| @@ -0,0 +1,176 | |||
|
|
1 | # http://pastie.textmate.org/50774/ | |
|
|
2 | module CodeRay module Scanners | |
|
|
3 | ||
|
|
4 | class JavaScript < Scanner | |
|
|
5 | ||
|
|
6 | register_for :javascript | |
|
|
7 | ||
|
|
8 | RESERVED_WORDS = [ | |
|
|
9 | 'asm', 'break', 'case', 'continue', 'default', 'do', 'else', | |
|
|
10 | 'for', 'goto', 'if', 'return', 'switch', 'while', | |
|
|
11 | # 'struct', 'union', 'enum', 'typedef', | |
|
|
12 | # 'static', 'register', 'auto', 'extern', | |
|
|
13 | # 'sizeof', | |
|
|
14 | 'typeof', | |
|
|
15 | # 'volatile', 'const', # C89 | |
|
|
16 | # 'inline', 'restrict', # C99 | |
|
|
17 | 'var', 'function','try','new','in', | |
|
|
18 | 'instanceof','throw','catch' | |
|
|
19 | ] | |
|
|
20 | ||
|
|
21 | PREDEFINED_CONSTANTS = [ | |
|
|
22 | 'void', 'null', 'this', | |
|
|
23 | 'true', 'false','undefined', | |
|
|
24 | ] | |
|
|
25 | ||
|
|
26 | IDENT_KIND = WordList.new(:ident). | |
|
|
27 | add(RESERVED_WORDS, :reserved). | |
|
|
28 | add(PREDEFINED_CONSTANTS, :pre_constant) | |
|
|
29 | ||
|
|
30 | ESCAPE = / [rbfnrtv\n\\\/'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x | |
|
|
31 | UNICODE_ESCAPE = / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x | |
|
|
32 | ||
|
|
33 | def scan_tokens tokens, options | |
|
|
34 | ||
|
|
35 | state = :initial | |
|
|
36 | string_type = nil | |
|
|
37 | regexp_allowed = true | |
|
|
38 | ||
|
|
39 | until eos? | |
|
|
40 | ||
|
|
41 | kind = :error | |
|
|
42 | match = nil | |
|
|
43 | ||
|
|
44 | if state == :initial | |
|
|
45 | ||
|
|
46 | if scan(/ \s+ | \\\n /x) | |
|
|
47 | kind = :space | |
|
|
48 | ||
|
|
49 | elsif scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx) | |
|
|
50 | kind = :comment | |
|
|
51 | regexp_allowed = false | |
|
|
52 | ||
|
|
53 | elsif match = scan(/ \# \s* if \s* 0 /x) | |
|
|
54 | match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos? | |
|
|
55 | kind = :comment | |
|
|
56 | regexp_allowed = false | |
|
|
57 | ||
|
|
58 | elsif regexp_allowed and scan(/\//) | |
|
|
59 | tokens << [:open, :regexp] | |
|
|
60 | state = :regex | |
|
|
61 | kind = :delimiter | |
|
|
62 | ||
|
|
63 | elsif scan(/ [-+*\/=<>?:;,!&^|()\[\]{}~%] | \.(?!\d) /x) | |
|
|
64 | kind = :operator | |
|
|
65 | regexp_allowed=true | |
|
|
66 | ||
|
|
67 | elsif match = scan(/ [$A-Za-z_][A-Za-z_0-9]* /x) | |
|
|
68 | kind = IDENT_KIND[match] | |
|
|
69 | # if kind == :ident and check(/:(?!:)/) | |
|
|
70 | # match << scan(/:/) | |
|
|
71 | # kind = :label | |
|
|
72 | # end | |
|
|
73 | regexp_allowed=false | |
|
|
74 | ||
|
|
75 | elsif match = scan(/["']/) | |
|
|
76 | tokens << [:open, :string] | |
|
|
77 | string_type = matched | |
|
|
78 | state = :string | |
|
|
79 | kind = :delimiter | |
|
|
80 | ||
|
|
81 | # elsif scan(/#\s*(\w*)/) | |
|
|
82 | # kind = :preprocessor # FIXME multiline preprocs | |
|
|
83 | # state = :include_expected if self[1] == 'include' | |
|
|
84 | # | |
|
|
85 | # elsif scan(/ L?' (?: [^\'\n\\] | \\ #{ESCAPE} )? '? /ox) | |
|
|
86 | # kind = :char | |
|
|
87 | ||
|
|
88 | elsif scan(/0[xX][0-9A-Fa-f]+/) | |
|
|
89 | kind = :hex | |
|
|
90 | regexp_allowed=false | |
|
|
91 | ||
|
|
92 | elsif scan(/(?:0[0-7]+)(?![89.eEfF])/) | |
|
|
93 | kind = :oct | |
|
|
94 | regexp_allowed=false | |
|
|
95 | ||
|
|
96 | elsif scan(/(?:\d+)(?![.eEfF])/) | |
|
|
97 | kind = :integer | |
|
|
98 | regexp_allowed=false | |
|
|
99 | ||
|
|
100 | elsif scan(/\d[fF]?|\d*\.\d+(?:[eE][+-]?\d+)?[fF]?|\d+[eE][+-]?\d+[fF]?/) | |
|
|
101 | kind = :float | |
|
|
102 | regexp_allowed=false | |
|
|
103 | ||
|
|
104 | else | |
|
|
105 | getch | |
|
|
106 | end | |
|
|
107 | ||
|
|
108 | elsif state == :regex | |
|
|
109 | if scan(/[^\\\/]+/) | |
|
|
110 | kind = :content | |
|
|
111 | elsif scan(/\\\/|\\\\/) | |
|
|
112 | kind = :content | |
|
|
113 | elsif scan(/\//) | |
|
|
114 | tokens << [matched, :delimiter] | |
|
|
115 | tokens << [:close, :regexp] | |
|
|
116 | state = :initial | |
|
|
117 | next | |
|
|
118 | else | |
|
|
119 | getch | |
|
|
120 | kind = :content | |
|
|
121 | end | |
|
|
122 | ||
|
|
123 | elsif state == :string | |
|
|
124 | if scan(/[^\\"']+/) | |
|
|
125 | kind = :content | |
|
|
126 | elsif scan(/["']/) | |
|
|
127 | if string_type==matched | |
|
|
128 | tokens << [matched, :delimiter] | |
|
|
129 | tokens << [:close, :string] | |
|
|
130 | state = :initial | |
|
|
131 | string_type=nil | |
|
|
132 | next | |
|
|
133 | else | |
|
|
134 | kind = :content | |
|
|
135 | end | |
|
|
136 | elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox) | |
|
|
137 | kind = :char | |
|
|
138 | elsif scan(/ \\ | $ /x) | |
|
|
139 | kind = :error | |
|
|
140 | state = :initial | |
|
|
141 | else | |
|
|
142 | raise "else case \" reached; %p not handled." % peek(1), tokens | |
|
|
143 | end | |
|
|
144 | ||
|
|
145 | # elsif state == :include_expected | |
|
|
146 | # if scan(/<[^>\n]+>?|"[^"\n\\]*(?:\\.[^"\n\\]*)*"?/) | |
|
|
147 | # kind = :include | |
|
|
148 | # state = :initial | |
|
|
149 | # | |
|
|
150 | # elsif match = scan(/\s+/) | |
|
|
151 | # kind = :space | |
|
|
152 | # state = :initial if match.index ?\n | |
|
|
153 | # | |
|
|
154 | # else | |
|
|
155 | # getch | |
|
|
156 | # | |
|
|
157 | # end | |
|
|
158 | # | |
|
|
159 | else | |
|
|
160 | raise 'else-case reached', tokens | |
|
|
161 | ||
|
|
162 | end | |
|
|
163 | ||
|
|
164 | match ||= matched | |
|
|
165 | # raise [match, kind], tokens if kind == :error | |
|
|
166 | ||
|
|
167 | tokens << [match, kind] | |
|
|
168 | ||
|
|
169 | end | |
|
|
170 | tokens | |
|
|
171 | ||
|
|
172 | end | |
|
|
173 | ||
|
|
174 | end | |
|
|
175 | ||
|
|
176 | end end No newline at end of file | |
General Comments 0
You need to be logged in to leave comments.
Login now