csvops 0.1.0.alpha → 0.3.0.alpha

data/lib/csvtool/domain/{extraction_session → column_session}/extraction_value.rb

@@ -2,7 +2,7 @@
 
 module Csvtool
   module Domain
-    module ExtractionSession
+    module ColumnSession
       class ExtractionValue
         attr_reader :value
 

data/lib/csvtool/domain/{extraction_session → column_session}/output_destination.rb

@@ -2,7 +2,7 @@
 
 module Csvtool
   module Domain
-    module ExtractionSession
+    module ColumnSession
       class OutputDestination
         attr_reader :mode, :path
 

data/lib/csvtool/domain/{extraction_session → column_session}/preview.rb

@@ -2,7 +2,7 @@
 
 module Csvtool
   module Domain
-    module ExtractionSession
+    module ColumnSession
       class Preview
         attr_reader :values
 

data/lib/csvtool/domain/{extraction_session → column_session}/separator.rb

@@ -2,7 +2,7 @@
 
 module Csvtool
   module Domain
-    module ExtractionSession
+    module ColumnSession
       class Separator
         attr_reader :value
 

data/lib/csvtool/domain/row_randomization_session/randomization_options.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowRandomizationSession
+      class RandomizationOptions
+        attr_reader :seed
+
+        def initialize(seed:)
+          raise ArgumentError, "seed must be an integer or nil" unless seed.nil? || seed.is_a?(Integer)
+
+          @seed = seed
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_randomization_session/randomization_output_destination.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowRandomizationSession
+      class RandomizationOutputDestination
+        attr_reader :mode, :path
+
+        def self.console
+          new(mode: :console)
+        end
+
+        def self.file(path:)
+          new(mode: :file, path: path)
+        end
+
+        def initialize(mode:, path: nil)
+          raise ArgumentError, "invalid output mode" unless %i[console file].include?(mode)
+          raise ArgumentError, "file output path cannot be empty" if mode == :file && path.to_s.empty?
+
+          @mode = mode
+          @path = path
+        end
+
+        def file?
+          @mode == :file
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_randomization_session/randomization_session.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowRandomizationSession
+      class RandomizationSession
+        attr_reader :source, :options, :output_destination
+
+        def self.start(source:, options:)
+          new(source: source, options: options)
+        end
+
+        def initialize(source:, options:, output_destination: nil)
+          @source = source
+          @options = options
+          @output_destination = output_destination
+        end
+
+        def with_output_destination(destination)
+          self.class.new(source: @source, options: @options, output_destination: destination)
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_randomization_session/randomization_source.rb

@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowRandomizationSession
+      class RandomizationSource
+        attr_reader :path, :separator
+
+        def initialize(path:, separator:, headers_present:)
+          raise ArgumentError, "separator cannot be empty" if separator.to_s.empty?
+
+          @path = path
+          @separator = separator
+          @headers_present = headers_present
+        end
+
+        def headers_present?
+          @headers_present
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_session/row_output_destination.rb

@@ -0,0 +1,31 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowSession
+      class RowOutputDestination
+        attr_reader :mode, :path
+
+        def self.console
+          new(mode: :console)
+        end
+
+        def self.file(path:)
+          new(mode: :file, path: path)
+        end
+
+        def initialize(mode:, path: nil)
+          raise ArgumentError, "invalid output mode" unless %i[console file].include?(mode)
+          raise ArgumentError, "file output path cannot be empty" if mode == :file && path.to_s.empty?
+
+          @mode = mode
+          @path = path
+        end
+
+        def file?
+          @mode == :file
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_session/row_range.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowSession
+      class InvalidStartRowError < StandardError; end
+      class InvalidEndRowError < StandardError; end
+      class InvalidRowRangeOrderError < StandardError; end
+
+      class RowRange
+        attr_reader :start_row, :end_row
+
+        def self.from_inputs(start_row_input:, end_row_input:)
+          unless /\A[1-9]\d*\z/.match?(start_row_input.to_s)
+            raise InvalidStartRowError, "invalid start row"
+          end
+          unless /\A[1-9]\d*\z/.match?(end_row_input.to_s)
+            raise InvalidEndRowError, "invalid end row"
+          end
+
+          start_row = start_row_input.to_i
+          end_row = end_row_input.to_i
+          raise InvalidRowRangeOrderError, "end row before start row" if end_row < start_row
+
+          new(start_row: start_row, end_row: end_row)
+        end
+
+        def initialize(start_row:, end_row:)
+          raise InvalidStartRowError, "invalid start row" unless start_row.to_i >= 1
+          raise InvalidEndRowError, "invalid end row" unless end_row.to_i >= 1
+          raise InvalidRowRangeOrderError, "end row before start row" if end_row < start_row
+
+          @start_row = start_row
+          @end_row = end_row
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_session/row_session.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowSession
+      class RowSession
+        attr_reader :source, :row_range, :output_destination
+
+        def self.start(source:, row_range:)
+          new(source: source, row_range: row_range)
+        end
+
+        def initialize(source:, row_range:, output_destination: nil)
+          @source = source
+          @row_range = row_range
+          @output_destination = output_destination
+        end
+
+        def with_output_destination(destination)
+          self.class.new(source: @source, row_range: @row_range, output_destination: destination)
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/domain/row_session/row_source.rb

@@ -0,0 +1,16 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Domain
+    module RowSession
+      class RowSource
+        attr_reader :path, :separator
+
+        def initialize(path:, separator:)
+          @path = path
+          @separator = separator
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/infrastructure/csv/row_randomizer.rb

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+require "csv"
+require "tempfile"
+
+module Csvtool
+  module Infrastructure
+    module CSV
+      class RowRandomizer
+        DEFAULT_CHUNK_SIZE = 10_000
+
+        def call(file_path:, col_sep:, headers:, seed: nil)
+          each(file_path: file_path, col_sep: col_sep, headers: headers, seed: seed).to_a
+        end
+
+        def each(file_path:, col_sep:, headers:, seed: nil, chunk_size: DEFAULT_CHUNK_SIZE)
+          chunk_paths = []
+          return enum_for(:each, file_path: file_path, col_sep: col_sep, headers: headers, seed: seed, chunk_size: chunk_size) unless block_given?
+
+          rng = seed.nil? ? Random.new : Random.new(seed)
+          sequence = 0
+          chunk_entries = []
+
+          ::CSV.foreach(file_path, headers: headers, col_sep: col_sep) do |row|
+            fields = headers ? row.fields : row
+            chunk_entries << [rng.rand, sequence, fields]
+            sequence += 1
+            flush_chunk(chunk_entries, chunk_paths) if chunk_entries.length >= chunk_size
+          end
+
+          flush_chunk(chunk_entries, chunk_paths) unless chunk_entries.empty?
+          merge_chunks(chunk_paths) { |fields| yield fields }
+        ensure
+          cleanup_chunks(chunk_paths)
+        end
+
+        private
+
+        def flush_chunk(entries, chunk_paths)
+          entries.sort_by! { |rand_key, seq, _fields| [rand_key, seq] }
+          file = Tempfile.new("csvtool-row-randomizer-chunk")
+          file.binmode
+          entries.each { |entry| Marshal.dump(entry, file) }
+          file.close
+          chunk_paths << file.path
+          entries.clear
+        end
+
+        def merge_chunks(chunk_paths)
+          readers = chunk_paths.map { |path| File.open(path, "rb") }
+          heads = readers.map { |reader| next_entry(reader) }
+
+          loop do
+            indexed = heads.each_with_index.select { |entry, _i| !entry.nil? }
+            break if indexed.empty?
+
+            min_entry, min_index = indexed.min_by { |entry, _i| [entry[0], entry[1]] }
+            yield min_entry[2]
+            heads[min_index] = next_entry(readers[min_index])
+          end
+        ensure
+          readers&.each(&:close)
+        end
+
+        def next_entry(reader)
+          Marshal.load(reader)
+        rescue EOFError
+          nil
+        end
+
+        def cleanup_chunks(chunk_paths)
+          return if chunk_paths.nil?
+
+          chunk_paths.each do |path|
+            File.delete(path) if File.exist?(path)
+          rescue Errno::EACCES, Errno::ENOENT
+            nil
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/infrastructure/csv/row_streamer.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Infrastructure
+    module CSV
+      class RowStreamer
+        def each_in_range(file_path:, col_sep:, start_row:, end_row:)
+          row_index = 0
+          matched = false
+
+          ::CSV.foreach(file_path, headers: true, col_sep: col_sep) do |row|
+            row_index += 1
+            next if row_index < start_row
+            break if row_index > end_row
+
+            matched = true
+            yield row.fields
+          end
+
+          { matched: matched, row_count: row_index }
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/infrastructure/output/csv_row_console_writer.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Infrastructure
+    module Output
+      class CsvRowConsoleWriter
+        def initialize(stdout:, row_streamer:)
+          @stdout = stdout
+          @row_streamer = row_streamer
+        end
+
+        def call(file_path:, col_sep:, headers:, start_row:, end_row:)
+          wrote_header = false
+          stats = @row_streamer.each_in_range(
+            file_path: file_path,
+            col_sep: col_sep,
+            start_row: start_row,
+            end_row: end_row
+          ) do |fields|
+            unless wrote_header
+              @stdout.puts ::CSV.generate_line(headers, row_sep: "").chomp
+              wrote_header = true
+            end
+            @stdout.puts ::CSV.generate_line(fields, row_sep: "").chomp
+          end
+
+          stats
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/infrastructure/output/csv_row_file_writer.rb

@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Infrastructure
+    module Output
+      class CsvRowFileWriter
+        def initialize(stdout:, errors:, row_streamer:)
+          @stdout = stdout
+          @errors = errors
+          @row_streamer = row_streamer
+        end
+
+        def call(file_path:, col_sep:, headers:, start_row:, end_row:, output_path:)
+          csv = nil
+          wrote_rows = false
+
+          stats = @row_streamer.each_in_range(
+            file_path: file_path,
+            col_sep: col_sep,
+            start_row: start_row,
+            end_row: end_row
+          ) do |fields|
+            unless wrote_rows
+              csv = ::CSV.open(output_path, "w")
+              csv << headers
+              wrote_rows = true
+            end
+            csv << fields
+          end
+
+          csv&.close
+          @stdout.puts "Wrote output to #{output_path}" if wrote_rows
+          stats
+        rescue Errno::EACCES, Errno::ENOENT => e
+          @errors.cannot_write_output_file(output_path, e.class)
+          nil
+        ensure
+          csv&.close unless csv&.closed?
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/errors/presenter.rb

@@ -49,9 +49,29 @@ module Csvtool
             @stdout.puts "Invalid separator choice."
           end
 
+          def invalid_seed
+            @stdout.puts "Seed must be an integer."
+          end
+
           def canceled
             @stdout.puts "Canceled."
           end
+
+          def invalid_start_row
+            @stdout.puts "Start row must be a positive integer."
+          end
+
+          def invalid_end_row
+            @stdout.puts "End row must be a positive integer."
+          end
+
+          def invalid_row_range_order
+            @stdout.puts "End row must be greater than or equal to start row."
+          end
+
+          def row_range_out_of_bounds(total_rows)
+            @stdout.puts "Row range is out of bounds. File has #{total_rows} data rows."
+          end
         end
       end
     end

data/lib/csvtool/interface/cli/menu_loop.rb

@@ -4,25 +4,33 @@ module Csvtool
   module Interface
     module CLI
       class MenuLoop
-        def initialize(stdin:, stdout:, menu_options:, extract_action:)
+        def initialize(stdin:, stdout:, menu_options:, extract_column_action:, extract_rows_action:, randomize_rows_action:)
           @stdin = stdin
           @stdout = stdout
           @menu_options = menu_options
-          @extract_action = extract_action
+          @extract_column_action = extract_column_action
+          @extract_rows_action = extract_rows_action
+          @randomize_rows_action = randomize_rows_action
         end
 
         def run
           loop do
             print_menu
             @stdout.print "> "
+            choice = @stdin.gets
+            return 0 if choice.nil?
 
-            case @stdin.gets&.strip
+            case choice.strip
             when "1"
-              @extract_action.call
+              @extract_column_action.call
             when "2"
+              @extract_rows_action.call
+            when "3"
+              @randomize_rows_action.call
+            when "4"
               return 0
             else
-              @stdout.puts "Please choose 1 or 2."
+              @stdout.puts "Please choose 1, 2, 3, or 4."
             end
           end
         end

data/lib/csvtool/interface/cli/prompts/headers_present_prompt.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Interface
+    module CLI
+      module Prompts
+        class HeadersPresentPrompt
+          def initialize(stdin:, stdout:)
+            @stdin = stdin
+            @stdout = stdout
+          end
+
+          def call
+            @stdout.print "Headers present? [Y/n]: "
+            answer = @stdin.gets&.strip.to_s.downcase
+            !%w[n no].include?(answer)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/prompts/seed_prompt.rb

@@ -0,0 +1,29 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Interface
+    module CLI
+      module Prompts
+        class SeedPrompt
+          INVALID = :invalid
+
+          def initialize(stdin:, stdout:, errors:)
+            @stdin = stdin
+            @stdout = stdout
+            @errors = errors
+          end
+
+          def call
+            @stdout.print "Random seed (optional integer): "
+            raw = @stdin.gets&.strip.to_s
+            return nil if raw.empty?
+            return raw.to_i if /\A-?\d+\z/.match?(raw)
+
+            @errors.invalid_seed
+            INVALID
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Csvtool
-  VERSION = "0.1.0.alpha"
+  VERSION = "0.3.0.alpha"
 end