Chris@909
|
1 module CodeRay
|
Chris@909
|
2 module Scanners
|
Chris@909
|
3
|
Chris@909
|
4 # Scanner for the Delphi language (Object Pascal).
|
Chris@909
|
5 #
|
Chris@909
|
6 # Alias: +pascal+
|
Chris@909
|
7 class Delphi < Scanner
|
Chris@909
|
8
|
Chris@909
|
9 register_for :delphi
|
Chris@909
|
10 file_extension 'pas'
|
Chris@909
|
11
|
Chris@909
|
12 KEYWORDS = [
|
Chris@909
|
13 'and', 'array', 'as', 'at', 'asm', 'at', 'begin', 'case', 'class',
|
Chris@909
|
14 'const', 'constructor', 'destructor', 'dispinterface', 'div', 'do',
|
Chris@909
|
15 'downto', 'else', 'end', 'except', 'exports', 'file', 'finalization',
|
Chris@909
|
16 'finally', 'for', 'function', 'goto', 'if', 'implementation', 'in',
|
Chris@909
|
17 'inherited', 'initialization', 'inline', 'interface', 'is', 'label',
|
Chris@909
|
18 'library', 'mod', 'nil', 'not', 'object', 'of', 'or', 'out', 'packed',
|
Chris@909
|
19 'procedure', 'program', 'property', 'raise', 'record', 'repeat',
|
Chris@909
|
20 'resourcestring', 'set', 'shl', 'shr', 'string', 'then', 'threadvar',
|
Chris@909
|
21 'to', 'try', 'type', 'unit', 'until', 'uses', 'var', 'while', 'with',
|
Chris@909
|
22 'xor', 'on',
|
Chris@909
|
23 ] # :nodoc:
|
Chris@909
|
24
|
Chris@909
|
25 DIRECTIVES = [
|
Chris@909
|
26 'absolute', 'abstract', 'assembler', 'at', 'automated', 'cdecl',
|
Chris@909
|
27 'contains', 'deprecated', 'dispid', 'dynamic', 'export',
|
Chris@909
|
28 'external', 'far', 'forward', 'implements', 'local',
|
Chris@909
|
29 'near', 'nodefault', 'on', 'overload', 'override',
|
Chris@909
|
30 'package', 'pascal', 'platform', 'private', 'protected', 'public',
|
Chris@909
|
31 'published', 'read', 'readonly', 'register', 'reintroduce',
|
Chris@909
|
32 'requires', 'resident', 'safecall', 'stdcall', 'stored', 'varargs',
|
Chris@909
|
33 'virtual', 'write', 'writeonly',
|
Chris@909
|
34 ] # :nodoc:
|
Chris@909
|
35
|
Chris@909
|
36 IDENT_KIND = WordList::CaseIgnoring.new(:ident).
|
Chris@909
|
37 add(KEYWORDS, :keyword).
|
Chris@909
|
38 add(DIRECTIVES, :directive) # :nodoc:
|
Chris@909
|
39
|
Chris@909
|
40 NAME_FOLLOWS = WordList::CaseIgnoring.new(false).
|
Chris@909
|
41 add(%w(procedure function .)) # :nodoc:
|
Chris@909
|
42
|
Chris@909
|
43 protected
|
Chris@909
|
44
|
Chris@909
|
45 def scan_tokens encoder, options
|
Chris@909
|
46
|
Chris@909
|
47 state = :initial
|
Chris@909
|
48 last_token = ''
|
Chris@909
|
49
|
Chris@909
|
50 until eos?
|
Chris@909
|
51
|
Chris@909
|
52 if state == :initial
|
Chris@909
|
53
|
Chris@909
|
54 if match = scan(/ \s+ /x)
|
Chris@909
|
55 encoder.text_token match, :space
|
Chris@909
|
56 next
|
Chris@909
|
57
|
Chris@909
|
58 elsif match = scan(%r! \{ \$ [^}]* \}? | \(\* \$ (?: .*? \*\) | .* ) !mx)
|
Chris@909
|
59 encoder.text_token match, :preprocessor
|
Chris@909
|
60 next
|
Chris@909
|
61
|
Chris@909
|
62 elsif match = scan(%r! // [^\n]* | \{ [^}]* \}? | \(\* (?: .*? \*\) | .* ) !mx)
|
Chris@909
|
63 encoder.text_token match, :comment
|
Chris@909
|
64 next
|
Chris@909
|
65
|
Chris@909
|
66 elsif match = scan(/ <[>=]? | >=? | :=? | [-+=*\/;,@\^|\(\)\[\]] | \.\. /x)
|
Chris@909
|
67 encoder.text_token match, :operator
|
Chris@909
|
68
|
Chris@909
|
69 elsif match = scan(/\./)
|
Chris@909
|
70 encoder.text_token match, :operator
|
Chris@909
|
71 next if last_token == 'end'
|
Chris@909
|
72
|
Chris@909
|
73 elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x)
|
Chris@909
|
74 encoder.text_token match, NAME_FOLLOWS[last_token] ? :ident : IDENT_KIND[match]
|
Chris@909
|
75
|
Chris@909
|
76 elsif match = skip(/ ' ( [^\n']|'' ) (?:'|$) /x)
|
Chris@909
|
77 encoder.begin_group :char
|
Chris@909
|
78 encoder.text_token "'", :delimiter
|
Chris@909
|
79 encoder.text_token self[1], :content
|
Chris@909
|
80 encoder.text_token "'", :delimiter
|
Chris@909
|
81 encoder.end_group :char
|
Chris@909
|
82 next
|
Chris@909
|
83
|
Chris@909
|
84 elsif match = scan(/ ' /x)
|
Chris@909
|
85 encoder.begin_group :string
|
Chris@909
|
86 encoder.text_token match, :delimiter
|
Chris@909
|
87 state = :string
|
Chris@909
|
88
|
Chris@909
|
89 elsif match = scan(/ \# (?: \d+ | \$[0-9A-Fa-f]+ ) /x)
|
Chris@909
|
90 encoder.text_token match, :char
|
Chris@909
|
91
|
Chris@909
|
92 elsif match = scan(/ \$ [0-9A-Fa-f]+ /x)
|
Chris@909
|
93 encoder.text_token match, :hex
|
Chris@909
|
94
|
Chris@909
|
95 elsif match = scan(/ (?: \d+ ) (?![eE]|\.[^.]) /x)
|
Chris@909
|
96 encoder.text_token match, :integer
|
Chris@909
|
97
|
Chris@909
|
98 elsif match = scan(/ \d+ (?: \.\d+ (?: [eE][+-]? \d+ )? | [eE][+-]? \d+ ) /x)
|
Chris@909
|
99 encoder.text_token match, :float
|
Chris@909
|
100
|
Chris@909
|
101 else
|
Chris@909
|
102 encoder.text_token getch, :error
|
Chris@909
|
103 next
|
Chris@909
|
104
|
Chris@909
|
105 end
|
Chris@909
|
106
|
Chris@909
|
107 elsif state == :string
|
Chris@909
|
108 if match = scan(/[^\n']+/)
|
Chris@909
|
109 encoder.text_token match, :content
|
Chris@909
|
110 elsif match = scan(/''/)
|
Chris@909
|
111 encoder.text_token match, :char
|
Chris@909
|
112 elsif match = scan(/'/)
|
Chris@909
|
113 encoder.text_token match, :delimiter
|
Chris@909
|
114 encoder.end_group :string
|
Chris@909
|
115 state = :initial
|
Chris@909
|
116 next
|
Chris@909
|
117 elsif match = scan(/\n/)
|
Chris@909
|
118 encoder.end_group :string
|
Chris@909
|
119 encoder.text_token match, :space
|
Chris@909
|
120 state = :initial
|
Chris@909
|
121 else
|
Chris@909
|
122 raise "else case \' reached; %p not handled." % peek(1), encoder
|
Chris@909
|
123 end
|
Chris@909
|
124
|
Chris@909
|
125 else
|
Chris@909
|
126 raise 'else-case reached', encoder
|
Chris@909
|
127
|
Chris@909
|
128 end
|
Chris@909
|
129
|
Chris@909
|
130 last_token = match
|
Chris@909
|
131
|
Chris@909
|
132 end
|
Chris@909
|
133
|
Chris@909
|
134 if state == :string
|
Chris@909
|
135 encoder.end_group state
|
Chris@909
|
136 end
|
Chris@909
|
137
|
Chris@909
|
138 encoder
|
Chris@909
|
139 end
|
Chris@909
|
140
|
Chris@909
|
141 end
|
Chris@909
|
142
|
Chris@909
|
143 end
|
Chris@909
|
144 end
|