csvops 0.3.0.alpha → 0.5.0.alpha

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

@@ -20,14 +20,16 @@ class MenuLoopTest < Minitest::Test
     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\n4\n"),
+      stdin: StringIO.new("1\n5\n"),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Exit"],
+      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
+      randomize_rows_action: randomize_rows_action,
+      dedupe_action: dedupe_action
     )
 
     status = menu.run
@@ -36,6 +38,7 @@ class MenuLoopTest < Minitest::Test
     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"
   end
 
@@ -43,14 +46,16 @@ class MenuLoopTest < Minitest::Test
     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\n4\n"),
+      stdin: StringIO.new("2\n5\n"),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Exit"],
+      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
+      randomize_rows_action: randomize_rows_action,
+      dedupe_action: dedupe_action
     )
 
     status = menu.run
@@ -59,20 +64,23 @@ class MenuLoopTest < Minitest::Test
     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
   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\n4\n"),
+      stdin: StringIO.new("3\n5\n"),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Exit"],
+      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
+      randomize_rows_action: randomize_rows_action,
+      dedupe_action: dedupe_action
     )
 
     status = menu.run
@@ -81,27 +89,56 @@ class MenuLoopTest < Minitest::Test
     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
+  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
+    )
+
+    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
   end
 
   def test_invalid_choice_shows_prompt
     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("x\n4\n"),
+      stdin: StringIO.new("x\n5\n"),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Exit"],
+      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
+      randomize_rows_action: randomize_rows_action,
+      dedupe_action: dedupe_action
     )
 
     menu.run
 
-    assert_includes stdout.string, "Please choose 1, 2, 3, or 4."
+    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
   end
 end

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

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/prompts/dedupe_key_selector_prompt"
+
+class DedupeKeySelectorPromptTest < Minitest::Test
+  def test_builds_name_selector_in_header_mode
+    prompt = Csvtool::Interface::CLI::Prompts::DedupeKeySelectorPrompt.new(stdin: StringIO.new("customer_id\n"), stdout: StringIO.new)
+
+    selector = prompt.call(label: "Source", headers_present: true)
+
+    assert_equal true, selector.headers_present?
+    assert_equal "customer_id", selector.value
+  end
+
+  def test_builds_index_selector_in_headerless_mode
+    prompt = Csvtool::Interface::CLI::Prompts::DedupeKeySelectorPrompt.new(stdin: StringIO.new("2\n"), stdout: StringIO.new)
+
+    selector = prompt.call(label: "Reference", headers_present: false)
+
+    assert_equal true, selector.index?
+    assert_equal 2, selector.value
+  end
+
+  def test_returns_nil_for_invalid_selector
+    prompt = Csvtool::Interface::CLI::Prompts::DedupeKeySelectorPrompt.new(stdin: StringIO.new("\n"), stdout: StringIO.new)
+
+    assert_nil prompt.call(label: "Source", headers_present: true)
+  end
+end

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

@@ -8,4 +8,13 @@ class FilePathPromptTest < Minitest::Test
     prompt = Csvtool::Interface::CLI::Prompts::FilePathPrompt.new(stdin: StringIO.new("  /tmp/a.csv  \n"), stdout: StringIO.new)
     assert_equal "/tmp/a.csv", prompt.call
   end
+
+  def test_supports_custom_label
+    out = StringIO.new
+    prompt = Csvtool::Interface::CLI::Prompts::FilePathPrompt.new(stdin: StringIO.new("/tmp/a.csv\n"), stdout: out)
+
+    prompt.call(label: "Reference CSV file path: ")
+
+    assert_includes out.string, "Reference CSV file path: "
+  end
 end

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

@@ -11,4 +11,14 @@ class HeadersPresentPromptTest < Minitest::Test
     assert_equal true, yes_prompt.call
     assert_equal false, no_prompt.call
   end
+
+  def test_supports_custom_label
+    out = StringIO.new
+    prompt = Csvtool::Interface::CLI::Prompts::HeadersPresentPrompt.new(stdin: StringIO.new("yes\n"), stdout: out)
+
+    result = prompt.call(label: "Source headers present? [Y/n]: ")
+
+    assert_equal true, result
+    assert_includes out.string, "Source headers present? [Y/n]: "
+  end
 end

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

@@ -28,4 +28,14 @@ class SeparatorPromptTest < Minitest::Test
     assert_nil prompt.call
     assert_includes errors.calls, :empty_custom_separator
   end
+
+  def test_supports_custom_label
+    errors = FakeErrors.new
+    out = StringIO.new
+    prompt = Csvtool::Interface::CLI::Prompts::SeparatorPrompt.new(stdin: StringIO.new("\n"), stdout: out, errors: errors)
+
+    prompt.call(label: "Reference CSV separator:")
+
+    assert_includes out.string, "Reference CSV separator:"
+  end
 end

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

@@ -0,0 +1,22 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/prompts/yes_no_prompt"
+
+class YesNoPromptTest < Minitest::Test
+  def test_uses_default_for_blank_or_invalid
+    prompt_blank = Csvtool::Interface::CLI::Prompts::YesNoPrompt.new(stdin: StringIO.new("\n"), stdout: StringIO.new)
+    prompt_invalid = Csvtool::Interface::CLI::Prompts::YesNoPrompt.new(stdin: StringIO.new("maybe\n"), stdout: StringIO.new)
+
+    assert_equal true, prompt_blank.call(label: "Q? ", default: true)
+    assert_equal false, prompt_invalid.call(label: "Q? ", default: false)
+  end
+
+  def test_accepts_yes_and_no_inputs
+    prompt_yes = Csvtool::Interface::CLI::Prompts::YesNoPrompt.new(stdin: StringIO.new("y\n"), stdout: StringIO.new)
+    prompt_no = Csvtool::Interface::CLI::Prompts::YesNoPrompt.new(stdin: StringIO.new("no\n"), stdout: StringIO.new)
+
+    assert_equal true, prompt_yes.call(label: "Q? ", default: false)
+    assert_equal false, prompt_no.call(label: "Q? ", default: true)
+  end
+end

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

@@ -0,0 +1,17 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/column_session_builder"
+
+class ColumnSessionBuilderTest < Minitest::Test
+  def test_builds_column_session
+    builder = Csvtool::Interface::CLI::Workflows::Builders::ColumnSessionBuilder.new
+
+    session = builder.call(file_path: "/tmp/data.csv", col_sep: ",", column_name: "name", skip_blanks: true)
+
+    assert_equal "/tmp/data.csv", session.source.path
+    assert_equal ",", session.source.separator.value
+    assert_equal "name", session.column_selection.name
+    assert_equal true, session.options.skip_blanks?
+  end
+end

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

@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/cross_csv_dedupe_session_builder"
+require "csvtool/domain/cross_csv_dedupe_session/csv_profile"
+require "csvtool/domain/cross_csv_dedupe_session/column_selector"
+require "csvtool/domain/shared/output_destination"
+
+class CrossCsvDedupeSessionBuilderTest < Minitest::Test
+  def test_builds_cross_csv_dedupe_session
+    builder = Csvtool::Interface::CLI::Workflows::Builders::CrossCsvDedupeSessionBuilder.new
+    source = Csvtool::Domain::CrossCsvDedupeSession::CsvProfile.new(path: "/tmp/source.csv", separator: ",", headers_present: true)
+    reference = Csvtool::Domain::CrossCsvDedupeSession::CsvProfile.new(path: "/tmp/reference.csv", separator: ",", headers_present: true)
+    source_selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: true, input: "id")
+    reference_selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: true, input: "rid")
+    destination = Csvtool::Domain::Shared::OutputDestination.console
+
+    session = builder.call(
+      source: source,
+      reference: reference,
+      source_selector: source_selector,
+      reference_selector: reference_selector,
+      trim_whitespace: true,
+      case_insensitive: false,
+      destination: destination
+    )
+
+    assert_equal "/tmp/source.csv", session.source.path
+    assert_equal "/tmp/reference.csv", session.reference.path
+    assert_equal "id", session.key_mapping.source_selector.value
+    assert_equal "rid", session.key_mapping.reference_selector.value
+    assert_equal true, session.match_options.trim_whitespace?
+    assert_equal false, session.match_options.case_insensitive?
+    assert_equal true, session.output_destination.console?
+  end
+end

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

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/row_extraction_session_builder"
+require "csvtool/domain/row_session/row_range"
+require "csvtool/domain/shared/output_destination"
+
+class RowExtractionSessionBuilderTest < Minitest::Test
+  def test_builds_row_extraction_session
+    builder = Csvtool::Interface::CLI::Workflows::Builders::RowExtractionSessionBuilder.new
+    row_range = Csvtool::Domain::RowSession::RowRange.new(start_row: 2, end_row: 4)
+    destination = Csvtool::Domain::Shared::OutputDestination.console
+
+    session = builder.call(file_path: "/tmp/data.csv", col_sep: ";", row_range: row_range, destination: destination)
+
+    assert_equal "/tmp/data.csv", session.source.path
+    assert_equal ";", session.source.separator
+    assert_equal 2, session.row_range.start_row
+    assert_equal true, session.output_destination.console?
+  end
+end

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

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/row_randomization_session_builder"
+require "csvtool/domain/shared/output_destination"
+
+class RowRandomizationSessionBuilderTest < Minitest::Test
+  def test_builds_row_randomization_session
+    builder = Csvtool::Interface::CLI::Workflows::Builders::RowRandomizationSessionBuilder.new
+    destination = Csvtool::Domain::Shared::OutputDestination.file(path: "/tmp/out.csv")
+
+    session = builder.call(
+      file_path: "/tmp/data.csv",
+      col_sep: "\t",
+      headers_present: false,
+      seed: 12,
+      destination: destination
+    )
+
+    assert_equal "/tmp/data.csv", session.source.path
+    assert_equal "\t", session.source.separator
+    assert_equal false, session.source.headers_present?
+    assert_equal 12, session.options.seed
+    assert_equal true, session.output_destination.file?
+  end
+end

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

@@ -0,0 +1,24 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/column_extraction_presenter"
+
+class ColumnExtractionPresenterTest < Minitest::Test
+  def test_prints_value
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::ColumnExtractionPresenter.new(stdout: out)
+
+    presenter.print_value("Alice")
+
+    assert_equal "Alice\n", out.string
+  end
+
+  def test_prints_file_written_message
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::ColumnExtractionPresenter.new(stdout: out)
+
+    presenter.print_file_written("/tmp/names.csv")
+
+    assert_includes out.string, "Wrote output to /tmp/names.csv"
+  end
+end

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

@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/cross_csv_dedupe_presenter"
+
+class CrossCsvDedupePresenterTest < Minitest::Test
+  def test_prints_header_row_and_summary
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CrossCsvDedupePresenter.new(stdout: out, col_sep: ",")
+
+    presenter.print_header(["id", "name"])
+    presenter.print_row(["1", "Alice"])
+    presenter.print_summary(source_rows: 5, removed_rows: 3, kept_rows_count: 2)
+
+    assert_includes out.string, "\nid,name\n"
+    assert_includes out.string, "1,Alice"
+    assert_includes out.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+  end
+
+  def test_prints_zero_and_all_removed_messages
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CrossCsvDedupePresenter.new(stdout: out, col_sep: ",")
+
+    presenter.print_summary(source_rows: 5, removed_rows: 0, kept_rows_count: 5)
+    presenter.print_summary(source_rows: 5, removed_rows: 5, kept_rows_count: 0)
+
+    assert_includes out.string, "No rows removed; no matching keys found."
+    assert_includes out.string, "All source rows were removed by dedupe."
+  end
+end

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

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/row_extraction_presenter"
+
+class RowExtractionPresenterTest < Minitest::Test
+  def test_prints_header_once_then_rows
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::RowExtractionPresenter.new(
+      stdout: out,
+      headers: ["name", "city"],
+      col_sep: ","
+    )
+
+    presenter.print_row(["Alice", "London"])
+    presenter.print_row(["Bob", "Paris"])
+
+    assert_equal "name,city\nAlice,London\nBob,Paris\n", out.string
+  end
+
+  def test_prints_file_written_message
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::RowExtractionPresenter.new(
+      stdout: out,
+      headers: ["name"],
+      col_sep: ","
+    )
+
+    presenter.print_file_written("/tmp/out.csv")
+
+    assert_includes out.string, "Wrote output to /tmp/out.csv"
+  end
+end

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

@@ -0,0 +1,33 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/row_randomization_presenter"
+
+class RowRandomizationPresenterTest < Minitest::Test
+  def test_prints_console_start_and_rows
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::RowRandomizationPresenter.new(
+      stdout: out,
+      headers: ["name", "city"],
+      col_sep: ","
+    )
+
+    presenter.print_console_start
+    presenter.print_row(["Alice", "London"])
+
+    assert_equal "\nname,city\nAlice,London\n", out.string
+  end
+
+  def test_prints_file_written_message
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::RowRandomizationPresenter.new(
+      stdout: out,
+      headers: nil,
+      col_sep: ","
+    )
+
+    presenter.print_file_written("/tmp/out.csv")
+
+    assert_includes out.string, "Wrote output to /tmp/out.csv"
+  end
+end