[b87a5ed] | 1 | /* |
---|
| 2 | * Cforall Version 1.0.0 Copyright (C) 2015 University of Waterloo |
---|
| 3 | * |
---|
| 4 | * The contents of this file are covered under the licence agreement in the |
---|
| 5 | * file "LICENCE" distributed with Cforall. |
---|
[8f60f0b] | 6 | * |
---|
[6016c87] | 7 | * lex.ll -- |
---|
[8f60f0b] | 8 | * |
---|
[51b7345] | 9 | * Author : Peter A. Buhr |
---|
| 10 | * Created On : Sat Sep 22 08:58:10 2001 |
---|
[926af74] | 11 | * Last Modified By : Peter A. Buhr |
---|
[f1da02c] | 12 | * Last Modified On : Sun Aug 4 20:53:47 2019 |
---|
| 13 | * Update Count : 719 |
---|
[51b7345] | 14 | */ |
---|
| 15 | |
---|
| 16 | %option yylineno |
---|
[4e9c7c1] | 17 | %option noyywrap |
---|
[5f2f2d7] | 18 | %option nounput |
---|
[51b7345] | 19 | |
---|
| 20 | %{ |
---|
[9ed4f94] | 21 | // The lexer assumes the program has been preprocessed by cpp. Hence, all user level preprocessor directive have been |
---|
[de62360d] | 22 | // performed and removed from the source. The only exceptions are preprocessor directives passed to the compiler (e.g., |
---|
| 23 | // line-number directives) and C/C++ style comments, which are ignored. |
---|
[51b7345] | 24 | |
---|
[8c17ab0] | 25 | //**************************** Includes and Defines **************************** |
---|
[51b7345] | 26 | |
---|
[4439008] | 27 | // trigger before each matching rule's action |
---|
| 28 | #define YY_USER_ACTION \ |
---|
| 29 | yylloc.first_line = yylineno; \ |
---|
| 30 | yylloc.first_column = column; \ |
---|
| 31 | column += yyleng; \ |
---|
| 32 | yylloc.last_column = column; \ |
---|
| 33 | yylloc.last_line = yylineno; \ |
---|
| 34 | yylloc.filename = yyfilename ? yyfilename : ""; |
---|
[9ed4f94] | 35 | unsigned int column = 0; // position of the end of the last token parsed |
---|
| 36 | |
---|
[51b7345] | 37 | #include <string> |
---|
[f487962] | 38 | #include <cstdio> // FILENAME_MAX |
---|
[9ed4f94] | 39 | using namespace std; |
---|
[51b7345] | 40 | |
---|
[ba01b14] | 41 | #include "config.h" // configure info |
---|
[984dce6] | 42 | #include "ParseNode.h" |
---|
| 43 | #include "TypedefTable.h" |
---|
[51b7345] | 44 | |
---|
| 45 | char *yyfilename; |
---|
[9ed4f94] | 46 | string *strtext; // accumulate parts of character and string constant value |
---|
[51b7345] | 47 | |
---|
[de62360d] | 48 | #define RETURN_LOCN(x) yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x ) |
---|
[9ed4f94] | 49 | #define RETURN_VAL(x) yylval.tok.str = new string( yytext ); RETURN_LOCN( x ) |
---|
[926af74] | 50 | #define RETURN_CHAR(x) yylval.tok.str = nullptr; RETURN_LOCN( x ) |
---|
[de62360d] | 51 | #define RETURN_STR(x) yylval.tok.str = strtext; RETURN_LOCN( x ) |
---|
[5f2f2d7] | 52 | |
---|
[f487962] | 53 | #define WHITE_RETURN(x) // do nothing |
---|
[9ed4f94] | 54 | #define NEWLINE_RETURN() column = 0; WHITE_RETURN( '\n' ) |
---|
[de62360d] | 55 | #define ASCIIOP_RETURN() RETURN_CHAR( (int)yytext[0] ) // single character operator |
---|
[f487962] | 56 | #define NAMEDOP_RETURN(x) RETURN_CHAR( x ) // multichar operator, with a name |
---|
[de62360d] | 57 | #define NUMERIC_RETURN(x) rm_underscore(); RETURN_VAL( x ) // numeric constant |
---|
| 58 | #define KEYWORD_RETURN(x) RETURN_CHAR( x ) // keyword |
---|
[4439008] | 59 | #define QKEYWORD_RETURN(x) RETURN_VAL(x); // quasi-keyword |
---|
[984dce6] | 60 | #define IDENTIFIER_RETURN() RETURN_VAL( typedefTable.isKind( yytext ) ) |
---|
[51b7345] | 61 | |
---|
[ba01b14] | 62 | #ifdef HAVE_KEYWORDS_FLOATXX // GCC >= 7 => keyword, otherwise typedef |
---|
| 63 | #define FLOATXX(v) KEYWORD_RETURN(v); |
---|
| 64 | #else |
---|
| 65 | #define FLOATXX(v) IDENTIFIER_RETURN(); |
---|
| 66 | #endif // HAVE_KEYWORDS_FLOATXX |
---|
| 67 | |
---|
[3848e0e] | 68 | void rm_underscore() { |
---|
[7e419e7] | 69 | // SKULLDUGGERY: remove underscores (ok to shorten?) |
---|
[e7aed49] | 70 | yyleng = 0; |
---|
[7e419e7] | 71 | for ( int i = 0; yytext[i] != '\0'; i += 1 ) { // copying non-underscore characters to front of string |
---|
[b87a5ed] | 72 | if ( yytext[i] != '_' ) { |
---|
[e7aed49] | 73 | yytext[yyleng] = yytext[i]; |
---|
| 74 | yyleng += 1; |
---|
[b87a5ed] | 75 | } // if |
---|
| 76 | } // for |
---|
| 77 | yytext[yyleng] = '\0'; |
---|
[7e419e7] | 78 | } // rm_underscore |
---|
[51b7345] | 79 | |
---|
[7812f1d1] | 80 | // Stop warning due to incorrectly generated flex code. |
---|
| 81 | #pragma GCC diagnostic ignored "-Wsign-compare" |
---|
[51b7345] | 82 | %} |
---|
| 83 | |
---|
[0a2168f] | 84 | binary [0-1] |
---|
[51b7345] | 85 | octal [0-7] |
---|
| 86 | nonzero [1-9] |
---|
| 87 | decimal [0-9] |
---|
| 88 | hex [0-9a-fA-F] |
---|
[3848e0e] | 89 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2})) |
---|
[51b7345] | 90 | |
---|
[b87a5ed] | 91 | // identifier, GCC: $ in identifier |
---|
[51b7345] | 92 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})* |
---|
| 93 | |
---|
[b87a5ed] | 94 | // numeric constants, CFA: '_' in constant |
---|
[3848e0e] | 95 | hex_quad {hex}("_"?{hex}){3} |
---|
[201aeb9] | 96 | size_opt (8|16|32|64|128)? |
---|
[0a616e0] | 97 | // CFA: explicit l8/l16/l32/l64/l128, char 'hh', short 'h', int 'n' |
---|
[f56c32e] | 98 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hHnN]) |
---|
[0a616e0] | 99 | // CFA: size_t 'z', pointer 'p', which define a sign and length |
---|
| 100 | integer_suffix_opt ("_"?(([uU]({length}?[iI]?)|([iI]{length}))|([iI]({length}?[uU]?)|([uU]{length}))|({length}([iI]?[uU]?)|([uU][iI]))|[zZ]|[pP]))? |
---|
[51b7345] | 101 | |
---|
| 102 | octal_digits ({octal})|({octal}({octal}|"_")*{octal}) |
---|
| 103 | octal_prefix "0""_"? |
---|
[930f69e] | 104 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt} |
---|
[51b7345] | 105 | |
---|
| 106 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal}) |
---|
[930f69e] | 107 | decimal_constant {nonzero_digits}{integer_suffix_opt} |
---|
[51b7345] | 108 | |
---|
[0a2168f] | 109 | binary_digits ({binary})|({binary}({binary}|"_")*{binary}) |
---|
| 110 | binary_prefix "0"[bB]"_"? |
---|
| 111 | binary_constant {binary_prefix}{binary_digits}{integer_suffix_opt} |
---|
| 112 | |
---|
[51b7345] | 113 | hex_digits ({hex})|({hex}({hex}|"_")*{hex}) |
---|
| 114 | hex_prefix "0"[xX]"_"? |
---|
[930f69e] | 115 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt} |
---|
[51b7345] | 116 | |
---|
[76c62b2] | 117 | // GCC: D (double) and iI (imaginary) suffixes, and DL (long double) |
---|
[51b7345] | 118 | exponent "_"?[eE]"_"?[+-]?{decimal_digits} |
---|
[ba01b14] | 119 | floating_size 16|32|32x|64|64x|80|128|128x |
---|
| 120 | floating_length ([fFdDlLwWqQ]|[fF]{floating_size}) |
---|
[201aeb9] | 121 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length}) |
---|
[dc7db63] | 122 | floating_suffix_opt ("_"?({floating_suffix}|"DL"))? |
---|
[f4b77f2] | 123 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal}) |
---|
[930f69e] | 124 | floating_decimal {decimal_digits}"."{exponent}?{floating_suffix_opt} |
---|
| 125 | floating_fraction "."{decimal_digits}{exponent}?{floating_suffix_opt} |
---|
| 126 | floating_constant ({decimal_digits}{exponent}{floating_suffix_opt})|({decimal_digits}{floating_fraction}) |
---|
[51b7345] | 127 | |
---|
| 128 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits} |
---|
[dc7db63] | 129 | hex_floating_suffix_opt ("_"?({floating_suffix}))? |
---|
[930f69e] | 130 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".") |
---|
[f4b77f2] | 131 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt} |
---|
[51b7345] | 132 | |
---|
[b87a5ed] | 133 | // character escape sequence, GCC: \e => esc character |
---|
[51b7345] | 134 | simple_escape "\\"[abefnrtv'"?\\] |
---|
[930f69e] | 135 | // ' stop editor highlighting |
---|
[3848e0e] | 136 | octal_escape "\\"{octal}("_"?{octal}){0,2} |
---|
| 137 | hex_escape "\\""x""_"?{hex_digits} |
---|
[51b7345] | 138 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char} |
---|
[59db689] | 139 | cwide_prefix "L"|"U"|"u" |
---|
| 140 | swide_prefix {cwide_prefix}|"u8" |
---|
[51b7345] | 141 | |
---|
[b87a5ed] | 142 | // display/white-space characters |
---|
[51b7345] | 143 | h_tab [\011] |
---|
| 144 | form_feed [\014] |
---|
| 145 | v_tab [\013] |
---|
| 146 | c_return [\015] |
---|
| 147 | h_white [ ]|{h_tab} |
---|
| 148 | |
---|
[e7aed49] | 149 | // overloadable operators |
---|
[51b7345] | 150 | op_unary_only "~"|"!" |
---|
| 151 | op_unary_binary "+"|"-"|"*" |
---|
| 152 | op_unary_pre_post "++"|"--" |
---|
| 153 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post} |
---|
| 154 | |
---|
[994d080] | 155 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>=" |
---|
[51b7345] | 156 | op_binary_over {op_unary_binary}|{op_binary_only} |
---|
[e7aed49] | 157 | // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@=" |
---|
| 158 | // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over} |
---|
[51b7345] | 159 | |
---|
| 160 | %x COMMENT |
---|
[3848e0e] | 161 | %x BKQUOTE |
---|
| 162 | %x QUOTE |
---|
| 163 | %x STRING |
---|
[51b7345] | 164 | |
---|
| 165 | %% |
---|
[994d080] | 166 | /* line directives */ |
---|
[4040425] | 167 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" { |
---|
[930f69e] | 168 | /* " stop editor highlighting */ |
---|
[c1c1112] | 169 | static char filename[FILENAME_MAX]; // temporarily store current source-file name |
---|
[51b7345] | 170 | char *end_num; |
---|
| 171 | char *begin_string, *end_string; |
---|
| 172 | long lineno, length; |
---|
| 173 | lineno = strtol( yytext + 1, &end_num, 0 ); |
---|
| 174 | begin_string = strchr( end_num, '"' ); |
---|
[f487962] | 175 | if ( begin_string ) { // file name ? |
---|
| 176 | end_string = strchr( begin_string + 1, '"' ); // look for ending delimiter |
---|
| 177 | assert( end_string ); // closing quote ? |
---|
| 178 | length = end_string - begin_string - 1; // file-name length without quotes or sentinel |
---|
| 179 | assert( length < FILENAME_MAX ); // room for sentinel ? |
---|
| 180 | memcpy( &filename, begin_string + 1, length ); // copy file name from yytext |
---|
| 181 | filename[ length ] = '\0'; // terminate string with sentinel |
---|
[9ed4f94] | 182 | //cout << "file " << filename << " line " << lineno << endl; |
---|
[f487962] | 183 | yylineno = lineno; |
---|
[c1c1112] | 184 | yyfilename = filename; |
---|
[b87a5ed] | 185 | } // if |
---|
[51b7345] | 186 | } |
---|
| 187 | |
---|
[637dd9c] | 188 | /* preprocessor-style directives */ |
---|
[61fc4f6] | 189 | ^{h_white}*"#"[^\n]*"\n" { RETURN_VAL( DIRECTIVE ); } |
---|
[51b7345] | 190 | |
---|
[cd623a4] | 191 | /* ignore C style comments (ALSO HANDLED BY CPP) */ |
---|
[3848e0e] | 192 | "/*" { BEGIN COMMENT; } |
---|
[cd623a4] | 193 | <COMMENT>.|\n ; |
---|
| 194 | <COMMENT>"*/" { BEGIN 0; } |
---|
[51b7345] | 195 | |
---|
[cd623a4] | 196 | /* ignore C++ style comments (ALSO HANDLED BY CPP) */ |
---|
| 197 | "//"[^\n]*"\n" ; |
---|
[51b7345] | 198 | |
---|
[b87a5ed] | 199 | /* ignore whitespace */ |
---|
[3848e0e] | 200 | {h_white}+ { WHITE_RETURN(' '); } |
---|
| 201 | ({v_tab}|{c_return}|{form_feed})+ { WHITE_RETURN(' '); } |
---|
| 202 | ({h_white}|{v_tab}|{c_return}|{form_feed})*"\n" { NEWLINE_RETURN(); } |
---|
[51b7345] | 203 | |
---|
[b87a5ed] | 204 | /* keywords */ |
---|
| 205 | _Alignas { KEYWORD_RETURN(ALIGNAS); } // C11 |
---|
| 206 | _Alignof { KEYWORD_RETURN(ALIGNOF); } // C11 |
---|
| 207 | __alignof { KEYWORD_RETURN(ALIGNOF); } // GCC |
---|
| 208 | __alignof__ { KEYWORD_RETURN(ALIGNOF); } // GCC |
---|
| 209 | asm { KEYWORD_RETURN(ASM); } |
---|
| 210 | __asm { KEYWORD_RETURN(ASM); } // GCC |
---|
| 211 | __asm__ { KEYWORD_RETURN(ASM); } // GCC |
---|
| 212 | _Atomic { KEYWORD_RETURN(ATOMIC); } // C11 |
---|
| 213 | __attribute { KEYWORD_RETURN(ATTRIBUTE); } // GCC |
---|
| 214 | __attribute__ { KEYWORD_RETURN(ATTRIBUTE); } // GCC |
---|
[3848e0e] | 215 | auto { KEYWORD_RETURN(AUTO); } |
---|
[f673c13c] | 216 | __auto_type { KEYWORD_RETURN(AUTO_TYPE); } |
---|
[b6ad601] | 217 | basetypeof { KEYWORD_RETURN(BASETYPEOF); } // CFA |
---|
[b87a5ed] | 218 | _Bool { KEYWORD_RETURN(BOOL); } // C99 |
---|
[3848e0e] | 219 | break { KEYWORD_RETURN(BREAK); } |
---|
| 220 | case { KEYWORD_RETURN(CASE); } |
---|
[b87a5ed] | 221 | catch { KEYWORD_RETURN(CATCH); } // CFA |
---|
[02e5ab6] | 222 | catchResume { KEYWORD_RETURN(CATCHRESUME); } // CFA |
---|
[3848e0e] | 223 | char { KEYWORD_RETURN(CHAR); } |
---|
[b87a5ed] | 224 | choose { KEYWORD_RETURN(CHOOSE); } // CFA |
---|
[1528a2c] | 225 | coerce { KEYWORD_RETURN(COERCE); } // CFA |
---|
[b87a5ed] | 226 | _Complex { KEYWORD_RETURN(COMPLEX); } // C99 |
---|
| 227 | __complex { KEYWORD_RETURN(COMPLEX); } // GCC |
---|
| 228 | __complex__ { KEYWORD_RETURN(COMPLEX); } // GCC |
---|
[3848e0e] | 229 | const { KEYWORD_RETURN(CONST); } |
---|
[b87a5ed] | 230 | __const { KEYWORD_RETURN(CONST); } // GCC |
---|
| 231 | __const__ { KEYWORD_RETURN(CONST); } // GCC |
---|
[3848e0e] | 232 | continue { KEYWORD_RETURN(CONTINUE); } |
---|
[e04b636] | 233 | coroutine { KEYWORD_RETURN(COROUTINE); } // CFA |
---|
[3848e0e] | 234 | default { KEYWORD_RETURN(DEFAULT); } |
---|
[02e5ab6] | 235 | disable { KEYWORD_RETURN(DISABLE); } // CFA |
---|
[b87a5ed] | 236 | do { KEYWORD_RETURN(DO); } |
---|
[3848e0e] | 237 | double { KEYWORD_RETURN(DOUBLE); } |
---|
[b87a5ed] | 238 | dtype { KEYWORD_RETURN(DTYPE); } // CFA |
---|
[3848e0e] | 239 | else { KEYWORD_RETURN(ELSE); } |
---|
[02e5ab6] | 240 | enable { KEYWORD_RETURN(ENABLE); } // CFA |
---|
[3848e0e] | 241 | enum { KEYWORD_RETURN(ENUM); } |
---|
[b87a5ed] | 242 | __extension__ { KEYWORD_RETURN(EXTENSION); } // GCC |
---|
[c27fb59] | 243 | exception { KEYWORD_RETURN(EXCEPTION); } // CFA |
---|
[3848e0e] | 244 | extern { KEYWORD_RETURN(EXTERN); } |
---|
[bd7f401] | 245 | fallthrough { KEYWORD_RETURN(FALLTHROUGH); } // CFA |
---|
[35f730f] | 246 | fallthru { KEYWORD_RETURN(FALLTHRU); } // CFA |
---|
[b87a5ed] | 247 | finally { KEYWORD_RETURN(FINALLY); } // CFA |
---|
[3848e0e] | 248 | float { KEYWORD_RETURN(FLOAT); } |
---|
[e15853c] | 249 | __float80 { KEYWORD_RETURN(uuFLOAT80); } // GCC |
---|
| 250 | float80 { KEYWORD_RETURN(uuFLOAT80); } // GCC |
---|
| 251 | __float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC |
---|
| 252 | float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC |
---|
| 253 | _Float16 { FLOATXX(uFLOAT16); } // GCC |
---|
| 254 | _Float32 { FLOATXX(uFLOAT32); } // GCC |
---|
| 255 | _Float32x { FLOATXX(uFLOAT32X); } // GCC |
---|
| 256 | _Float64 { FLOATXX(uFLOAT64); } // GCC |
---|
| 257 | _Float64x { FLOATXX(uFLOAT64X); } // GCC |
---|
| 258 | _Float128 { FLOATXX(uFLOAT128); } // GCC |
---|
| 259 | _Float128x { FLOATXX(uFLOAT128); } // GCC |
---|
[b87a5ed] | 260 | for { KEYWORD_RETURN(FOR); } |
---|
| 261 | forall { KEYWORD_RETURN(FORALL); } // CFA |
---|
[3848e0e] | 262 | fortran { KEYWORD_RETURN(FORTRAN); } |
---|
[b87a5ed] | 263 | ftype { KEYWORD_RETURN(FTYPE); } // CFA |
---|
[553772b] | 264 | generator { KEYWORD_RETURN(GENERATOR); } // CFA |
---|
[b87a5ed] | 265 | _Generic { KEYWORD_RETURN(GENERIC); } // C11 |
---|
[3848e0e] | 266 | goto { KEYWORD_RETURN(GOTO); } |
---|
[b87a5ed] | 267 | if { KEYWORD_RETURN(IF); } |
---|
| 268 | _Imaginary { KEYWORD_RETURN(IMAGINARY); } // C99 |
---|
| 269 | __imag { KEYWORD_RETURN(IMAGINARY); } // GCC |
---|
| 270 | __imag__ { KEYWORD_RETURN(IMAGINARY); } // GCC |
---|
| 271 | inline { KEYWORD_RETURN(INLINE); } // C99 |
---|
| 272 | __inline { KEYWORD_RETURN(INLINE); } // GCC |
---|
| 273 | __inline__ { KEYWORD_RETURN(INLINE); } // GCC |
---|
| 274 | int { KEYWORD_RETURN(INT); } |
---|
[f1da02c] | 275 | int128 { KEYWORD_RETURN(INT128); } // CFA |
---|
[201aeb9] | 276 | __int128 { KEYWORD_RETURN(INT128); } // GCC |
---|
[f1da02c] | 277 | __int128_t { KEYWORD_RETURN(INT128); } // GCC |
---|
[b87a5ed] | 278 | __label__ { KEYWORD_RETURN(LABEL); } // GCC |
---|
[3848e0e] | 279 | long { KEYWORD_RETURN(LONG); } |
---|
[6016c87] | 280 | monitor { KEYWORD_RETURN(MONITOR); } // CFA |
---|
[a7c90d4] | 281 | mutex { KEYWORD_RETURN(MUTEX); } // CFA |
---|
[b87a5ed] | 282 | _Noreturn { KEYWORD_RETURN(NORETURN); } // C11 |
---|
[5721a6d] | 283 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); } // GCC |
---|
[3a2128f] | 284 | one_t { NUMERIC_RETURN(ONE_T); } // CFA |
---|
[ce28c7b] | 285 | or { QKEYWORD_RETURN(WOR); } // CFA |
---|
[4040425] | 286 | otype { KEYWORD_RETURN(OTYPE); } // CFA |
---|
[3848e0e] | 287 | register { KEYWORD_RETURN(REGISTER); } |
---|
[b87a5ed] | 288 | restrict { KEYWORD_RETURN(RESTRICT); } // C99 |
---|
| 289 | __restrict { KEYWORD_RETURN(RESTRICT); } // GCC |
---|
| 290 | __restrict__ { KEYWORD_RETURN(RESTRICT); } // GCC |
---|
[3848e0e] | 291 | return { KEYWORD_RETURN(RETURN); } |
---|
[f673c13c] | 292 | /* resume { KEYWORD_RETURN(RESUME); } // CFA */ |
---|
[3848e0e] | 293 | short { KEYWORD_RETURN(SHORT); } |
---|
| 294 | signed { KEYWORD_RETURN(SIGNED); } |
---|
[b87a5ed] | 295 | __signed { KEYWORD_RETURN(SIGNED); } // GCC |
---|
| 296 | __signed__ { KEYWORD_RETURN(SIGNED); } // GCC |
---|
[3848e0e] | 297 | sizeof { KEYWORD_RETURN(SIZEOF); } |
---|
| 298 | static { KEYWORD_RETURN(STATIC); } |
---|
[b87a5ed] | 299 | _Static_assert { KEYWORD_RETURN(STATICASSERT); } // C11 |
---|
[3848e0e] | 300 | struct { KEYWORD_RETURN(STRUCT); } |
---|
[f673c13c] | 301 | /* suspend { KEYWORD_RETURN(SUSPEND); } // CFA */ |
---|
[3848e0e] | 302 | switch { KEYWORD_RETURN(SWITCH); } |
---|
[bd4d011] | 303 | thread { KEYWORD_RETURN(THREAD); } // C11 |
---|
[b87a5ed] | 304 | _Thread_local { KEYWORD_RETURN(THREADLOCAL); } // C11 |
---|
| 305 | throw { KEYWORD_RETURN(THROW); } // CFA |
---|
[02e5ab6] | 306 | throwResume { KEYWORD_RETURN(THROWRESUME); } // CFA |
---|
[5b2edbc] | 307 | timeout { QKEYWORD_RETURN(TIMEOUT); } // CFA |
---|
[4040425] | 308 | trait { KEYWORD_RETURN(TRAIT); } // CFA |
---|
[b87a5ed] | 309 | try { KEYWORD_RETURN(TRY); } // CFA |
---|
[8f60f0b] | 310 | ttype { KEYWORD_RETURN(TTYPE); } // CFA |
---|
[3848e0e] | 311 | typedef { KEYWORD_RETURN(TYPEDEF); } |
---|
[b87a5ed] | 312 | typeof { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
| 313 | __typeof { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
| 314 | __typeof__ { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
[3848e0e] | 315 | union { KEYWORD_RETURN(UNION); } |
---|
[f1da02c] | 316 | __uint128_t { KEYWORD_RETURN(UINT128); } // GCC |
---|
[3848e0e] | 317 | unsigned { KEYWORD_RETURN(UNSIGNED); } |
---|
[90c3b1c] | 318 | __builtin_va_list { KEYWORD_RETURN(VALIST); } // GCC |
---|
[72457b6] | 319 | virtual { KEYWORD_RETURN(VIRTUAL); } // CFA |
---|
[3848e0e] | 320 | void { KEYWORD_RETURN(VOID); } |
---|
| 321 | volatile { KEYWORD_RETURN(VOLATILE); } |
---|
[b87a5ed] | 322 | __volatile { KEYWORD_RETURN(VOLATILE); } // GCC |
---|
| 323 | __volatile__ { KEYWORD_RETURN(VOLATILE); } // GCC |
---|
[5b2edbc] | 324 | waitfor { KEYWORD_RETURN(WAITFOR); } |
---|
| 325 | when { KEYWORD_RETURN(WHEN); } |
---|
[3848e0e] | 326 | while { KEYWORD_RETURN(WHILE); } |
---|
[8b47e50] | 327 | with { KEYWORD_RETURN(WITH); } // CFA |
---|
[3a2128f] | 328 | zero_t { NUMERIC_RETURN(ZERO_T); } // CFA |
---|
[51b7345] | 329 | |
---|
[b87a5ed] | 330 | /* identifier */ |
---|
| 331 | {identifier} { IDENTIFIER_RETURN(); } |
---|
[7e419e7] | 332 | "`"{identifier}"`" { // CFA |
---|
| 333 | yytext[yyleng - 1] = '\0'; yytext += 1; // SKULLDUGGERY: remove backquotes (ok to shorten?) |
---|
[dc7db63] | 334 | IDENTIFIER_RETURN(); |
---|
| 335 | } |
---|
[51b7345] | 336 | |
---|
[b87a5ed] | 337 | /* numeric constants */ |
---|
[0a2168f] | 338 | {binary_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[59db689] | 339 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[0a2168f] | 340 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[59db689] | 341 | {hex_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[930f69e] | 342 | {floating_decimal} { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant |
---|
| 343 | {floating_fraction} { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant |
---|
[3848e0e] | 344 | {floating_constant} { NUMERIC_RETURN(FLOATINGconstant); } |
---|
| 345 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); } |
---|
[51b7345] | 346 | |
---|
[b87a5ed] | 347 | /* character constant, allows empty value */ |
---|
[9ed4f94] | 348 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); } |
---|
[c1c1112] | 349 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); } |
---|
[dc7db63] | 350 | <QUOTE>['\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); } |
---|
[930f69e] | 351 | /* ' stop editor highlighting */ |
---|
[51b7345] | 352 | |
---|
[b87a5ed] | 353 | /* string constant */ |
---|
[9ed4f94] | 354 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); } |
---|
[c1c1112] | 355 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); } |
---|
[dc7db63] | 356 | <STRING>["\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); } |
---|
[930f69e] | 357 | /* " stop editor highlighting */ |
---|
[51b7345] | 358 | |
---|
[59db689] | 359 | /* common character/string constant */ |
---|
[c1c1112] | 360 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); } |
---|
[cd623a4] | 361 | <QUOTE,STRING>"\\"{h_white}*"\n" {} // continuation (ALSO HANDLED BY CPP) |
---|
[c1c1112] | 362 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character |
---|
[3848e0e] | 363 | |
---|
[b87a5ed] | 364 | /* punctuation */ |
---|
[615a096] | 365 | "@" { ASCIIOP_RETURN(); } |
---|
[dc7db63] | 366 | "`" { ASCIIOP_RETURN(); } |
---|
[b87a5ed] | 367 | "[" { ASCIIOP_RETURN(); } |
---|
| 368 | "]" { ASCIIOP_RETURN(); } |
---|
| 369 | "(" { ASCIIOP_RETURN(); } |
---|
| 370 | ")" { ASCIIOP_RETURN(); } |
---|
| 371 | "{" { ASCIIOP_RETURN(); } |
---|
| 372 | "}" { ASCIIOP_RETURN(); } |
---|
| 373 | "," { ASCIIOP_RETURN(); } // also operator |
---|
| 374 | ":" { ASCIIOP_RETURN(); } |
---|
| 375 | ";" { ASCIIOP_RETURN(); } |
---|
| 376 | "." { ASCIIOP_RETURN(); } // also operator |
---|
[3848e0e] | 377 | "..." { NAMEDOP_RETURN(ELLIPSIS); } |
---|
| 378 | |
---|
[b87a5ed] | 379 | /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */ |
---|
[3848e0e] | 380 | "<:" { RETURN_VAL('['); } |
---|
| 381 | ":>" { RETURN_VAL(']'); } |
---|
| 382 | "<%" { RETURN_VAL('{'); } |
---|
| 383 | "%>" { RETURN_VAL('}'); } |
---|
[51b7345] | 384 | |
---|
[b87a5ed] | 385 | /* operators */ |
---|
| 386 | "!" { ASCIIOP_RETURN(); } |
---|
| 387 | "+" { ASCIIOP_RETURN(); } |
---|
| 388 | "-" { ASCIIOP_RETURN(); } |
---|
| 389 | "*" { ASCIIOP_RETURN(); } |
---|
[e5f2a67] | 390 | "\\" { ASCIIOP_RETURN(); } // CFA, exponentiation |
---|
[b87a5ed] | 391 | "/" { ASCIIOP_RETURN(); } |
---|
| 392 | "%" { ASCIIOP_RETURN(); } |
---|
| 393 | "^" { ASCIIOP_RETURN(); } |
---|
| 394 | "~" { ASCIIOP_RETURN(); } |
---|
| 395 | "&" { ASCIIOP_RETURN(); } |
---|
| 396 | "|" { ASCIIOP_RETURN(); } |
---|
| 397 | "<" { ASCIIOP_RETURN(); } |
---|
| 398 | ">" { ASCIIOP_RETURN(); } |
---|
| 399 | "=" { ASCIIOP_RETURN(); } |
---|
| 400 | "?" { ASCIIOP_RETURN(); } |
---|
[3848e0e] | 401 | |
---|
| 402 | "++" { NAMEDOP_RETURN(ICR); } |
---|
| 403 | "--" { NAMEDOP_RETURN(DECR); } |
---|
| 404 | "==" { NAMEDOP_RETURN(EQ); } |
---|
| 405 | "!=" { NAMEDOP_RETURN(NE); } |
---|
| 406 | "<<" { NAMEDOP_RETURN(LS); } |
---|
| 407 | ">>" { NAMEDOP_RETURN(RS); } |
---|
| 408 | "<=" { NAMEDOP_RETURN(LE); } |
---|
| 409 | ">=" { NAMEDOP_RETURN(GE); } |
---|
| 410 | "&&" { NAMEDOP_RETURN(ANDAND); } |
---|
| 411 | "||" { NAMEDOP_RETURN(OROR); } |
---|
| 412 | "->" { NAMEDOP_RETURN(ARROW); } |
---|
| 413 | "+=" { NAMEDOP_RETURN(PLUSassign); } |
---|
| 414 | "-=" { NAMEDOP_RETURN(MINUSassign); } |
---|
[e5f2a67] | 415 | "\\=" { NAMEDOP_RETURN(EXPassign); } // CFA, exponentiation |
---|
[3848e0e] | 416 | "*=" { NAMEDOP_RETURN(MULTassign); } |
---|
| 417 | "/=" { NAMEDOP_RETURN(DIVassign); } |
---|
| 418 | "%=" { NAMEDOP_RETURN(MODassign); } |
---|
| 419 | "&=" { NAMEDOP_RETURN(ANDassign); } |
---|
| 420 | "|=" { NAMEDOP_RETURN(ORassign); } |
---|
| 421 | "^=" { NAMEDOP_RETURN(ERassign); } |
---|
| 422 | "<<=" { NAMEDOP_RETURN(LSassign); } |
---|
| 423 | ">>=" { NAMEDOP_RETURN(RSassign); } |
---|
[51b7345] | 424 | |
---|
[08061589] | 425 | "@=" { NAMEDOP_RETURN(ATassign); } // CFA |
---|
[d69f4bb4] | 426 | "~=" { NAMEDOP_RETURN(ErangeUpEq); } // CFA |
---|
| 427 | "-~" { NAMEDOP_RETURN(ErangeDown); } // CFA |
---|
| 428 | "-~=" { NAMEDOP_RETURN(ErangeDownEq); } // CFA |
---|
[097e2b0] | 429 | |
---|
[b87a5ed] | 430 | /* CFA, operator identifier */ |
---|
| 431 | {op_unary}"?" { IDENTIFIER_RETURN(); } // unary |
---|
[a61fea9a] | 432 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); } |
---|
[76c62b2] | 433 | "^?{}" { IDENTIFIER_RETURN(); } |
---|
[7e419e7] | 434 | "?`"{identifier} { IDENTIFIER_RETURN(); } // postfix operator |
---|
[b87a5ed] | 435 | "?"{op_binary_over}"?" { IDENTIFIER_RETURN(); } // binary |
---|
[51b7345] | 436 | /* |
---|
[daf9671] | 437 | This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?" can be |
---|
| 438 | lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier, |
---|
| 439 | e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is |
---|
| 440 | a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then |
---|
| 441 | reparsing the trailing operator identifier. Otherwise a space is needed between the unary operator and operator |
---|
| 442 | identifier to disambiguate this common case. |
---|
| 443 | |
---|
| 444 | A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers. The ambiguity |
---|
| 445 | occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace |
---|
| 446 | look-ahead for the routine-call parameter-list to disambiguate. However, the dereference operator must have a |
---|
| 447 | parameter/argument to dereference *?(...). Hence, always interpreting the string *?() as * ?() does not preclude |
---|
| 448 | any meaningful program. |
---|
| 449 | |
---|
| 450 | The remaining cases are with the increment/decrement operators and conditional expression: |
---|
| 451 | |
---|
| 452 | i++? ...(...); |
---|
| 453 | i?++ ...(...); |
---|
| 454 | |
---|
| 455 | requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an |
---|
| 456 | incorrect expression (juxtaposed identifiers). Therefore, it is necessary to disambiguate these cases with a |
---|
| 457 | space: |
---|
| 458 | |
---|
| 459 | i++ ? i : 0; |
---|
| 460 | i? ++i : 0; |
---|
[51b7345] | 461 | */ |
---|
[daf9671] | 462 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") { |
---|
[b87a5ed] | 463 | // 1 or 2 character unary operator ? |
---|
| 464 | int i = yytext[1] == '?' ? 1 : 2; |
---|
| 465 | yyless( i ); // put back characters up to first '?' |
---|
| 466 | if ( i > 1 ) { |
---|
| 467 | NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR ); |
---|
| 468 | } else { |
---|
| 469 | ASCIIOP_RETURN(); |
---|
| 470 | } // if |
---|
| 471 | } |
---|
| 472 | |
---|
[9ed4f94] | 473 | /* unknown character */ |
---|
| 474 | . { yyerror( "unknown character" ); } |
---|
[51b7345] | 475 | |
---|
| 476 | %% |
---|
[8551b859] | 477 | |
---|
[9ed4f94] | 478 | // ----end of lexer---- |
---|
| 479 | |
---|
| 480 | void yyerror( const char * errmsg ) { |
---|
[8551b859] | 481 | SemanticErrorThrow = true; |
---|
[a8615fd1] | 482 | cerr << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1 |
---|
[a16764a6] | 483 | << ": " << ErrorHelpers::error_str() << errmsg << " at token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl; |
---|
[9ed4f94] | 484 | } |
---|
[51b7345] | 485 | |
---|
[b87a5ed] | 486 | // Local Variables: // |
---|
| 487 | // mode: c++ // |
---|
[de62360d] | 488 | // tab-width: 4 // |
---|
[b87a5ed] | 489 | // compile-command: "make install" // |
---|
| 490 | // End: // |
---|