csvops 0.4.0.alpha → 0.5.0.alpha

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

@@ -5,13 +5,15 @@ module Csvtool
     module CLI
       module Prompts
         class FilePathPrompt
+          DEFAULT_LABEL = "CSV file path: "
+
           def initialize(stdin:, stdout:)
             @stdin = stdin
             @stdout = stdout
           end
 
-          def call
-            @stdout.print "CSV file path: "
+          def call(label: DEFAULT_LABEL)
+            @stdout.print label
             @stdin.gets&.strip.to_s
           end
         end

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

@@ -5,13 +5,15 @@ module Csvtool
     module CLI
       module Prompts
         class HeadersPresentPrompt
+          DEFAULT_LABEL = "Headers present? [Y/n]: "
+
           def initialize(stdin:, stdout:)
             @stdin = stdin
             @stdout = stdout
           end
 
-          def call
-            @stdout.print "Headers present? [Y/n]: "
+          def call(label: DEFAULT_LABEL)
+            @stdout.print label
             answer = @stdin.gets&.strip.to_s.downcase
             !%w[n no].include?(answer)
           end

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

@@ -5,14 +5,16 @@ module Csvtool
     module CLI
       module Prompts
         class SeparatorPrompt
+          DEFAULT_LABEL = "Choose separator:"
+
           def initialize(stdin:, stdout:, errors:)
             @stdin = stdin
             @stdout = stdout
             @errors = errors
           end
 
-          def call
-            @stdout.puts "Choose separator:"
+          def call(label: DEFAULT_LABEL)
+            @stdout.puts label
             @stdout.puts "1. comma (,)"
             @stdout.puts "2. tab (\\t)"
             @stdout.puts "3. semicolon (;)"

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

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Interface
+    module CLI
+      module Prompts
+        class YesNoPrompt
+          def initialize(stdin:, stdout:)
+            @stdin = stdin
+            @stdout = stdout
+          end
+
+          def call(label:, default:)
+            @stdout.print label
+            answer = @stdin.gets&.strip.to_s.downcase
+            return default if answer.empty?
+            return true if %w[y yes].include?(answer)
+            return false if %w[n no].include?(answer)
+
+            default
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/builders/column_session_builder.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+require "csvtool/domain/column_session/separator"
+require "csvtool/domain/column_session/csv_source"
+require "csvtool/domain/column_session/column_selection"
+require "csvtool/domain/column_session/extraction_options"
+require "csvtool/domain/column_session/column_session"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Builders
+          class ColumnSessionBuilder
+            def call(file_path:, col_sep:, column_name:, skip_blanks:)
+              separator = Domain::ColumnSession::Separator.new(col_sep)
+              source = Domain::ColumnSession::CsvSource.new(path: file_path, separator: separator)
+              column_selection = Domain::ColumnSession::ColumnSelection.new(name: column_name)
+              options = Domain::ColumnSession::ExtractionOptions.new(skip_blanks: skip_blanks, preview_limit: 10)
+
+              Domain::ColumnSession::ColumnSession.start(
+                source: source,
+                column_selection: column_selection,
+                options: options
+              )
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/builders/cross_csv_dedupe_session_builder.rb

@@ -0,0 +1,35 @@
+# frozen_string_literal: true
+
+require "csvtool/domain/cross_csv_dedupe_session/key_mapping"
+require "csvtool/domain/cross_csv_dedupe_session/match_options"
+require "csvtool/domain/cross_csv_dedupe_session/cross_csv_dedupe_session"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Builders
+          class CrossCsvDedupeSessionBuilder
+            def call(source:, reference:, source_selector:, reference_selector:, trim_whitespace:, case_insensitive:, destination:)
+              key_mapping = Domain::CrossCsvDedupeSession::KeyMapping.new(
+                source_selector: source_selector,
+                reference_selector: reference_selector
+              )
+              match_options = Domain::CrossCsvDedupeSession::MatchOptions.new(
+                trim_whitespace: trim_whitespace,
+                case_insensitive: case_insensitive
+              )
+
+              Domain::CrossCsvDedupeSession::CrossCsvDedupeSession.start(
+                source: source,
+                reference: reference,
+                key_mapping: key_mapping,
+                match_options: match_options
+              ).with_output_destination(destination)
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/builders/row_extraction_session_builder.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+require "csvtool/domain/row_session/row_source"
+require "csvtool/domain/row_session/row_session"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Builders
+          class RowExtractionSessionBuilder
+            def call(file_path:, col_sep:, row_range:, destination:)
+              source = Domain::RowSession::RowSource.new(path: file_path, separator: col_sep)
+              session = Domain::RowSession::RowSession.start(source: source, row_range: row_range)
+              session.with_output_destination(destination)
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/builders/row_randomization_session_builder.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+require "csvtool/domain/row_randomization_session/randomization_source"
+require "csvtool/domain/row_randomization_session/randomization_options"
+require "csvtool/domain/row_randomization_session/randomization_session"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Builders
+          class RowRandomizationSessionBuilder
+            def call(file_path:, col_sep:, headers_present:, seed:, destination:)
+              source = Domain::RowRandomizationSession::RandomizationSource.new(
+                path: file_path,
+                separator: col_sep,
+                headers_present: headers_present
+              )
+              options = Domain::RowRandomizationSession::RandomizationOptions.new(seed: seed)
+              session = Domain::RowRandomizationSession::RandomizationSession.start(source: source, options: options)
+              session.with_output_destination(destination)
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/presenters/column_extraction_presenter.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Presenters
+          class ColumnExtractionPresenter
+            def initialize(stdout:)
+              @stdout = stdout
+            end
+
+            def print_value(value)
+              @stdout.puts value
+            end
+
+            def print_file_written(path)
+              @stdout.puts "Wrote output to #{path}"
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/presenters/cross_csv_dedupe_presenter.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Presenters
+          class CrossCsvDedupePresenter
+            def initialize(stdout:, col_sep:)
+              @stdout = stdout
+              @col_sep = col_sep
+            end
+
+            def print_header(headers)
+              @stdout.puts
+              @stdout.puts ::CSV.generate_line(headers, row_sep: "", col_sep: @col_sep).chomp
+            end
+
+            def print_row(fields)
+              @stdout.puts ::CSV.generate_line(fields, row_sep: "", col_sep: @col_sep).chomp
+            end
+
+            def print_file_written(path)
+              @stdout.puts "Wrote output to #{path}"
+            end
+
+            def print_summary(stats)
+              @stdout.puts "Summary: source_rows=#{stats[:source_rows]} removed_rows=#{stats[:removed_rows]} kept_rows=#{stats[:kept_rows_count]}"
+              @stdout.puts "No rows removed; no matching keys found." if stats[:removed_rows].zero?
+              @stdout.puts "All source rows were removed by dedupe." if stats[:source_rows].positive? && stats[:kept_rows_count].zero?
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/presenters/row_extraction_presenter.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Presenters
+          class RowExtractionPresenter
+            def initialize(stdout:, headers:, col_sep:)
+              @stdout = stdout
+              @headers = headers
+              @col_sep = col_sep
+              @printed_header = false
+            end
+
+            def print_row(fields)
+              unless @printed_header
+                @stdout.puts ::CSV.generate_line(@headers, row_sep: "", col_sep: @col_sep).chomp
+                @printed_header = true
+              end
+              @stdout.puts ::CSV.generate_line(fields, row_sep: "", col_sep: @col_sep).chomp
+            end
+
+            def print_file_written(path)
+              @stdout.puts "Wrote output to #{path}"
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/presenters/row_randomization_presenter.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+require "csv"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        module Presenters
+          class RowRandomizationPresenter
+            def initialize(stdout:, headers:, col_sep:)
+              @stdout = stdout
+              @headers = headers
+              @col_sep = col_sep
+            end
+
+            def print_console_start
+              @stdout.puts
+              @stdout.puts ::CSV.generate_line(@headers, row_sep: "", col_sep: @col_sep).chomp if @headers
+            end
+
+            def print_row(fields)
+              @stdout.puts ::CSV.generate_line(fields, row_sep: "", col_sep: @col_sep).chomp
+            end
+
+            def print_file_written(path)
+              @stdout.puts "Wrote output to #{path}"
+            end
+          end
+        end
+      end
+    end
+  end
+end

data/lib/csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow.rb

@@ -1,18 +1,23 @@
 # frozen_string_literal: true
 
-require "csv"
 require "csvtool/application/use_cases/run_cross_csv_dedupe"
 require "csvtool/interface/cli/errors/presenter"
 require "csvtool/interface/cli/prompts/file_path_prompt"
 require "csvtool/interface/cli/prompts/separator_prompt"
 require "csvtool/interface/cli/prompts/output_destination_prompt"
+require "csvtool/interface/cli/prompts/headers_present_prompt"
+require "csvtool/interface/cli/prompts/yes_no_prompt"
+require "csvtool/interface/cli/prompts/dedupe_key_selector_prompt"
+require "csvtool/interface/cli/workflows/builders/cross_csv_dedupe_session_builder"
+require "csvtool/interface/cli/workflows/presenters/cross_csv_dedupe_presenter"
+require "csvtool/interface/cli/workflows/support/output_destination_mapper"
+require "csvtool/interface/cli/workflows/support/result_error_handler"
+require "csvtool/interface/cli/workflows/steps/workflow_step_pipeline"
+require "csvtool/interface/cli/workflows/steps/cross_csv_dedupe/collect_profiles_step"
+require "csvtool/interface/cli/workflows/steps/cross_csv_dedupe/collect_options_step"
+require "csvtool/interface/cli/workflows/steps/cross_csv_dedupe/execute_step"
 require "csvtool/domain/cross_csv_dedupe_session/csv_profile"
 require "csvtool/domain/cross_csv_dedupe_session/column_selector"
-require "csvtool/domain/cross_csv_dedupe_session/key_mapping"
-require "csvtool/domain/cross_csv_dedupe_session/match_options"
-require "csvtool/domain/cross_csv_dedupe_session/cross_csv_dedupe_session"
-require "csvtool/domain/shared/output_destination"
-
 module Csvtool
   module Interface
     module CLI
@@ -23,89 +28,40 @@ module Csvtool
             @stdout = stdout
             @use_case = use_case
             @errors = Interface::CLI::Errors::Presenter.new(stdout: stdout)
+            @session_builder = Builders::CrossCsvDedupeSessionBuilder.new
+            @output_destination_mapper = Support::OutputDestinationMapper.new
+            @result_error_handler = Support::ResultErrorHandler.new(errors: @errors)
           end
 
           def call
-            source_path = Interface::CLI::Prompts::FilePathPrompt.new(stdin: @stdin, stdout: @stdout).call
-            return @errors.file_not_found(source_path) unless File.file?(source_path)
-
-            @stdout.puts "Source CSV separator:"
-            source_col_sep = Interface::CLI::Prompts::SeparatorPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors).call
-            return if source_col_sep.nil?
-            @stdout.print "Source headers present? [Y/n]: "
-            source_headers_present = !%w[n no].include?(@stdin.gets&.strip.to_s.downcase)
-            source = Domain::CrossCsvDedupeSession::CsvProfile.new(
-              path: source_path,
-              separator: source_col_sep,
-              headers_present: source_headers_present
-            )
-
-            @stdout.print "Reference CSV file path: "
-            reference_path = @stdin.gets&.strip.to_s
-            return @errors.file_not_found(reference_path) unless File.file?(reference_path)
-
-            @stdout.puts "Reference CSV separator:"
-            reference_col_sep = Interface::CLI::Prompts::SeparatorPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors).call
-            return if reference_col_sep.nil?
-            @stdout.print "Reference headers present? [Y/n]: "
-            reference_headers_present = !%w[n no].include?(@stdin.gets&.strip.to_s.downcase)
-            reference = Domain::CrossCsvDedupeSession::CsvProfile.new(
-              path: reference_path,
-              separator: reference_col_sep,
-              headers_present: reference_headers_present
-            )
-
-            source_selector = prompt_selector("Source", source.headers_present?)
-            return @errors.column_not_found if source_selector.nil?
-            reference_selector = prompt_selector("Reference", reference.headers_present?)
-            return @errors.column_not_found if reference_selector.nil?
-
-            @stdout.print "Trim whitespace before matching? [Y/n]: "
-            trim_whitespace = read_yes_no(default: true)
-            @stdout.print "Case-insensitive matching? [y/N]: "
-            case_insensitive = read_yes_no(default: false)
-
-            key_mapping = Domain::CrossCsvDedupeSession::KeyMapping.new(
-              source_selector: source_selector,
-              reference_selector: reference_selector
-            )
-            match_options = Domain::CrossCsvDedupeSession::MatchOptions.new(
-              trim_whitespace: trim_whitespace,
-              case_insensitive: case_insensitive
-            )
-            session = Domain::CrossCsvDedupeSession::CrossCsvDedupeSession.start(
-              source: source,
-              reference: reference,
-              key_mapping: key_mapping,
-              match_options: match_options
-            )
-
-            output_destination = Interface::CLI::Prompts::OutputDestinationPrompt.new(
-              stdin: @stdin,
-              stdout: @stdout,
-              errors: @errors
-            ).call
-            return if output_destination.nil?
-            session = session.with_output_destination(
-              if output_destination[:mode] == :file
-                Domain::Shared::OutputDestination.file(path: output_destination[:path])
-              else
-                Domain::Shared::OutputDestination.console
-              end
-            )
-
-            result = @use_case.call(
-              session: session,
-              on_header: ->(headers) { print_header(headers, col_sep: session.source.separator) },
-              on_row: ->(fields) { print_row(fields, col_sep: session.source.separator) }
-            )
-            return handle_error(result) unless result.ok?
-
-            @stdout.puts "Wrote output to #{result.data[:output_path]}" if session.output_destination.file?
-            stats = result.data[:stats]
-            @stdout.puts "Summary: source_rows=#{stats[:source_rows]} removed_rows=#{stats[:removed_rows]} kept_rows=#{stats[:kept_rows_count]}"
-            @stdout.puts "No rows removed; no matching keys found." if stats[:removed_rows].zero?
-            @stdout.puts "All source rows were removed by dedupe." if stats[:source_rows].positive? && stats[:kept_rows_count].zero?
+            context = {
+              use_case: @use_case,
+              session_builder: @session_builder,
+              output_destination_mapper: @output_destination_mapper,
+              presenter_factory: ->(col_sep:) { Presenters::CrossCsvDedupePresenter.new(stdout: @stdout, col_sep: col_sep) },
+              handle_error: method(:handle_error)
+            }
+
+            pipeline = Steps::WorkflowStepPipeline.new(steps: [
+              Steps::CrossCsvDedupe::CollectProfilesStep.new(
+                file_path_prompt: Interface::CLI::Prompts::FilePathPrompt.new(stdin: @stdin, stdout: @stdout),
+                separator_prompt: Interface::CLI::Prompts::SeparatorPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors),
+                headers_present_prompt: Interface::CLI::Prompts::HeadersPresentPrompt.new(stdin: @stdin, stdout: @stdout),
+                errors: @errors
+              ),
+              Steps::CrossCsvDedupe::CollectOptionsStep.new(
+                selector_prompt: Interface::CLI::Prompts::DedupeKeySelectorPrompt.new(stdin: @stdin, stdout: @stdout),
+                yes_no_prompt: Interface::CLI::Prompts::YesNoPrompt.new(stdin: @stdin, stdout: @stdout),
+                output_destination_prompt: Interface::CLI::Prompts::OutputDestinationPrompt.new(
+                  stdin: @stdin,
+                  stdout: @stdout,
+                  errors: @errors
+                ),
+                errors: @errors
+              ),
+              Steps::CrossCsvDedupe::ExecuteStep.new
+            ])
+            pipeline.call(context)
           rescue ArgumentError => e
             return @errors.empty_output_path if e.message == "file output path cannot be empty"
 
@@ -114,48 +70,15 @@ module Csvtool
 
           private
 
-          def prompt_selector(label, headers_present)
-            if headers_present
-              @stdout.print "#{label} key column name: "
-            else
-              @stdout.print "#{label} key column index (1-based): "
-            end
-            input = @stdin.gets&.strip.to_s
-            Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: headers_present, input: input)
-          rescue ArgumentError
-            nil
-          end
-
-          def print_header(headers, col_sep:)
-            @stdout.puts
-            @stdout.puts ::CSV.generate_line(headers, row_sep: "", col_sep: col_sep).chomp
-          end
-
-          def print_row(fields, col_sep:)
-            @stdout.puts ::CSV.generate_line(fields, row_sep: "", col_sep: col_sep).chomp
-          end
-
           def handle_error(result)
-            case result.error
-            when :column_not_found
-              @errors.column_not_found
-            when :could_not_parse_csv
-              @errors.could_not_parse_csv
-            when :cannot_read_file
-              @errors.cannot_read_file(result.data[:path])
-            when :cannot_write_output_file
-              @errors.cannot_write_output_file(result.data[:path], result.data[:error_class])
-            end
+            @result_error_handler.call(result, {
+              column_not_found: ->(_r, errors) { errors.column_not_found },
+              could_not_parse_csv: ->(_r, errors) { errors.could_not_parse_csv },
+              cannot_read_file: ->(r, errors) { errors.cannot_read_file(r.data[:path]) },
+              cannot_write_output_file: ->(r, errors) { errors.cannot_write_output_file(r.data[:path], r.data[:error_class]) }
+            })
           end
 
-          def read_yes_no(default:)
-            answer = @stdin.gets&.strip.to_s.downcase
-            return default if answer.empty?
-            return true if %w[y yes].include?(answer)
-            return false if %w[n no].include?(answer)
-
-            default
-          end
         end
       end
     end

data/lib/csvtool/interface/cli/workflows/run_extraction_workflow.rb

@@ -0,0 +1,88 @@
+# frozen_string_literal: true
+
+require "csvtool/application/use_cases/run_extraction"
+require "csvtool/interface/cli/errors/presenter"
+require "csvtool/interface/cli/prompts/file_path_prompt"
+require "csvtool/interface/cli/prompts/separator_prompt"
+require "csvtool/interface/cli/prompts/column_selector_prompt"
+require "csvtool/interface/cli/prompts/skip_blanks_prompt"
+require "csvtool/interface/cli/prompts/confirm_prompt"
+require "csvtool/interface/cli/prompts/output_destination_prompt"
+require "csvtool/interface/cli/workflows/builders/column_session_builder"
+require "csvtool/interface/cli/workflows/presenters/column_extraction_presenter"
+require "csvtool/interface/cli/workflows/support/output_destination_mapper"
+require "csvtool/interface/cli/workflows/support/result_error_handler"
+require "csvtool/interface/cli/workflows/steps/workflow_step_pipeline"
+require "csvtool/interface/cli/workflows/steps/extraction/collect_inputs_step"
+require "csvtool/interface/cli/workflows/steps/extraction/build_preview_step"
+require "csvtool/interface/cli/workflows/steps/extraction/collect_destination_step"
+require "csvtool/interface/cli/workflows/steps/extraction/execute_step"
+
+module Csvtool
+  module Interface
+    module CLI
+      module Workflows
+        class RunExtractionWorkflow
+          def initialize(stdin:, stdout:, use_case: Application::UseCases::RunExtraction.new)
+            @stdin = stdin
+            @stdout = stdout
+            @use_case = use_case
+            @errors = Interface::CLI::Errors::Presenter.new(stdout: stdout)
+            @session_builder = Builders::ColumnSessionBuilder.new
+            @presenter = Presenters::ColumnExtractionPresenter.new(stdout: @stdout)
+            @output_destination_mapper = Support::OutputDestinationMapper.new
+            @result_error_handler = Support::ResultErrorHandler.new(errors: @errors)
+          end
+
+          def call
+            context = {
+              use_case: @use_case,
+              session_builder: @session_builder,
+              output_destination_mapper: @output_destination_mapper,
+              presenter: @presenter,
+              handle_error: method(:handle_error)
+            }
+
+            pipeline = Steps::WorkflowStepPipeline.new(steps: [
+              Steps::Extraction::CollectInputsStep.new(
+                file_path_prompt: Interface::CLI::Prompts::FilePathPrompt.new(stdin: @stdin, stdout: @stdout),
+                separator_prompt: Interface::CLI::Prompts::SeparatorPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors),
+                column_selector_prompt: Interface::CLI::Prompts::ColumnSelectorPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors),
+                skip_blanks_prompt: Interface::CLI::Prompts::SkipBlanksPrompt.new(stdin: @stdin, stdout: @stdout)
+              ),
+              Steps::Extraction::BuildPreviewStep.new(
+                confirm_prompt: Interface::CLI::Prompts::ConfirmPrompt.new(stdin: @stdin, stdout: @stdout, errors: @errors)
+              ),
+              Steps::Extraction::CollectDestinationStep.new(
+                output_destination_prompt: Interface::CLI::Prompts::OutputDestinationPrompt.new(
+                  stdin: @stdin,
+                  stdout: @stdout,
+                  errors: @errors
+                )
+              ),
+              Steps::Extraction::ExecuteStep.new
+            ])
+            pipeline.call(context)
+          rescue ArgumentError => e
+            return @errors.empty_output_path if e.message == "file output path cannot be empty"
+
+            raise e
+          end
+
+          private
+
+          def handle_error(result)
+            @result_error_handler.call(result, {
+              file_not_found: ->(r, errors) { errors.file_not_found(r.data[:path]) },
+              no_headers: ->(_r, errors) { errors.no_headers },
+              column_not_found: ->(_r, errors) { errors.column_not_found },
+              could_not_parse_csv: ->(_r, errors) { errors.could_not_parse_csv },
+              cannot_read_file: ->(r, errors) { errors.cannot_read_file(r.data[:path]) },
+              cannot_write_output_file: ->(r, errors) { errors.cannot_write_output_file(r.data[:path], r.data[:error_class]) }
+            })
+          end
+        end
+      end
+    end
+  end
+end