markdown_exec 2.3.0 → 2.4.0

data/lib/table_extractor.rb

@@ -0,0 +1,166 @@
+# frozen_string_literal: true
+
+class TableExtractor
+  # Extract tables from an array of text lines formatted in Markdown style
+  # @param [Array<String>] lines The array of text lines
+  # @return [Array<Hash>] An array of tables with row count, column count, and start index
+  def self.extract_tables(lines)
+    tables = []
+    inside_table = false
+    table_start = nil
+    row_count = 0
+    column_count = 0
+
+    separator_regexp = /^[ \t]*\|? *(?::?-+:?) *( *\| *(?::?-+:?) *)+\|? *$/
+
+    lines.each_with_index do |line, index|
+      # Match line separators with at least 2 columns
+      if line.strip.match?(separator_regexp)
+        if inside_table
+          # Add the current table before starting a new one
+          tables << {
+            rows: row_count,
+            columns: column_count,
+            start_index: table_start
+          }
+        end
+        # Start a new table
+        table_start = index - 1 if table_start.nil?
+        column_count = line.split('|').count - 1
+        row_count = 2 # Reset to 2 to account for the header and separator rows
+        inside_table = true
+      elsif inside_table && (line.strip.start_with?('|') || line.include?('|'))
+        row_count += 1
+      elsif inside_table
+        # Add the current table and reset the state
+        tables << {
+          rows: row_count,
+          columns: column_count,
+          start_index: table_start
+        }
+        inside_table = false
+        table_start = nil
+        row_count = 0
+        column_count = 0
+      end
+    end
+
+    # Handle case where table ends at the last line
+    if inside_table
+      tables << {
+        rows: row_count,
+        columns: column_count,
+        start_index: table_start
+      }
+    end
+
+    tables
+  end
+end
+
+return if $PROGRAM_NAME != __FILE__
+
+require 'minitest/autorun'
+
+class TestTableExtractor < Minitest::Test
+  def test_single_table
+    lines = [
+      '| Species| Genus| Family',
+      '|-|-|-',
+      '| Pongo tapanuliensis| Pongo| Hominidae',
+      '| | Histiophryne| Antennariidae'
+    ]
+    expected = [{ rows: 4, columns: 3, start_index: 0 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_indented_table
+    lines = [
+      "\t | Species| Genus| Family",
+      "\t |-|-|-",
+      "\t | Pongo tapanuliensis| Pongo| Hominidae",
+      "\t | | Histiophryne| Antennariidae"
+    ]
+    expected = [{ rows: 4, columns: 3, start_index: 0 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_multiple_tables
+    lines = [
+      '| Species| Genus| Family',
+      '|-|-|-',
+      '| Pongo tapanuliensis| Pongo| Hominidae',
+      '| | Histiophryne| Antennariidae',
+      '',
+      '| Name| Species',
+      '|-|-',
+      '| Tapanuli Orangutan| Pongo tapanuliensis'
+    ]
+    expected = [
+      { rows: 4, columns: 3, start_index: 0 },
+      { rows: 3, columns: 2, start_index: 5 }
+    ]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_no_tables
+    lines = [
+      'This is a regular line.',
+      'Another regular line.'
+    ]
+    expected = []
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_inconsistent_columns
+    lines = [
+      '| Species| Genus| Family',
+      '|-|-',
+      '| Pongo tapanuliensis| Pongo| Hominidae',
+      '| | Histiophryne| Antennariidae',
+      '',
+      '| Name| Species',
+      '|-|-|-',
+      '| Tapanuli Orangutan| Pongo tapanuliensis'
+    ]
+    # number of columns determined from row of dividers
+    expected = [{ rows: 4, columns: 2, start_index: 0 },
+                { rows: 3, columns: 3, start_index: 5 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_table_at_end_of_lines
+    lines = [
+      'Some introductory text.',
+      '| Species| Genus| Family',
+      '|-|-|-',
+      '| Pongo tapanuliensis| Pongo| Hominidae',
+      '| | Histiophryne| Antennariidae'
+    ]
+    expected = [{ rows: 4, columns: 3, start_index: 1 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_table_without_starting_pipe
+    lines = [
+      'Some introductory text.',
+      'Platform| Target Environment| Command',
+      '|-|-|-',
+      '| Pongo tapanuliensis| Pongo| Hominidae',
+      '| | Histiophryne| Antennariidae'
+    ]
+    expected = [{ rows: 4, columns: 3, start_index: 1 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+
+  def test_table_with_colon_hyphens
+    lines = [
+      '| Name| Age| City',
+      '|:-:|:-|:-:',
+      '| John Doe| 30| New York',
+      '| Jane Doe| 25| Los Angeles'
+    ]
+    expected = [{ rows: 4, columns: 3, start_index: 0 }]
+    assert_equal expected, TableExtractor.extract_tables(lines)
+  end
+end

data/lib/tap.rb

@@ -35,8 +35,6 @@ unless defined?(Env)
   end
 end
 
-# rubocop:disable Metrics/ParameterLists
-
 ## application-level debug control
 #
 # :reek:TooManyConstants
@@ -98,8 +96,8 @@ module Tap
 
   # :reek:ControlParameter
   # :reek:LongParameterList
-  def tap_inspect(name_ = nil, caller_first: nil, mask: TDD, name: DN, source: nil,
-                  type: nil)
+  def tap_inspect(name_ = nil, caller_first: nil, mask: TDD, name: DN,
+                  source: nil, type: nil)
     return self unless $tap_enable
     return self unless (mask & $tap_mask).positive?
 
@@ -107,7 +105,8 @@ module Tap
     outs = []
     outs.push(source.to_s) if source.present?
 
-    vs = (caller_first || caller[0]).scan(/in `?(\S+)'$/).fetch(0, []).fetch(0, '')
+    vs = (caller_first || caller[0]).scan(/in `?(\S+)'$/)
+                                    .fetch(0, []).fetch(0, '')
     outs.push("#{vs}()") if vs.present?
 
     outs.push(tap_named_value(name_ || name, method(fn).call))
@@ -143,7 +142,7 @@ module Tap
   # :reek:ControlParameter
   # :reek:LongParameterList
   def tap_yaml(name_ = nil, caller_first: nil, mask: TDD, name: DN, source: nil)
-    tap_inspect name_, caller_first: (caller_first || caller[0]),
+    tap_inspect name_, caller_first: caller_first || caller[0],
                        mask: mask, name: name, source: source, type: :yaml
   end
 

data/lib/text_analyzer.rb

@@ -11,13 +11,17 @@ module TextAnalyzer
   # @return [Array<Hash>, Array<Array<Hash>>] an array or nested arrays of highlighted segments
   #
   # @raise [ArgumentError] if the hierarchy structure is neither a String nor an Array
-  def self.analyze_hierarchy(hierarchy, pattern, default_color, match_color)
+  def self.analyze_hierarchy(
+    hierarchy, pattern, default_color, match_color,
+    text_sym: :text, style_sym: :color
+  )
     case hierarchy
     when String
       highlight_segments(hierarchy, pattern, default_color, match_color)
+
     when Hash
-      decorated = highlight_segments(hierarchy[:text], pattern,
-                                     hierarchy[:color], match_color)
+      decorated = highlight_segments(hierarchy[text_sym], pattern,
+                                     hierarchy[style_sym], match_color)
 
       case decorated
       when String
@@ -31,18 +35,19 @@ module TextAnalyzer
       else
         decorated
       end
+
     when Array
       hierarchy.map do |element|
         analyze_hierarchy(element, pattern, default_color, match_color)
       end
-    when HierarchyString
 
+    when HierarchyString
       hierarchy.replace_text! do |substring|
-        substring ### no change
+        substring # no change
       end
 
     else
-      binding.irb
+      warn [hierarchy.class, hierarchy].inspect
       raise ArgumentError, 'Invalid hierarchy structure'
     end
   end
@@ -77,14 +82,14 @@ module TextAnalyzer
   # @yieldparam segment [String] a segment of the text
   # @yieldparam is_match [Boolean] true if the segment matches the pattern, false otherwise
   def self.yield_matches_and_non_matches(text, pattern)
-    last_end = 0
+    last_end = nil
 
     text.scan(pattern) do |match|
       match_start = Regexp.last_match.begin(0)
       match_end = Regexp.last_match.end(0)
 
       # Yield the non-matching segment before the match
-      yield text[last_end...match_start], false if match_start > last_end
+      yield text[(last_end || 0)...match_start], false if last_end.nil? || match_start > last_end
 
       # Yield the matching segment
       yield match.first, true
@@ -92,9 +97,140 @@ module TextAnalyzer
       last_end = match_end
     end
 
+    last_end ||= 0
     # Yield any remaining non-matching segment after the last match
     return unless last_end < text.length
 
     yield text[last_end..-1], false
   end
 end
+
+return if $PROGRAM_NAME != __FILE__
+
+require 'minitest/autorun'
+require_relative 'hierarchy_string'
+
+$default_color = :upcase
+$match_color = :downcase
+
+class TestTextAnalyzer < Minitest::Test
+  def test_analyze_hierarchy_with_string
+    text = 'This is a test string.'
+    pattern = /(test)/
+
+    expected_output = [[[
+      { text: 'This is a ', color: $default_color },
+      { text: 'test', color: $match_color },
+      { text: ' string.', color: $default_color }
+    ]]]
+
+    tree = HierarchyString.new([{ text: text, color: $default_color }])
+    assert_equal expected_output,
+                 TextAnalyzer.analyze_hierarchy(tree.substrings, pattern, $default_color,
+                                                $match_color)
+  end
+
+  def test_analyze_hierarchy_with_array
+    hierarchy = [
+      'This is a test string.',
+      'Another test line.'
+    ]
+    pattern = /(test)/
+
+    expected_output = [
+      [
+        { text: 'This is a ', color: $default_color },
+        { text: 'test', color: $match_color },
+        { text: ' string.', color: $default_color }
+      ],
+      [
+        { text: 'Another ', color: $default_color },
+        { text: 'test', color: $match_color },
+        { text: ' line.', color: $default_color }
+      ]
+    ]
+
+    assert_equal expected_output,
+                 TextAnalyzer.analyze_hierarchy(hierarchy, pattern,
+                                                $default_color, $match_color)
+  end
+
+  def test_analyze_hierarchy_with_nested_array
+    hierarchy = [
+      'This is a test string.',
+      ['Another test line.', 'Yet another test.']
+    ]
+    pattern = /(test)/
+
+    expected_output = [
+      [
+        { text: 'This is a ', color: $default_color },
+        { text: 'test', color: $match_color },
+        { text: ' string.', color: $default_color }
+      ],
+      [
+        [
+          { text: 'Another ', color: $default_color },
+          { text: 'test', color: $match_color },
+          { text: ' line.', color: $default_color }
+        ],
+        [
+          { text: 'Yet another ', color: $default_color },
+          { text: 'test', color: $match_color },
+          { text: '.', color: $default_color }
+        ]
+      ]
+    ]
+
+    assert_equal expected_output,
+                 TextAnalyzer.analyze_hierarchy(hierarchy, pattern,
+                                                $default_color, $match_color)
+  end
+
+  def test_analyze_hierarchy_with_invalid_type
+    hierarchy = 12_345
+    # hierarchy = HierarchyString.new([{ text: '12345', color: $default_color }])
+    pattern = /(test)/
+
+    assert_raises(ArgumentError) do
+      TextAnalyzer.analyze_hierarchy(hierarchy, pattern, $default_color,
+                                     $match_color)
+    end
+  end
+
+  def test_highlight_segments
+    text = 'This is a test string.'
+    pattern = /(test)/
+
+    expected_output = [
+      { text: 'This is a ', color: $default_color },
+      { text: 'test', color: $match_color },
+      { text: ' string.', color: $default_color }
+    ]
+
+    assert_equal expected_output,
+                 TextAnalyzer.highlight_segments(text, pattern, $default_color,
+                                                 $match_color)
+  end
+
+  def test_yield_matches_and_non_matches
+    text = 'This is a test string with multiple tests.'
+    pattern = /(test)/
+    segments = []
+
+    TextAnalyzer.yield_matches_and_non_matches(text,
+                                               pattern) do |segment, is_match|
+      segments << { text: segment, is_match: is_match }
+    end
+
+    expected_output = [
+      { text: 'This is a ', is_match: false },
+      { text: 'test', is_match: true },
+      { text: ' string with multiple ', is_match: false },
+      { text: 'test', is_match: true },
+      { text: 's.', is_match: false }
+    ]
+
+    assert_equal expected_output, segments
+  end
+end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: markdown_exec
 version: !ruby/object:Gem::Version
-  version: 2.3.0
+  version: 2.4.0
 platform: ruby
 authors:
 - Fareed Stevenson
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-08-05 00:00:00.000000000 Z
+date: 2024-09-22 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: clipboard
@@ -106,6 +106,15 @@ files:
 - assets/output_of_execution.png
 - assets/select_a_block.png
 - assets/select_a_file.png
+- bats/bats.bats
+- bats/block-types.bats
+- bats/cli.bats
+- bats/fail.bats
+- bats/history.bats
+- bats/markup.bats
+- bats/mde.bats
+- bats/options.bats
+- bats/test_helper.bash
 - bin/bmde
 - bin/colorize_env_vars.sh
 - bin/console
@@ -113,6 +122,13 @@ files:
 - bin/setup
 - bin/tab_completion.sh
 - bin/tab_completion.sh.erb
+- docs/dev/block-type-opts.md
+- docs/dev/block-type-port.md
+- docs/dev/block-type-vars.md
+- docs/dev/pass-through-arguments.md
+- docs/dev/specs-import.md
+- docs/dev/specs.md
+- docs/dev/text-decoration.md
 - examples/bash-blocks.md
 - examples/block-names.md
 - examples/block_names.md
@@ -149,9 +165,13 @@ files:
 - examples/port-blocks.md
 - examples/save.md
 - examples/search.md
+- examples/table-markup.md
+- examples/text-markup.md
 - examples/vars-blocks.md
 - examples/wrap.md
 - lib/ansi_formatter.rb
+- lib/ansi_string.rb
+- lib/argument_processor.rb
 - lib/array.rb
 - lib/array_util.rb
 - lib/block_label.rb
@@ -163,17 +183,20 @@ files:
 - lib/colorize.rb
 - lib/constants.rb
 - lib/directory_searcher.rb
+- lib/doh.rb
 - lib/env.rb
 - lib/exceptions.rb
 - lib/fcb.rb
 - lib/filter.rb
 - lib/find_files.rb
+- lib/format_table.rb
 - lib/fout.rb
 - lib/hash.rb
 - lib/hash_delegator.rb
 - lib/hierarchy_string.rb
 - lib/input_sequencer.rb
 - lib/instance_method_wrapper.rb
+- lib/layered_hash.rb
 - lib/link_history.rb
 - lib/markdown_exec.rb
 - lib/markdown_exec/version.rb
@@ -190,9 +213,9 @@ files:
 - lib/saved_assets.rb
 - lib/saved_files_matcher.rb
 - lib/shared.rb
-- lib/std_out_err_logger.rb
 - lib/streams_out.rb
 - lib/string_util.rb
+- lib/table_extractor.rb
 - lib/tap.rb
 - lib/text_analyzer.rb
 homepage: https://rubygems.org/gems/markdown_exec

data/lib/std_out_err_logger.rb

@@ -1,119 +0,0 @@
-#!/usr/bin/env -S bundle exec ruby
-# frozen_string_literal: true
-
-# encoding=utf-8
-require 'logger'
-
-# Logger::LogDevice is used by Logger, the parent class of StdOutErrLogger
-class Logger::LogDevice
-  # remove header
-  def add_log_header(file); end
-end
-
-# Custom logger to direct info to stdout and warn and above to stderr
-#
-class StdOutErrLogger < Logger
-  attr_reader :file
-
-  # def initialize(file = nil)
-  def initialize(file = "#{__dir__}/../tmp/hash_delegator_next_link_state.yaml")
-    @file = file
-    super(file || $stdout)
-    self.formatter = proc do |_severity, _datetime, _progname, msg|
-      "#{msg}\n"
-    end
-  end
-
-  def add(severity, message = nil, progname = nil, &block)
-    message = (message || block&.call || progname) if message.nil?
-    message = "- #{message.to_json}\n"
-    ### message = message.join("\n") if message.is_a? Array
-    out = format_message(format_severity(severity), Time.now, progname, message)
-    if severity == Logger::UNKNOWN # does not follow spec, outputs to stderr for IO
-      # $stderr.puts(out)
-      super
-    elsif severity >= Logger::WARN
-      if @file
-        super
-      else
-        warn(out)
-      end
-    elsif @file
-      super
-    else
-      $stdout.puts(out)
-    end
-  end
-end
-
-return if $PROGRAM_NAME != __FILE__
-
-require 'minitest/autorun'
-
-class StdOutErrLoggerTest < Minitest::Test
-  # Redirect STDOUT and STDERR to capture them for assertions
-  def setup
-    @original_stdout = $stdout
-    @original_stderr = $stderr
-    $stdout = StringIO.new
-    $stderr = StringIO.new
-  end
-
-  def teardown
-    $stdout = @original_stdout
-    $stderr = @original_stderr
-  end
-
-  def test_initialize_without_file
-    logger = StdOutErrLogger.new
-    assert_nil logger.file
-    assert_equal Logger::DEBUG, logger.level
-  end
-
-  def test_initialize_with_file
-    Tempfile.open do |file|
-      logger = StdOutErrLogger.new(file.path)
-      assert_equal file.path, logger.file
-    end
-  end
-
-  def test_logging_info
-    logger = StdOutErrLogger.new
-    logger.info('Info message')
-    assert_equal "Info message\n", $stdout.string
-    assert_empty $stderr.string
-  end
-
-  def test_logging_warning
-    logger = StdOutErrLogger.new
-    logger.warn('Warning message')
-    assert_empty $stdout.string
-    assert_equal "Warning message\n", $stderr.string
-  end
-
-  def test_logging_error
-    logger = StdOutErrLogger.new
-    logger.error('Error message')
-    assert_empty $stdout.string
-    assert_equal "Error message\n", $stderr.string
-  end
-
-  def test_logging_with_array
-    logger = StdOutErrLogger.new
-    logger.info(['Message line 1', 'Message line 2'])
-    assert_equal "Message line 1\nMessage line 2\n", $stdout.string
-  end
-
-  def test_logging_with_block
-    logger = StdOutErrLogger.new
-    logger.info { 'Block message' }
-    assert_equal "Block message\n", $stdout.string
-  end
-
-  def test_logging_unknown_severity
-    logger = StdOutErrLogger.new
-    logger.add(Logger::UNKNOWN, 'Unknown severity message')
-    assert_empty $stdout.string
-    assert_equal "Unknown severity message\n", $stderr.string
-  end
-end