regexp_parser 2.7.0 → 2.8.0

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

@@ -12,6 +12,7 @@ module Regexp::Expression
           level:             exp.level,
           set_level:         exp.set_level,
           conditional_level: params[:conditional_level] || exp.conditional_level,
+          ts:                params[:ts],
         )
         sequence.options = active_opts
         exp.expressions << sequence
@@ -19,17 +20,12 @@ module Regexp::Expression
       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
@@ -76,14 +97,6 @@ module Regexp::Expression
       '@%d+%d' % offset
     end
 
-    def terminal?
-      true # overridden to be false in Expression::Subexpression
-    end
-
-    def referential?
-      false # overridden to be true e.g. in Expression::Backreference::Base
-    end
-
     def nesting_level=(lvl)
       @nesting_level = lvl
       quantifier && quantifier.nesting_level = lvl

data/lib/regexp_parser/expression/subexpression.rb

@@ -11,16 +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
-        expressions << exp
-      end
+      exp.parent = self
+      expressions << exp
     end
 
     %w[[] at each empty? fetch index join last length values_at].each do |method|
@@ -38,11 +37,7 @@ module Regexp::Expression
     end
 
     def te
-      ts + to_s.length
-    end
-
-    def parts
-      expressions
+      ts + base_length
     end
 
     def to_h
@@ -52,14 +47,21 @@ module Regexp::Expression
       )
     end
 
-    def terminal?
-      false
-    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

@@ -29,6 +29,8 @@ 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/parts'
+require 'regexp_parser/expression/methods/printing'
 require 'regexp_parser/expression/methods/strfregexp'
 require 'regexp_parser/expression/methods/tests'
 require 'regexp_parser/expression/methods/traverse'

data/lib/regexp_parser/lexer.rb

@@ -6,7 +6,7 @@ class Regexp::Lexer
 
   OPENING_TOKENS = %i[
     capture passive lookahead nlookahead lookbehind nlookbehind
-    atomic options options_switch named absence
+    atomic options options_switch named absence open
   ].freeze
 
   CLOSING_TOKENS = %i[close].freeze
@@ -89,24 +89,32 @@ class Regexp::Lexer
                 :nesting, :set_nesting, :conditional_nesting, :shift
 
   def ascend(type, token)
+    return unless CLOSING_TOKENS.include?(token)
+
     case type
     when :group, :assertion
-      self.nesting = nesting - 1 if CLOSING_TOKENS.include?(token)
+      self.nesting = nesting - 1
     when :set
-      self.set_nesting = set_nesting - 1 if token == :close
+      self.set_nesting = set_nesting - 1
     when :conditional
-      self.conditional_nesting = conditional_nesting - 1 if token == :close
+      self.conditional_nesting = conditional_nesting - 1
+    else
+      raise "unhandled nesting type #{type}"
     end
   end
 
   def descend(type, token)
+    return unless OPENING_TOKENS.include?(token)
+
     case type
     when :group, :assertion
-      self.nesting = nesting + 1 if OPENING_TOKENS.include?(token)
+      self.nesting = nesting + 1
     when :set
-      self.set_nesting = set_nesting + 1 if token == :open
+      self.set_nesting = set_nesting + 1
     when :conditional
-      self.conditional_nesting = conditional_nesting + 1 if token == :open
+      self.conditional_nesting = conditional_nesting + 1
+    else
+      raise "unhandled nesting type #{type}"
     end
   end
 

data/lib/regexp_parser/parser.rb

@@ -232,7 +232,7 @@ class Regexp::Parser
       node << Backreference::NameRecursionLevel.new(token, active_opts)
     when :name_call
       node << Backreference::NameCall.new(token, active_opts)
-    when :number, :number_ref
+    when :number, :number_ref # TODO: split in v3.0.0
       node << Backreference::Number.new(token, active_opts)
     when :number_recursion_ref
       node << Backreference::NumberRecursionLevel.new(token, active_opts).tap do |exp|
@@ -272,9 +272,9 @@ class Regexp::Parser
       nest_conditional(Conditional::Expression.new(token, active_opts))
     when :condition
       conditional_nesting.last.condition = Conditional::Condition.new(token, active_opts)
-      conditional_nesting.last.add_sequence(active_opts)
+      conditional_nesting.last.add_sequence(active_opts, { ts: token.te })
     when :separator
-      conditional_nesting.last.add_sequence(active_opts)
+      conditional_nesting.last.add_sequence(active_opts, { ts: token.te })
       self.node = conditional_nesting.last.branches.last
     when :close
       conditional_nesting.pop
@@ -322,6 +322,7 @@ class Regexp::Parser
 
     when :control
       if token.text =~ /\A(?:\\C-\\M|\\c\\M)/
+        # TODO: emit :meta_control_sequence token in v3.0.0
         node << EscapeSequence::MetaControl.new(token, active_opts)
       else
         node << EscapeSequence::Control.new(token, active_opts)
@@ -329,6 +330,7 @@ class Regexp::Parser
 
     when :meta_sequence
       if token.text =~ /\A\\M-\\[Cc]/
+        # TODO: emit :meta_control_sequence token in v3.0.0:
         node << EscapeSequence::MetaControl.new(token, active_opts)
       else
         node << EscapeSequence::Meta.new(token, active_opts)
@@ -349,11 +351,7 @@ class Regexp::Parser
     when :comment
       node << Comment.new(token, active_opts)
     when :whitespace
-      if node.last.is_a?(WhiteSpace)
-        node.last.merge(WhiteSpace.new(token, active_opts))
-      else
-        node << WhiteSpace.new(token, active_opts)
-      end
+      node << WhiteSpace.new(token, active_opts)
     else
       raise UnknownTokenError.new('FreeSpace', token)
     end
@@ -381,96 +379,96 @@ class Regexp::Parser
   def sequence_operation(klass, token)
     unless node.instance_of?(klass)
       operator = klass.new(token, active_opts)
-      sequence = operator.add_sequence(active_opts)
+      sequence = operator.add_sequence(active_opts, { ts: token.ts })
       sequence.expressions = node.expressions
       node.expressions = []
       nest(operator)
     end
-    node.add_sequence(active_opts)
+    node.add_sequence(active_opts, { ts: token.te })
   end
 
   def posixclass(token)
     node << PosixClass.new(token, active_opts)
   end
 
-  include Regexp::Expression::UnicodeProperty
-  UPTokens = Regexp::Syntax::Token::UnicodeProperty
+  UP = Regexp::Expression::Property
+  UPTokens = Regexp::Syntax::Token::Property
 
   def property(token)
     case token.token
-    when :alnum;                  node << Alnum.new(token, active_opts)
-    when :alpha;                  node << Alpha.new(token, active_opts)
-    when :ascii;                  node << Ascii.new(token, active_opts)
-    when :blank;                  node << Blank.new(token, active_opts)
-    when :cntrl;                  node << Cntrl.new(token, active_opts)
-    when :digit;                  node << Digit.new(token, active_opts)
-    when :graph;                  node << Graph.new(token, active_opts)
-    when :lower;                  node << Lower.new(token, active_opts)
-    when :print;                  node << Print.new(token, active_opts)
-    when :punct;                  node << Punct.new(token, active_opts)
-    when :space;                  node << Space.new(token, active_opts)
-    when :upper;                  node << Upper.new(token, active_opts)
-    when :word;                   node << Word.new(token, active_opts)
-    when :xdigit;                 node << Xdigit.new(token, active_opts)
-    when :xposixpunct;            node << XPosixPunct.new(token, active_opts)
+    when :alnum;                  node << UP::Alnum.new(token, active_opts)
+    when :alpha;                  node << UP::Alpha.new(token, active_opts)
+    when :ascii;                  node << UP::Ascii.new(token, active_opts)
+    when :blank;                  node << UP::Blank.new(token, active_opts)
+    when :cntrl;                  node << UP::Cntrl.new(token, active_opts)
+    when :digit;                  node << UP::Digit.new(token, active_opts)
+    when :graph;                  node << UP::Graph.new(token, active_opts)
+    when :lower;                  node << UP::Lower.new(token, active_opts)
+    when :print;                  node << UP::Print.new(token, active_opts)
+    when :punct;                  node << UP::Punct.new(token, active_opts)
+    when :space;                  node << UP::Space.new(token, active_opts)
+    when :upper;                  node << UP::Upper.new(token, active_opts)
+    when :word;                   node << UP::Word.new(token, active_opts)
+    when :xdigit;                 node << UP::Xdigit.new(token, active_opts)
+    when :xposixpunct;            node << UP::XPosixPunct.new(token, active_opts)
 
     # only in Oniguruma (old rubies)
-    when :newline;                node << Newline.new(token, active_opts)
-
-    when :any;                    node << Any.new(token, active_opts)
-    when :assigned;               node << Assigned.new(token, active_opts)
-
-    when :letter;                 node << Letter::Any.new(token, active_opts)
-    when :cased_letter;           node << Letter::Cased.new(token, active_opts)
-    when :uppercase_letter;       node << Letter::Uppercase.new(token, active_opts)
-    when :lowercase_letter;       node << Letter::Lowercase.new(token, active_opts)
-    when :titlecase_letter;       node << Letter::Titlecase.new(token, active_opts)
-    when :modifier_letter;        node << Letter::Modifier.new(token, active_opts)
-    when :other_letter;           node << Letter::Other.new(token, active_opts)
-
-    when :mark;                   node << Mark::Any.new(token, active_opts)
-    when :combining_mark;         node << Mark::Combining.new(token, active_opts)
-    when :nonspacing_mark;        node << Mark::Nonspacing.new(token, active_opts)
-    when :spacing_mark;           node << Mark::Spacing.new(token, active_opts)
-    when :enclosing_mark;         node << Mark::Enclosing.new(token, active_opts)
-
-    when :number;                 node << Number::Any.new(token, active_opts)
-    when :decimal_number;         node << Number::Decimal.new(token, active_opts)
-    when :letter_number;          node << Number::Letter.new(token, active_opts)
-    when :other_number;           node << Number::Other.new(token, active_opts)
-
-    when :punctuation;            node << Punctuation::Any.new(token, active_opts)
-    when :connector_punctuation;  node << Punctuation::Connector.new(token, active_opts)
-    when :dash_punctuation;       node << Punctuation::Dash.new(token, active_opts)
-    when :open_punctuation;       node << Punctuation::Open.new(token, active_opts)
-    when :close_punctuation;      node << Punctuation::Close.new(token, active_opts)
-    when :initial_punctuation;    node << Punctuation::Initial.new(token, active_opts)
-    when :final_punctuation;      node << Punctuation::Final.new(token, active_opts)
-    when :other_punctuation;      node << Punctuation::Other.new(token, active_opts)
-
-    when :separator;              node << Separator::Any.new(token, active_opts)
-    when :space_separator;        node << Separator::Space.new(token, active_opts)
-    when :line_separator;         node << Separator::Line.new(token, active_opts)
-    when :paragraph_separator;    node << Separator::Paragraph.new(token, active_opts)
-
-    when :symbol;                 node << Symbol::Any.new(token, active_opts)
-    when :math_symbol;            node << Symbol::Math.new(token, active_opts)
-    when :currency_symbol;        node << Symbol::Currency.new(token, active_opts)
-    when :modifier_symbol;        node << Symbol::Modifier.new(token, active_opts)
-    when :other_symbol;           node << Symbol::Other.new(token, active_opts)
-
-    when :other;                  node << Codepoint::Any.new(token, active_opts)
-    when :control;                node << Codepoint::Control.new(token, active_opts)
-    when :format;                 node << Codepoint::Format.new(token, active_opts)
-    when :surrogate;              node << Codepoint::Surrogate.new(token, active_opts)
-    when :private_use;            node << Codepoint::PrivateUse.new(token, active_opts)
-    when :unassigned;             node << Codepoint::Unassigned.new(token, active_opts)
-
-    when *UPTokens::Age;          node << Age.new(token, active_opts)
-    when *UPTokens::Derived;      node << Derived.new(token, active_opts)
-    when *UPTokens::Emoji;        node << Emoji.new(token, active_opts)
-    when *UPTokens::Script;       node << Script.new(token, active_opts)
-    when *UPTokens::UnicodeBlock; node << Block.new(token, active_opts)
+    when :newline;                node << UP::Newline.new(token, active_opts)
+
+    when :any;                    node << UP::Any.new(token, active_opts)
+    when :assigned;               node << UP::Assigned.new(token, active_opts)
+
+    when :letter;                 node << UP::Letter::Any.new(token, active_opts)
+    when :cased_letter;           node << UP::Letter::Cased.new(token, active_opts)
+    when :uppercase_letter;       node << UP::Letter::Uppercase.new(token, active_opts)
+    when :lowercase_letter;       node << UP::Letter::Lowercase.new(token, active_opts)
+    when :titlecase_letter;       node << UP::Letter::Titlecase.new(token, active_opts)
+    when :modifier_letter;        node << UP::Letter::Modifier.new(token, active_opts)
+    when :other_letter;           node << UP::Letter::Other.new(token, active_opts)
+
+    when :mark;                   node << UP::Mark::Any.new(token, active_opts)
+    when :combining_mark;         node << UP::Mark::Combining.new(token, active_opts)
+    when :nonspacing_mark;        node << UP::Mark::Nonspacing.new(token, active_opts)
+    when :spacing_mark;           node << UP::Mark::Spacing.new(token, active_opts)
+    when :enclosing_mark;         node << UP::Mark::Enclosing.new(token, active_opts)
+
+    when :number;                 node << UP::Number::Any.new(token, active_opts)
+    when :decimal_number;         node << UP::Number::Decimal.new(token, active_opts)
+    when :letter_number;          node << UP::Number::Letter.new(token, active_opts)
+    when :other_number;           node << UP::Number::Other.new(token, active_opts)
+
+    when :punctuation;            node << UP::Punctuation::Any.new(token, active_opts)
+    when :connector_punctuation;  node << UP::Punctuation::Connector.new(token, active_opts)
+    when :dash_punctuation;       node << UP::Punctuation::Dash.new(token, active_opts)
+    when :open_punctuation;       node << UP::Punctuation::Open.new(token, active_opts)
+    when :close_punctuation;      node << UP::Punctuation::Close.new(token, active_opts)
+    when :initial_punctuation;    node << UP::Punctuation::Initial.new(token, active_opts)
+    when :final_punctuation;      node << UP::Punctuation::Final.new(token, active_opts)
+    when :other_punctuation;      node << UP::Punctuation::Other.new(token, active_opts)
+
+    when :separator;              node << UP::Separator::Any.new(token, active_opts)
+    when :space_separator;        node << UP::Separator::Space.new(token, active_opts)
+    when :line_separator;         node << UP::Separator::Line.new(token, active_opts)
+    when :paragraph_separator;    node << UP::Separator::Paragraph.new(token, active_opts)
+
+    when :symbol;                 node << UP::Symbol::Any.new(token, active_opts)
+    when :math_symbol;            node << UP::Symbol::Math.new(token, active_opts)
+    when :currency_symbol;        node << UP::Symbol::Currency.new(token, active_opts)
+    when :modifier_symbol;        node << UP::Symbol::Modifier.new(token, active_opts)
+    when :other_symbol;           node << UP::Symbol::Other.new(token, active_opts)
+
+    when :other;                  node << UP::Codepoint::Any.new(token, active_opts)
+    when :control;                node << UP::Codepoint::Control.new(token, active_opts)
+    when :format;                 node << UP::Codepoint::Format.new(token, active_opts)
+    when :surrogate;              node << UP::Codepoint::Surrogate.new(token, active_opts)
+    when :private_use;            node << UP::Codepoint::PrivateUse.new(token, active_opts)
+    when :unassigned;             node << UP::Codepoint::Unassigned.new(token, active_opts)
+
+    when *UPTokens::Age;          node << UP::Age.new(token, active_opts)
+    when *UPTokens::Derived;      node << UP::Derived.new(token, active_opts)
+    when *UPTokens::Emoji;        node << UP::Emoji.new(token, active_opts)
+    when *UPTokens::Script;       node << UP::Script.new(token, active_opts)
+    when *UPTokens::UnicodeBlock; node << UP::Block.new(token, active_opts)
 
     else
       raise UnknownTokenError.new('UnicodeProperty', token)
@@ -478,8 +476,7 @@ class Regexp::Parser
   end
 
   def quantifier(token)
-    target_node = node.expressions.reverse.find { |exp| !exp.is_a?(FreeSpace) }
-    target_node or raise ParserError, "No valid target found for '#{token.text}'"
+    target_node = node.extract_quantifier_target(token.text)
 
     # in case of chained quantifiers, wrap target in an implicit passive group
     # description of the problem: https://github.com/ammar/regexp_parser/issues/3
@@ -527,6 +524,8 @@ class Regexp::Parser
   end
 
   def open_set(token)
+    # TODO: this and Quantifier are the only cases where Expression#token
+    # does not match the scanner/lexer output. Fix in v3.0.0.
     token.token = :character
     nest(CharacterSet.new(token, active_opts))
   end
@@ -590,7 +589,7 @@ class Regexp::Parser
     # (in a second iteration because there might be forward references)
     referrers.each do |exp|
       exp.referenced_expression = targets[exp.reference] ||
-        raise(ParserError, "Invalid reference: #{exp.reference}")
+        raise(ParserError, "Invalid reference #{exp.reference} at pos #{exp.ts}")
     end
   end
 end # module Regexp::Parser

data/lib/regexp_parser/scanner/errors/premature_end_error.rb

@@ -0,0 +1,8 @@
+class Regexp::Scanner
+  # Unexpected end of pattern
+  class PrematureEndError < ScannerError
+    def initialize(where = '')
+      super "Premature end of pattern at #{where}"
+    end
+  end
+end

data/lib/regexp_parser/scanner/errors/scanner_error.rb

@@ -0,0 +1,6 @@
+require 'regexp_parser/error'
+
+class Regexp::Scanner
+  # General scanner error (catch all)
+  class ScannerError < Regexp::Parser::Error; end
+end