Chris@210: # encoding: utf-8
Chris@210: module CodeRay
Chris@210: module Scanners
Chris@210: 
Chris@210:   module Ruby::Patterns  # :nodoc:
Chris@210: 
Chris@210:     RESERVED_WORDS = %w[
Chris@210:       and def end in or unless begin
Chris@210:       defined? ensure module redo super until
Chris@210:       BEGIN break do next rescue then
Chris@210:       when END case else for retry
Chris@210:       while alias class elsif if not return
Chris@210:       undef yield
Chris@210:     ]
Chris@210: 
Chris@210:     DEF_KEYWORDS = %w[ def ]
Chris@210:     UNDEF_KEYWORDS = %w[ undef ]
Chris@210:     ALIAS_KEYWORDS = %w[ alias ]
Chris@210:     MODULE_KEYWORDS = %w[ class module ]
Chris@210:     DEF_NEW_STATE = WordList.new(:initial).
Chris@210:       add(DEF_KEYWORDS, :def_expected).
Chris@210:       add(UNDEF_KEYWORDS, :undef_expected).
Chris@210:       add(ALIAS_KEYWORDS, :alias_expected).
Chris@210:       add(MODULE_KEYWORDS, :module_expected)
Chris@210: 
Chris@210:     PREDEFINED_CONSTANTS = %w[
Chris@210:       nil true false self
Chris@210:       DATA ARGV ARGF
Chris@210:       __FILE__ __LINE__ __ENCODING__
Chris@210:     ]
Chris@210: 
Chris@210:     IDENT_KIND = WordList.new(:ident).
Chris@210:       add(RESERVED_WORDS, :reserved).
Chris@210:       add(PREDEFINED_CONSTANTS, :pre_constant)
Chris@210: 
Chris@210:     if /\w/u === '∑'
Chris@210:       # MRI 1.8.6, 1.8.7
Chris@210:       IDENT = /[^\W\d]\w*/
Chris@210:     else
Chris@210:       if //.respond_to? :encoding
Chris@210:         # MRI 1.9.1, 1.9.2
Chris@210:         IDENT = Regexp.new '[\p{L}\p{M}\p{Pc}\p{Sm}&&[^\x00-\x40\x5b-\x5e\x60\x7b-\x7f]][\p{L}\p{M}\p{N}\p{Pc}\p{Sm}&&[^\x00-\x2f\x3a-\x40\x5b-\x5e\x60\x7b-\x7f]]*'
Chris@210:       else
Chris@210:         # JRuby, Rubinius
Chris@210:         IDENT = /[^\x00-\x40\x5b-\x5e\x60\x7b-\x7f][^\x00-\x2f\x3a-\x40\x5b-\x5e\x60\x7b-\x7f]*/
Chris@210:       end
Chris@210:     end
Chris@210: 
Chris@210:     METHOD_NAME = / #{IDENT} [?!]? /ox
Chris@210:     METHOD_NAME_OPERATOR = /
Chris@210:       \*\*?           # multiplication and power
Chris@210:       | [-+~]@?       # plus, minus, tilde with and without at sign
Chris@210:       | [\/%&|^`]     # division, modulo or format strings, and, or, xor, system
Chris@210:       | \[\]=?        # array getter and setter
Chris@210:       | << | >>       # append or shift left, shift right
Chris@210:       | <=?>? | >=?   # comparison, rocket operator
Chris@210:       | ===? | =~     # simple equality, case equality, match
Chris@210:       | ![~=@]?       # negation with and without at sign, not-equal and not-match
Chris@210:     /ox
Chris@210:     METHOD_NAME_EX = / #{IDENT} (?:[?!]|=(?!>))? | #{METHOD_NAME_OPERATOR} /ox
Chris@210:     INSTANCE_VARIABLE = / @ #{IDENT} /ox
Chris@210:     CLASS_VARIABLE = / @@ #{IDENT} /ox
Chris@210:     OBJECT_VARIABLE = / @@? #{IDENT} /ox
Chris@210:     GLOBAL_VARIABLE = / \$ (?: #{IDENT} | [1-9]\d* | 0\w* | [~&+`'=\/,;_.<>!@$?*":\\] | -[a-zA-Z_0-9] ) /ox
Chris@210:     PREFIX_VARIABLE = / #{GLOBAL_VARIABLE} | #{OBJECT_VARIABLE} /ox
Chris@210:     VARIABLE = / @?@? #{IDENT} | #{GLOBAL_VARIABLE} /ox
Chris@210: 
Chris@210:     QUOTE_TO_TYPE = {
Chris@210:       '`' => :shell,
Chris@210:       '/'=> :regexp,
Chris@210:     }
Chris@210:     QUOTE_TO_TYPE.default = :string
Chris@210: 
Chris@210:     REGEXP_MODIFIERS = /[mixounse]*/
Chris@210:     REGEXP_SYMBOLS = /[|?*+(){}\[\].^$]/
Chris@210: 
Chris@210:     DECIMAL = /\d+(?:_\d+)*/
Chris@210:     OCTAL = /0_?[0-7]+(?:_[0-7]+)*/
Chris@210:     HEXADECIMAL = /0x[0-9A-Fa-f]+(?:_[0-9A-Fa-f]+)*/
Chris@210:     BINARY = /0b[01]+(?:_[01]+)*/
Chris@210: 
Chris@210:     EXPONENT = / [eE] [+-]? #{DECIMAL} /ox
Chris@210:     FLOAT_SUFFIX = / #{EXPONENT} | \. #{DECIMAL} #{EXPONENT}? /ox
Chris@210:     FLOAT_OR_INT = / #{DECIMAL} (?: #{FLOAT_SUFFIX} () )? /ox
Chris@210:     NUMERIC = / (?: (?=0) (?: #{OCTAL} | #{HEXADECIMAL} | #{BINARY} ) | #{FLOAT_OR_INT} ) /ox
Chris@210: 
Chris@210:     SYMBOL = /
Chris@210:       :
Chris@210:       (?:
Chris@210:         #{METHOD_NAME_EX}
Chris@210:       | #{PREFIX_VARIABLE}
Chris@210:       | ['"]
Chris@210:       )
Chris@210:     /ox
Chris@210:     METHOD_NAME_OR_SYMBOL = / #{METHOD_NAME_EX} | #{SYMBOL} /ox
Chris@210: 
Chris@210:     SIMPLE_ESCAPE = /
Chris@210:         [abefnrstv]
Chris@210:       |  [0-7]{1,3}
Chris@210:       | x[0-9A-Fa-f]{1,2}
Chris@210:       | .?
Chris@210:     /mx
Chris@210:     
Chris@210:     CONTROL_META_ESCAPE = /
Chris@210:       (?: M-|C-|c )
Chris@210:       (?: \\ (?: M-|C-|c ) )*
Chris@210:       (?: [^\\] | \\ #{SIMPLE_ESCAPE} )?
Chris@210:     /mox
Chris@210:     
Chris@210:     ESCAPE = /
Chris@210:       #{CONTROL_META_ESCAPE} | #{SIMPLE_ESCAPE}
Chris@210:     /mox
Chris@210:     
Chris@210:     CHARACTER = /
Chris@210:       \?
Chris@210:       (?:
Chris@210:         [^\s\\]
Chris@210:       | \\ #{ESCAPE}
Chris@210:       )
Chris@210:     /mox
Chris@210: 
Chris@210:     # NOTE: This is not completely correct, but
Chris@210:     # nobody needs heredoc delimiters ending with \n.
Chris@210:     # Also, delimiters starting with numbers are allowed.
Chris@210:     # but they are more often than not a false positive.
Chris@210:     HEREDOC_OPEN = /
Chris@210:       << (-)?              # $1 = float
Chris@210:       (?:
Chris@210:         ( #{IDENT} )       # $2 = delim
Chris@210:       |
Chris@210:         ( ["'`\/] )        # $3 = quote, type
Chris@210:         ( [^\n]*? ) \3     # $4 = delim
Chris@210:       )
Chris@210:     /mx
Chris@210: 
Chris@210:     RUBYDOC = /
Chris@210:       =begin (?!\S)
Chris@210:       .*?
Chris@210:       (?: \Z | ^=end (?!\S) [^\n]* )
Chris@210:     /mx
Chris@210: 
Chris@210:     DATA = /
Chris@210:       __END__$
Chris@210:       .*?
Chris@210:       (?: \Z | (?=^\#CODE) )
Chris@210:     /mx
Chris@210:     
Chris@210:     # Checks for a valid value to follow. This enables
Chris@210:     # value_expected in method calls without parentheses.
Chris@210:     VALUE_FOLLOWS = /
Chris@210:       (?>[ \t\f\v]+)
Chris@210:       (?:
Chris@210:         [%\/][^\s=]
Chris@210:       | <<-?\S
Chris@210:       | [-+] \d
Chris@210:       | #{CHARACTER}
Chris@210:       )
Chris@210:     /x
Chris@210:     KEYWORDS_EXPECTING_VALUE = WordList.new.add(%w[
Chris@210:       and end in or unless begin
Chris@210:       defined? ensure redo super until
Chris@210:       break do next rescue then
Chris@210:       when case else for retry
Chris@210:       while elsif if not return
Chris@210:       yield
Chris@210:     ])
Chris@210: 
Chris@210:     RUBYDOC_OR_DATA = / #{RUBYDOC} | #{DATA} /xo
Chris@210: 
Chris@210:     RDOC_DATA_START = / ^=begin (?!\S) | ^__END__$ /x
Chris@210: 
Chris@210:     FANCY_START_CORRECT = / % ( [qQwWxsr] | (?![a-zA-Z0-9]) ) ([^a-zA-Z0-9]) /mx
Chris@210: 
Chris@210:     FancyStringType = {
Chris@210:       'q' => [:string, false],
Chris@210:       'Q' => [:string, true],
Chris@210:       'r' => [:regexp, true],
Chris@210:       's' => [:symbol, false],
Chris@210:       'x' => [:shell, true]
Chris@210:     }
Chris@210:     FancyStringType['w'] = FancyStringType['q']
Chris@210:     FancyStringType['W'] = FancyStringType[''] = FancyStringType['Q']
Chris@210: 
Chris@210:     class StringState < Struct.new :type, :interpreted, :delim, :heredoc,
Chris@210:       :paren, :paren_depth, :pattern, :next_state
Chris@210: 
Chris@210:       CLOSING_PAREN = Hash[ *%w[
Chris@210:         ( )
Chris@210:         [ ]
Chris@210:         < >
Chris@210:         { }
Chris@210:       ] ]
Chris@210: 
Chris@210:       CLOSING_PAREN.each { |k,v| k.freeze; v.freeze }  # debug, if I try to change it with <<
Chris@210:       OPENING_PAREN = CLOSING_PAREN.invert
Chris@210: 
Chris@210:       STRING_PATTERN = Hash.new do |h, k|
Chris@210:         delim, interpreted = *k
Chris@210:         delim_pattern = Regexp.escape(delim.dup)  # dup: workaround for old Ruby
Chris@210:         if closing_paren = CLOSING_PAREN[delim]
Chris@210:           delim_pattern = delim_pattern[0..-1] if defined? JRUBY_VERSION  # JRuby fix
Chris@210:           delim_pattern << Regexp.escape(closing_paren)
Chris@210:         end
Chris@210:         delim_pattern << '\\\\' unless delim == '\\'
Chris@210:         
Chris@210:         special_escapes =
Chris@210:           case interpreted
Chris@210:           when :regexp_symbols
Chris@210:             '| ' + REGEXP_SYMBOLS.source
Chris@210:           when :words
Chris@210:             '| \s'
Chris@210:           end
Chris@210:         
Chris@210:         h[k] =
Chris@210:           if interpreted and not delim == '#'
Chris@210:             / (?= [#{delim_pattern}] | \# [{$@] #{special_escapes} ) /mx
Chris@210:           else
Chris@210:             / (?= [#{delim_pattern}] #{special_escapes} ) /mx
Chris@210:           end
Chris@210:       end
Chris@210: 
Chris@210:       HEREDOC_PATTERN = Hash.new do |h, k|
Chris@210:         delim, interpreted, indented = *k
Chris@210:         delim_pattern = Regexp.escape(delim.dup)  # dup: workaround for old Ruby
Chris@210:         delim_pattern = / \n #{ '(?>[\ \t]*)' if indented } #{ Regexp.new delim_pattern } $ /x
Chris@210:         h[k] =
Chris@210:           if interpreted
Chris@210:             / (?= #{delim_pattern}() | \\ | \# [{$@] ) /mx  # $1 set == end of heredoc
Chris@210:           else
Chris@210:             / (?= #{delim_pattern}() | \\ ) /mx
Chris@210:           end
Chris@210:       end
Chris@210: 
Chris@210:       def initialize kind, interpreted, delim, heredoc = false
Chris@210:         if heredoc
Chris@210:           pattern = HEREDOC_PATTERN[ [delim, interpreted, heredoc == :indented] ]
Chris@210:           delim = nil
Chris@210:         else
Chris@210:           pattern = STRING_PATTERN[ [delim, interpreted] ]
Chris@210:           if paren = CLOSING_PAREN[delim]
Chris@210:             delim, paren = paren, delim
Chris@210:             paren_depth = 1
Chris@210:           end
Chris@210:         end
Chris@210:         super kind, interpreted, delim, heredoc, paren, paren_depth, pattern, :initial
Chris@210:       end
Chris@210:     end unless defined? StringState
Chris@210: 
Chris@210:   end
Chris@210: 
Chris@210: end
Chris@210: end