1 | %{ |
---|
2 | #include <stdio.h> |
---|
3 | #include "JavaTokenTypes.hpp" |
---|
4 | extern void newLine(); |
---|
5 | |
---|
6 | char string_buf[1024]; |
---|
7 | char* string_buf_ptr = 0; |
---|
8 | |
---|
9 | %} |
---|
10 | |
---|
11 | HexDigit [0-9a-fA-F] |
---|
12 | Digit [0-9] |
---|
13 | OctalDigit [0-7] |
---|
14 | NonZeroDigit [1-9] |
---|
15 | AnyButSlash [^\/] |
---|
16 | AnyButAstr [^\*] |
---|
17 | |
---|
18 | Unicode [\\][u]{HexDigit}{HexDigit}{HexDigit}{HexDigit} |
---|
19 | Octal [\\]{OctalDigit}({Digit}({Digit})?)? |
---|
20 | Escape [\\]([r]|[n]|[b]|[f]|[t]|[\\]|[']|["])|{Unicode}|{Octal} |
---|
21 | |
---|
22 | IntSuffix ([l]|[L]) |
---|
23 | DecimalNum {NonZeroDigit}{Digit}*{IntSuffix}? |
---|
24 | OctalNum [0]{OctalDigit}*{IntSuffix}? |
---|
25 | HexNum [0]([x]|[X]){HexDigit}{HexDigit}*{IntSuffix}? |
---|
26 | IntegerLiteral ({DecimalNum}|{OctalNum}|{HexNum}) |
---|
27 | |
---|
28 | Sign ([\+]|[\-]) |
---|
29 | FlSuffix ([f]|[F]|[d][D]) |
---|
30 | SignedInt {Sign}?{Digit}+ |
---|
31 | Expo ([e]|[E]) |
---|
32 | ExponentPart {Expo}{SignedInt}? |
---|
33 | Float1 {Digit}+[\.]{Digit}+?{ExponentPart}?{FlSuffix}? |
---|
34 | Float2 [\.]{Digit}+{ExponentPart}?{FlSuffix}? |
---|
35 | Float3 {Digit}+{ExponentPart}{FlSuffix}? |
---|
36 | Float4 {Digit}+{FlSuffix} |
---|
37 | FloatingPoint ({Float1}|{Float2}|{Float3}|{Float4}) |
---|
38 | |
---|
39 | %x string comment |
---|
40 | %% |
---|
41 | "?" return QUESTION; |
---|
42 | "(" return LPAREN; |
---|
43 | ")" return RPAREN; |
---|
44 | "[" return LBRACK; |
---|
45 | "]" return RBRACK; |
---|
46 | "{" return LCURLY; |
---|
47 | "}" return RCURLY; |
---|
48 | ":" return COLON; |
---|
49 | "," return COMMA; |
---|
50 | "." return DOT; |
---|
51 | "==" return EQUAL; |
---|
52 | "=" return ASSIGN; |
---|
53 | "!" return LNOT; |
---|
54 | "~" return BNOT; |
---|
55 | "!=" return NOT_EQUAL; |
---|
56 | "/=" return DIV_ASSIGN; |
---|
57 | "/" return DIV; |
---|
58 | "+=" return PLUS_ASSIGN; |
---|
59 | "++" return INC; |
---|
60 | "+" return PLUS; |
---|
61 | "-=" return MINUS_ASSIGN; |
---|
62 | "--" return DEC; |
---|
63 | "-" return MINUS; |
---|
64 | "*=" return STAR_ASSIGN; |
---|
65 | "*" return STAR; |
---|
66 | "%=" return MOD_ASSIGN; |
---|
67 | "%" return MOD; |
---|
68 | ">>>=" return BSR_ASSIGN; |
---|
69 | ">>>" return BSR; |
---|
70 | ">>=" return SR_ASSIGN; |
---|
71 | ">>" return SR; |
---|
72 | ">=" return GE; |
---|
73 | ">" return GT; |
---|
74 | "<<=" return SL_ASSIGN; |
---|
75 | "<<" return SL; |
---|
76 | "<=" return LE; |
---|
77 | "<" return LT_; |
---|
78 | "^=" return BXOR_ASSIGN; |
---|
79 | "^" return BXOR; |
---|
80 | "|=" return BOR_ASSIGN; |
---|
81 | "||" return LOR; |
---|
82 | "|" return BOR; |
---|
83 | "&=" return BAND_ASSIGN; |
---|
84 | "&&" return LAND; |
---|
85 | "&" return BAND; |
---|
86 | ";" return SEMI; |
---|
87 | |
---|
88 | [ \f\t] ; |
---|
89 | "\r\n"|\r|\n { newLine(); } |
---|
90 | |
---|
91 | "//".* ; |
---|
92 | |
---|
93 | [\']({Escape}|[^\'])[\'] return CHAR_LITERAL; |
---|
94 | |
---|
95 | "/*" BEGIN(comment); |
---|
96 | <comment>{ |
---|
97 | [^*\n]* /* eat anything that's not a '*' */ |
---|
98 | "*"+[^*/\n]* /* eat up '*'s not followed by '/'s */ |
---|
99 | \n { newLine(); } |
---|
100 | "*"+"/" BEGIN(INITIAL); |
---|
101 | } |
---|
102 | |
---|
103 | [\"] { |
---|
104 | BEGIN(string); |
---|
105 | string_buf_ptr = string_buf; |
---|
106 | } |
---|
107 | <string>{ |
---|
108 | \" { |
---|
109 | BEGIN(INITIAL); |
---|
110 | *string_buf_ptr = '\0'; |
---|
111 | return STRING_LITERAL; |
---|
112 | } |
---|
113 | {Octal} { |
---|
114 | int result; |
---|
115 | sscanf( yytext + 1, "%o", &result ); |
---|
116 | *string_buf_ptr++ = result; |
---|
117 | } |
---|
118 | {Unicode} { |
---|
119 | int result; |
---|
120 | sscanf( yytext + 1, "%x", &result ); |
---|
121 | *string_buf_ptr++ = result; |
---|
122 | } |
---|
123 | \\n *string_buf_ptr++ = '\n'; |
---|
124 | \\t *string_buf_ptr++ = '\t'; |
---|
125 | \\r *string_buf_ptr++ = '\r'; |
---|
126 | \\b *string_buf_ptr++ = '\b'; |
---|
127 | \\f *string_buf_ptr++ = '\f'; |
---|
128 | \\(.|\n) *string_buf_ptr++ = yytext[1]; |
---|
129 | [^\\\n\"]+ { |
---|
130 | char *yptr = yytext; |
---|
131 | |
---|
132 | while ( *yptr ) |
---|
133 | *string_buf_ptr++ = *yptr++; |
---|
134 | } |
---|
135 | } // end string states |
---|
136 | |
---|
137 | while return LITERAL_while; |
---|
138 | volatile return LITERAL_volatile; |
---|
139 | void return LITERAL_void; |
---|
140 | try return LITERAL_try; |
---|
141 | true return LITERAL_true; |
---|
142 | transient return LITERAL_transient; |
---|
143 | throws return LITERAL_throws; |
---|
144 | throw return LITERAL_throw; |
---|
145 | threadsafe return LITERAL_threadsafe; |
---|
146 | this return LITERAL_this; |
---|
147 | synchronized return LITERAL_synchronized; |
---|
148 | switch return LITERAL_switch; |
---|
149 | super return LITERAL_super; |
---|
150 | static return LITERAL_static; |
---|
151 | short return LITERAL_short; |
---|
152 | return return LITERAL_return; |
---|
153 | public return LITERAL_public; |
---|
154 | protected return LITERAL_protected; |
---|
155 | private return LITERAL_private; |
---|
156 | package return LITERAL_package; |
---|
157 | null return LITERAL_null; |
---|
158 | new return LITERAL_new; |
---|
159 | native return LITERAL_native; |
---|
160 | long return LITERAL_long; |
---|
161 | interface return LITERAL_interface; |
---|
162 | int return LITERAL_int; |
---|
163 | instanceof return LITERAL_instanceof; |
---|
164 | import return LITERAL_import; |
---|
165 | implements return LITERAL_implements; |
---|
166 | if return LITERAL_if; |
---|
167 | for return LITERAL_for; |
---|
168 | float return LITERAL_float; |
---|
169 | finally return LITERAL_finally; |
---|
170 | final return FINAL; |
---|
171 | false return LITERAL_false; |
---|
172 | extends return LITERAL_extends; |
---|
173 | else return LITERAL_else; |
---|
174 | double return LITERAL_double; |
---|
175 | do return LITERAL_do; |
---|
176 | default return LITERAL_default; |
---|
177 | continue return LITERAL_continue; |
---|
178 | class return LITERAL_class; |
---|
179 | char return LITERAL_char; |
---|
180 | catch return LITERAL_catch; |
---|
181 | case return LITERAL_case; |
---|
182 | byte return LITERAL_byte; |
---|
183 | break return LITERAL_break; |
---|
184 | boolean return LITERAL_boolean; |
---|
185 | abstract return ABSTRACT; |
---|
186 | |
---|
187 | [a-zA-Z_$][a-zA-Z_0-9$]* return IDENT; |
---|
188 | |
---|
189 | {IntegerLiteral} return NUM_INT; |
---|
190 | {FloatingPoint} return NUM_FLOAT; |
---|
191 | |
---|
192 | %% |
---|