csvops 0.7.0.alpha → 0.8.0.alpha

data/test/csvtool/interface/cli/workflows/run_csv_stats_workflow_test.rb

@@ -0,0 +1,146 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_csv_stats_workflow"
+require "tmpdir"
+
+class RunCsvStatsWorkflowTest < Minitest::Test
+  def fixture_path(name)
+    File.expand_path("../../../../fixtures/#{name}", __dir__)
+  end
+
+  def test_workflow_prints_core_stats_summary
+    out = StringIO.new
+    input = [fixture_path("sample_people.csv"), "", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "CSV Stats Summary"
+    assert_includes out.string, "Rows: 3"
+    assert_includes out.string, "Columns: 2"
+    assert_includes out.string, "Headers: name, city"
+    assert_includes out.string, "Column completeness:"
+    assert_includes out.string, "name: non_blank=3 blank=0"
+    assert_includes out.string, "city: non_blank=3 blank=0"
+  end
+
+  def test_workflow_supports_tsv_separator
+    out = StringIO.new
+    input = [fixture_path("sample_people.tsv"), "2", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Rows: 3"
+    assert_includes out.string, "Columns: 2"
+    assert_includes out.string, "Headers: name, city"
+  end
+
+  def test_workflow_supports_headerless_mode
+    out = StringIO.new
+    input = [fixture_path("sample_people_no_headers.csv"), "", "n"].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Rows: 3"
+    assert_includes out.string, "Columns: 2"
+    refute_includes out.string, "Headers:"
+    assert_includes out.string, "column_1: non_blank=3 blank=0"
+    assert_includes out.string, "column_2: non_blank=3 blank=0"
+  end
+
+  def test_workflow_supports_custom_separator
+    out = StringIO.new
+    input = [fixture_path("sample_people_colon.txt"), "5", ":", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Rows: 3"
+    assert_includes out.string, "Columns: 2"
+    assert_includes out.string, "Headers: name, city"
+  end
+
+  def test_workflow_prints_column_completeness_for_blank_values
+    out = StringIO.new
+    input = [fixture_path("sample_people_blanks.csv"), "", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "name: non_blank=3 blank=2"
+    assert_includes out.string, "city: non_blank=4 blank=1"
+  end
+
+  def test_workflow_reports_missing_file
+    out = StringIO.new
+    input = ["/tmp/does-not-exist.csv", "", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "File not found: /tmp/does-not-exist.csv"
+    refute_includes out.string, "Traceback"
+  end
+
+  def test_workflow_reports_parse_error
+    out = StringIO.new
+    input = [fixture_path("sample_people_bad_tail.csv"), "", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Could not parse CSV file."
+    refute_includes out.string, "Traceback"
+  end
+
+  def test_workflow_can_write_stats_to_file
+    out = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "stats.csv")
+      input = [fixture_path("sample_people.csv"), "", "", "2", output_path].join("\n") + "\n"
+
+      Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+        stdin: StringIO.new(input),
+        stdout: out
+      ).call
+
+      assert_includes out.string, "Wrote output to #{output_path}"
+      csv_text = File.read(output_path)
+      assert_includes csv_text, "metric,value"
+      assert_includes csv_text, "row_count,3"
+      assert_includes csv_text, "column_count,2"
+    end
+  end
+
+  def test_workflow_reports_cannot_write_output_file
+    out = StringIO.new
+    output_path = "/tmp/does-not-exist-dir/stats.csv"
+    input = [fixture_path("sample_people.csv"), "", "", "2", output_path].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCsvStatsWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Cannot write output file: #{output_path} (Errno::ENOENT)"
+    refute_includes out.string, "Traceback"
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/csv_stats/build_session_step_test.rb

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/csv_stats/build_session_step"
+
+class CsvStatsBuildSessionStepTest < Minitest::Test
+  class FakeBuilder
+    attr_reader :params
+
+    def call(**params)
+      @params = params
+      :session
+    end
+  end
+
+  def test_builds_session_from_context
+    builder = FakeBuilder.new
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::BuildSessionStep.new
+    context = {
+      session_builder: builder,
+      file_path: "/tmp/data.csv",
+      col_sep: "\t",
+      headers_present: true,
+      output_destination: :destination
+    }
+
+    result = step.call(context)
+
+    assert_nil result
+    assert_equal :session, context[:session]
+    assert_equal "/tmp/data.csv", builder.params[:file_path]
+    assert_equal "\t", builder.params[:col_sep]
+    assert_equal true, builder.params[:headers_present]
+    assert_equal :destination, builder.params[:destination]
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/csv_stats/collect_destination_step_test.rb

@@ -0,0 +1,49 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/csv_stats/collect_destination_step"
+
+class CsvStatsCollectDestinationStepTest < Minitest::Test
+  class FakePrompt
+    def initialize(result)
+      @result = result
+    end
+
+    def call
+      @result
+    end
+  end
+
+  class FakeMapper
+    attr_reader :input
+
+    def call(input)
+      @input = input
+      :mapped_destination
+    end
+  end
+
+  def test_collects_and_maps_destination
+    mapper = FakeMapper.new
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::CollectDestinationStep.new(
+      output_destination_prompt: FakePrompt.new({ mode: :file, path: "/tmp/out.csv" })
+    )
+    context = { output_destination_mapper: mapper }
+
+    result = step.call(context)
+
+    assert_nil result
+    assert_equal({ mode: :file, path: "/tmp/out.csv" }, mapper.input)
+    assert_equal :mapped_destination, context[:output_destination]
+  end
+
+  def test_halts_when_destination_prompt_returns_nil
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::CollectDestinationStep.new(
+      output_destination_prompt: FakePrompt.new(nil)
+    )
+
+    result = step.call(output_destination_mapper: FakeMapper.new)
+
+    assert_equal :halt, result
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/csv_stats/collect_inputs_step_test.rb

@@ -0,0 +1,61 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/csv_stats/collect_inputs_step"
+
+class CsvStatsCollectInputsStepTest < Minitest::Test
+  class FakeFilePathPrompt
+    attr_reader :label
+
+    def call(label:)
+      @label = label
+      "/tmp/input.csv"
+    end
+  end
+
+  class FakeSeparatorPrompt
+    def initialize(result)
+      @result = result
+    end
+
+    def call
+      @result
+    end
+  end
+
+  class FakeHeadersPresentPrompt
+    def call
+      false
+    end
+  end
+
+  def test_collects_inputs_into_context
+    file_prompt = FakeFilePathPrompt.new
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::CollectInputsStep.new(
+      file_path_prompt: file_prompt,
+      separator_prompt: FakeSeparatorPrompt.new(";"),
+      headers_present_prompt: FakeHeadersPresentPrompt.new
+    )
+    context = {}
+
+    result = step.call(context)
+
+    assert_nil result
+    assert_equal "CSV file path: ", file_prompt.label
+    assert_equal "/tmp/input.csv", context[:file_path]
+    assert_equal ";", context[:col_sep]
+    assert_equal false, context[:headers_present]
+  end
+
+  def test_halts_when_separator_prompt_returns_nil
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::CollectInputsStep.new(
+      file_path_prompt: FakeFilePathPrompt.new,
+      separator_prompt: FakeSeparatorPrompt.new(nil),
+      headers_present_prompt: FakeHeadersPresentPrompt.new
+    )
+
+    result = step.call({})
+
+    assert_equal :halt, result
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/csv_stats/execute_step_test.rb

@@ -0,0 +1,65 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/csv_stats/execute_step"
+
+class CsvStatsExecuteStepTest < Minitest::Test
+  Result = Struct.new(:ok, :data) do
+    def ok? = ok
+  end
+
+  class FakeUseCase
+    def initialize(result)
+      @result = result
+    end
+
+    def call(session:)
+      @result
+    end
+  end
+
+  class FakePresenter
+    attr_reader :summary_data, :written_path
+
+    def print_summary(data)
+      @summary_data = data
+    end
+
+    def print_file_written(path)
+      @written_path = path
+    end
+  end
+
+  def test_prints_summary_and_file_path_when_present
+    presenter = FakePresenter.new
+    result = Result.new(true, { row_count: 2, output_path: "/tmp/stats.csv" })
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::ExecuteStep.new
+
+    outcome = step.call(
+      session: :session,
+      use_case: FakeUseCase.new(result),
+      presenter: presenter,
+      handle_error: ->(_result) { raise "unexpected" }
+    )
+
+    assert_nil outcome
+    assert_equal result.data, presenter.summary_data
+    assert_equal "/tmp/stats.csv", presenter.written_path
+  end
+
+  def test_halts_on_use_case_failure
+    fail_result = Result.new(false, { reason: :bad })
+    handled = []
+    step = Csvtool::Interface::CLI::Workflows::Steps::CsvStats::ExecuteStep.new
+
+    outcome = step.call(
+      session: :session,
+      use_case: FakeUseCase.new(fail_result),
+      presenter: FakePresenter.new,
+      handle_error: ->(result) { handled << result }
+    )
+
+    assert_equal :halt, outcome
+    assert_equal [fail_result], handled
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: csvops
 version: !ruby/object:Gem::Version
-  version: 0.7.0.alpha
+  version: 0.8.0.alpha
 platform: ruby
 authors:
 - Robert Hall
@@ -75,10 +75,12 @@ files:
 - docs/release-v0.5.0-alpha.md
 - docs/release-v0.6.0-alpha.md
 - docs/release-v0.7.0-alpha.md
+- docs/release-v0.8.0-alpha.md
 - exe/csvtool
 - lib/csvtool/application/use_cases/run_cross_csv_dedupe.rb
 - lib/csvtool/application/use_cases/run_csv_parity.rb
 - lib/csvtool/application/use_cases/run_csv_split.rb
+- lib/csvtool/application/use_cases/run_csv_stats.rb
 - lib/csvtool/application/use_cases/run_extraction.rb
 - lib/csvtool/application/use_cases/run_row_extraction.rb
 - lib/csvtool/application/use_cases/run_row_randomization.rb
@@ -101,6 +103,9 @@ files:
 - lib/csvtool/domain/csv_split_session/split_options.rb
 - lib/csvtool/domain/csv_split_session/split_session.rb
 - lib/csvtool/domain/csv_split_session/split_source.rb
+- lib/csvtool/domain/csv_stats_session/stats_options.rb
+- lib/csvtool/domain/csv_stats_session/stats_session.rb
+- lib/csvtool/domain/csv_stats_session/stats_source.rb
 - lib/csvtool/domain/row_randomization_session/randomization_options.rb
 - lib/csvtool/domain/row_randomization_session/randomization_session.rb
 - lib/csvtool/domain/row_randomization_session/randomization_source.rb
@@ -111,6 +116,7 @@ files:
 - lib/csvtool/infrastructure/csv/cross_csv_deduper.rb
 - lib/csvtool/infrastructure/csv/csv_parity_comparator.rb
 - lib/csvtool/infrastructure/csv/csv_splitter.rb
+- lib/csvtool/infrastructure/csv/csv_stats_scanner.rb
 - lib/csvtool/infrastructure/csv/header_reader.rb
 - lib/csvtool/infrastructure/csv/row_randomizer.rb
 - lib/csvtool/infrastructure/csv/row_streamer.rb
@@ -123,6 +129,7 @@ files:
 - lib/csvtool/infrastructure/output/csv_row_console_writer.rb
 - lib/csvtool/infrastructure/output/csv_row_file_writer.rb
 - lib/csvtool/infrastructure/output/csv_split_manifest_writer.rb
+- lib/csvtool/infrastructure/output/csv_stats_file_writer.rb
 - lib/csvtool/interface/cli/errors/presenter.rb
 - lib/csvtool/interface/cli/menu_loop.rb
 - lib/csvtool/interface/cli/prompts/chunk_size_prompt.rb
@@ -142,17 +149,20 @@ files:
 - lib/csvtool/interface/cli/workflows/builders/cross_csv_dedupe_session_builder.rb
 - lib/csvtool/interface/cli/workflows/builders/csv_parity_session_builder.rb
 - lib/csvtool/interface/cli/workflows/builders/csv_split_session_builder.rb
+- lib/csvtool/interface/cli/workflows/builders/csv_stats_session_builder.rb
 - lib/csvtool/interface/cli/workflows/builders/row_extraction_session_builder.rb
 - lib/csvtool/interface/cli/workflows/builders/row_randomization_session_builder.rb
 - lib/csvtool/interface/cli/workflows/presenters/column_extraction_presenter.rb
 - lib/csvtool/interface/cli/workflows/presenters/cross_csv_dedupe_presenter.rb
 - lib/csvtool/interface/cli/workflows/presenters/csv_parity_presenter.rb
 - lib/csvtool/interface/cli/workflows/presenters/csv_split_presenter.rb
+- lib/csvtool/interface/cli/workflows/presenters/csv_stats_presenter.rb
 - lib/csvtool/interface/cli/workflows/presenters/row_extraction_presenter.rb
 - lib/csvtool/interface/cli/workflows/presenters/row_randomization_presenter.rb
 - lib/csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow.rb
 - lib/csvtool/interface/cli/workflows/run_csv_parity_workflow.rb
 - lib/csvtool/interface/cli/workflows/run_csv_split_workflow.rb
+- lib/csvtool/interface/cli/workflows/run_csv_stats_workflow.rb
 - lib/csvtool/interface/cli/workflows/run_extraction_workflow.rb
 - lib/csvtool/interface/cli/workflows/run_row_extraction_workflow.rb
 - lib/csvtool/interface/cli/workflows/run_row_randomization_workflow.rb
@@ -164,6 +174,10 @@ files:
 - lib/csvtool/interface/cli/workflows/steps/csv_split/collect_manifest_step.rb
 - lib/csvtool/interface/cli/workflows/steps/csv_split/collect_output_step.rb
 - lib/csvtool/interface/cli/workflows/steps/csv_split/execute_step.rb
+- lib/csvtool/interface/cli/workflows/steps/csv_stats/build_session_step.rb
+- lib/csvtool/interface/cli/workflows/steps/csv_stats/collect_destination_step.rb
+- lib/csvtool/interface/cli/workflows/steps/csv_stats/collect_inputs_step.rb
+- lib/csvtool/interface/cli/workflows/steps/csv_stats/execute_step.rb
 - lib/csvtool/interface/cli/workflows/steps/extraction/build_preview_step.rb
 - lib/csvtool/interface/cli/workflows/steps/extraction/collect_destination_step.rb
 - lib/csvtool/interface/cli/workflows/steps/extraction/collect_inputs_step.rb
@@ -188,6 +202,7 @@ files:
 - test/csvtool/application/use_cases/run_cross_csv_dedupe_test.rb
 - test/csvtool/application/use_cases/run_csv_parity_test.rb
 - test/csvtool/application/use_cases/run_csv_split_test.rb
+- test/csvtool/application/use_cases/run_csv_stats_test.rb
 - test/csvtool/application/use_cases/run_extraction_test.rb
 - test/csvtool/application/use_cases/run_row_extraction_test.rb
 - test/csvtool/application/use_cases/run_row_randomization_test.rb
@@ -218,6 +233,7 @@ files:
 - test/csvtool/infrastructure/csv/cross_csv_deduper_test.rb
 - test/csvtool/infrastructure/csv/csv_parity_comparator_test.rb
 - test/csvtool/infrastructure/csv/csv_splitter_test.rb
+- test/csvtool/infrastructure/csv/csv_stats_scanner_test.rb
 - test/csvtool/infrastructure/csv/header_reader_test.rb
 - test/csvtool/infrastructure/csv/row_randomizer_test.rb
 - test/csvtool/infrastructure/csv/row_streamer_test.rb
@@ -230,6 +246,7 @@ files:
 - test/csvtool/infrastructure/output/csv_row_console_writer_test.rb
 - test/csvtool/infrastructure/output/csv_row_file_writer_test.rb
 - test/csvtool/infrastructure/output/csv_split_manifest_writer_test.rb
+- test/csvtool/infrastructure/output/csv_stats_file_writer_test.rb
 - test/csvtool/interface/cli/errors/presenter_test.rb
 - test/csvtool/interface/cli/menu_loop_test.rb
 - test/csvtool/interface/cli/prompts/chunk_size_prompt_test.rb
@@ -249,17 +266,20 @@ files:
 - test/csvtool/interface/cli/workflows/builders/cross_csv_dedupe_session_builder_test.rb
 - test/csvtool/interface/cli/workflows/builders/csv_parity_session_builder_test.rb
 - test/csvtool/interface/cli/workflows/builders/csv_split_session_builder_test.rb
+- test/csvtool/interface/cli/workflows/builders/csv_stats_session_builder_test.rb
 - test/csvtool/interface/cli/workflows/builders/row_extraction_session_builder_test.rb
 - test/csvtool/interface/cli/workflows/builders/row_randomization_session_builder_test.rb
 - test/csvtool/interface/cli/workflows/presenters/column_extraction_presenter_test.rb
 - test/csvtool/interface/cli/workflows/presenters/cross_csv_dedupe_presenter_test.rb
 - test/csvtool/interface/cli/workflows/presenters/csv_parity_presenter_test.rb
 - test/csvtool/interface/cli/workflows/presenters/csv_split_presenter_test.rb
+- test/csvtool/interface/cli/workflows/presenters/csv_stats_presenter_test.rb
 - test/csvtool/interface/cli/workflows/presenters/row_extraction_presenter_test.rb
 - test/csvtool/interface/cli/workflows/presenters/row_randomization_presenter_test.rb
 - test/csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow_test.rb
 - test/csvtool/interface/cli/workflows/run_csv_parity_workflow_test.rb
 - test/csvtool/interface/cli/workflows/run_csv_split_workflow_test.rb
+- test/csvtool/interface/cli/workflows/run_csv_stats_workflow_test.rb
 - test/csvtool/interface/cli/workflows/run_extraction_workflow_test.rb
 - test/csvtool/interface/cli/workflows/run_row_extraction_workflow_test.rb
 - test/csvtool/interface/cli/workflows/run_row_randomization_workflow_test.rb
@@ -269,6 +289,10 @@ files:
 - test/csvtool/interface/cli/workflows/steps/csv_split/collect_manifest_step_test.rb
 - test/csvtool/interface/cli/workflows/steps/csv_split/collect_output_step_test.rb
 - test/csvtool/interface/cli/workflows/steps/csv_split/execute_step_test.rb
+- test/csvtool/interface/cli/workflows/steps/csv_stats/build_session_step_test.rb
+- test/csvtool/interface/cli/workflows/steps/csv_stats/collect_destination_step_test.rb
+- test/csvtool/interface/cli/workflows/steps/csv_stats/collect_inputs_step_test.rb
+- test/csvtool/interface/cli/workflows/steps/csv_stats/execute_step_test.rb
 - test/csvtool/interface/cli/workflows/steps/extraction/collect_inputs_step_test.rb
 - test/csvtool/interface/cli/workflows/steps/parity/build_session_step_test.rb
 - test/csvtool/interface/cli/workflows/steps/parity/collect_inputs_step_test.rb