regexp_parser 2.6.0 → 2.10.0

data/lib/regexp_parser/expression/methods/tests.rb

@@ -95,12 +95,49 @@ module Regexp::Expression
     end
 
     # Deep-compare two expressions for equality.
+    #
+    # When changing the conditions, please make sure to update
+    # #pretty_print_instance_variables so that it includes all relevant values.
     def ==(other)
-      other.class == self.class &&
-        other.to_s == to_s &&
-        other.options == options
+      self.class   == other.class &&
+        text       == other.text &&
+        quantifier == other.quantifier &&
+        options    == other.options &&
+        (terminal? || expressions == other.expressions)
     end
     alias :=== :==
     alias :eql? :==
+
+    def optional?
+      quantified? && quantifier.min == 0
+    end
+
+    def quantified?
+      !quantifier.nil?
+    end
   end
+
+  Shared.class_eval                     { def terminal?; self.class.terminal? end }
+  Shared::ClassMethods.class_eval       { def terminal?; true  end }
+  Subexpression.instance_eval           { def terminal?; false end }
+
+  Shared.class_eval                     { def capturing?; self.class.capturing? end }
+  Shared::ClassMethods.class_eval       { def capturing?; false end }
+  Group::Capture.instance_eval          { def capturing?; true  end }
+
+  Shared.class_eval                     { def comment?; self.class.comment? end }
+  Shared::ClassMethods.class_eval       { def comment?; false end }
+  Comment.instance_eval                 { def comment?; true  end }
+  Group::Comment.instance_eval          { def comment?; true  end }
+
+  Shared.class_eval                     { def decorative?; self.class.decorative? end }
+  Shared::ClassMethods.class_eval       { def decorative?; false end }
+  FreeSpace.instance_eval               { def decorative?; true  end }
+  Group::Comment.instance_eval          { def decorative?; true  end }
+
+  Shared.class_eval                     { def referential?; self.class.referential? end }
+  Shared::ClassMethods.class_eval       { def referential?; false end }
+  Backreference::Base.instance_eval     { def referential?; true  end }
+  Conditional::Condition.instance_eval  { def referential?; true  end }
+  Conditional::Expression.instance_eval { def referential?; true  end }
 end

data/lib/regexp_parser/expression/methods/traverse.rb

@@ -1,6 +1,22 @@
 module Regexp::Expression
   class Subexpression < Regexp::Expression::Base
 
+    # Traverses the expression, passing each recursive child to the
+    # given block.
+    # If the block takes two arguments, the indices of the children within
+    # their parents are also passed to it.
+    def each_expression(include_self = false, &block)
+      return enum_for(__method__, include_self) unless block
+
+      if block.arity == 1
+        block.call(self) if include_self
+        each_expression_without_index(&block)
+      else
+        block.call(self, 0) if include_self
+        each_expression_with_index(&block)
+      end
+    end
+
     # Traverses the subexpression (depth-first, pre-order) and calls the given
     # block for each expression with three arguments; the traversal event,
     # the expression, and the index of the expression within its parent.
@@ -34,31 +50,31 @@ module Regexp::Expression
     end
     alias :walk :traverse
 
-    # Iterates over the expressions of this expression as an array, passing
-    # the expression and its index within its parent to the given block.
-    def each_expression(include_self = false)
-      return enum_for(__method__, include_self) unless block_given?
-
-      traverse(include_self) do |event, exp, index|
-        yield(exp, index) unless event == :exit
-      end
-    end
-
     # Returns a new array with the results of calling the given block once
     # for every expression. If a block is not given, returns an array with
     # each expression and its level index as an array.
-    def flat_map(include_self = false)
-      result = []
+    def flat_map(include_self = false, &block)
+      case block && block.arity
+      when nil then each_expression(include_self).to_a
+      when 2   then each_expression(include_self).map(&block)
+      else          each_expression(include_self).map { |exp| block.call(exp) }
+      end
+    end
 
-      each_expression(include_self) do |exp, index|
-        if block_given?
-          result << yield(exp, index)
-        else
-          result << [exp, index]
-        end
+    protected
+
+    def each_expression_with_index(&block)
+      each_with_index do |exp, index|
+        block.call(exp, index)
+        exp.each_expression_with_index(&block) unless exp.terminal?
       end
+    end
 
-      result
+    def each_expression_without_index(&block)
+      each do |exp|
+        block.call(exp)
+        exp.each_expression_without_index(&block) unless exp.terminal?
+      end
     end
   end
 end

data/lib/regexp_parser/expression/quantifier.rb

@@ -8,14 +8,10 @@ module Regexp::Expression
 
     MODES = %i[greedy possessive reluctant]
 
-    attr_reader :min, :max, :mode
-
     def initialize(*args)
       deprecated_old_init(*args) and return if args.count == 4 || args.count == 5
 
       init_from_token_and_options(*args)
-      @mode = (token.to_s[/greedy|reluctant|possessive/] || :greedy).to_sym
-      @min, @max = minmax
       # TODO: remove in v3.0.0, stop removing parts of #token (?)
       self.token = token.to_s.sub(/_(greedy|possessive|reluctant)/, '').to_sym
     end
@@ -39,9 +35,21 @@ module Regexp::Expression
     end
     alias :lazy? :reluctant?
 
+    def min
+      derived_data[:min]
+    end
+
+    def max
+      derived_data[:max]
+    end
+
+    def mode
+      derived_data[:mode]
+    end
+
     private
 
-    def deprecated_old_init(token, text, min, max, mode = :greedy)
+    def deprecated_old_init(token, text, _min, _max, _mode = :greedy)
       warn "Calling `Expression::Base#quantify` or `#{self.class}.new` with 4+ arguments "\
            "is deprecated.\nIt will no longer be supported in regexp_parser v3.0.0.\n"\
            "Please pass a Regexp::Token instead, e.g. replace `token, text, min, max, mode` "\
@@ -51,20 +59,25 @@ module Regexp::Expression
            "This is consistent with how Expression::Base instances are created. "
       @token = token
       @text  = text
-      @min   = min
-      @max   = max
-      @mode  = mode
     end
 
-    def minmax
-      case token
-      when /zero_or_one/  then [0, 1]
-      when /zero_or_more/ then [0, -1]
-      when /one_or_more/  then [1, -1]
-      when :interval
-        int_min = text[/\{(\d*)/, 1]
-        int_max = text[/,?(\d*)\}/, 1]
-        [int_min.to_i, (int_max.empty? ? -1 : int_max.to_i)]
+    def derived_data
+      @derived_data ||= begin
+        min, max =
+          case text[0]
+          when '?'; [0, 1]
+          when '*'; [0, -1]
+          when '+'; [1, -1]
+          else
+            int_min = text[/\{(\d*)/, 1]
+            int_max = text[/,?(\d*)\}/, 1]
+            [int_min.to_i, (int_max.empty? ? -1 : int_max.to_i)]
+          end
+
+        mod = text[/.([?+])/, 1]
+        mode = (mod == '?' && :reluctant) || (mod == '+' && :possessive) || :greedy
+
+        { min: min, max: max, mode: mode }
       end
     end
   end

data/lib/regexp_parser/expression/sequence.rb

@@ -12,25 +12,20 @@ module Regexp::Expression
           level:             exp.level,
           set_level:         exp.set_level,
           conditional_level: params[:conditional_level] || exp.conditional_level,
+          ts:                params[:ts],
         )
-        sequence.nesting_level = exp.nesting_level + 1
         sequence.options = active_opts
         exp.expressions << sequence
         sequence
       end
     end
 
-    def starts_at
-      expressions.first.starts_at
+    def ts
+      (head = expressions.first) ? head.ts : @ts
     end
-    alias :ts :starts_at
 
-    def quantify(*args)
-      target = expressions.reverse.find { |exp| !exp.is_a?(FreeSpace) }
-      target or raise Regexp::Parser::Error,
-        "No valid target found for '#{text}' quantifier"
-
-      target.quantify(*args)
+    def quantify(token, *args)
+      extract_quantifier_target(token.text).quantify(token, *args)
     end
   end
 end

data/lib/regexp_parser/expression/sequence_operation.rb

@@ -5,21 +5,16 @@ module Regexp::Expression
     alias :operands :expressions
     alias :operator :text
 
-    def starts_at
-      expressions.first.starts_at
+    def ts
+      (head = expressions.first) ? head.ts : @ts
     end
-    alias :ts :starts_at
 
     def <<(exp)
       expressions.last << exp
     end
 
-    def add_sequence(active_opts = {})
-      self.class::OPERAND.add_to(self, {}, active_opts)
-    end
-
-    def parts
-      intersperse(expressions, text.dup)
+    def add_sequence(active_opts = {}, params = { ts: 0 })
+      self.class::OPERAND.add_to(self, params, active_opts)
     end
   end
 end

data/lib/regexp_parser/expression/shared.rb

@@ -8,7 +8,8 @@ module Regexp::Expression
 
         attr_accessor :type, :token, :text, :ts, :te,
                       :level, :set_level, :conditional_level,
-                      :options
+                      :options, :parent,
+                      :custom_to_s_handling, :pre_quantifier_decorations
 
         attr_reader   :nesting_level, :quantifier
       end
@@ -32,6 +33,10 @@ module Regexp::Expression
       self.text       = orig.text.dup         if orig.text
       self.options    = orig.options.dup      if orig.options
       self.quantifier = orig.quantifier.clone if orig.quantifier
+      self.parent     = nil # updated by Subexpression#initialize_copy
+      if orig.pre_quantifier_decorations
+        self.pre_quantifier_decorations = orig.pre_quantifier_decorations.map(&:dup)
+      end
       super
     end
 
@@ -39,35 +44,51 @@ module Regexp::Expression
       ts
     end
 
+    def ends_at(include_quantifier = true)
+      ts + (include_quantifier ? full_length : base_length)
+    end
+
     def base_length
       to_s(:base).length
     end
 
     def full_length
-      to_s.length
-    end
-
+      to_s(:original).length
+    end
+
+    # #to_s reproduces the original source, as an unparser would.
+    #
+    # It takes an optional format argument.
+    #
+    # Example:
+    #
+    # lit = Regexp::Parser.parse(/a +/x)[0]
+    #
+    # lit.to_s            # => 'a+'  # default; with quantifier
+    # lit.to_s(:full)     # => 'a+'  # default; with quantifier
+    # lit.to_s(:base)     # => 'a'   # without quantifier
+    # lit.to_s(:original) # => 'a +' # with quantifier AND intermittent decorations
+    #
     def to_s(format = :full)
-      "#{parts.join}#{quantifier_affix(format)}"
+      base = parts.each_with_object(''.dup) do |part, buff|
+        if part.instance_of?(String)
+          buff << part
+        elsif !part.custom_to_s_handling
+          buff << part.to_s(:original)
+        end
+      end
+      "#{base}#{pre_quantifier_decoration(format)}#{quantifier_affix(format)}"
     end
     alias :to_str :to_s
 
-    def parts
-      [text.dup]
+    def pre_quantifier_decoration(expression_format = :original)
+      pre_quantifier_decorations.to_a.join if expression_format == :original
     end
 
-    def quantifier_affix(expression_format)
+    def quantifier_affix(expression_format = :full)
       quantifier.to_s if quantified? && expression_format != :base
     end
 
-    def quantified?
-      !quantifier.nil?
-    end
-
-    def optional?
-      quantified? && quantifier.min == 0
-    end
-
     def offset
       [starts_at, full_length]
     end
@@ -76,10 +97,6 @@ module Regexp::Expression
       '@%d+%d' % offset
     end
 
-    def terminal?
-      !respond_to?(:expressions)
-    end
-
     def nesting_level=(lvl)
       @nesting_level = lvl
       quantifier && quantifier.nesting_level = lvl

data/lib/regexp_parser/expression/subexpression.rb

@@ -11,17 +11,15 @@ module Regexp::Expression
 
     # Override base method to clone the expressions as well.
     def initialize_copy(orig)
-      self.expressions = orig.expressions.map(&:clone)
+      self.expressions = orig.expressions.map do |exp|
+        exp.clone.tap { |copy| copy.parent = self }
+      end
       super
     end
 
     def <<(exp)
-      if exp.is_a?(WhiteSpace) && last && last.is_a?(WhiteSpace)
-        last.merge(exp)
-      else
-        exp.nesting_level = nesting_level + 1
-        expressions << exp
-      end
+      exp.parent = self
+      expressions << exp
     end
 
     %w[[] at each empty? fetch index join last length values_at].each do |method|
@@ -39,11 +37,7 @@ module Regexp::Expression
     end
 
     def te
-      ts + to_s.length
-    end
-
-    def parts
-      expressions
+      ts + base_length
     end
 
     def to_h
@@ -53,10 +47,21 @@ module Regexp::Expression
       )
     end
 
-    private
+    def extract_quantifier_target(quantifier_description)
+      pre_quantifier_decorations = []
+      target = expressions.reverse.find do |exp|
+        if exp.decorative?
+          exp.custom_to_s_handling = true
+          pre_quantifier_decorations << exp.text
+          next
+        end
+        exp
+      end
+      target or raise Regexp::Parser::ParserError,
+        "No valid target found for '#{quantifier_description}' quantifier"
 
-    def intersperse(expressions, separator)
-      expressions.flat_map { |exp| [exp, separator] }.slice(0...-1)
+      target.pre_quantifier_decorations = pre_quantifier_decorations
+      target
     end
   end
 end

data/lib/regexp_parser/expression.rb

@@ -1,34 +1,40 @@
-require 'regexp_parser/error'
+require_relative 'error'
 
-require 'regexp_parser/expression/shared'
-require 'regexp_parser/expression/base'
-require 'regexp_parser/expression/quantifier'
-require 'regexp_parser/expression/subexpression'
-require 'regexp_parser/expression/sequence'
-require 'regexp_parser/expression/sequence_operation'
+require_relative 'expression/shared'
+require_relative 'expression/base'
+require_relative 'expression/quantifier'
+require_relative 'expression/subexpression'
+require_relative 'expression/sequence'
+require_relative 'expression/sequence_operation'
 
-require 'regexp_parser/expression/classes/alternation'
-require 'regexp_parser/expression/classes/anchor'
-require 'regexp_parser/expression/classes/backreference'
-require 'regexp_parser/expression/classes/character_set'
-require 'regexp_parser/expression/classes/character_set/intersection'
-require 'regexp_parser/expression/classes/character_set/range'
-require 'regexp_parser/expression/classes/character_type'
-require 'regexp_parser/expression/classes/conditional'
-require 'regexp_parser/expression/classes/escape_sequence'
-require 'regexp_parser/expression/classes/free_space'
-require 'regexp_parser/expression/classes/group'
-require 'regexp_parser/expression/classes/keep'
-require 'regexp_parser/expression/classes/literal'
-require 'regexp_parser/expression/classes/posix_class'
-require 'regexp_parser/expression/classes/root'
-require 'regexp_parser/expression/classes/unicode_property'
+require_relative 'expression/classes/alternation'
+require_relative 'expression/classes/anchor'
+require_relative 'expression/classes/backreference'
+require_relative 'expression/classes/character_set'
+require_relative 'expression/classes/character_set/intersection'
+require_relative 'expression/classes/character_set/range'
+require_relative 'expression/classes/character_type'
+require_relative 'expression/classes/conditional'
+require_relative 'expression/classes/escape_sequence'
+require_relative 'expression/classes/free_space'
+require_relative 'expression/classes/group'
+require_relative 'expression/classes/keep'
+require_relative 'expression/classes/literal'
+require_relative 'expression/classes/posix_class'
+require_relative 'expression/classes/root'
+require_relative 'expression/classes/unicode_property'
 
-require 'regexp_parser/expression/methods/construct'
-require 'regexp_parser/expression/methods/human_name'
-require 'regexp_parser/expression/methods/match'
-require 'regexp_parser/expression/methods/match_length'
-require 'regexp_parser/expression/methods/options'
-require 'regexp_parser/expression/methods/strfregexp'
-require 'regexp_parser/expression/methods/tests'
-require 'regexp_parser/expression/methods/traverse'
+require_relative 'expression/methods/construct'
+require_relative 'expression/methods/escape_sequence_char'
+require_relative 'expression/methods/escape_sequence_codepoint'
+require_relative 'expression/methods/human_name'
+require_relative 'expression/methods/match'
+require_relative 'expression/methods/match_length'
+require_relative 'expression/methods/negative'
+require_relative 'expression/methods/options'
+require_relative 'expression/methods/parts'
+require_relative 'expression/methods/printing'
+require_relative 'expression/methods/referenced_expressions'
+require_relative 'expression/methods/strfregexp'
+require_relative 'expression/methods/tests'
+require_relative 'expression/methods/traverse'