Chris@909: # encoding: utf-8
Chris@909: module CodeRay
Chris@909: module Scanners
Chris@909:   
Chris@909:   class Ruby
Chris@909:     
Chris@909:     class StringState < Struct.new :type, :interpreted, :delim, :heredoc,
Chris@909:       :opening_paren, :paren_depth, :pattern, :next_state  # :nodoc: all
Chris@909:       
Chris@909:       CLOSING_PAREN = Hash[ *%w[
Chris@909:         ( )
Chris@909:         [ ]
Chris@909:         < >
Chris@909:         { }
Chris@909:       ] ].each { |k,v| k.freeze; v.freeze }  # debug, if I try to change it with <<
Chris@909:       
Chris@909:       STRING_PATTERN = Hash.new do |h, k|
Chris@909:         delim, interpreted = *k
Chris@909:         # delim = delim.dup  # workaround for old Ruby
Chris@909:         delim_pattern = Regexp.escape(delim)
Chris@909:         if closing_paren = CLOSING_PAREN[delim]
Chris@909:           delim_pattern << Regexp.escape(closing_paren)
Chris@909:         end
Chris@909:         delim_pattern << '\\\\' unless delim == '\\'
Chris@909:         
Chris@909:         # special_escapes =
Chris@909:         #   case interpreted
Chris@909:         #   when :regexp_symbols
Chris@909:         #     '| [|?*+(){}\[\].^$]'
Chris@909:         #   end
Chris@909:         
Chris@909:         h[k] =
Chris@909:           if interpreted && delim != '#'
Chris@909:             / (?= [#{delim_pattern}] | \# [{$@] ) /mx
Chris@909:           else
Chris@909:             / (?= [#{delim_pattern}] ) /mx
Chris@909:           end
Chris@909:       end
Chris@909:       
Chris@909:       def initialize kind, interpreted, delim, heredoc = false
Chris@909:         if heredoc
Chris@909:           pattern = heredoc_pattern delim, interpreted, heredoc == :indented
Chris@909:           delim = nil
Chris@909:         else
Chris@909:           pattern = STRING_PATTERN[ [delim, interpreted] ]
Chris@909:           if closing_paren = CLOSING_PAREN[delim]
Chris@909:             opening_paren = delim
Chris@909:             delim = closing_paren
Chris@909:             paren_depth = 1
Chris@909:           end
Chris@909:         end
Chris@909:         super kind, interpreted, delim, heredoc, opening_paren, paren_depth, pattern, :initial
Chris@909:       end
Chris@909:       
Chris@909:       def heredoc_pattern delim, interpreted, indented
Chris@909:         # delim = delim.dup  # workaround for old Ruby
Chris@909:         delim_pattern = Regexp.escape(delim)
Chris@909:         delim_pattern = / (?:\A|\n) #{ '(?>[ \t]*)' if indented } #{ Regexp.new delim_pattern } $ /x
Chris@909:         if interpreted
Chris@909:           / (?= #{delim_pattern}() | \\ | \# [{$@] ) /mx  # $1 set == end of heredoc
Chris@909:         else
Chris@909:           / (?= #{delim_pattern}() | \\ ) /mx
Chris@909:         end
Chris@909:       end
Chris@909:       
Chris@909:     end
Chris@909:     
Chris@909:   end
Chris@909:   
Chris@909: end
Chris@909: end