csvops 0.3.0.alpha → 0.5.0.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/csvtool/interface/cli/workflows/run_extraction_workflow_test.rb

@@ -0,0 +1,56 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_extraction_workflow"
+require "tmpdir"
+
+class RunExtractionWorkflowTest < Minitest::Test
+  def fixture_path(name)
+    File.expand_path("../../../../fixtures/#{name}", __dir__)
+  end
+
+  def test_missing_file_path_reports_error
+    out = StringIO.new
+    workflow = Csvtool::Interface::CLI::Workflows::RunExtractionWorkflow.new(
+      stdin: StringIO.new("/tmp/not-present.csv\n\n"),
+      stdout: out
+    )
+
+    workflow.call
+
+    assert_includes out.string, "File not found: /tmp/not-present.csv"
+  end
+
+  def test_workflow_can_run_console_happy_path
+    out = StringIO.new
+    fixture = fixture_path("sample_people.csv")
+    input = ["#{fixture}", "1", "", "1", "", "y", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunExtractionWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Alice"
+    assert_includes out.string, "Bob"
+    assert_includes out.string, "Cara"
+  end
+
+  def test_workflow_can_write_output_file
+    out = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "names.csv")
+      fixture = fixture_path("sample_people.csv")
+      input = ["#{fixture}", "1", "", "1", "", "y", "2", output_path].join("\n") + "\n"
+
+      Csvtool::Interface::CLI::Workflows::RunExtractionWorkflow.new(
+        stdin: StringIO.new(input),
+        stdout: out
+      ).call
+
+      assert_includes out.string, "Wrote output to #{output_path}"
+      assert_equal "name\nAlice\nBob\nCara\n", File.read(output_path)
+    end
+  end
+end

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

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_row_extraction_workflow"
+require "tmpdir"
+
+class RunRowExtractionWorkflowTest < Minitest::Test
+  def fixture_path(name)
+    File.expand_path("../../../../fixtures/#{name}", __dir__)
+  end
+
+  def test_missing_file_path_reports_error
+    out = StringIO.new
+    workflow = Csvtool::Interface::CLI::Workflows::RunRowExtractionWorkflow.new(
+      stdin: StringIO.new("/tmp/not-present.csv\n\n"),
+      stdout: out
+    )
+
+    workflow.call
+
+    assert_includes out.string, "File not found: /tmp/not-present.csv"
+  end
+
+  def test_workflow_can_run_console_happy_path
+    out = StringIO.new
+    fixture = fixture_path("sample_people.csv")
+    input = [fixture, "", "2", "3", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunRowExtractionWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "name,city"
+    assert_includes out.string, "Bob,Paris"
+    assert_includes out.string, "Cara,Berlin"
+    refute_includes out.string, "Alice,London"
+  end
+
+  def test_workflow_can_write_output_file
+    out = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "rows.csv")
+      fixture = fixture_path("sample_people.csv")
+      input = [fixture, "", "2", "3", "2", output_path].join("\n") + "\n"
+
+      Csvtool::Interface::CLI::Workflows::RunRowExtractionWorkflow.new(
+        stdin: StringIO.new(input),
+        stdout: out
+      ).call
+
+      assert_includes out.string, "Wrote output to #{output_path}"
+      assert_equal "name,city\nBob,Paris\nCara,Berlin\n", File.read(output_path)
+    end
+  end
+
+  def test_rejects_non_numeric_start_row
+    out = StringIO.new
+    fixture = fixture_path("sample_people.csv")
+    input = [fixture, "", "abc", "3", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunRowExtractionWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Start row must be a positive integer."
+  end
+
+  def test_reports_out_of_bounds_range
+    out = StringIO.new
+    fixture = fixture_path("sample_people.csv")
+    input = [fixture, "", "10", "12", ""].join("\n") + "\n"
+
+    Csvtool::Interface::CLI::Workflows::RunRowExtractionWorkflow.new(
+      stdin: StringIO.new(input),
+      stdout: out
+    ).call
+
+    assert_includes out.string, "Row range is out of bounds. File has 3 data rows."
+  end
+end

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

@@ -0,0 +1,69 @@
+# frozen_string_literal: true
+
+require_relative "../../../../test_helper"
+require "csvtool/interface/cli/workflows/run_row_randomization_workflow"
+require "tmpdir"
+
+class RunRowRandomizationWorkflowTest < Minitest::Test
+  def fixture_path(name)
+    File.expand_path("../../../../fixtures/#{name}", __dir__)
+  end
+
+  def test_missing_file_shows_friendly_error
+    output = StringIO.new
+    input = StringIO.new("/tmp/does-not-exist.csv\n\n")
+
+    Csvtool::Interface::CLI::Workflows::RunRowRandomizationWorkflow.new(stdin: input, stdout: output).call
+
+    assert_includes output.string, "File not found: /tmp/does-not-exist.csv"
+  end
+
+  def test_workflow_prints_header_then_all_randomized_rows
+    output = StringIO.new
+    input = StringIO.new([fixture_path("sample_people.csv"), "", "", "", ""].join("\n") + "\n")
+
+    Csvtool::Interface::CLI::Workflows::RunRowRandomizationWorkflow.new(stdin: input, stdout: output).call
+
+    assert_includes output.string, "name,city"
+    assert_includes output.string, "Alice,London"
+    assert_includes output.string, "Bob,Paris"
+    assert_includes output.string, "Cara,Berlin"
+  end
+
+  def test_workflow_can_write_randomized_rows_to_file
+    output = StringIO.new
+
+    Dir.mktmpdir do |dir|
+      output_path = File.join(dir, "randomized.csv")
+      input = StringIO.new([fixture_path("sample_people.csv"), "", "", "", "2", output_path].join("\n") + "\n")
+
+      Csvtool::Interface::CLI::Workflows::RunRowRandomizationWorkflow.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
+  end
+
+  def test_workflow_supports_headerless_mode
+    output = StringIO.new
+    input = StringIO.new([fixture_path("sample_people_no_headers.csv"), "", "n", "", ""].join("\n") + "\n")
+
+    Csvtool::Interface::CLI::Workflows::RunRowRandomizationWorkflow.new(stdin: input, stdout: output).call
+
+    refute_includes output.string, "name,city"
+    assert_includes output.string, "Alice,London"
+    assert_includes output.string, "Bob,Paris"
+    assert_includes output.string, "Cara,Berlin"
+  end
+
+  def test_invalid_seed_shows_friendly_error
+    output = StringIO.new
+    input = StringIO.new([fixture_path("sample_people.csv"), "", "", "abc"].join("\n") + "\n")
+
+    Csvtool::Interface::CLI::Workflows::RunRowRandomizationWorkflow.new(stdin: input, stdout: output).call
+
+    assert_includes output.string, "Seed must be an integer."
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/cross_csv_dedupe/collect_options_step_test.rb

@@ -0,0 +1,41 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/cross_csv_dedupe/collect_options_step"
+require "csvtool/domain/cross_csv_dedupe_session/csv_profile"
+
+class CrossCsvDedupeCollectOptionsStepTest < Minitest::Test
+  class FakeErrors
+    attr_reader :column_not_found_called
+
+    def column_not_found
+      @column_not_found_called = true
+    end
+  end
+
+  def test_halts_when_source_selector_invalid
+    selector_prompt = Object.new
+    yes_no_prompt = Object.new
+    output_destination_prompt = Object.new
+    session_builder = Object.new
+    mapper = Object.new
+    errors = FakeErrors.new
+
+    def selector_prompt.call(label:, headers_present:) = nil
+
+    step = Csvtool::Interface::CLI::Workflows::Steps::CrossCsvDedupe::CollectOptionsStep.new(
+      selector_prompt: selector_prompt,
+      yes_no_prompt: yes_no_prompt,
+      output_destination_prompt: output_destination_prompt,
+      errors: errors
+    )
+
+    source = Csvtool::Domain::CrossCsvDedupeSession::CsvProfile.new(path: "/tmp/a.csv", separator: ",", headers_present: true)
+    reference = Csvtool::Domain::CrossCsvDedupeSession::CsvProfile.new(path: "/tmp/b.csv", separator: ",", headers_present: true)
+
+    result = step.call(source: source, reference: reference, session_builder: session_builder, output_destination_mapper: mapper)
+
+    assert_equal :halt, result
+    assert_equal true, errors.column_not_found_called
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/extraction/collect_inputs_step_test.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/extraction/collect_inputs_step"
+
+class ExtractionCollectInputsStepTest < Minitest::Test
+  Result = Struct.new(:ok, :data) do
+    def ok? = ok
+  end
+
+  class FakeUseCase
+    def initialize(result)
+      @result = result
+    end
+
+    def read_headers(file_path:, col_sep:)
+      @result
+    end
+  end
+
+  def test_halts_when_separator_missing
+    file_prompt = Object.new
+    separator_prompt = Object.new
+    selector_prompt = Object.new
+    skip_prompt = Object.new
+    def file_prompt.call = "/tmp/data.csv"
+    def separator_prompt.call = nil
+
+    step = Csvtool::Interface::CLI::Workflows::Steps::Extraction::CollectInputsStep.new(
+      file_path_prompt: file_prompt,
+      separator_prompt: separator_prompt,
+      column_selector_prompt: selector_prompt,
+      skip_blanks_prompt: skip_prompt
+    )
+
+    assert_equal :halt, step.call(
+      use_case: FakeUseCase.new(Result.new(true, { headers: [] })),
+      session_builder: Object.new,
+      handle_error: ->(_r) {}
+    )
+  end
+
+  def test_halts_when_header_read_fails
+    file_prompt = Object.new
+    separator_prompt = Object.new
+    selector_prompt = Object.new
+    skip_prompt = Object.new
+    builder = Object.new
+    handled = []
+    def file_prompt.call = "/tmp/data.csv"
+    def separator_prompt.call = ","
+
+    step = Csvtool::Interface::CLI::Workflows::Steps::Extraction::CollectInputsStep.new(
+      file_path_prompt: file_prompt,
+      separator_prompt: separator_prompt,
+      column_selector_prompt: selector_prompt,
+      skip_blanks_prompt: skip_prompt
+    )
+
+    fail_result = Result.new(false, {})
+    result = step.call(use_case: FakeUseCase.new(fail_result), session_builder: builder, handle_error: ->(r) { handled << r })
+
+    assert_equal :halt, result
+    assert_equal [fail_result], handled
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/row_extraction/collect_source_step_test.rb

@@ -0,0 +1,39 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/row_extraction/collect_source_step"
+
+class CollectSourceStepTest < Minitest::Test
+  def test_collects_file_and_separator
+    file_prompt = Object.new
+    separator_prompt = Object.new
+    def file_prompt.call = "/tmp/data.csv"
+    def separator_prompt.call = ","
+
+    step = Csvtool::Interface::CLI::Workflows::Steps::RowExtraction::CollectSourceStep.new(
+      file_path_prompt: file_prompt,
+      separator_prompt: separator_prompt
+    )
+    context = {}
+
+    result = step.call(context)
+
+    assert_nil result
+    assert_equal "/tmp/data.csv", context[:file_path]
+    assert_equal ",", context[:col_sep]
+  end
+
+  def test_halts_when_separator_missing
+    file_prompt = Object.new
+    separator_prompt = Object.new
+    def file_prompt.call = "/tmp/data.csv"
+    def separator_prompt.call = nil
+
+    step = Csvtool::Interface::CLI::Workflows::Steps::RowExtraction::CollectSourceStep.new(
+      file_path_prompt: file_prompt,
+      separator_prompt: separator_prompt
+    )
+
+    assert_equal :halt, step.call({})
+  end
+end

data/test/csvtool/interface/cli/workflows/steps/row_extraction/execute_step_test.rb

@@ -0,0 +1,91 @@
+# frozen_string_literal: true
+
+require_relative "../../../../../../test_helper"
+require "csvtool/interface/cli/workflows/steps/row_extraction/execute_step"
+
+class ExecuteStepTest < Minitest::Test
+  Result = Struct.new(:ok, :data) do
+    def ok? = ok
+  end
+
+  class FakeUseCase
+    def initialize(result)
+      @result = result
+    end
+
+    def extract(session:, headers:, on_row:)
+      @called = true
+      on_row.call(["Bob", "Paris"]) if @result.ok?
+      @result
+    end
+
+    attr_reader :called
+  end
+
+  class FakePresenter
+    attr_reader :rows, :written
+
+    def initialize(stdout:, headers:, col_sep:)
+      @rows = []
+      @written = nil
+    end
+
+    def print_row(fields)
+      @rows << fields
+    end
+
+    def print_file_written(path)
+      @written = path
+    end
+  end
+
+  class FakeErrors
+    attr_reader :out_of_bounds
+
+    def row_range_out_of_bounds(count)
+      @out_of_bounds = count
+    end
+  end
+
+  def test_prints_rows_and_reports_out_of_bounds
+    errors = FakeErrors.new
+    step = Csvtool::Interface::CLI::Workflows::Steps::RowExtraction::ExecuteStep.new(
+      stdout: StringIO.new,
+      errors: errors,
+      presenter_class: FakePresenter
+    )
+    use_case = FakeUseCase.new(Result.new(true, { matched: false, row_count: 3, wrote_rows: false }))
+    context = {
+      session: Object.new,
+      headers: ["name", "city"],
+      use_case: use_case,
+      handle_error: ->(_r) { raise "unexpected" }
+    }
+
+    result = step.call(context)
+
+    assert_nil result
+    assert_equal 3, errors.out_of_bounds
+  end
+
+  def test_halts_on_use_case_failure
+    handled = []
+    step = Csvtool::Interface::CLI::Workflows::Steps::RowExtraction::ExecuteStep.new(
+      stdout: StringIO.new,
+      errors: FakeErrors.new,
+      presenter_class: FakePresenter
+    )
+    fail_result = Result.new(false, {})
+    use_case = FakeUseCase.new(fail_result)
+
+    result = step.call(
+      session: Object.new,
+      headers: ["name", "city"],
+      use_case: use_case,
+      handle_error: ->(r) { handled << r }
+    )
+
+    assert_equal :halt, result
+    assert_equal [fail_result], handled
+  end
+end