csvops 0.3.0.alpha → 0.4.0.alpha

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

@@ -0,0 +1,246 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow"
+require "tmpdir"
+
+class RunCrossCsvDedupeWorkflowTest < Minitest::Test
+  def fixture_path(name)
+    File.expand_path("../../../../fixtures/#{name}", __dir__)
+  end
+
+  def test_dedupes_source_rows_by_reference_column
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, "CSV file path:"
+    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"
+    refute_includes output.string, "2,Bob"
+    refute_includes output.string, "4,Dan"
+    assert_includes output.string, "Summary: source_rows=5 removed_rows=3 kept_rows=2"
+  end
+
+  def test_can_write_deduped_rows_to_file
+    output = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "deduped.csv")
+      input = [
+        fixture_path("dedupe_source.csv"),
+        "",
+        "",
+        fixture_path("dedupe_reference.csv"),
+        "",
+        "",
+        "customer_id",
+        "external_id",
+        "",
+        "",
+        "2",
+        output_path
+      ].join("\n") + "\n"
+
+      Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+        .new(stdin: StringIO.new(input), stdout: output)
+        .call
+
+      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_supports_tsv_separators
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source.tsv"),
+      "2",
+      "",
+      fixture_path("dedupe_reference.tsv"),
+      "2",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, "customer_id\tname"
+    assert_includes output.string, "1\tAlice"
+    assert_includes output.string, "3\tCara"
+  end
+
+  def test_headerless_mode_supports_column_index
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source_no_headers.csv"),
+      "",
+      "n",
+      fixture_path("dedupe_reference_no_headers.csv"),
+      "",
+      "n",
+      "1",
+      "1",
+      "",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    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_reports_column_not_found_when_missing
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference.csv"),
+      "",
+      "",
+      "missing",
+      "external_id",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, "Column not found."
+  end
+
+  def test_reports_when_no_rows_were_removed
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference_none.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, "Summary: source_rows=5 removed_rows=0 kept_rows=5"
+    assert_includes output.string, "No rows removed; no matching keys found."
+  end
+
+  def test_reports_when_all_rows_were_removed
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference_all.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, "Summary: source_rows=5 removed_rows=5 kept_rows=0"
+    assert_includes output.string, "All source rows were removed by dedupe."
+  end
+
+  def test_normalization_trim_on_and_case_insensitive_on_matches_equivalent_keys
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source_normalization.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference_normalization.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "",
+      "y",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    refute_includes output.string, " A1 ,Alice"
+    refute_includes output.string, "c3,Cara"
+    assert_includes output.string, "B2,Bob"
+    assert_includes output.string, "Summary: source_rows=3 removed_rows=2 kept_rows=1"
+  end
+
+  def test_normalization_disabled_preserves_exact_match_behavior
+    output = StringIO.new
+    input = [
+      fixture_path("dedupe_source_normalization.csv"),
+      "",
+      "",
+      fixture_path("dedupe_reference_normalization.csv"),
+      "",
+      "",
+      "customer_id",
+      "external_id",
+      "n",
+      "n",
+      ""
+    ].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunCrossCsvDedupeWorkflow
+      .new(stdin: StringIO.new(input), stdout: output)
+      .call
+
+    assert_includes output.string, " A1 ,Alice"
+    assert_includes output.string, "B2,Bob"
+    assert_includes output.string, "c3,Cara"
+    assert_includes output.string, "Summary: source_rows=3 removed_rows=0 kept_rows=3"
+  end
+end

data/test/fixtures/dedupe_reference.csv

@@ -0,0 +1,3 @@
+external_id
+2
+4

data/test/fixtures/dedupe_reference.tsv

@@ -0,0 +1,3 @@
+external_id
+2
+4

data/test/fixtures/dedupe_reference_all.csv

@@ -0,0 +1,5 @@
+external_id
+1
+2
+3
+4

data/test/fixtures/dedupe_reference_no_headers.csv

@@ -0,0 +1,2 @@
+2
+4

data/test/fixtures/dedupe_reference_none.csv

@@ -0,0 +1,2 @@
+external_id
+999

data/test/fixtures/dedupe_reference_normalization.csv

@@ -0,0 +1,3 @@
+external_id
+a1
+C3

data/test/fixtures/dedupe_source.csv

@@ -0,0 +1,6 @@
+customer_id,name
+1,Alice
+2,Bob
+3,Cara
+4,Dan
+2,Bob Duplicate

data/test/fixtures/dedupe_source.tsv

@@ -0,0 +1,6 @@
+customer_id	name
+1	Alice
+2	Bob
+3	Cara
+4	Dan
+2	Bob Duplicate

data/test/fixtures/dedupe_source_no_headers.csv

@@ -0,0 +1,5 @@
+1,Alice
+2,Bob
+3,Cara
+4,Dan
+2,Bob Duplicate

data/test/fixtures/dedupe_source_normalization.csv

@@ -0,0 +1,4 @@
+customer_id,name
+ A1 ,Alice
+B2,Bob
+c3,Cara

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: csvops
 version: !ruby/object:Gem::Version
-  version: 0.3.0.alpha
+  version: 0.4.0.alpha
 platform: ruby
 authors:
 - Robert Hall
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-02-21 00:00:00.000000000 Z
+date: 2026-02-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: csv
@@ -67,10 +67,13 @@ files:
 - bin/csvtool
 - bin/tool
 - csvops.gemspec
+- docs/architecture.md
 - docs/release-v0.1.0-alpha.md
 - docs/release-v0.2.0-alpha.md
 - docs/release-v0.3.0-alpha.md
+- docs/release-v0.4.0-alpha.md
 - exe/csvtool
+- lib/csvtool/application/use_cases/run_cross_csv_dedupe.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
@@ -80,20 +83,25 @@ files:
 - lib/csvtool/domain/column_session/csv_source.rb
 - lib/csvtool/domain/column_session/extraction_options.rb
 - lib/csvtool/domain/column_session/extraction_value.rb
-- lib/csvtool/domain/column_session/output_destination.rb
 - lib/csvtool/domain/column_session/preview.rb
 - lib/csvtool/domain/column_session/separator.rb
+- lib/csvtool/domain/cross_csv_dedupe_session/column_selector.rb
+- lib/csvtool/domain/cross_csv_dedupe_session/cross_csv_dedupe_session.rb
+- lib/csvtool/domain/cross_csv_dedupe_session/csv_profile.rb
+- lib/csvtool/domain/cross_csv_dedupe_session/key_mapping.rb
+- lib/csvtool/domain/cross_csv_dedupe_session/match_options.rb
 - lib/csvtool/domain/row_randomization_session/randomization_options.rb
-- lib/csvtool/domain/row_randomization_session/randomization_output_destination.rb
 - lib/csvtool/domain/row_randomization_session/randomization_session.rb
 - lib/csvtool/domain/row_randomization_session/randomization_source.rb
-- lib/csvtool/domain/row_session/row_output_destination.rb
 - lib/csvtool/domain/row_session/row_range.rb
 - lib/csvtool/domain/row_session/row_session.rb
 - lib/csvtool/domain/row_session/row_source.rb
+- lib/csvtool/domain/shared/output_destination.rb
+- lib/csvtool/infrastructure/csv/cross_csv_deduper.rb
 - lib/csvtool/infrastructure/csv/header_reader.rb
 - lib/csvtool/infrastructure/csv/row_randomizer.rb
 - lib/csvtool/infrastructure/csv/row_streamer.rb
+- lib/csvtool/infrastructure/csv/selector_validator.rb
 - lib/csvtool/infrastructure/csv/value_streamer.rb
 - lib/csvtool/infrastructure/output/console_writer.rb
 - lib/csvtool/infrastructure/output/csv_file_writer.rb
@@ -109,8 +117,10 @@ files:
 - lib/csvtool/interface/cli/prompts/seed_prompt.rb
 - lib/csvtool/interface/cli/prompts/separator_prompt.rb
 - lib/csvtool/interface/cli/prompts/skip_blanks_prompt.rb
+- lib/csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow.rb
 - lib/csvtool/services/preview_builder.rb
 - lib/csvtool/version.rb
+- test/csvtool/application/use_cases/run_cross_csv_dedupe_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
@@ -121,20 +131,25 @@ files:
 - test/csvtool/domain/column_session/csv_source_test.rb
 - test/csvtool/domain/column_session/extraction_options_test.rb
 - test/csvtool/domain/column_session/extraction_value_test.rb
-- test/csvtool/domain/column_session/output_destination_test.rb
 - test/csvtool/domain/column_session/preview_test.rb
 - test/csvtool/domain/column_session/separator_test.rb
+- test/csvtool/domain/cross_csv_dedupe_session/column_selector_test.rb
+- test/csvtool/domain/cross_csv_dedupe_session/cross_csv_dedupe_session_test.rb
+- test/csvtool/domain/cross_csv_dedupe_session/csv_profile_test.rb
+- test/csvtool/domain/cross_csv_dedupe_session/key_mapping_test.rb
+- test/csvtool/domain/cross_csv_dedupe_session/match_options_test.rb
 - test/csvtool/domain/row_randomization_session/randomization_options_test.rb
-- test/csvtool/domain/row_randomization_session/randomization_output_destination_test.rb
 - test/csvtool/domain/row_randomization_session/randomization_session_test.rb
 - test/csvtool/domain/row_randomization_session/randomization_source_test.rb
-- test/csvtool/domain/row_session/row_output_destination_test.rb
 - test/csvtool/domain/row_session/row_range_test.rb
 - test/csvtool/domain/row_session/row_session_test.rb
 - test/csvtool/domain/row_session/row_source_test.rb
+- test/csvtool/domain/shared/output_destination_test.rb
+- test/csvtool/infrastructure/csv/cross_csv_deduper_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
+- test/csvtool/infrastructure/csv/selector_validator_test.rb
 - test/csvtool/infrastructure/csv/value_streamer_test.rb
 - test/csvtool/infrastructure/output/console_writer_test.rb
 - test/csvtool/infrastructure/output/csv_file_writer_test.rb
@@ -150,7 +165,18 @@ files:
 - test/csvtool/interface/cli/prompts/seed_prompt_test.rb
 - test/csvtool/interface/cli/prompts/separator_prompt_test.rb
 - test/csvtool/interface/cli/prompts/skip_blanks_prompt_test.rb
+- test/csvtool/interface/cli/workflows/run_cross_csv_dedupe_workflow_test.rb
 - test/csvtool/services/preview_builder_test.rb
+- test/fixtures/dedupe_reference.csv
+- test/fixtures/dedupe_reference.tsv
+- test/fixtures/dedupe_reference_all.csv
+- test/fixtures/dedupe_reference_no_headers.csv
+- test/fixtures/dedupe_reference_none.csv
+- test/fixtures/dedupe_reference_normalization.csv
+- test/fixtures/dedupe_source.csv
+- test/fixtures/dedupe_source.tsv
+- test/fixtures/dedupe_source_no_headers.csv
+- test/fixtures/dedupe_source_normalization.csv
 - test/fixtures/empty.csv
 - test/fixtures/sample_people.csv
 - test/fixtures/sample_people.tsv

data/lib/csvtool/domain/row_randomization_session/randomization_output_destination.rb

@@ -1,31 +0,0 @@
-# frozen_string_literal: true
-
-module Csvtool
-  module Domain
-    module RowRandomizationSession
-      class RandomizationOutputDestination
-        attr_reader :mode, :path
-
-        def self.console
-          new(mode: :console)
-        end
-
-        def self.file(path:)
-          new(mode: :file, path: path)
-        end
-
-        def initialize(mode:, path: nil)
-          raise ArgumentError, "invalid output mode" unless %i[console file].include?(mode)
-          raise ArgumentError, "file output path cannot be empty" if mode == :file && path.to_s.empty?
-
-          @mode = mode
-          @path = path
-        end
-
-        def file?
-          @mode == :file
-        end
-      end
-    end
-  end
-end

data/lib/csvtool/domain/row_session/row_output_destination.rb

@@ -1,31 +0,0 @@
-# frozen_string_literal: true
-
-module Csvtool
-  module Domain
-    module RowSession
-      class RowOutputDestination
-        attr_reader :mode, :path
-
-        def self.console
-          new(mode: :console)
-        end
-
-        def self.file(path:)
-          new(mode: :file, path: path)
-        end
-
-        def initialize(mode:, path: nil)
-          raise ArgumentError, "invalid output mode" unless %i[console file].include?(mode)
-          raise ArgumentError, "file output path cannot be empty" if mode == :file && path.to_s.empty?
-
-          @mode = mode
-          @path = path
-        end
-
-        def file?
-          @mode == :file
-        end
-      end
-    end
-  end
-end

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

@@ -1,18 +0,0 @@
-# frozen_string_literal: true
-
-require_relative "../../../test_helper"
-require "csvtool/domain/column_session/output_destination"
-
-class OutputDestinationTest < Minitest::Test
-  def test_console_factory
-    destination = Csvtool::Domain::ColumnSession::OutputDestination.console
-    assert_equal true, destination.console?
-    assert_equal false, destination.file?
-  end
-
-  def test_file_factory
-    destination = Csvtool::Domain::ColumnSession::OutputDestination.file(path: "/tmp/out.csv")
-    assert_equal true, destination.file?
-    assert_equal "/tmp/out.csv", destination.path
-  end
-end

data/test/csvtool/domain/row_randomization_session/randomization_output_destination_test.rb

@@ -1,21 +0,0 @@
-# frozen_string_literal: true
-
-require_relative "../../../test_helper"
-require "csvtool/domain/row_randomization_session/randomization_output_destination"
-
-class RandomizationOutputDestinationTest < Minitest::Test
-  def test_console_and_file_modes
-    console = Csvtool::Domain::RowRandomizationSession::RandomizationOutputDestination.console
-    file = Csvtool::Domain::RowRandomizationSession::RandomizationOutputDestination.file(path: "/tmp/out.csv")
-
-    assert_equal false, console.file?
-    assert_equal true, file.file?
-    assert_equal "/tmp/out.csv", file.path
-  end
-
-  def test_rejects_empty_file_path
-    assert_raises(ArgumentError) do
-      Csvtool::Domain::RowRandomizationSession::RandomizationOutputDestination.file(path: "")
-    end
-  end
-end

data/test/csvtool/domain/row_session/row_output_destination_test.rb

@@ -1,23 +0,0 @@
-# frozen_string_literal: true
-
-require_relative "../../../test_helper"
-require "csvtool/domain/row_session/row_output_destination"
-
-class RowRangeOutputDestinationTest < Minitest::Test
-  def test_console_destination
-    destination = Csvtool::Domain::RowSession::RowOutputDestination.console
-    refute destination.file?
-  end
-
-  def test_file_destination
-    destination = Csvtool::Domain::RowSession::RowOutputDestination.file(path: "/tmp/out.csv")
-    assert destination.file?
-    assert_equal "/tmp/out.csv", destination.path
-  end
-
-  def test_rejects_empty_file_path
-    assert_raises(ArgumentError) do
-      Csvtool::Domain::RowSession::RowOutputDestination.file(path: "")
-    end
-  end
-end