csvops 0.6.0.alpha → 0.8.0.alpha

data/test/csvtool/infrastructure/output/csv_stats_file_writer_test.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+require_relative "../../../test_helper"
+require "csvtool/infrastructure/output/csv_stats_file_writer"
+require "tmpdir"
+
+class InfrastructureCsvStatsFileWriterTest < Minitest::Test
+  def test_writes_stats_as_metric_value_csv
+    writer = Csvtool::Infrastructure::Output::CsvStatsFileWriter.new
+
+    Dir.mktmpdir do |dir|
+      path = File.join(dir, "stats.csv")
+      writer.call(
+        path: path,
+        data: {
+          row_count: 3,
+          column_count: 2,
+          headers: ["name", "city"],
+          column_stats: [
+            { name: "name", non_blank_count: 3, blank_count: 0 },
+            { name: "city", non_blank_count: 2, blank_count: 1 }
+          ]
+        }
+      )
+
+      assert_equal [
+        "metric,value",
+        "row_count,3",
+        "column_count,2",
+        "headers,name|city",
+        "column.name.non_blank,3",
+        "column.name.blank,0",
+        "column.city.non_blank,2",
+        "column.city.blank,1"
+      ], File.read(path).lines.map(&:strip)
+    end
+  end
+end

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

@@ -17,171 +17,145 @@ 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
-    parity_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("1\n6\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
-    )
-
+    menu, actions, = build_menu("1\n8\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_equal 0, parity_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
+    assert_equal 0, actions[:stats].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
-    parity_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("2\n6\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
-    )
-
+    menu, actions, = build_menu("2\n8\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, parity_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
+    assert_equal 0, actions[:stats].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
-    parity_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("3\n6\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
-    )
-
+    menu, actions, = build_menu("3\n8\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, parity_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
+    assert_equal 0, actions[:stats].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
-    parity_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("4\n6\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
-    )
-
+    menu, actions, = build_menu("4\n8\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, parity_action.runs
+    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
+    assert_equal 0, actions[:stats].runs
   end
 
   def test_routes_parity_then_exit
-    column_action = FakeAction.new
-    rows_action = FakeAction.new
-    randomize_rows_action = FakeAction.new
-    dedupe_action = FakeAction.new
-    parity_action = FakeAction.new
-    stdout = StringIO.new
-    menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("5\n6\n"),
-      stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
-    )
+    menu, actions, = build_menu("5\n8\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
+    assert_equal 0, actions[:stats].runs
+  end
+
+  def test_routes_split_then_exit
+    menu, actions, stdout = build_menu("6\n8\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 0, dedupe_action.runs
-    assert_equal 1, parity_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_equal 0, actions[:stats].runs
+    assert_includes stdout.string, "CSV Tool Menu"
+  end
+
+  def test_routes_stats_then_exit
+    menu, actions, = build_menu("7\n8\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 0, actions[:parity].runs
+    assert_equal 0, actions[:split].runs
+    assert_equal 1, actions[:stats].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
-    parity_action = FakeAction.new
+    menu, actions, stdout = build_menu("x\n8\n")
+    menu.run
+
+    assert_includes stdout.string, "Please choose 1, 2, 3, 4, 5, 6, 7, or 8."
+    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
+    assert_equal 0, actions[:stats].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,
+      stats: FakeAction.new
+    }
     stdout = StringIO.new
+
     menu = Csvtool::Interface::CLI::MenuLoop.new(
-      stdin: StringIO.new("x\n6\n"),
+      stdin: StringIO.new(input),
       stdout: stdout,
-      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Exit"],
-      extract_column_action: column_action,
-      extract_rows_action: rows_action,
-      randomize_rows_action: randomize_rows_action,
-      dedupe_action: dedupe_action,
-      parity_action: parity_action
+      menu_options: ["Extract column", "Extract rows (range)", "Randomize rows", "Dedupe using another CSV", "Validate parity", "Split CSV into chunks", "CSV stats summary", "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],
+      stats_action: actions[:stats]
     )
 
-    menu.run
-
-    assert_includes stdout.string, "Please choose 1, 2, 3, 4, 5, or 6."
-    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
-    assert_equal 0, parity_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_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/builders/csv_stats_session_builder_test.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/builders/csv_stats_session_builder"
+require "csvtool/domain/shared/output_destination"
+
+class CsvStatsSessionBuilderTest < Minitest::Test
+  def test_builds_stats_session
+    builder = Csvtool::Interface::CLI::Workflows::Builders::CsvStatsSessionBuilder.new
+    destination = Csvtool::Domain::Shared::OutputDestination.console
+
+    session = builder.call(file_path: "/tmp/data.csv", col_sep: ";", headers_present: false, destination: destination)
+
+    assert_equal "/tmp/data.csv", session.source.path
+    assert_equal ";", session.source.separator
+    assert_equal false, session.source.headers_present
+    assert_equal true, session.output_destination.console?
+  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/presenters/csv_stats_presenter_test.rb

@@ -0,0 +1,37 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../test_helper"
+require "csvtool/interface/cli/workflows/presenters/csv_stats_presenter"
+
+class CsvStatsPresenterTest < Minitest::Test
+  def test_prints_summary_with_headers_and_column_stats
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CsvStatsPresenter.new(stdout: out)
+
+    presenter.print_summary(
+      row_count: 3,
+      column_count: 2,
+      headers: ["name", "city"],
+      column_stats: [
+        { name: "name", non_blank_count: 3, blank_count: 0 },
+        { name: "city", non_blank_count: 2, blank_count: 1 }
+      ]
+    )
+
+    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, "name: non_blank=3 blank=0"
+    assert_includes out.string, "city: non_blank=2 blank=1"
+  end
+
+  def test_prints_file_written_message
+    out = StringIO.new
+    presenter = Csvtool::Interface::CLI::Workflows::Presenters::CsvStatsPresenter.new(stdout: out)
+
+    presenter.print_file_written("/tmp/stats.csv")
+
+    assert_includes out.string, "Wrote output to /tmp/stats.csv"
+  end
+end