Chris@0: module CodeRay
Chris@0: module Scanners
Chris@0: 
Chris@0:   class C < Scanner
Chris@0: 
Chris@0:     include Streamable
Chris@0:     
Chris@0:     register_for :c
Chris@0:     file_extension 'c'
Chris@0: 
Chris@0:     RESERVED_WORDS = [
Chris@0:       'asm', 'break', 'case', 'continue', 'default', 'do',
Chris@0:       'else', 'enum', 'for', 'goto', 'if', 'return',
Chris@0:       'sizeof', 'struct', 'switch', 'typedef', 'union', 'while',
Chris@0:       'restrict',  # added in C99
Chris@0:     ]
Chris@0: 
Chris@0:     PREDEFINED_TYPES = [
Chris@0:       'int', 'long', 'short', 'char',
Chris@0:       'signed', 'unsigned', 'float', 'double',
Chris@0:       'bool', 'complex',  # added in C99
Chris@0:     ]
Chris@0: 
Chris@0:     PREDEFINED_CONSTANTS = [
Chris@0:       'EOF', 'NULL',
Chris@0:       'true', 'false',  # added in C99
Chris@0:     ]
Chris@0:     DIRECTIVES = [
Chris@0:       'auto', 'extern', 'register', 'static', 'void',
Chris@0:       'const', 'volatile',  # added in C89
Chris@0:       'inline',  # added in C99
Chris@0:     ]
Chris@0: 
Chris@0:     IDENT_KIND = WordList.new(:ident).
Chris@0:       add(RESERVED_WORDS, :reserved).
Chris@0:       add(PREDEFINED_TYPES, :pre_type).
Chris@0:       add(DIRECTIVES, :directive).
Chris@0:       add(PREDEFINED_CONSTANTS, :pre_constant)
Chris@0: 
Chris@0:     ESCAPE = / [rbfntv\n\\'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x
Chris@0:     UNICODE_ESCAPE =  / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x
Chris@0: 
Chris@0:     def scan_tokens tokens, options
Chris@0: 
Chris@0:       state = :initial
Chris@0:       label_expected = true
Chris@0:       case_expected = false
Chris@0:       label_expected_before_preproc_line = nil
Chris@0:       in_preproc_line = 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: 
Chris@0:           if match = scan(/ \s+ | \\\n /x)
Chris@0:             if in_preproc_line && match != "\\\n" && match.index(?\n)
Chris@0:               in_preproc_line = false
Chris@0:               label_expected = label_expected_before_preproc_line
Chris@0:             end
Chris@0:             tokens << [match, :space]
Chris@0:             next
Chris@0: 
Chris@0:           elsif scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx)
Chris@0:             kind = :comment
Chris@0: 
Chris@0:           elsif match = scan(/ \# \s* if \s* 0 /x)
Chris@0:             match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos?
Chris@0:             kind = :comment
Chris@0: 
Chris@0:           elsif match = scan(/ [-+*=<>?:;,!&^|()\[\]{}~%]+ | \/=? | \.(?!\d) /x)
Chris@0:             label_expected = match =~ /[;\{\}]/
Chris@0:             if case_expected
Chris@0:               label_expected = true if match == ':'
Chris@0:               case_expected = false
Chris@0:             end
Chris@0:             kind = :operator
Chris@0: 
Chris@0:           elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
Chris@0:             kind = IDENT_KIND[match]
Chris@0:             if kind == :ident && label_expected && !in_preproc_line && scan(/:(?!:)/)
Chris@0:               kind = :label
Chris@0:               match << matched
Chris@0:             else
Chris@0:               label_expected = false
Chris@0:               if kind == :reserved
Chris@0:                 case match
Chris@0:                 when 'case', 'default'
Chris@0:                   case_expected = true
Chris@0:                 end
Chris@0:               end
Chris@0:             end
Chris@0: 
Chris@0:           elsif scan(/\$/)
Chris@0:             kind = :ident
Chris@0:           
Chris@0:           elsif match = scan(/L?"/)
Chris@0:             tokens << [:open, :string]
Chris@0:             if match[0] == ?L
Chris@0:               tokens << ['L', :modifier]
Chris@0:               match = '"'
Chris@0:             end
Chris@0:             state = :string
Chris@0:             kind = :delimiter
Chris@0: 
Chris@0:           elsif scan(/#[ \t]*(\w*)/)
Chris@0:             kind = :preprocessor
Chris@0:             in_preproc_line = true
Chris@0:             label_expected_before_preproc_line = label_expected
Chris@0:             state = :include_expected if self[1] == 'include'
Chris@0: 
Chris@0:           elsif scan(/ L?' (?: [^\'\n\\] | \\ #{ESCAPE} )? '? /ox)
Chris@0:             label_expected = false
Chris@0:             kind = :char
Chris@0: 
Chris@0:           elsif scan(/0[xX][0-9A-Fa-f]+/)
Chris@0:             label_expected = false
Chris@0:             kind = :hex
Chris@0: 
Chris@0:           elsif scan(/(?:0[0-7]+)(?![89.eEfF])/)
Chris@0:             label_expected = false
Chris@0:             kind = :oct
Chris@0: 
Chris@0:           elsif scan(/(?:\d+)(?![.eEfF])L?L?/)
Chris@0:             label_expected = false
Chris@0:             kind = :integer
Chris@0: 
Chris@0:           elsif scan(/\d[fF]?|\d*\.\d+(?:[eE][+-]?\d+)?[fF]?|\d+[eE][+-]?\d+[fF]?/)
Chris@0:             label_expected = false
Chris@0:             kind = :float
Chris@0: 
Chris@0:           else
Chris@0:             getch
Chris@0:             kind = :error
Chris@0: 
Chris@0:           end
Chris@0: 
Chris@0:         when :string
Chris@0:           if scan(/[^\\\n"]+/)
Chris@0:             kind = :content
Chris@0:           elsif scan(/"/)
Chris@0:             tokens << ['"', :delimiter]
Chris@0:             tokens << [:close, :string]
Chris@0:             state = :initial
Chris@0:             label_expected = false
Chris@0:             next
Chris@0:           elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox)
Chris@0:             kind = :char
Chris@0:           elsif scan(/ \\ | $ /x)
Chris@0:             tokens << [:close, :string]
Chris@0:             kind = :error
Chris@0:             state = :initial
Chris@0:             label_expected = false
Chris@0:           else
Chris@0:             raise_inspect "else case \" reached; %p not handled." % peek(1), tokens
Chris@0:           end
Chris@0: 
Chris@0:         when :include_expected
Chris@0:           if scan(/<[^>\n]+>?|"[^"\n\\]*(?:\\.[^"\n\\]*)*"?/)
Chris@0:             kind = :include
Chris@0:             state = :initial
Chris@0: 
Chris@0:           elsif match = scan(/\s+/)
Chris@0:             kind = :space
Chris@0:             state = :initial if match.index ?\n
Chris@0: 
Chris@0:           else
Chris@0:             state = :initial
Chris@0:             next
Chris@0: 
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 state == :string
Chris@0:         tokens << [:close, :string]
Chris@0:       end
Chris@0: 
Chris@0:       tokens
Chris@0:     end
Chris@0: 
Chris@0:   end
Chris@0: 
Chris@0: end
Chris@0: end