csvops 0.3.0.alpha → 0.5.0.alpha

data/test/csvtool/application/use_cases/run_row_randomization_test.rb

@@ -2,123 +2,133 @@
 
 require_relative "../../../test_helper"
 require "csvtool/application/use_cases/run_row_randomization"
+require "csvtool/domain/row_randomization_session/randomization_source"
+require "csvtool/domain/row_randomization_session/randomization_options"
+require "csvtool/domain/row_randomization_session/randomization_session"
+require "csvtool/domain/shared/output_destination"
 require "tmpdir"
 
 class RunRowRandomizationTest < Minitest::Test
-  def test_prints_header_then_all_randomized_rows
-    fixture = File.expand_path("../../../fixtures/sample_people.csv", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n\n\n\n\n")
-
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
-
-    assert_includes output.string, "CSV file path:"
-    header_index = output.string.index("name,city")
-    assert header_index
-    %w[Alice,London Bob,Paris Cara,Berlin].each do |row|
-      row_index = output.string.index(row)
-      assert row_index
-      assert_operator header_index, :<, row_index
+  class RaisingWriter
+    def call(**_kwargs)
+      raise Errno::ENOENT
     end
   end
 
-  def test_missing_file_shows_friendly_error
-    output = StringIO.new
-    input = StringIO.new("/tmp/does-not-exist.csv\n")
-
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
+  def fixture_path(name)
+    File.expand_path("../../../fixtures/#{name}", __dir__)
+  end
 
-    assert_includes output.string, "File not found: /tmp/does-not-exist.csv"
+  def build_session(file_path:, separator: ",", headers_present: true, seed: nil, output: :console, output_path: nil)
+    source = Csvtool::Domain::RowRandomizationSession::RandomizationSource.new(
+      path: file_path,
+      separator: separator,
+      headers_present: headers_present
+    )
+    options = Csvtool::Domain::RowRandomizationSession::RandomizationOptions.new(seed: seed)
+    session = Csvtool::Domain::RowRandomizationSession::RandomizationSession.start(source: source, options: options)
+
+    session.with_output_destination(
+      if output == :file
+        Csvtool::Domain::Shared::OutputDestination.file(path: output_path)
+      else
+        Csvtool::Domain::Shared::OutputDestination.console
+      end
+    )
   end
 
-  def test_can_write_randomized_rows_to_file
-    fixture = File.expand_path("../../../fixtures/sample_people.csv", __dir__)
-    output = StringIO.new
+  def test_read_headers_returns_headers_when_enabled
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
 
-    Dir.mktmpdir do |dir|
-      output_path = File.join(dir, "randomized.csv")
-      input = StringIO.new("#{fixture}\n\n\n\n2\n#{output_path}\n")
+    result = use_case.read_headers(file_path: fixture_path("sample_people.csv"), col_sep: ",", headers_present: true)
 
-      Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
-
-      written = File.read(output_path).lines.map(&:strip)
-      assert_equal "name,city", written.first
-      assert_equal ["Alice,London", "Bob,Paris", "Cara,Berlin"].sort, written[1..].sort
-      assert_includes output.string, "Wrote output to #{output_path}"
-    end
+    assert result.ok?
+    assert_equal ["name", "city"], result.data[:headers]
   end
 
-  def test_supports_tsv_separator
-    fixture = File.expand_path("../../../fixtures/sample_people.tsv", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n2\n\n\n\n")
+  def test_read_headers_returns_nil_when_headers_disabled
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
 
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
+    result = use_case.read_headers(file_path: fixture_path("sample_people_no_headers.csv"), col_sep: ",", headers_present: false)
 
-    assert_includes output.string, "name\tcity"
-    assert_includes output.string, "Alice\tLondon"
-    assert_includes output.string, "Bob\tParis"
-    assert_includes output.string, "Cara\tBerlin"
+    assert result.ok?
+    assert_nil result.data[:headers]
   end
 
-  def test_supports_custom_separator
-    fixture = File.expand_path("../../../fixtures/sample_people_colon.txt", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n5\n:\n\n\n\n")
+  def test_read_headers_fails_for_missing_file
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
 
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
+    result = use_case.read_headers(file_path: "/tmp/not-present.csv", col_sep: ",", headers_present: true)
 
-    assert_includes output.string, "name:city"
-    assert_includes output.string, "Alice:London"
-    assert_includes output.string, "Bob:Paris"
-    assert_includes output.string, "Cara:Berlin"
+    refute result.ok?
+    assert_equal :file_not_found, result.error
   end
 
-  def test_headerless_mode_randomizes_all_rows
-    fixture = File.expand_path("../../../fixtures/sample_people_no_headers.csv", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n\nn\n\n\n")
+  def test_randomize_streams_rows_for_console_mode
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
+    session = build_session(file_path: fixture_path("sample_people.csv"), seed: 123)
+    rows = []
 
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
+    result = use_case.randomize(session: session, headers: ["name", "city"], on_row: ->(fields) { rows << fields })
 
-    refute_includes output.string, "name,city"
-    assert_includes output.string, "Alice,London"
-    assert_includes output.string, "Bob,Paris"
-    assert_includes output.string, "Cara,Berlin"
+    assert result.ok?
+    assert_equal 3, rows.length
+    assert_equal [["Alice", "London"], ["Bob", "Paris"], ["Cara", "Berlin"]].sort, rows.sort
   end
 
-  def test_same_seed_produces_same_output_order
-    fixture = File.expand_path("../../../fixtures/sample_people_many.csv", __dir__)
-    input_data = "#{fixture}\n\n\n123\n\n"
-
-    out1 = StringIO.new
-    out2 = StringIO.new
+  def test_randomize_writes_rows_to_file
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
 
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: StringIO.new(input_data), stdout: out1).call
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: StringIO.new(input_data), stdout: out2).call
-
-    rows1 = out1.string.lines.map(&:strip).select { |line| line.include?(",") && !line.start_with?("name,city") }
-    rows2 = out2.string.lines.map(&:strip).select { |line| line.include?(",") && !line.start_with?("name,city") }
-    assert_equal rows1, rows2
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "randomized.csv")
+      session = build_session(
+        file_path: fixture_path("sample_people.csv"),
+        seed: 123,
+        output: :file,
+        output_path: output_path
+      )
+
+      result = use_case.randomize(session: session, headers: ["name", "city"])
+
+      assert result.ok?
+      assert_equal output_path, result.data[:output_path]
+      lines = File.read(output_path).lines.map(&:strip)
+      assert_equal "name,city", lines.first
+      assert_equal ["Alice,London", "Bob,Paris", "Cara,Berlin"].sort, lines[1..].sort
+    end
   end
 
-  def test_invalid_seed_shows_friendly_error
-    fixture = File.expand_path("../../../fixtures/sample_people.csv", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n\n\nabc\n")
+  def test_same_seed_produces_stable_order
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new
+    session_1 = build_session(file_path: fixture_path("sample_people_many.csv"), seed: 123)
+    session_2 = build_session(file_path: fixture_path("sample_people_many.csv"), seed: 123)
+    rows_1 = []
+    rows_2 = []
 
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
+    result_1 = use_case.randomize(session: session_1, headers: ["name", "city"], on_row: ->(fields) { rows_1 << fields })
+    result_2 = use_case.randomize(session: session_2, headers: ["name", "city"], on_row: ->(fields) { rows_2 << fields })
 
-    assert_includes output.string, "Seed must be an integer."
+    assert result_1.ok?
+    assert result_2.ok?
+    assert_equal rows_1, rows_2
   end
 
-  def test_malformed_csv_shows_friendly_error
-    fixture = File.expand_path("../../../fixtures/sample_people_bad_tail.csv", __dir__)
-    output = StringIO.new
-    input = StringIO.new("#{fixture}\n\n\n\n\n")
-
-    Csvtool::Application::UseCases::RunRowRandomization.new(stdin: input, stdout: output).call
-
-    assert_includes output.string, "Could not parse CSV file."
+  def test_randomize_returns_cannot_write_output_file_when_writer_fails
+    use_case = Csvtool::Application::UseCases::RunRowRandomization.new(
+      csv_randomized_row_file_writer: RaisingWriter.new
+    )
+    session = build_session(
+      file_path: fixture_path("sample_people.csv"),
+      seed: 123,
+      output: :file,
+      output_path: "/tmp/randomized.csv"
+    )
+
+    result = use_case.randomize(session: session, headers: ["name", "city"])
+
+    refute result.ok?
+    assert_equal :cannot_write_output_file, result.error
+    assert_equal "/tmp/randomized.csv", result.data[:path]
+    assert_equal Errno::ENOENT, result.data[:error_class]
   end
 end

data/test/csvtool/cli_test.rb

@@ -11,7 +11,7 @@ class TestCli < Minitest::Test
 
   def test_menu_can_exit_cleanly
     output = StringIO.new
-    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new("4\n"), stdout: output, stderr: StringIO.new)
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new("5\n"), stdout: output, stderr: StringIO.new)
     assert_equal 0, status
     assert_includes output.string, "CSV Tool Menu"
   end
@@ -26,7 +26,7 @@ class TestCli < Minitest::Test
       "",
       "y",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     output = StringIO.new
@@ -58,7 +58,7 @@ class TestCli < Minitest::Test
       "2",
       "3",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -79,7 +79,7 @@ class TestCli < Minitest::Test
       "0",
       "3",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -98,7 +98,7 @@ class TestCli < Minitest::Test
       "2",
       "3",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -119,7 +119,7 @@ class TestCli < Minitest::Test
       "2",
       "3",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -144,7 +144,7 @@ class TestCli < Minitest::Test
         "3",
         "2",
         output_path,
-        "4"
+        "5"
       ].join("\n") + "\n"
 
       status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -164,7 +164,7 @@ class TestCli < Minitest::Test
       "1",
       "2",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -184,7 +184,7 @@ class TestCli < Minitest::Test
       "",
       "",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -209,7 +209,7 @@ class TestCli < Minitest::Test
         "",
         "2",
         output_path,
-        "4"
+        "5"
       ].join("\n") + "\n"
 
       status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -231,7 +231,7 @@ class TestCli < Minitest::Test
       "",
       "",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -250,7 +250,7 @@ class TestCli < Minitest::Test
       "n",
       "",
       "",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -270,7 +270,7 @@ class TestCli < Minitest::Test
       "",
       "",
       "abc",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -280,6 +280,120 @@ class TestCli < Minitest::Test
     assert_operator output.string.scan("CSV Tool Menu").length, :>=, 2
   end
 
+  def test_dedupe_workflow_shell_prompts_and_returns_to_menu
+    output = StringIO.new
+    input = [
+      "4",
+      fixture_path("dedupe_source.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      "",
+      "5"
+    ].join("\n") + "\n"
+
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
+
+    assert_equal 0, status
+    assert_includes output.string, "Reference CSV file path:"
+    assert_includes output.string, "Source key column name:"
+    assert_includes output.string, "Reference key column name:"
+    assert_includes output.string, "customer_id,name"
+    assert_includes output.string, "1,Alice"
+    assert_includes output.string, "3,Cara"
+    assert_includes output.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+  end
+
+  def test_dedupe_workflow_can_write_to_file
+    output = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "deduped.csv")
+      input = [
+        "4",
+        fixture_path("dedupe_source.csv"),
+        "",
+        "",
+        fixture_path("dedupe_reference.csv"),
+        "",
+        "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      "2",
+      output_path,
+      "5"
+      ].join("\n") + "\n"
+
+      status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
+
+      assert_equal 0, status
+      assert_includes output.string, "Wrote output to #{output_path}"
+      assert_equal "customer_id,name\n1,Alice\n3,Cara\n", File.read(output_path)
+      assert_includes output.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+    end
+  end
+
+  def test_dedupe_workflow_supports_tsv_separators
+    output = StringIO.new
+    input = [
+      "4",
+      fixture_path("dedupe_source.tsv"),
+      "2",
+      "",
+      fixture_path("dedupe_reference.tsv"),
+      "2",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      "",
+      "5"
+    ].join("\n") + "\n"
+
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
+
+    assert_equal 0, status
+    assert_includes output.string, "customer_id\tname"
+    assert_includes output.string, "1\tAlice"
+    assert_includes output.string, "3\tCara"
+  end
+
+  def test_dedupe_workflow_headerless_mode_supports_index
+    output = StringIO.new
+    input = [
+      "4",
+      fixture_path("dedupe_source_no_headers.csv"),
+      "",
+      "n",
+      fixture_path("dedupe_reference_no_headers.csv"),
+      "",
+      "n",
+      "1",
+      "1",
+      "",
+      "",
+      "",
+      "5"
+    ].join("\n") + "\n"
+
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
+
+    assert_equal 0, status
+    refute_includes output.string, "customer_id,name"
+    assert_includes output.string, "1,Alice"
+    assert_includes output.string, "3,Cara"
+    assert_includes output.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+  end
+
   def test_end_to_end_file_output_writes_expected_csv
     output = StringIO.new
     output_path = nil
@@ -296,7 +410,7 @@ class TestCli < Minitest::Test
         "y",
         "2",
         output_path,
-        "4"
+        "5"
       ].join("\n") + "\n"
 
       status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: output, stderr: StringIO.new)
@@ -316,7 +430,7 @@ class TestCli < Minitest::Test
       "1",
       "",
       "n",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     output = StringIO.new
@@ -352,7 +466,7 @@ class TestCli < Minitest::Test
       "y",
       "2",
       "/tmp/not-a-dir/out.csv",
-      "4"
+      "5"
     ].join("\n") + "\n"
 
     output = StringIO.new

data/test/csvtool/cli_unit_test.rb

@@ -16,7 +16,7 @@ class CliUnitTest < Minitest::Test
   end
 
   def test_menu_command_can_exit_zero
-    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new("4\n"), stdout: StringIO.new, stderr: StringIO.new)
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new("5\n"), stdout: StringIO.new, stderr: StringIO.new)
     assert_equal 0, status
   end
 
@@ -28,7 +28,7 @@ class CliUnitTest < Minitest::Test
   def test_menu_routes_to_row_range_shell
     stdout = StringIO.new
     fixture = File.expand_path("../fixtures/sample_people.csv", __dir__)
-    input = ["2", fixture, "", "2", "3", "", "4"].join("\n") + "\n"
+    input = ["2", fixture, "", "2", "3", "", "5"].join("\n") + "\n"
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: stdout, stderr: StringIO.new)
     assert_equal 0, status
     assert_includes stdout.string, "name,city"
@@ -39,7 +39,7 @@ class CliUnitTest < Minitest::Test
   def test_menu_routes_to_randomize_rows_shell
     stdout = StringIO.new
     fixture = File.expand_path("../fixtures/sample_people.csv", __dir__)
-    input = ["3", fixture, "", "", "", "", "4"].join("\n") + "\n"
+    input = ["3", fixture, "", "", "", "", "5"].join("\n") + "\n"
     status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: stdout, stderr: StringIO.new)
     assert_equal 0, status
     assert_includes stdout.string, "name,city"
@@ -47,4 +47,17 @@ class CliUnitTest < Minitest::Test
     assert_includes stdout.string, "Bob,Paris"
     assert_includes stdout.string, "Cara,Berlin"
   end
+
+  def test_menu_routes_to_dedupe_shell
+    stdout = StringIO.new
+    source_fixture = File.expand_path("../fixtures/dedupe_source.csv", __dir__)
+    reference_fixture = File.expand_path("../fixtures/dedupe_reference.csv", __dir__)
+    input = ["4", source_fixture, "", "", reference_fixture, "", "", "customer_id", "external_id", "", "", "", "5"].join("\n") + "\n"
+    status = Csvtool::CLI.start(["menu"], stdin: StringIO.new(input), stdout: stdout, stderr: StringIO.new)
+    assert_equal 0, status
+    assert_includes stdout.string, "customer_id,name"
+    assert_includes stdout.string, "1,Alice"
+    assert_includes stdout.string, "3,Cara"
+    assert_includes stdout.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+  end
 end

data/test/csvtool/domain/column_session/column_session_test.rb

@@ -8,7 +8,7 @@ require "csvtool/domain/column_session/column_selection"
 require "csvtool/domain/column_session/extraction_options"
 require "csvtool/domain/column_session/preview"
 require "csvtool/domain/column_session/extraction_value"
-require "csvtool/domain/column_session/output_destination"
+require "csvtool/domain/shared/output_destination"
 
 class ColumnSessionTest < Minitest::Test
   def test_state_transitions
@@ -25,7 +25,7 @@ class ColumnSessionTest < Minitest::Test
       values: [Csvtool::Domain::ColumnSession::ExtractionValue.new("Alice")]
     )
     session = session.with_preview(preview).confirm!.with_output_destination(
-      Csvtool::Domain::ColumnSession::OutputDestination.console
+      Csvtool::Domain::Shared::OutputDestination.console
     )
 
     assert_equal true, session.confirmed?

data/test/csvtool/domain/column_session/csv_source_test.rb

@@ -11,4 +11,14 @@ class CsvSourceTest < Minitest::Test
     assert_equal "/tmp/a.csv", source.path
     assert_equal separator, source.separator
   end
+
+  def test_rejects_empty_path
+    separator = Csvtool::Domain::ColumnSession::Separator.new(",")
+
+    error = assert_raises(ArgumentError) do
+      Csvtool::Domain::ColumnSession::CsvSource.new(path: "", separator: separator)
+    end
+
+    assert_equal "path cannot be empty", error.message
+  end
 end

data/test/csvtool/domain/cross_csv_dedupe_session/column_selector_test.rb

@@ -0,0 +1,42 @@
+# frozen_string_literal: true
+
+require_relative "../../../test_helper"
+require "csvtool/domain/cross_csv_dedupe_session/column_selector"
+
+class CrossCsvDedupeColumnSelectorTest < Minitest::Test
+  def test_builds_header_selector_from_input
+    selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: true, input: "customer_id")
+
+    assert_equal "customer_id", selector.value
+    assert_equal true, selector.headers_present?
+  end
+
+  def test_builds_index_selector_from_input
+    selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: false, input: "2")
+
+    assert_equal 2, selector.value
+    assert_equal true, selector.index?
+  end
+
+  def test_rejects_invalid_index_input
+    error = assert_raises(ArgumentError) do
+      Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: false, input: "0")
+    end
+
+    assert_equal "column index must be a positive integer", error.message
+  end
+
+  def test_extracts_from_headered_row
+    selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: true, input: "customer_id")
+    row = { "customer_id" => "42" }
+
+    assert_equal "42", selector.extract_from(row)
+  end
+
+  def test_extracts_from_headerless_row_by_index
+    selector = Csvtool::Domain::CrossCsvDedupeSession::ColumnSelector.from_input(headers_present: false, input: "2")
+    row = ["a", "b", "c"]
+
+    assert_equal "b", selector.extract_from(row)
+  end
+end