Chris@909: module CodeRay
Chris@909: module Scanners
Chris@909: 
Chris@909:   # Scanner for C++.
Chris@909:   # 
Chris@909:   # Aliases: +cplusplus+, c++
Chris@909:   class CPlusPlus < Scanner
Chris@909: 
Chris@909:     register_for :cpp
Chris@909:     file_extension 'cpp'
Chris@909:     title 'C++'
Chris@909:     
Chris@909:     #-- http://www.cppreference.com/wiki/keywords/start
Chris@909:     KEYWORDS = [
Chris@909:       'and', 'and_eq', 'asm', 'bitand', 'bitor', 'break',
Chris@909:       'case', 'catch', 'class', 'compl', 'const_cast',
Chris@909:       'continue', 'default', 'delete', 'do', 'dynamic_cast', 'else',
Chris@909:       'enum', 'export', 'for', 'goto', 'if', 'namespace', 'new',
Chris@909:       'not', 'not_eq', 'or', 'or_eq', 'reinterpret_cast', 'return',
Chris@909:       'sizeof', 'static_cast', 'struct', 'switch', 'template',
Chris@909:       'throw', 'try', 'typedef', 'typeid', 'typename', 'union',
Chris@909:       'while', 'xor', 'xor_eq',
Chris@909:     ]  # :nodoc:
Chris@909:     
Chris@909:     PREDEFINED_TYPES = [
Chris@909:       'bool', 'char', 'double', 'float', 'int', 'long',
Chris@909:       'short', 'signed', 'unsigned', 'wchar_t', 'string',
Chris@909:     ]  # :nodoc:
Chris@909:     PREDEFINED_CONSTANTS = [
Chris@909:       'false', 'true',
Chris@909:       'EOF', 'NULL',
Chris@909:     ]  # :nodoc:
Chris@909:     PREDEFINED_VARIABLES = [
Chris@909:       'this',
Chris@909:     ]  # :nodoc:
Chris@909:     DIRECTIVES = [
Chris@909:       'auto', 'const', 'explicit', 'extern', 'friend', 'inline', 'mutable', 'operator',
Chris@909:       'private', 'protected', 'public', 'register', 'static', 'using', 'virtual', 'void',
Chris@909:       'volatile',
Chris@909:     ]  # :nodoc:
Chris@909:     
Chris@909:     IDENT_KIND = WordList.new(:ident).
Chris@909:       add(KEYWORDS, :keyword).
Chris@909:       add(PREDEFINED_TYPES, :predefined_type).
Chris@909:       add(PREDEFINED_VARIABLES, :local_variable).
Chris@909:       add(DIRECTIVES, :directive).
Chris@909:       add(PREDEFINED_CONSTANTS, :predefined_constant)  # :nodoc:
Chris@909: 
Chris@909:     ESCAPE = / [rbfntv\n\\'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x  # :nodoc:
Chris@909:     UNICODE_ESCAPE =  / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x  # :nodoc:
Chris@909:     
Chris@909:   protected
Chris@909:     
Chris@909:     def scan_tokens encoder, options
Chris@909: 
Chris@909:       state = :initial
Chris@909:       label_expected = true
Chris@909:       case_expected = false
Chris@909:       label_expected_before_preproc_line = nil
Chris@909:       in_preproc_line = false
Chris@909: 
Chris@909:       until eos?
Chris@909: 
Chris@909:         case state
Chris@909: 
Chris@909:         when :initial
Chris@909: 
Chris@909:           if match = scan(/ \s+ | \\\n /x)
Chris@909:             if in_preproc_line && match != "\\\n" && match.index(?\n)
Chris@909:               in_preproc_line = false
Chris@909:               label_expected = label_expected_before_preproc_line
Chris@909:             end
Chris@909:             encoder.text_token match, :space
Chris@909: 
Chris@909:           elsif match = scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx)
Chris@909:             encoder.text_token match, :comment
Chris@909: 
Chris@909:           elsif match = scan(/ \# \s* if \s* 0 /x)
Chris@909:             match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos?
Chris@909:             encoder.text_token match, :comment
Chris@909: 
Chris@909:           elsif match = scan(/ [-+*=<>?:;,!&^|()\[\]{}~%]+ | \/=? | \.(?!\d) /x)
Chris@909:             label_expected = match =~ /[;\{\}]/
Chris@909:             if case_expected
Chris@909:               label_expected = true if match == ':'
Chris@909:               case_expected = false
Chris@909:             end
Chris@909:             encoder.text_token match, :operator
Chris@909: 
Chris@909:           elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
Chris@909:             kind = IDENT_KIND[match]
Chris@909:             if kind == :ident && label_expected && !in_preproc_line && scan(/:(?!:)/)
Chris@909:               kind = :label
Chris@909:               match << matched
Chris@909:             else
Chris@909:               label_expected = false
Chris@909:               if kind == :keyword
Chris@909:                 case match
Chris@909:                 when 'class'
Chris@909:                   state = :class_name_expected
Chris@909:                 when 'case', 'default'
Chris@909:                   case_expected = true
Chris@909:                 end
Chris@909:               end
Chris@909:             end
Chris@909:             encoder.text_token match, kind
Chris@909: 
Chris@909:           elsif match = scan(/\$/)
Chris@909:             encoder.text_token match, :ident
Chris@909:           
Chris@909:           elsif match = scan(/L?"/)
Chris@909:             encoder.begin_group :string
Chris@909:             if match[0] == ?L
Chris@909:               encoder.text_token match, 'L', :modifier
Chris@909:               match = '"'
Chris@909:             end
Chris@909:             state = :string
Chris@909:             encoder.text_token match, :delimiter
Chris@909: 
Chris@909:           elsif match = scan(/#[ \t]*(\w*)/)
Chris@909:             encoder.text_token match, :preprocessor
Chris@909:             in_preproc_line = true
Chris@909:             label_expected_before_preproc_line = label_expected
Chris@909:             state = :include_expected if self[1] == 'include'
Chris@909: 
Chris@909:           elsif match = scan(/ L?' (?: [^\'\n\\] | \\ #{ESCAPE} )? '? /ox)
Chris@909:             label_expected = false
Chris@909:             encoder.text_token match, :char
Chris@909: 
Chris@909:           elsif match = scan(/0[xX][0-9A-Fa-f]+/)
Chris@909:             label_expected = false
Chris@909:             encoder.text_token match, :hex
Chris@909: 
Chris@909:           elsif match = scan(/(?:0[0-7]+)(?![89.eEfF])/)
Chris@909:             label_expected = false
Chris@909:             encoder.text_token match, :octal
Chris@909: 
Chris@909:           elsif match = scan(/(?:\d+)(?![.eEfF])L?L?/)
Chris@909:             label_expected = false
Chris@909:             encoder.text_token match, :integer
Chris@909: 
Chris@909:           elsif match = scan(/\d[fF]?|\d*\.\d+(?:[eE][+-]?\d+)?[fF]?|\d+[eE][+-]?\d+[fF]?/)
Chris@909:             label_expected = false
Chris@909:             encoder.text_token match, :float
Chris@909: 
Chris@909:           else
Chris@909:             encoder.text_token getch, :error
Chris@909: 
Chris@909:           end
Chris@909: 
Chris@909:         when :string
Chris@909:           if match = scan(/[^\\"]+/)
Chris@909:             encoder.text_token match, :content
Chris@909:           elsif match = scan(/"/)
Chris@909:             encoder.text_token match, :delimiter
Chris@909:             encoder.end_group :string
Chris@909:             state = :initial
Chris@909:             label_expected = false
Chris@909:           elsif match = scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox)
Chris@909:             encoder.text_token match, :char
Chris@909:           elsif match = scan(/ \\ | $ /x)
Chris@909:             encoder.end_group :string
Chris@909:             encoder.text_token match, :error
Chris@909:             state = :initial
Chris@909:             label_expected = false
Chris@909:           else
Chris@909:             raise_inspect "else case \" reached; %p not handled." % peek(1), encoder
Chris@909:           end
Chris@909: 
Chris@909:         when :include_expected
Chris@909:           if match = scan(/<[^>\n]+>?|"[^"\n\\]*(?:\\.[^"\n\\]*)*"?/)
Chris@909:             encoder.text_token match, :include
Chris@909:             state = :initial
Chris@909: 
Chris@909:           elsif match = scan(/\s+/)
Chris@909:             encoder.text_token match, :space
Chris@909:             state = :initial if match.index ?\n
Chris@909: 
Chris@909:           else
Chris@909:             state = :initial
Chris@909: 
Chris@909:           end
Chris@909:         
Chris@909:         when :class_name_expected
Chris@909:           if match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
Chris@909:             encoder.text_token match, :class
Chris@909:             state = :initial
Chris@909: 
Chris@909:           elsif match = scan(/\s+/)
Chris@909:             encoder.text_token match, :space
Chris@909: 
Chris@909:           else
Chris@909:             encoder.text_token getch, :error
Chris@909:             state = :initial
Chris@909: 
Chris@909:           end
Chris@909:           
Chris@909:         else
Chris@909:           raise_inspect 'Unknown state', encoder
Chris@909: 
Chris@909:         end
Chris@909: 
Chris@909:       end
Chris@909: 
Chris@909:       if state == :string
Chris@909:         encoder.end_group :string
Chris@909:       end
Chris@909: 
Chris@909:       encoder
Chris@909:     end
Chris@909: 
Chris@909:   end
Chris@909: 
Chris@909: end
Chris@909: end