Chris@0: module CodeRay
Chris@0: module Scanners
Chris@0:   
Chris@0:   class Delphi < Scanner
Chris@0: 
Chris@0:     register_for :delphi
Chris@0:     file_extension 'pas'
Chris@0:     
Chris@0:     RESERVED_WORDS = [
Chris@0:       'and', 'array', 'as', 'at', 'asm', 'at', 'begin', 'case', 'class',
Chris@0:       'const', 'constructor', 'destructor', 'dispinterface', 'div', 'do',
Chris@0:       'downto', 'else', 'end', 'except', 'exports', 'file', 'finalization',
Chris@0:       'finally', 'for', 'function', 'goto', 'if', 'implementation', 'in',
Chris@0:       'inherited', 'initialization', 'inline', 'interface', 'is', 'label',
Chris@0:       'library', 'mod', 'nil', 'not', 'object', 'of', 'or', 'out', 'packed',
Chris@0:       'procedure', 'program', 'property', 'raise', 'record', 'repeat',
Chris@0:       'resourcestring', 'set', 'shl', 'shr', 'string', 'then', 'threadvar',
Chris@0:       'to', 'try', 'type', 'unit', 'until', 'uses', 'var', 'while', 'with',
Chris@0:       'xor', 'on'
Chris@0:     ]
Chris@0: 
Chris@0:     DIRECTIVES = [
Chris@0:       'absolute', 'abstract', 'assembler', 'at', 'automated', 'cdecl',
Chris@0:       'contains', 'deprecated', 'dispid', 'dynamic', 'export',
Chris@0:       'external', 'far', 'forward', 'implements', 'local', 
Chris@0:       'near', 'nodefault', 'on', 'overload', 'override',
Chris@0:       'package', 'pascal', 'platform', 'private', 'protected', 'public',
Chris@0:       'published', 'read', 'readonly', 'register', 'reintroduce',
Chris@0:       'requires', 'resident', 'safecall', 'stdcall', 'stored', 'varargs',
Chris@0:       'virtual', 'write', 'writeonly'
Chris@0:     ]
Chris@0: 
Chris@0:     IDENT_KIND = CaseIgnoringWordList.new(:ident).
Chris@0:       add(RESERVED_WORDS, :reserved).
Chris@0:       add(DIRECTIVES, :directive)
Chris@0:     
Chris@0:     NAME_FOLLOWS = CaseIgnoringWordList.new(false).
Chris@0:       add(%w(procedure function .))
Chris@0: 
Chris@0:   private
Chris@0:     def scan_tokens tokens, options
Chris@0: 
Chris@0:       state = :initial
Chris@0:       last_token = ''
Chris@0: 
Chris@0:       until eos?
Chris@0: 
Chris@0:         kind = nil
Chris@0:         match = nil
Chris@0: 
Chris@0:         if state == :initial
Chris@0:           
Chris@0:           if scan(/ \s+ /x)
Chris@0:             tokens << [matched, :space]
Chris@0:             next
Chris@0:             
Chris@0:           elsif scan(%r! \{ \$ [^}]* \}? | \(\* \$ (?: .*? \*\) | .* ) !mx)
Chris@0:             tokens << [matched, :preprocessor]
Chris@0:             next
Chris@0:             
Chris@0:           elsif scan(%r! // [^\n]* | \{ [^}]* \}? | \(\* (?: .*? \*\) | .* ) !mx)
Chris@0:             tokens << [matched, :comment]
Chris@0:             next
Chris@0:             
Chris@0:           elsif match = scan(/ <[>=]? | >=? | :=? | [-+=*\/;,@\^|\(\)\[\]] | \.\. /x)
Chris@0:             kind = :operator
Chris@0:           
Chris@0:           elsif match = scan(/\./)
Chris@0:             kind = :operator
Chris@0:             if last_token == 'end'
Chris@0:               tokens << [match, kind]
Chris@0:               next
Chris@0:             end
Chris@0:             
Chris@0:           elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
Chris@0:             kind = NAME_FOLLOWS[last_token] ? :ident : IDENT_KIND[match]
Chris@0:             
Chris@0:           elsif match = scan(/ ' ( [^\n']|'' ) (?:'|$) /x)
Chris@0:             tokens << [:open, :char]
Chris@0:             tokens << ["'", :delimiter]
Chris@0:             tokens << [self[1], :content]
Chris@0:             tokens << ["'", :delimiter]
Chris@0:             tokens << [:close, :char]
Chris@0:             next
Chris@0:             
Chris@0:           elsif match = scan(/ ' /x)
Chris@0:             tokens << [:open, :string]
Chris@0:             state = :string
Chris@0:             kind = :delimiter
Chris@0:             
Chris@0:           elsif scan(/ \# (?: \d+ | \$[0-9A-Fa-f]+ ) /x)
Chris@0:             kind = :char
Chris@0:             
Chris@0:           elsif scan(/ \$ [0-9A-Fa-f]+ /x)
Chris@0:             kind = :hex
Chris@0:             
Chris@0:           elsif scan(/ (?: \d+ ) (?![eE]|\.[^.]) /x)
Chris@0:             kind = :integer
Chris@0:             
Chris@0:           elsif scan(/ \d+ (?: \.\d+ (?: [eE][+-]? \d+ )? | [eE][+-]? \d+ ) /x)
Chris@0:             kind = :float
Chris@0: 
Chris@0:           else
Chris@0:             kind = :error
Chris@0:             getch
Chris@0: 
Chris@0:           end
Chris@0:           
Chris@0:         elsif state == :string
Chris@0:           if scan(/[^\n']+/)
Chris@0:             kind = :content
Chris@0:           elsif scan(/''/)
Chris@0:             kind = :char
Chris@0:           elsif scan(/'/)
Chris@0:             tokens << ["'", :delimiter]
Chris@0:             tokens << [:close, :string]
Chris@0:             state = :initial
Chris@0:             next
Chris@0:           elsif scan(/\n/)
Chris@0:             tokens << [:close, :string]
Chris@0:             kind = :error
Chris@0:             state = :initial
Chris@0:           else
Chris@0:             raise "else case \' reached; %p not handled." % peek(1), tokens
Chris@0:           end
Chris@0:           
Chris@0:         else
Chris@0:           raise 'else-case reached', 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, state
Chris@0:         end
Chris@0:         raise_inspect 'Empty token', tokens unless match
Chris@0: 
Chris@0:         last_token = match
Chris@0:         tokens << [match, kind]
Chris@0:         
Chris@0:       end
Chris@0:       
Chris@0:       tokens
Chris@0:     end
Chris@0: 
Chris@0:   end
Chris@0: 
Chris@0: end
Chris@0: end