1 lexer grammar CTFLexer;
8 package org.eclipse.linuxtools.ctf.parser;
21 DOUBLETOK : 'double' ;
24 FLOATINGPOINTTOK : 'floating_point' ;
26 INTEGERTOK : 'integer' ;
30 SIGNEDTOK : 'signed' ;
31 STREAMTOK : 'stream' ;
32 STRINGTOK : 'string' ;
33 STRUCTTOK : 'struct' ;
35 TYPEALIASTOK : 'typealias' ;
36 TYPEDEFTOK : 'typedef' ;
37 UNSIGNEDTOK : 'unsigned' ;
38 VARIANTTOK : 'variant' ;
41 COMPLEXTOK : '_Complex' ;
42 IMAGINARYTOK : '_Imaginary' ;
46 * Callsite tokens (CTF v1.9)
48 CALLSITETOK : 'callsite' ;
52 * These tokens are not part of the CTF standard.
53 * There are planned to be in CTF v1.9
55 NANNUMBERTOK : 'NaN' ;
56 INFINITYTOK : '+inf' ;
57 NINFINITYTOK : '-inf' ;
66 TYPE_ASSIGNMENT : ':=' ;
80 fragment BACKSLASH : '\\' ;
82 /* Helpers for integer literals */
83 fragment DIGIT : '0'..'9' ;
84 fragment OCT_DIGIT : '0'..'7' ;
85 fragment OCT_PREFIX : '0' ;
86 fragment NONZERO_DIGIT : '1'..'9' ;
87 fragment HEX_DIGIT : DIGIT | ('a'..'f') | ('A'..'F') ;
88 fragment HEX_PREFIX : '0' ('x' | 'X') ;
93 OCTAL_LITERAL : OCT_PREFIX (OCT_DIGIT)+ INTEGER_TYPES_SUFFIX? ;
94 DECIMAL_LITERAL : DIGIT+ INTEGER_TYPES_SUFFIX? ;
95 HEX_LITERAL : HEX_PREFIX HEX_DIGIT+ INTEGER_TYPES_SUFFIX? ;
98 * Integer suffix for long, long long and unsigned.
100 * Matches all possible combination of L, LL and U.
102 fragment INTEGER_TYPES_SUFFIX
103 : ('l' ('l')? | 'L' ('L')?) // l, ll
105 | ('u' | 'U') ('l' ('l')? | 'L' ('L')?) // ul, ull
106 | ('l' ('l')? | 'L' ('L')?) ('u'| 'U') // lu, llu
112 fragment ESCAPE_SEQUENCE
113 : BACKSLASH ('\'' | '"' | '?' | BACKSLASH | 'a' | 'b' | 'f' | 'n' | 'r' | 't' | 'v' )
120 * Octal escape sequence
122 fragment OCTAL_ESCAPE
123 : BACKSLASH ('0'..'3') ('0'..'7') ('0'..'7')
124 | BACKSLASH ('0'..'7') ('0'..'7')
125 | BACKSLASH ('0'..'7')
129 * Hexadecimal escape sequence
131 fragment HEXADECIMAL_ESCAPE : BACKSLASH 'x' HEX_DIGIT+ ;
134 * Unicode escape sequence
136 fragment UNICODE_ESCAPE
137 : BACKSLASH 'u' HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT
138 | BACKSLASH 'U' HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT HEX_DIGIT
142 /* Used in both character and string literal */
143 fragment STRINGPREFIX : 'L';
148 CHARACTER_LITERAL : STRINGPREFIX? SINGLEQUOTE CHAR_CONTENT+ SINGLEQUOTE ;
149 fragment CHAR_CONTENT : (ESCAPE_SEQUENCE | ~(BACKSLASH | SINGLEQUOTE)) ;
150 fragment SINGLEQUOTE : '\'';
155 STRING_LITERAL : STRINGPREFIX? DOUBLEQUOTE STRING_CONTENT* DOUBLEQUOTE ;
156 fragment STRING_CONTENT : (ESCAPE_SEQUENCE | ~(BACKSLASH | DOUBLEQUOTE)) ;
157 fragment DOUBLEQUOTE : '"' ;
162 WS : (' ' | '\r' | '\t' | '\u000C' | '\n') { $channel = HIDDEN; } ;
167 // About the greedy option: see page 100-101 of The Definitive ANTLR reference
168 // COMMENT : '/*' ( options { greedy = false; } : . )* '*/' { $channel = HIDDEN; } ;
169 COMMENT : COMMENT_OPEN .* COMMENT_CLOSE { $channel = HIDDEN; } ;
170 fragment COMMENT_OPEN : '/*';
171 fragment COMMENT_CLOSE : '*/';
174 * Single line comment
176 LINE_COMMENT : '//' ~('\n')* '\n' { $channel = HIDDEN; } ;
181 IDENTIFIER : NONDIGIT (NONDIGIT | DIGIT)* ;
182 fragment NONDIGIT : ('_') | ('A'..'Z') | ('a'..'z') ;