@@ -1,165 +1,165 | |||
|
1 | 1 | module CodeRay |
|
2 | 2 | module Scanners |
|
3 | 3 | |
|
4 | 4 | class C < Scanner |
|
5 | 5 | |
|
6 | 6 | register_for :c |
|
7 | 7 | |
|
8 | 8 | include Streamable |
|
9 | 9 | |
|
10 | 10 | RESERVED_WORDS = [ |
|
11 | 11 | 'asm', 'break', 'case', 'continue', 'default', 'do', 'else', |
|
12 | 12 | 'for', 'goto', 'if', 'return', 'switch', 'while', |
|
13 | 13 | 'struct', 'union', 'enum', 'typedef', |
|
14 | 14 | 'static', 'register', 'auto', 'extern', |
|
15 | 15 | 'sizeof', |
|
16 | 16 | 'volatile', 'const', # C89 |
|
17 | 17 | 'inline', 'restrict', # C99 |
|
18 | 18 | ] |
|
19 | 19 | |
|
20 | 20 | PREDEFINED_TYPES = [ |
|
21 | 21 | 'int', 'long', 'short', 'char', 'void', |
|
22 | 22 | 'signed', 'unsigned', 'float', 'double', |
|
23 | 23 | 'bool', 'complex', # C99 |
|
24 | 24 | ] |
|
25 | 25 | |
|
26 | 26 | PREDEFINED_CONSTANTS = [ |
|
27 | 27 | 'EOF', 'NULL', |
|
28 | 28 | 'true', 'false', # C99 |
|
29 | 29 | ] |
|
30 | 30 | |
|
31 | 31 | IDENT_KIND = WordList.new(:ident). |
|
32 | 32 | add(RESERVED_WORDS, :reserved). |
|
33 | 33 | add(PREDEFINED_TYPES, :pre_type). |
|
34 | 34 | add(PREDEFINED_CONSTANTS, :pre_constant) |
|
35 | 35 | |
|
36 | 36 | ESCAPE = / [rbfnrtv\n\\'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x |
|
37 | 37 | UNICODE_ESCAPE = / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x |
|
38 | 38 | |
|
39 | 39 | def scan_tokens tokens, options |
|
40 | 40 | |
|
41 | 41 | state = :initial |
|
42 | 42 | |
|
43 | 43 | until eos? |
|
44 | 44 | |
|
45 | 45 | kind = nil |
|
46 | 46 | match = nil |
|
47 | 47 | |
|
48 | 48 | case state |
|
49 | 49 | |
|
50 | 50 | when :initial |
|
51 | 51 | |
|
52 | 52 | if scan(/ \s+ | \\\n /x) |
|
53 | 53 | kind = :space |
|
54 | 54 | |
|
55 | 55 | elsif scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx) |
|
56 | 56 | kind = :comment |
|
57 | 57 | |
|
58 | 58 | elsif match = scan(/ \# \s* if \s* 0 /x) |
|
59 | 59 | match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos? |
|
60 | 60 | kind = :comment |
|
61 | 61 | |
|
62 | 62 | elsif scan(/ [-+*\/=<>?:;,!&^|()\[\]{}~%]+ | \.(?!\d) /x) |
|
63 | 63 | kind = :operator |
|
64 | 64 | |
|
65 | 65 | elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x) |
|
66 | 66 | kind = IDENT_KIND[match] |
|
67 | 67 | if kind == :ident and check(/:(?!:)/) |
|
68 | 68 | match << scan(/:/) |
|
69 | 69 | kind = :label |
|
70 | 70 | end |
|
71 | 71 | |
|
72 | 72 | elsif match = scan(/L?"/) |
|
73 | 73 | tokens << [:open, :string] |
|
74 | 74 | if match[0] == ?L |
|
75 | 75 | tokens << ['L', :modifier] |
|
76 | 76 | match = '"' |
|
77 | 77 | end |
|
78 | 78 | state = :string |
|
79 | 79 | kind = :delimiter |
|
80 | 80 | |
|
81 | 81 | elsif scan(/#\s*(\w*)/) |
|
82 | 82 | kind = :preprocessor # FIXME multiline preprocs |
|
83 | 83 | state = :include_expected if self[1] == 'include' |
|
84 | 84 | |
|
85 | 85 | elsif scan(/ L?' (?: [^\'\n\\] | \\ #{ESCAPE} )? '? /ox) |
|
86 | 86 | kind = :char |
|
87 | 87 | |
|
88 | 88 | elsif scan(/0[xX][0-9A-Fa-f]+/) |
|
89 | 89 | kind = :hex |
|
90 | 90 | |
|
91 | 91 | elsif scan(/(?:0[0-7]+)(?![89.eEfF])/) |
|
92 | 92 | kind = :oct |
|
93 | 93 | |
|
94 | 94 | elsif scan(/(?:\d+)(?![.eEfF])/) |
|
95 | 95 | kind = :integer |
|
96 | 96 | |
|
97 | 97 | elsif scan(/\d[fF]?|\d*\.\d+(?:[eE][+-]?\d+)?[fF]?|\d+[eE][+-]?\d+[fF]?/) |
|
98 | 98 | kind = :float |
|
99 | 99 | |
|
100 | 100 | else |
|
101 | 101 | getch |
|
102 | 102 | kind = :error |
|
103 | 103 | |
|
104 | 104 | end |
|
105 | 105 | |
|
106 | 106 | when :string |
|
107 | 107 | if scan(/[^\\\n"]+/) |
|
108 | 108 | kind = :content |
|
109 | 109 | elsif scan(/"/) |
|
110 | 110 | tokens << ['"', :delimiter] |
|
111 | 111 | tokens << [:close, :string] |
|
112 | 112 | state = :initial |
|
113 | 113 | next |
|
114 | 114 | elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox) |
|
115 | 115 | kind = :char |
|
116 | 116 | elsif scan(/ \\ | $ /x) |
|
117 | 117 | tokens << [:close, :string] |
|
118 | 118 | kind = :error |
|
119 | 119 | state = :initial |
|
120 | 120 | else |
|
121 | 121 | raise_inspect "else case \" reached; %p not handled." % peek(1), tokens |
|
122 | 122 | end |
|
123 | 123 | |
|
124 | 124 | when :include_expected |
|
125 |
if scan(/ |
|
|
125 | if scan(/[^\n]+/) | |
|
126 | 126 | kind = :include |
|
127 | 127 | state = :initial |
|
128 | 128 | |
|
129 | 129 | elsif match = scan(/\s+/) |
|
130 | 130 | kind = :space |
|
131 | 131 | state = :initial if match.index ?\n |
|
132 | 132 | |
|
133 | 133 | else |
|
134 | 134 | getch |
|
135 | 135 | kind = :error |
|
136 | 136 | |
|
137 | 137 | end |
|
138 | 138 | |
|
139 | 139 | else |
|
140 | 140 | raise_inspect 'Unknown state', tokens |
|
141 | 141 | |
|
142 | 142 | end |
|
143 | 143 | |
|
144 | 144 | match ||= matched |
|
145 | 145 | if $DEBUG and not kind |
|
146 | 146 | raise_inspect 'Error token %p in line %d' % |
|
147 | 147 | [[match, kind], line], tokens |
|
148 | 148 | end |
|
149 | 149 | raise_inspect 'Empty token', tokens unless match |
|
150 | 150 | |
|
151 | 151 | tokens << [match, kind] |
|
152 | 152 | |
|
153 | 153 | end |
|
154 | 154 | |
|
155 | 155 | if state == :string |
|
156 | 156 | tokens << [:close, :string] |
|
157 | 157 | end |
|
158 | 158 | |
|
159 | 159 | tokens |
|
160 | 160 | end |
|
161 | 161 | |
|
162 | 162 | end |
|
163 | 163 | |
|
164 | 164 | end |
|
165 | 165 | end |
General Comments 0
You need to be logged in to leave comments.
Login now