@@ -0,0 +1,130 | |||
|
1 | module CodeRay | |
|
2 | module Scanners | |
|
3 | class Java < Scanner | |
|
4 | ||
|
5 | register_for :java | |
|
6 | ||
|
7 | RESERVED_WORDS = %w(abstract assert break case catch class | |
|
8 | const continue default do else enum extends final finally for | |
|
9 | goto if implements import instanceof interface native new | |
|
10 | package private protected public return static strictfp super switch | |
|
11 | synchronized this throw throws transient try void volatile while) | |
|
12 | ||
|
13 | PREDEFINED_TYPES = %w(boolean byte char double float int long short) | |
|
14 | ||
|
15 | PREDEFINED_CONSTANTS = %w(true false null) | |
|
16 | ||
|
17 | IDENT_KIND = WordList.new(:ident). | |
|
18 | add(RESERVED_WORDS, :reserved). | |
|
19 | add(PREDEFINED_TYPES, :pre_type). | |
|
20 | add(PREDEFINED_CONSTANTS, :pre_constant) | |
|
21 | ||
|
22 | ESCAPE = / [rbfnrtv\n\\'"] | x[a-fA-F0-9]{1,2} | [0-7]{1,3} /x | |
|
23 | UNICODE_ESCAPE = / u[a-fA-F0-9]{4} | U[a-fA-F0-9]{8} /x | |
|
24 | ||
|
25 | def scan_tokens tokens, options | |
|
26 | state = :initial | |
|
27 | ||
|
28 | until eos? | |
|
29 | kind = nil | |
|
30 | match = nil | |
|
31 | ||
|
32 | case state | |
|
33 | when :initial | |
|
34 | ||
|
35 | if scan(/ \s+ | \\\n /x) | |
|
36 | kind = :space | |
|
37 | ||
|
38 | elsif scan(%r! // [^\n\\]* (?: \\. [^\n\\]* )* | /\* (?: .*? \*/ | .* ) !mx) | |
|
39 | kind = :comment | |
|
40 | ||
|
41 | elsif match = scan(/ \# \s* if \s* 0 /x) | |
|
42 | match << scan_until(/ ^\# (?:elif|else|endif) .*? $ | \z /xm) unless eos? | |
|
43 | kind = :comment | |
|
44 | ||
|
45 | elsif scan(/ [-+*\/=<>?:;,!&^|()\[\]{}~%]+ | \.(?!\d) /x) | |
|
46 | kind = :operator | |
|
47 | ||
|
48 | elsif match = scan(/ [A-Za-z_][A-Za-z_0-9]* /x) | |
|
49 | kind = IDENT_KIND[match] | |
|
50 | if kind == :ident and check(/:(?!:)/) | |
|
51 | match << scan(/:/) | |
|
52 | kind = :label | |
|
53 | end | |
|
54 | ||
|
55 | elsif match = scan(/L?"/) | |
|
56 | tokens << [:open, :string] | |
|
57 | if match[0] == ?L | |
|
58 | tokens << ['L', :modifier] | |
|
59 | match = '"' | |
|
60 | end | |
|
61 | state = :string | |
|
62 | kind = :delimiter | |
|
63 | ||
|
64 | elsif scan(%r! \@ .* !x) | |
|
65 | kind = :preprocessor | |
|
66 | ||
|
67 | elsif scan(/ L?' (?: [^\'\n\\] | \\ #{ESCAPE} )? '? /ox) | |
|
68 | kind = :char | |
|
69 | ||
|
70 | elsif scan(/0[xX][0-9A-Fa-f]+/) | |
|
71 | kind = :hex | |
|
72 | ||
|
73 | elsif scan(/(?:0[0-7]+)(?![89.eEfF])/) | |
|
74 | kind = :oct | |
|
75 | ||
|
76 | elsif scan(/(?:\d+)(?![.eEfF])/) | |
|
77 | kind = :integer | |
|
78 | ||
|
79 | elsif scan(/\d[fF]?|\d*\.\d+(?:[eE][+-]?\d+)?[fF]?|\d+[eE][+-]?\d+[fF]?/) | |
|
80 | kind = :float | |
|
81 | ||
|
82 | else | |
|
83 | getch | |
|
84 | kind = :error | |
|
85 | ||
|
86 | end | |
|
87 | ||
|
88 | when :string | |
|
89 | if scan(/[^\\\n"]+/) | |
|
90 | kind = :content | |
|
91 | elsif scan(/"/) | |
|
92 | tokens << ['"', :delimiter] | |
|
93 | tokens << [:close, :string] | |
|
94 | state = :initial | |
|
95 | next | |
|
96 | elsif scan(/ \\ (?: #{ESCAPE} | #{UNICODE_ESCAPE} ) /mox) | |
|
97 | kind = :char | |
|
98 | elsif scan(/ \\ | $ /x) | |
|
99 | tokens << [:close, :string] | |
|
100 | kind = :error | |
|
101 | state = :initial | |
|
102 | else | |
|
103 | raise_inspect "else case \" reached; %p not handled." % peek(1), tokens | |
|
104 | end | |
|
105 | ||
|
106 | else | |
|
107 | raise_inspect 'Unknown state', tokens | |
|
108 | ||
|
109 | end | |
|
110 | ||
|
111 | match ||= matched | |
|
112 | if $DEBUG and not kind | |
|
113 | raise_inspect 'Error token %p in line %d' % | |
|
114 | [[match, kind], line], tokens | |
|
115 | end | |
|
116 | raise_inspect 'Empty token', tokens unless match | |
|
117 | ||
|
118 | tokens << [match, kind] | |
|
119 | ||
|
120 | end | |
|
121 | ||
|
122 | if state == :string | |
|
123 | tokens << [:close, :string] | |
|
124 | end | |
|
125 | ||
|
126 | tokens | |
|
127 | end | |
|
128 | end | |
|
129 | end | |
|
130 | end |
General Comments 0
You need to be logged in to leave comments.
Login now