Chris@0: module CodeRay
Chris@0: module Scanners
Chris@0:   
Chris@0:   class JSON < Scanner
Chris@0:     
Chris@0:     include Streamable
Chris@0:     
Chris@0:     register_for :json
Chris@0:     file_extension 'json'
Chris@0:     
Chris@0:     KINDS_NOT_LOC = [
Chris@0:       :float, :char, :content, :delimiter,
Chris@0:       :error, :integer, :operator, :value,
Chris@0:     ]
Chris@0:     
Chris@0:     CONSTANTS = %w( true false null )
Chris@0:     IDENT_KIND = WordList.new(:key).add(CONSTANTS, :value)
Chris@0:     
Chris@0:     ESCAPE = / [bfnrt\\"\/] /x
Chris@0:     UNICODE_ESCAPE =  / u[a-fA-F0-9]{4} /x
Chris@0:     
Chris@0:     def scan_tokens tokens, options
Chris@0:       
Chris@0:       state = :initial
Chris@0:       stack = []
Chris@0:       string_delimiter = nil
Chris@0:       key_expected = false
Chris@0:       
Chris@0:       until eos?
Chris@0:         
Chris@0:         kind = nil
Chris@0:         match = nil
Chris@0:         
Chris@0:         case state
Chris@0:         
Chris@0:         when :initial
Chris@0:           if match = scan(/ \s+ | \\\n /x)
Chris@0:             tokens << [match, :space]
Chris@0:             next
Chris@0:           elsif match = scan(/ [:,\[{\]}] /x)
Chris@0:             kind = :operator
Chris@0:             case match
Chris@0:             when '{' then stack << :object; key_expected = true
Chris@0:             when '[' then stack << :array
Chris@0:             when ':' then key_expected = false
Chris@0:             when ',' then key_expected = true if stack.last == :object
Chris@0:             when '}', ']' then stack.pop  # no error recovery, but works for valid JSON
Chris@0:             end
Chris@0:           elsif match = scan(/ true | false | null /x)
Chris@0:             kind = IDENT_KIND[match]
Chris@0:           elsif match = scan(/-?(?:0|[1-9]\d*)/)
Chris@0:             kind = :integer
Chris@0:             if scan(/\.\d+(?:[eE][-+]?\d+)?|[eE][-+]?\d+/)
Chris@0:               match << matched
Chris@0:               kind = :float
Chris@0:             end
Chris@0:           elsif match = scan(/"/)
Chris@0:             state = key_expected ? :key : :string
Chris@0:             tokens << [:open, state]
Chris@0:             kind = :delimiter
Chris@0:           else
Chris@0:             getch
Chris@0:             kind = :error
Chris@0:           end
Chris@0:           
Chris@0:         when :string, :key
Chris@0:           if scan(/[^\\"]+/)
Chris@0:             kind = :content
Chris@0:           elsif scan(/"/)
Chris@0:             tokens << ['"', :delimiter]
Chris@0:             tokens << [:close, state]
Chris@0:             state = :initial
Chris@0:             next
Chris@0:           elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox)
Chris@0:             kind = :char
Chris@0:           elsif scan(/\\./m)
Chris@0:             kind = :content
Chris@0:           elsif scan(/ \\ | $ /x)
Chris@0:             tokens << [:close, :delimiter]
Chris@0:             kind = :error
Chris@0:             state = :initial
Chris@0:           else
Chris@0:             raise_inspect "else case \" reached; %p not handled." % peek(1), tokens
Chris@0:           end
Chris@0:           
Chris@0:         else
Chris@0:           raise_inspect 'Unknown state', tokens
Chris@0:           
Chris@0:         end
Chris@0:         
Chris@0:         match ||= matched
Chris@0:         if $CODERAY_DEBUG and not kind
Chris@0:           raise_inspect 'Error token %p in line %d' %
Chris@0:             [[match, kind], line], tokens
Chris@0:         end
Chris@0:         raise_inspect 'Empty token', tokens unless match
Chris@0:         
Chris@0:         tokens << [match, kind]
Chris@0:         
Chris@0:       end
Chris@0:       
Chris@0:       if [:string, :key].include? state
Chris@0:         tokens << [:close, state]
Chris@0:       end
Chris@0:       
Chris@0:       tokens
Chris@0:     end
Chris@0:     
Chris@0:   end
Chris@0:   
Chris@0: end
Chris@0: end