Chris@909: module CodeRay
Chris@909: module Scanners
Chris@909:   
Chris@909:   # Scanner for the Delphi language (Object Pascal).
Chris@909:   # 
Chris@909:   # Alias: +pascal+
Chris@909:   class Delphi < Scanner
Chris@909:     
Chris@909:     register_for :delphi
Chris@909:     file_extension 'pas'
Chris@909:     
Chris@909:     KEYWORDS = [
Chris@909:       'and', 'array', 'as', 'at', 'asm', 'at', 'begin', 'case', 'class',
Chris@909:       'const', 'constructor', 'destructor', 'dispinterface', 'div', 'do',
Chris@909:       'downto', 'else', 'end', 'except', 'exports', 'file', 'finalization',
Chris@909:       'finally', 'for', 'function', 'goto', 'if', 'implementation', 'in',
Chris@909:       'inherited', 'initialization', 'inline', 'interface', 'is', 'label',
Chris@909:       'library', 'mod', 'nil', 'not', 'object', 'of', 'or', 'out', 'packed',
Chris@909:       'procedure', 'program', 'property', 'raise', 'record', 'repeat',
Chris@909:       'resourcestring', 'set', 'shl', 'shr', 'string', 'then', 'threadvar',
Chris@909:       'to', 'try', 'type', 'unit', 'until', 'uses', 'var', 'while', 'with',
Chris@909:       'xor', 'on',
Chris@909:     ]  # :nodoc:
Chris@909:     
Chris@909:     DIRECTIVES = [
Chris@909:       'absolute', 'abstract', 'assembler', 'at', 'automated', 'cdecl',
Chris@909:       'contains', 'deprecated', 'dispid', 'dynamic', 'export',
Chris@909:       'external', 'far', 'forward', 'implements', 'local', 
Chris@909:       'near', 'nodefault', 'on', 'overload', 'override',
Chris@909:       'package', 'pascal', 'platform', 'private', 'protected', 'public',
Chris@909:       'published', 'read', 'readonly', 'register', 'reintroduce',
Chris@909:       'requires', 'resident', 'safecall', 'stdcall', 'stored', 'varargs',
Chris@909:       'virtual', 'write', 'writeonly',
Chris@909:     ]  # :nodoc:
Chris@909:     
Chris@909:     IDENT_KIND = WordList::CaseIgnoring.new(:ident).
Chris@909:       add(KEYWORDS, :keyword).
Chris@909:       add(DIRECTIVES, :directive)  # :nodoc:
Chris@909:     
Chris@909:     NAME_FOLLOWS = WordList::CaseIgnoring.new(false).
Chris@909:       add(%w(procedure function .))  # :nodoc:
Chris@909:     
Chris@909:   protected
Chris@909:     
Chris@909:     def scan_tokens encoder, options
Chris@909:       
Chris@909:       state = :initial
Chris@909:       last_token = ''
Chris@909:       
Chris@909:       until eos?
Chris@909:         
Chris@909:         if state == :initial
Chris@909:           
Chris@909:           if match = scan(/ \s+ /x)
Chris@909:             encoder.text_token match, :space
Chris@909:             next
Chris@909:             
Chris@909:           elsif match = scan(%r! \{ \$ [^}]* \}? | \(\* \$ (?: .*? \*\) | .* ) !mx)
Chris@909:             encoder.text_token match, :preprocessor
Chris@909:             next
Chris@909:             
Chris@909:           elsif match = scan(%r! // [^\n]* | \{ [^}]* \}? | \(\* (?: .*? \*\) | .* ) !mx)
Chris@909:             encoder.text_token match, :comment
Chris@909:             next
Chris@909:             
Chris@909:           elsif match = scan(/ <[>=]? | >=? | :=? | [-+=*\/;,@\^|\(\)\[\]] | \.\. /x)
Chris@909:             encoder.text_token match, :operator
Chris@909:           
Chris@909:           elsif match = scan(/\./)
Chris@909:             encoder.text_token match, :operator
Chris@909:             next if last_token == 'end'
Chris@909:             
Chris@909:           elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
Chris@909:             encoder.text_token match, NAME_FOLLOWS[last_token] ? :ident : IDENT_KIND[match]
Chris@909:             
Chris@909:           elsif match = skip(/ ' ( [^\n']|'' ) (?:'|$) /x)
Chris@909:             encoder.begin_group :char
Chris@909:             encoder.text_token "'", :delimiter
Chris@909:             encoder.text_token self[1], :content
Chris@909:             encoder.text_token "'", :delimiter
Chris@909:             encoder.end_group :char
Chris@909:             next
Chris@909:             
Chris@909:           elsif match = scan(/ ' /x)
Chris@909:             encoder.begin_group :string
Chris@909:             encoder.text_token match, :delimiter
Chris@909:             state = :string
Chris@909:             
Chris@909:           elsif match = scan(/ \# (?: \d+ | \$[0-9A-Fa-f]+ ) /x)
Chris@909:             encoder.text_token match, :char
Chris@909:             
Chris@909:           elsif match = scan(/ \$ [0-9A-Fa-f]+ /x)
Chris@909:             encoder.text_token match, :hex
Chris@909:             
Chris@909:           elsif match = scan(/ (?: \d+ ) (?![eE]|\.[^.]) /x)
Chris@909:             encoder.text_token match, :integer
Chris@909:             
Chris@909:           elsif match = scan(/ \d+ (?: \.\d+ (?: [eE][+-]? \d+ )? | [eE][+-]? \d+ ) /x)
Chris@909:             encoder.text_token match, :float
Chris@909:             
Chris@909:           else
Chris@909:             encoder.text_token getch, :error
Chris@909:             next
Chris@909:             
Chris@909:           end
Chris@909:           
Chris@909:         elsif state == :string
Chris@909:           if match = scan(/[^\n']+/)
Chris@909:             encoder.text_token match, :content
Chris@909:           elsif match = scan(/''/)
Chris@909:             encoder.text_token match, :char
Chris@909:           elsif match = scan(/'/)
Chris@909:             encoder.text_token match, :delimiter
Chris@909:             encoder.end_group :string
Chris@909:             state = :initial
Chris@909:             next
Chris@909:           elsif match = scan(/\n/)
Chris@909:             encoder.end_group :string
Chris@909:             encoder.text_token match, :space
Chris@909:             state = :initial
Chris@909:           else
Chris@909:             raise "else case \' reached; %p not handled." % peek(1), encoder
Chris@909:           end
Chris@909:           
Chris@909:         else
Chris@909:           raise 'else-case reached', encoder
Chris@909:           
Chris@909:         end
Chris@909:         
Chris@909:         last_token = match
Chris@909:         
Chris@909:       end
Chris@909:       
Chris@909:       if state == :string
Chris@909:         encoder.end_group state
Chris@909:       end
Chris@909:       
Chris@909:       encoder
Chris@909:     end
Chris@909: 
Chris@909:   end
Chris@909: 
Chris@909: end
Chris@909: end