regexp_parser 1.7.0 → 2.8.3

data/lib/regexp_parser/syntax/token/virtual.rb

@@ -0,0 +1,11 @@
+module Regexp::Syntax
+  module Token
+    module Virtual
+      Root     = %i[root]
+      Sequence = %i[sequence]
+
+      All  = %i[root sequence]
+      Type = :expression
+    end
+  end
+end

data/lib/regexp_parser/syntax/token.rb

@@ -0,0 +1,45 @@
+# Define the base module and the simplest of tokens.
+module Regexp::Syntax
+  module Token
+    Map = {}
+
+    module Literal
+      All = %i[literal]
+      Type = :literal
+    end
+
+    module FreeSpace
+      All  = %i[comment whitespace]
+      Type = :free_space
+    end
+
+    Map[FreeSpace::Type] = FreeSpace::All
+    Map[Literal::Type]   = Literal::All
+  end
+end
+
+
+# Load all the token files, they will populate the Map constant.
+require 'regexp_parser/syntax/token/anchor'
+require 'regexp_parser/syntax/token/assertion'
+require 'regexp_parser/syntax/token/backreference'
+require 'regexp_parser/syntax/token/posix_class'
+require 'regexp_parser/syntax/token/character_set'
+require 'regexp_parser/syntax/token/character_type'
+require 'regexp_parser/syntax/token/conditional'
+require 'regexp_parser/syntax/token/escape'
+require 'regexp_parser/syntax/token/group'
+require 'regexp_parser/syntax/token/keep'
+require 'regexp_parser/syntax/token/meta'
+require 'regexp_parser/syntax/token/quantifier'
+require 'regexp_parser/syntax/token/unicode_property'
+
+
+# After loading all the tokens the map is full. Extract all tokens and types
+# into the All and Types constants.
+module Regexp::Syntax
+  module Token
+    All   = Map.values.flatten.uniq.sort.freeze
+    Types = Map.keys.freeze
+  end
+end

data/lib/regexp_parser/syntax/version_lookup.rb

@@ -3,13 +3,13 @@ module Regexp::Syntax
   VERSION_REGEXP = /#{VERSION_FORMAT}/
   VERSION_CONST_REGEXP = /\AV\d+_\d+(?:_\d+)?\z/
 
-  class InvalidVersionNameError < SyntaxError
+  class InvalidVersionNameError < Regexp::Syntax::SyntaxError
     def initialize(name)
       super "Invalid version name '#{name}'. Expected format is '#{VERSION_FORMAT}'"
     end
   end
 
-  class UnknownSyntaxNameError < SyntaxError
+  class UnknownSyntaxNameError < Regexp::Syntax::SyntaxError
     def initialize(name)
       super "Unknown syntax name '#{name}'."
     end
@@ -17,29 +17,30 @@ module Regexp::Syntax
 
   module_function
 
-  # Loads and instantiates an instance of the syntax specification class for
-  # the given syntax version name. The special names 'any' and '*' return an
-  # instance of Syntax::Any.
+  # Returns the syntax specification class for the given syntax
+  # version name. The special names 'any' and '*' return Syntax::Any.
+  def for(name)
+    (@alias_map ||= {})[name] ||= version_class(name)
+  end
+
   def new(name)
-    return Regexp::Syntax::Any.new if ['*', 'any'].include?(name.to_s)
-    version_class(name).new
+    warn 'Regexp::Syntax.new is deprecated in favor of Regexp::Syntax.for. '\
+         'It does not return distinct instances and will be removed in v3.0.0.'
+    self.for(name)
   end
 
   def supported?(name)
-    name =~ VERSION_REGEXP &&
-      comparable_version(name) >= comparable_version('1.8.6')
+    name =~ VERSION_REGEXP && comparable(name) >= comparable('1.8.6')
   end
 
   def version_class(version)
+    return Regexp::Syntax::Any if ['*', 'any'].include?(version.to_s)
+
     version =~ VERSION_REGEXP || raise(InvalidVersionNameError, version)
-    version_const_name = version_const_name(version)
+    version_const_name = "V#{version.to_s.scan(/\d+/).join('_')}"
     const_get(version_const_name) || raise(UnknownSyntaxNameError, version)
   end
 
-  def version_const_name(version_string)
-    "V#{version_string.to_s.scan(/\d+/).join('_')}"
-  end
-
   def const_missing(const_name)
     if const_name =~ VERSION_CONST_REGEXP
       return fallback_version_class(const_name)
@@ -48,35 +49,17 @@ module Regexp::Syntax
   end
 
   def fallback_version_class(version)
-    sorted_versions = (specified_versions + [version])
-                      .sort_by { |name| comparable_version(name) }
-    return if (version_index = sorted_versions.index(version)) < 1
-
-    next_lower_version = sorted_versions[version_index - 1]
-    inherit_from_version(next_lower_version, version)
-  end
-
-  def inherit_from_version(parent_version, new_version)
-    new_const = version_const_name(new_version)
-    parent = const_get(version_const_name(parent_version))
-    const_defined?(new_const) || const_set(new_const, Class.new(parent))
-    warn_if_future_version(new_const)
-    const_get(new_const)
+    sorted = (specified_versions + [version]).sort_by { |ver| comparable(ver) }
+    index = sorted.index(version)
+    index > 0 && const_get(sorted[index - 1])
   end
 
   def specified_versions
     constants.select { |const_name| const_name =~ VERSION_CONST_REGEXP }
   end
 
-  def comparable_version(name)
+  def comparable(name)
     # add .99 to treat versions without a patch value as latest patch version
     Gem::Version.new((name.to_s.scan(/\d+/) << 99).join('.'))
   end
-
-  def warn_if_future_version(const_name)
-    return if comparable_version(const_name) < comparable_version('3.0.0')
-
-    warn('This library has only been tested up to Ruby 2.x, '\
-         "but you are running with #{const_get(const_name).inspect}")
-  end
 end

data/lib/regexp_parser/syntax/versions/1.8.6.rb

@@ -1,21 +1,14 @@
-module Regexp::Syntax
-  class V1_8_6 < Regexp::Syntax::Base
-    def initialize
-      super
-
-      implements :anchor, Anchor::All
-      implements :assertion, Assertion::Lookahead
-      implements :backref, [:number]
-      implements :posixclass, PosixClass::Standard
-      implements :group, Group::All
-      implements :meta, Meta::Extended
-      implements :set, CharacterSet::All
-      implements :type, CharacterType::Extended
-      implements :escape,
-        Escape::Basic + Escape::ASCII + Escape::Meta + Escape::Control
-      implements :quantifier,
-        Quantifier::Greedy + Quantifier::Reluctant +
-        Quantifier::Interval + Quantifier::IntervalReluctant
-    end
-  end
+class Regexp::Syntax::V1_8_6 < Regexp::Syntax::Base
+  implements :anchor,     Anchor::All
+  implements :assertion,  Assertion::Lookahead
+  implements :backref,    Backreference::V1_8_6
+  implements :escape,     Escape::Basic + Escape::ASCII + Escape::Meta + Escape::Control
+  implements :free_space, FreeSpace::All
+  implements :group,      Group::V1_8_6
+  implements :literal,    Literal::All
+  implements :meta,       Meta::Extended
+  implements :posixclass, PosixClass::Standard
+  implements :quantifier, Quantifier::V1_8_6
+  implements :set,        CharacterSet::All
+  implements :type,       CharacterType::Extended
 end

data/lib/regexp_parser/syntax/versions/1.9.1.rb

@@ -1,18 +1,11 @@
-module Regexp::Syntax
-  class V1_9_1 < Regexp::Syntax::V1_8_6
-    def initialize
-      super
-
-      implements :assertion, Assertion::Lookbehind
-      implements :backref, Backreference::All + SubexpressionCall::All
-      implements :posixclass, PosixClass::Extensions
-      implements :nonposixclass, PosixClass::All
-      implements :escape, Escape::Unicode + Escape::Hex + Escape::Octal
-      implements :type, CharacterType::Hex
-      implements :property, UnicodeProperty::V1_9_0
-      implements :nonproperty, UnicodeProperty::V1_9_0
-      implements :quantifier,
-        Quantifier::Possessive + Quantifier::IntervalPossessive
-    end
-  end
+class Regexp::Syntax::V1_9_1 < Regexp::Syntax::V1_8_6
+  implements :assertion,     Assertion::Lookbehind
+  implements :backref,       Backreference::V1_9_1 + SubexpressionCall::All
+  implements :escape,        Escape::Unicode + Escape::Hex + Escape::Octal
+  implements :posixclass,    PosixClass::Extensions
+  implements :nonposixclass, PosixClass::All
+  implements :property,      UnicodeProperty::V1_9_0
+  implements :nonproperty,   UnicodeProperty::V1_9_0
+  implements :quantifier,    Quantifier::Possessive + Quantifier::IntervalPossessive
+  implements :type,          CharacterType::Hex
 end

data/lib/regexp_parser/syntax/versions/1.9.3.rb

@@ -1,11 +1,4 @@
-module Regexp::Syntax
-  class V1_9_3 < Regexp::Syntax::V1_9_1
-    def initialize
-      super
-
-      # these were added with update of Oniguruma to Unicode 6.0
-      implements :property,    UnicodeProperty::V1_9_3
-      implements :nonproperty, UnicodeProperty::V1_9_3
-    end
-  end
+class Regexp::Syntax::V1_9_3 < Regexp::Syntax::V1_9_1
+  implements :property,    UnicodeProperty::V1_9_3
+  implements :nonproperty, UnicodeProperty::V1_9_3
 end

data/lib/regexp_parser/syntax/versions/2.0.0.rb

@@ -1,17 +1,10 @@
-module Regexp::Syntax
-  # use the last 1.9 release as the base
-  class V2_0_0 < Regexp::Syntax::V1_9
-    def initialize
-      super
+class Regexp::Syntax::V2_0_0 < Regexp::Syntax::V1_9_3
+  implements :keep,        Keep::All
+  implements :conditional, Conditional::All
+  implements :property,    UnicodeProperty::V2_0_0
+  implements :nonproperty, UnicodeProperty::V2_0_0
+  implements :type,        CharacterType::Clustered
 
-      implements :keep,        Keep::All
-      implements :conditional, Conditional::All
-      implements :property,    UnicodeProperty::V2_0_0
-      implements :nonproperty, UnicodeProperty::V2_0_0
-      implements :type,        CharacterType::Clustered
-
-      excludes   :property,    :newline
-      excludes   :nonproperty, :newline
-    end
-  end
+  excludes   :property,    %i[newline]
+  excludes   :nonproperty, %i[newline]
 end

data/lib/regexp_parser/syntax/versions/2.2.0.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_2_0 < Regexp::Syntax::V2_1
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_2_0
-      implements :nonproperty, UnicodeProperty::V2_2_0
-    end
-  end
+class Regexp::Syntax::V2_2_0 < Regexp::Syntax::V2_0_0
+  implements :property,    UnicodeProperty::V2_2_0
+  implements :nonproperty, UnicodeProperty::V2_2_0
 end

data/lib/regexp_parser/syntax/versions/2.3.0.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_3_0 < Regexp::Syntax::V2_2
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_3_0
-      implements :nonproperty, UnicodeProperty::V2_3_0
-    end
-  end
+class Regexp::Syntax::V2_3_0 < Regexp::Syntax::V2_2_0
+  implements :property,    UnicodeProperty::V2_3_0
+  implements :nonproperty, UnicodeProperty::V2_3_0
 end

data/lib/regexp_parser/syntax/versions/2.4.0.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_4_0 < Regexp::Syntax::V2_3
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_4_0
-      implements :nonproperty, UnicodeProperty::V2_4_0
-    end
-  end
+class Regexp::Syntax::V2_4_0 < Regexp::Syntax::V2_3_0
+  implements :property,    UnicodeProperty::V2_4_0
+  implements :nonproperty, UnicodeProperty::V2_4_0
 end

data/lib/regexp_parser/syntax/versions/2.4.1.rb

@@ -1,9 +1,3 @@
-module Regexp::Syntax
-  class V2_4_1 < Regexp::Syntax::V2_4_0
-    def initialize
-      super
-
-      implements :group, Group::V2_4_1
-    end
-  end
+class Regexp::Syntax::V2_4_1 < Regexp::Syntax::V2_4_0
+  implements :group, Group::V2_4_1
 end

data/lib/regexp_parser/syntax/versions/2.5.0.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_5_0 < Regexp::Syntax::V2_4
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_5_0
-      implements :nonproperty, UnicodeProperty::V2_5_0
-    end
-  end
+class Regexp::Syntax::V2_5_0 < Regexp::Syntax::V2_4_1
+  implements :property,    UnicodeProperty::V2_5_0
+  implements :nonproperty, UnicodeProperty::V2_5_0
 end

data/lib/regexp_parser/syntax/versions/2.6.0.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_6_0 < Regexp::Syntax::V2_5
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_6_0
-      implements :nonproperty, UnicodeProperty::V2_6_0
-    end
-  end
+class Regexp::Syntax::V2_6_0 < Regexp::Syntax::V2_5_0
+  implements :property,    UnicodeProperty::V2_6_0
+  implements :nonproperty, UnicodeProperty::V2_6_0
 end

data/lib/regexp_parser/syntax/versions/2.6.2.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_6_2 < Regexp::Syntax::V2_6_0
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_6_2
-      implements :nonproperty, UnicodeProperty::V2_6_2
-    end
-  end
+class Regexp::Syntax::V2_6_2 < Regexp::Syntax::V2_6_0
+  implements :property,    UnicodeProperty::V2_6_2
+  implements :nonproperty, UnicodeProperty::V2_6_2
 end

data/lib/regexp_parser/syntax/versions/2.6.3.rb

@@ -1,10 +1,4 @@
-module Regexp::Syntax
-  class V2_6_3 < Regexp::Syntax::V2_6_2
-    def initialize
-      super
-
-      implements :property,    UnicodeProperty::V2_6_3
-      implements :nonproperty, UnicodeProperty::V2_6_3
-    end
-  end
+class Regexp::Syntax::V2_6_3 < Regexp::Syntax::V2_6_2
+  implements :property,    UnicodeProperty::V2_6_3
+  implements :nonproperty, UnicodeProperty::V2_6_3
 end

data/lib/regexp_parser/syntax/versions/3.1.0.rb

@@ -0,0 +1,4 @@
+class Regexp::Syntax::V3_1_0 < Regexp::Syntax::V2_6_3
+  implements :property,    UnicodeProperty::V3_1_0
+  implements :nonproperty, UnicodeProperty::V3_1_0
+end

data/lib/regexp_parser/syntax/versions/3.2.0.rb

@@ -0,0 +1,4 @@
+class Regexp::Syntax::V3_2_0 < Regexp::Syntax::V3_1_0
+  implements :property,    UnicodeProperty::V3_2_0
+  implements :nonproperty, UnicodeProperty::V3_2_0
+end

data/lib/regexp_parser/syntax/versions.rb

@@ -1,6 +1,8 @@
-# Ruby 1.8.x is no longer a supported runtime,
+# Ruby 1.x is no longer a supported runtime,
 # but its regex features are still recognized.
 #
 # Aliases for the latest patch version are provided as 'ruby/n.n',
 # e.g. 'ruby/1.9' refers to Ruby v1.9.3.
 Dir[File.expand_path('../versions/*.rb', __FILE__)].sort.each { |f| require f }
+
+Regexp::Syntax::CURRENT = Regexp::Syntax.for("ruby/#{RUBY_VERSION}")

data/lib/regexp_parser/syntax.rb

@@ -1,9 +1,11 @@
-require File.expand_path('../syntax/tokens', __FILE__)
-require File.expand_path('../syntax/base', __FILE__)
-require File.expand_path('../syntax/any', __FILE__)
-require File.expand_path('../syntax/version_lookup', __FILE__)
-require File.expand_path('../syntax/versions', __FILE__)
+require 'regexp_parser/error'
 
 module Regexp::Syntax
-  class SyntaxError < StandardError; end
+  class SyntaxError < Regexp::Parser::Error; end
 end
+
+require_relative 'syntax/token'
+require_relative 'syntax/base'
+require_relative 'syntax/any'
+require_relative 'syntax/version_lookup'
+require_relative 'syntax/versions'

data/lib/regexp_parser/token.rb

@@ -1,14 +1,13 @@
 class Regexp
-
-  TOKEN_KEYS = [
-    :type,
-    :token,
-    :text,
-    :ts,
-    :te,
-    :level,
-    :set_level,
-    :conditional_level
+  TOKEN_KEYS = %i[
+    type
+    token
+    text
+    ts
+    te
+    level
+    set_level
+    conditional_level
   ].freeze
 
   Token = Struct.new(*TOKEN_KEYS) do
@@ -21,15 +20,5 @@ class Regexp
     def length
       te - ts
     end
-
-    if RUBY_VERSION < '2.0.0'
-      def to_h
-        members.inject({}) do |hash, member|
-          hash[member.to_sym] = self[member]
-          hash
-        end
-      end
-    end
   end
-
 end

data/lib/regexp_parser/version.rb

@@ -1,5 +1,5 @@
 class Regexp
   class Parser
-    VERSION = '1.7.0'
+    VERSION = '2.8.3'
   end
 end

data/lib/regexp_parser.rb

@@ -1,5 +1,3 @@
-# encoding: utf-8
-
 require 'regexp_parser/version'
 require 'regexp_parser/token'
 require 'regexp_parser/scanner'

data/regexp_parser.gemspec

@@ -2,35 +2,31 @@ $:.unshift File.join(File.dirname(__FILE__), 'lib')
 
 require 'regexp_parser/version'
 
-Gem::Specification.new do |gem|
-  gem.name          = 'regexp_parser'
-  gem.version       = ::Regexp::Parser::VERSION
+Gem::Specification.new do |spec|
+  spec.name          = 'regexp_parser'
+  spec.version       = ::Regexp::Parser::VERSION
 
-  gem.summary       = "Scanner, lexer, parser for ruby's regular expressions"
-  gem.description   = 'A library for tokenizing, lexing, and parsing Ruby regular expressions.'
-  gem.homepage      = 'https://github.com/ammar/regexp_parser'
+  spec.summary       = "Scanner, lexer, parser for ruby's regular expressions"
+  spec.description   = 'A library for tokenizing, lexing, and parsing Ruby regular expressions.'
+  spec.homepage      = 'https://github.com/ammar/regexp_parser'
 
-  if gem.respond_to?(:metadata)
-    gem.metadata    = { 'issue_tracker' => 'https://github.com/ammar/regexp_parser/issues' }
-  end
+  spec.metadata['bug_tracker_uri'] = "#{spec.homepage}/issues"
+  spec.metadata['changelog_uri']   = "#{spec.homepage}/blob/master/CHANGELOG.md"
+  spec.metadata['homepage_uri']    = spec.homepage
+  spec.metadata['source_code_uri'] = spec.homepage
+  spec.metadata['wiki_uri']        = "#{spec.homepage}/wiki"
 
-  gem.authors       = ['Ammar Ali']
-  gem.email         = ['ammarabuali@gmail.com']
+  spec.authors       = ['Ammar Ali', 'Janosch Müller']
+  spec.email         = ['ammarabuali@gmail.com', 'janosch84@gmail.com']
 
-  gem.license       = 'MIT'
+  spec.license       = 'MIT'
 
-  gem.require_paths = ['lib']
+  spec.require_paths = ['lib']
 
-  gem.files         = Dir.glob('{lib,spec}/**/*.rb') +
-                      Dir.glob('lib/**/*.rl') +
-                      Dir.glob('lib/**/*.yml') +
-                      %w(Gemfile Rakefile LICENSE README.md CHANGELOG.md regexp_parser.gemspec)
+  spec.files         = Dir.glob('lib/**/*.{csv,rb,rl}') +
+                       %w[Gemfile Rakefile LICENSE regexp_parser.gemspec]
 
-  gem.test_files    = Dir.glob('spec/**/*.rb')
+  spec.platform      = Gem::Platform::RUBY
 
-  gem.rdoc_options  = ["--inline-source", "--charset=UTF-8"]
-
-  gem.platform      = Gem::Platform::RUBY
-
-  gem.required_ruby_version = '>= 1.9.1'
+  spec.required_ruby_version = '>= 2.0.0'
 end