csvops 0.5.0.alpha → 0.7.0.alpha

data/test/csvtool/interface/cli/menu_loop_test.rb

@@ -17,128 +17,122 @@ class MenuLoopTest < Minitest::Test
   end
 
   def test_routes_extract_column_then_exit
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("1\n5\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action
-    )
-
+    menu, actions, = build_menu("1\n7\n")
     status = menu.run
 
     assert_equal 0, status
-    assert_equal 1, column_action.runs
-    assert_equal 0, rows_action.runs
-    assert_equal 0, randomize_rows_action.runs
-    assert_equal 0, dedupe_action.runs
-    assert_includes stdout.string, "CSV Tool Menu"
+    assert_equal 1, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
   end
 
   def test_routes_extract_rows_then_exit
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("2\n5\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action
-    )
-
+    menu, actions, = build_menu("2\n7\n")
     status = menu.run
 
     assert_equal 0, status
-    assert_equal 0, column_action.runs
-    assert_equal 1, rows_action.runs
-    assert_equal 0, randomize_rows_action.runs
-    assert_equal 0, dedupe_action.runs
+    assert_equal 0, actions[:column].runs
+    assert_equal 1, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
   end
 
   def test_routes_randomize_rows_then_exit
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("3\n5\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action
-    )
-
+    menu, actions, = build_menu("3\n7\n")
     status = menu.run
 
     assert_equal 0, status
-    assert_equal 0, column_action.runs
-    assert_equal 0, rows_action.runs
-    assert_equal 1, randomize_rows_action.runs
-    assert_equal 0, dedupe_action.runs
+    assert_equal 0, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 1, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
   end
 
   def test_routes_dedupe_then_exit
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("4\n5\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action
-    )
+    menu, actions, = build_menu("4\n7\n")
+    status = menu.run
+
+    assert_equal 0, status
+    assert_equal 0, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 1, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
+  end
+
+  def test_routes_parity_then_exit
+    menu, actions, = build_menu("5\n7\n")
+    status = menu.run
+
+    assert_equal 0, status
+    assert_equal 0, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 1, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
+  end
 
+  def test_routes_split_then_exit
+    menu, actions, stdout = build_menu("6\n7\n")
     status = menu.run
 
     assert_equal 0, status
-    assert_equal 0, column_action.runs
-    assert_equal 0, rows_action.runs
-    assert_equal 0, randomize_rows_action.runs
-    assert_equal 1, dedupe_action.runs
+    assert_equal 0, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 1, actions[:split].runs
+    assert_includes stdout.string, "CSV Tool Menu"
   end
 
   def test_invalid_choice_shows_prompt
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
+    menu, actions, stdout = build_menu("x\n7\n")
+    menu.run
+
+    assert_includes stdout.string, "Please choose 1, 2, 3, 4, 5, 6, or 7."
+    assert_equal 0, actions[:column].runs
+    assert_equal 0, actions[:rows].runs
+    assert_equal 0, actions[:randomize].runs
+    assert_equal 0, actions[:dedupe].runs
+    assert_equal 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
+  end
+
+  private
+
+  def build_menu(input)
+    actions = {
+      column: FakeAction.new,
+      rows: FakeAction.new,
+      randomize: FakeAction.new,
+      dedupe: FakeAction.new,
+      parity: FakeAction.new,
+      split: FakeAction.new
+    }
     stdout = StringIO.new
+
     menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("x\n5\n"),
+      stdin: StringIO.new(input),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action
+      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Split CSV into chunks", "Exit"],
+      extract_column_action: actions[:column],
+      extract_rows_action: actions[:rows],
+      randomize_rows_action: actions[:randomize],
+      dedupe_action: actions[:dedupe],
+      parity_action: actions[:parity],
+      split_action: actions[:split]
     )
 
-    menu.run
-
-    assert_includes stdout.string, "Please choose 1, 2, 3, 4, or 5."
-    assert_equal 0, column_action.runs
-    assert_equal 0, rows_action.runs
-    assert_equal 0, randomize_rows_action.runs
-    assert_equal 0, dedupe_action.runs
+    [menu, actions, stdout]
   end
 end

data/test/csvtool/interface/cli/prompts/chunk_size_prompt_test.rb

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/prompts/chunk_size_prompt"
+
+class ChunkSizePromptTest < Minitest::Test
+  def test_returns_entered_value
+    out = StringIO.new
+    prompt = Csvtool::Interface::CLI::Prompts::ChunkSizePrompt.new(
+      stdin: StringIO.new("25\n"),
+      stdout: out
+    )
+
+    assert_equal "25", prompt.call
+    assert_includes out.string, "Rows per chunk: "
+  end
+end

data/test/csvtool/interface/cli/prompts/split_manifest_prompt_test.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/prompts/split_manifest_prompt"
+
+class SplitManifestPromptTest < Minitest::Test
+  class FakeYesNoPrompt
+    def initialize(value)
+      @value = value
+    end
+
+    def call(label:, default:)
+      @value.nil? ? default : @value
+    end
+  end
+
+  def test_returns_disabled_when_user_declines_manifest
+    prompt = Csvtool::Interface::CLI::Prompts::SplitManifestPrompt.new(
+      stdin: StringIO.new,
+      stdout: StringIO.new,
+      yes_no_prompt: FakeYesNoPrompt.new(false)
+    )
+
+    result = prompt.call(default_path: "/tmp/manifest.csv")
+
+    assert_equal false, result[:write_manifest]
+    assert_nil result[:manifest_path]
+  end
+
+  def test_uses_default_manifest_path_when_blank
+    prompt = Csvtool::Interface::CLI::Prompts::SplitManifestPrompt.new(
+      stdin: StringIO.new("\n"),
+      stdout: StringIO.new,
+      yes_no_prompt: FakeYesNoPrompt.new(true)
+    )
+
+    result = prompt.call(default_path: "/tmp/manifest.csv")
+
+    assert_equal true, result[:write_manifest]
+    assert_equal "/tmp/manifest.csv", result[:manifest_path]
+  end
+end

data/test/csvtool/interface/cli/prompts/split_output_prompt_test.rb

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/prompts/split_output_prompt"
+require "csvtool/interface/cli/prompts/yes_no_prompt"
+
+class SplitOutputPromptTest < Minitest::Test
+  def test_uses_defaults_for_blank_values
+    out = StringIO.new
+    prompt = Csvtool::Interface::CLI::Prompts::SplitOutputPrompt.new(
+      stdin: StringIO.new("\n\n\n"),
+      stdout: out,
+      yes_no_prompt: Csvtool::Interface::CLI::Prompts::YesNoPrompt.new(stdin: StringIO.new("\n"), stdout: StringIO.new)
+    )
+
+    result = prompt.call(default_directory: "/tmp/out", default_prefix: "people")
+
+    assert_equal "/tmp/out", result[:output_directory]
+    assert_equal "people", result[:file_prefix]
+    assert_equal false, result[:overwrite_existing]
+  end
+end

data/test/csvtool/interface/cli/workflows/builders/csv_parity_session_builder_test.rb

@@ -0,0 +1,20 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/csv_parity_session_builder"
+
+class CsvParitySessionBuilderTest < Minitest::Test
+  def test_builds_parity_session
+    session = Csvtool::Interface::CLI::Workflows::Builders::CsvParitySessionBuilder.new.call(
+      left_path: "/tmp/left.csv",
+      right_path: "/tmp/right.csv",
+      col_sep: "\t",
+      headers_present: false
+    )
+
+    assert_equal "/tmp/left.csv", session.source_pair.left_path
+    assert_equal "/tmp/right.csv", session.source_pair.right_path
+    assert_equal "\t", session.options.separator
+    assert_equal false, session.options.headers_present?
+  end
+end

data/test/csvtool/interface/cli/workflows/builders/csv_split_session_builder_test.rb

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/csv_split_session_builder"
+
+class CsvSplitSessionBuilderTest < Minitest::Test
+  def test_builds_split_session
+    session = Csvtool::Interface::CLI::Workflows::Builders::CsvSplitSessionBuilder.new.call(
+      file_path: "/tmp/people.csv",
+      col_sep: ",",
+      headers_present: true,
+      chunk_size: 10,
+      output_directory: "/tmp/out",
+      file_prefix: "batch",
+      overwrite_existing: true,
+      write_manifest: true,
+      manifest_path: "/tmp/out/manifest.csv"
+    )
+
+    assert_equal "/tmp/people.csv", session.source.path
+    assert_equal ",", session.source.separator
+    assert_equal true, session.source.headers_present
+    assert_equal 10, session.options.chunk_size
+    assert_equal "/tmp/out", session.options.output_directory
+    assert_equal "batch", session.options.file_prefix
+    assert_equal true, session.options.overwrite_existing
+    assert_equal true, session.options.write_manifest
+    assert_equal "/tmp/out/manifest.csv", session.options.manifest_path
+  end
+end

data/test/csvtool/interface/cli/workflows/presenters/csv_parity_presenter_test.rb

@@ -0,0 +1,43 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/csv_parity_presenter"
+
+class CsvParityPresenterTest < Minitest::Test
+  def test_prints_match_summary
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CsvParityPresenter.new(stdout: out)
+
+    presenter.print_summary(
+      match: true,
+      left_rows: 3,
+      right_rows: 3,
+      left_only_count: 0,
+      right_only_count: 0
+    )
+
+    assert_includes out.string, "MATCH"
+    assert_includes out.string, "Summary: left_rows=3 right_rows=3 left_only=0 right_only=0"
+  end
+
+  def test_prints_mismatch_examples
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CsvParityPresenter.new(stdout: out)
+
+    presenter.print_summary(
+      match: false,
+      left_rows: 3,
+      right_rows: 3,
+      left_only_count: 1,
+      right_only_count: 1,
+      left_only_examples: [{ row: "Cara,Berlin", count_delta: 1 }],
+      right_only_examples: [{ row: "Dina,Rome", count_delta: 1 }]
+    )
+
+    assert_includes out.string, "MISMATCH"
+    assert_includes out.string, "Left-only examples:"
+    assert_includes out.string, "Cara,Berlin (count +1)"
+    assert_includes out.string, "Right-only examples:"
+    assert_includes out.string, "Dina,Rome (count +1)"
+  end
+end

data/test/csvtool/interface/cli/workflows/presenters/csv_split_presenter_test.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/csv_split_presenter"
+
+class CsvSplitPresenterTest < Minitest::Test
+  def test_prints_summary_and_chunk_paths
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CsvSplitPresenter.new(stdout: out)
+
+    presenter.print_summary(
+      chunk_size: 10,
+      data_rows: 25,
+      chunk_count: 3,
+      manifest_path: "/tmp/manifest.csv",
+      chunk_paths: ["/tmp/people_part_001.csv", "/tmp/people_part_002.csv", "/tmp/people_part_003.csv"]
+    )
+
+    assert_includes out.string, "Split complete."
+    assert_includes out.string, "Chunk size: 10"
+    assert_includes out.string, "Data rows: 25"
+    assert_includes out.string, "Chunks written: 3"
+    assert_includes out.string, "Manifest: /tmp/manifest.csv"
+    assert_includes out.string, "/tmp/people_part_001.csv"
+  end
+end

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

@@ -0,0 +1,94 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_csv_parity_workflow"
+
+class RunCsvParityWorkflowTest < Minitest::Test
+  class FakeUseCase
+    attr_reader :calls
+
+    def initialize
+      @calls = []
+    end
+
+    def call(session:)
+      @calls << session
+      Struct.new(:ok?, :data).new(true, {
+        match: true,
+        left_rows: 3,
+        right_rows: 3,
+        left_only_count: 0,
+        right_only_count: 0
+      })
+    end
+  end
+
+  class MismatchUseCase
+    def call(session:)
+      Struct.new(:ok?, :data).new(true, {
+        match: false,
+        left_rows: 3,
+        right_rows: 3,
+        left_only_count: 1,
+        right_only_count: 1,
+        left_only_examples: [{ row: "Cara,Berlin", count_delta: 1 }],
+        right_only_examples: [{ row: "Dina,Rome", count_delta: 1 }]
+      })
+    end
+  end
+
+  class CannotReadUseCase
+    def call(session:)
+      Struct.new(:ok?, :error, :data).new(false, :cannot_read_file, { path: "/tmp/protected.csv" })
+    end
+  end
+
+  def test_prompts_for_paths_and_calls_use_case
+    stdout = StringIO.new
+    use_case = FakeUseCase.new
+    input = StringIO.new("/tmp/left.csv\n/tmp/right.csv\n2\ny\n")
+
+    Csvtool::Interface::CLI::Workflows::RunCsvParityWorkflow
+      .new(stdin: input, stdout: stdout, use_case: use_case)
+      .call
+
+    call = use_case.calls.first
+    assert_equal "/tmp/left.csv", call.source_pair.left_path
+    assert_equal "/tmp/right.csv", call.source_pair.right_path
+    assert_equal "\t", call.options.separator
+    assert_equal true, call.options.headers_present?
+    assert_includes stdout.string, "Left CSV file path: "
+    assert_includes stdout.string, "Right CSV file path: "
+    assert_includes stdout.string, "Choose separator:"
+    assert_includes stdout.string, "Headers present? [Y/n]: "
+    assert_includes stdout.string, "MATCH"
+    assert_includes stdout.string, "Summary: left_rows=3 right_rows=3 left_only=0 right_only=0"
+  end
+
+  def test_prints_mismatch_examples_when_not_equal
+    stdout = StringIO.new
+    input = StringIO.new("/tmp/left.csv\n/tmp/right.csv\n\ny\n")
+
+    Csvtool::Interface::CLI::Workflows::RunCsvParityWorkflow
+      .new(stdin: input, stdout: stdout, use_case: MismatchUseCase.new)
+      .call
+
+    assert_includes stdout.string, "MISMATCH"
+    assert_includes stdout.string, "Left-only examples:"
+    assert_includes stdout.string, "Cara,Berlin (count +1)"
+    assert_includes stdout.string, "Right-only examples:"
+    assert_includes stdout.string, "Dina,Rome (count +1)"
+  end
+
+  def test_prints_cannot_read_error_without_stacktrace
+    stdout = StringIO.new
+    input = StringIO.new("/tmp/left.csv\n/tmp/right.csv\n\ny\n")
+
+    Csvtool::Interface::CLI::Workflows::RunCsvParityWorkflow
+      .new(stdin: input, stdout: stdout, use_case: CannotReadUseCase.new)
+      .call
+
+    assert_includes stdout.string, "Cannot read file: /tmp/protected.csv"
+    refute_includes stdout.string, "Traceback"
+  end
+end