3
 signing_key:
 specification_version: 4
-summary: Parqueteur - A Ruby gem that convert JSON to Parquet
+summary: Parqueteur - A Ruby gem that convert data to Parquet
 test_files: []

data/example.rb

@@ -1,39 +0,0 @@
-require 'bundler/setup'
-require 'parqueteur'
-
-class Foo < Parqueteur::Converter
-  column :id, :bigint
-  column :reference, :string
-  column :hash, :map, key: :string, value: :string
-  column :valid, :boolean
-  column :total, :integer
-  column :numbers, :array, elements: :integer
-  column :my_struct, :struct do
-    field :test, :string
-    field :mon_nombre, :integer
-  end
-end
-
-LETTERS = ('a'..'z').to_a
-
-data = 1000.times.collect do |i|
-  {
-    'id' => i + 1,
-    'reference' => "coucou:#{i}",
-    'hash' => { 'a' => LETTERS.sample },
-    'valid' => rand < 0.5,
-    'total' => rand(100..500),
-    'numbers' => [1, 2, 3],
-    'my_struct' => {
-      'test' => 'super'
-    }
-  }
-end
-
-# chunked_converter = Parqueteur::ChunkedConverter.new(data, Foo)
-# pp chunked_converter.write_files('test')
-puts Foo.convert(data, output: 'tmp/test.parquet')
-table = Arrow::Table.load('tmp/test.parquet')
-table.each_record do |record|
-  puts record.to_h
-end