[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 |
---|
[01fac29] | 12 | * Last Modified On : Fri Jun 9 10:04:00 2023 |
---|
| 13 | * Update Count : 770 |
---|
[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 | |
---|
[bb7422a] | 25 | // *************************** Includes and Defines **************************** |
---|
[51b7345] | 26 | |
---|
[0bd46fd] | 27 | #ifdef __clang__ |
---|
| 28 | #pragma GCC diagnostic ignored "-Wnull-conversion" |
---|
| 29 | #endif |
---|
| 30 | |
---|
[4439008] | 31 | // trigger before each matching rule's action |
---|
| 32 | #define YY_USER_ACTION \ |
---|
| 33 | yylloc.first_line = yylineno; \ |
---|
| 34 | yylloc.first_column = column; \ |
---|
| 35 | column += yyleng; \ |
---|
| 36 | yylloc.last_column = column; \ |
---|
| 37 | yylloc.last_line = yylineno; \ |
---|
| 38 | yylloc.filename = yyfilename ? yyfilename : ""; |
---|
[9ed4f94] | 39 | unsigned int column = 0; // position of the end of the last token parsed |
---|
| 40 | |
---|
[51b7345] | 41 | #include <string> |
---|
[f487962] | 42 | #include <cstdio> // FILENAME_MAX |
---|
[9ed4f94] | 43 | using namespace std; |
---|
[51b7345] | 44 | |
---|
[ba01b14] | 45 | #include "config.h" // configure info |
---|
[c468150] | 46 | #include "DeclarationNode.h" // for DeclarationNode |
---|
| 47 | #include "ExpressionNode.h" // for LabelNode |
---|
| 48 | #include "InitializerNode.h" // for InitializerNode |
---|
[984dce6] | 49 | #include "ParseNode.h" |
---|
[c468150] | 50 | #include "ParserTypes.h" // for Token |
---|
| 51 | #include "StatementNode.h" // for CondCtl, ForCtrl |
---|
[984dce6] | 52 | #include "TypedefTable.h" |
---|
[c468150] | 53 | // This (generated) header must come late as it is missing includes. |
---|
| 54 | #include "parser.hh" // generated info |
---|
[51b7345] | 55 | |
---|
[60a8062] | 56 | string * build_postfix_name( string * name ); |
---|
| 57 | |
---|
[51b7345] | 58 | char *yyfilename; |
---|
[9ed4f94] | 59 | string *strtext; // accumulate parts of character and string constant value |
---|
[51b7345] | 60 | |
---|
[de62360d] | 61 | #define RETURN_LOCN(x) yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x ) |
---|
[9ed4f94] | 62 | #define RETURN_VAL(x) yylval.tok.str = new string( yytext ); RETURN_LOCN( x ) |
---|
[926af74] | 63 | #define RETURN_CHAR(x) yylval.tok.str = nullptr; RETURN_LOCN( x ) |
---|
[de62360d] | 64 | #define RETURN_STR(x) yylval.tok.str = strtext; RETURN_LOCN( x ) |
---|
[5f2f2d7] | 65 | |
---|
[f487962] | 66 | #define WHITE_RETURN(x) // do nothing |
---|
[9ed4f94] | 67 | #define NEWLINE_RETURN() column = 0; WHITE_RETURN( '\n' ) |
---|
[de62360d] | 68 | #define ASCIIOP_RETURN() RETURN_CHAR( (int)yytext[0] ) // single character operator |
---|
[f487962] | 69 | #define NAMEDOP_RETURN(x) RETURN_CHAR( x ) // multichar operator, with a name |
---|
[de62360d] | 70 | #define NUMERIC_RETURN(x) rm_underscore(); RETURN_VAL( x ) // numeric constant |
---|
| 71 | #define KEYWORD_RETURN(x) RETURN_CHAR( x ) // keyword |
---|
[4439008] | 72 | #define QKEYWORD_RETURN(x) RETURN_VAL(x); // quasi-keyword |
---|
[984dce6] | 73 | #define IDENTIFIER_RETURN() RETURN_VAL( typedefTable.isKind( yytext ) ) |
---|
[51b7345] | 74 | |
---|
[25744d2] | 75 | #ifdef HAVE_KEYWORDS_FLOATXX // GCC >= 7 => keyword, otherwise typedef |
---|
[ba01b14] | 76 | #define FLOATXX(v) KEYWORD_RETURN(v); |
---|
| 77 | #else |
---|
[37cdd97] | 78 | #define FLOATXX(v) IDENTIFIER_RETURN(); |
---|
[ba01b14] | 79 | #endif // HAVE_KEYWORDS_FLOATXX |
---|
| 80 | |
---|
[3848e0e] | 81 | void rm_underscore() { |
---|
[7e419e7] | 82 | // SKULLDUGGERY: remove underscores (ok to shorten?) |
---|
[e7aed49] | 83 | yyleng = 0; |
---|
[7e419e7] | 84 | for ( int i = 0; yytext[i] != '\0'; i += 1 ) { // copying non-underscore characters to front of string |
---|
[b87a5ed] | 85 | if ( yytext[i] != '_' ) { |
---|
[e7aed49] | 86 | yytext[yyleng] = yytext[i]; |
---|
| 87 | yyleng += 1; |
---|
[b87a5ed] | 88 | } // if |
---|
| 89 | } // for |
---|
| 90 | yytext[yyleng] = '\0'; |
---|
[7e419e7] | 91 | } // rm_underscore |
---|
[51b7345] | 92 | |
---|
[7812f1d1] | 93 | // Stop warning due to incorrectly generated flex code. |
---|
| 94 | #pragma GCC diagnostic ignored "-Wsign-compare" |
---|
[51b7345] | 95 | %} |
---|
| 96 | |
---|
[0a2168f] | 97 | binary [0-1] |
---|
[51b7345] | 98 | octal [0-7] |
---|
| 99 | nonzero [1-9] |
---|
| 100 | decimal [0-9] |
---|
| 101 | hex [0-9a-fA-F] |
---|
[3848e0e] | 102 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2})) |
---|
[51b7345] | 103 | |
---|
[b87a5ed] | 104 | // identifier, GCC: $ in identifier |
---|
[51b7345] | 105 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})* |
---|
| 106 | |
---|
[b87a5ed] | 107 | // numeric constants, CFA: '_' in constant |
---|
[3848e0e] | 108 | hex_quad {hex}("_"?{hex}){3} |
---|
[201aeb9] | 109 | size_opt (8|16|32|64|128)? |
---|
[0a616e0] | 110 | // CFA: explicit l8/l16/l32/l64/l128, char 'hh', short 'h', int 'n' |
---|
[f56c32e] | 111 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hHnN]) |
---|
[0a616e0] | 112 | // CFA: size_t 'z', pointer 'p', which define a sign and length |
---|
| 113 | integer_suffix_opt ("_"?(([uU]({length}?[iI]?)|([iI]{length}))|([iI]({length}?[uU]?)|([uU]{length}))|({length}([iI]?[uU]?)|([uU][iI]))|[zZ]|[pP]))? |
---|
[51b7345] | 114 | |
---|
| 115 | octal_digits ({octal})|({octal}({octal}|"_")*{octal}) |
---|
| 116 | octal_prefix "0""_"? |
---|
[930f69e] | 117 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt} |
---|
[51b7345] | 118 | |
---|
| 119 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal}) |
---|
[930f69e] | 120 | decimal_constant {nonzero_digits}{integer_suffix_opt} |
---|
[51b7345] | 121 | |
---|
[0a2168f] | 122 | binary_digits ({binary})|({binary}({binary}|"_")*{binary}) |
---|
| 123 | binary_prefix "0"[bB]"_"? |
---|
| 124 | binary_constant {binary_prefix}{binary_digits}{integer_suffix_opt} |
---|
| 125 | |
---|
[51b7345] | 126 | hex_digits ({hex})|({hex}({hex}|"_")*{hex}) |
---|
| 127 | hex_prefix "0"[xX]"_"? |
---|
[930f69e] | 128 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt} |
---|
[51b7345] | 129 | |
---|
[d3b5a14] | 130 | // GCC: floating D (double), imaginary iI, and decimal floating DF, DD, DL |
---|
[51b7345] | 131 | exponent "_"?[eE]"_"?[+-]?{decimal_digits} |
---|
[ba01b14] | 132 | floating_size 16|32|32x|64|64x|80|128|128x |
---|
| 133 | floating_length ([fFdDlLwWqQ]|[fF]{floating_size}) |
---|
[201aeb9] | 134 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length}) |
---|
[d3b5a14] | 135 | decimal_floating_suffix [dD][fFdDlL] |
---|
| 136 | floating_suffix_opt ("_"?({floating_suffix}|{decimal_floating_suffix}))? |
---|
[f4b77f2] | 137 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal}) |
---|
[930f69e] | 138 | floating_decimal {decimal_digits}"."{exponent}?{floating_suffix_opt} |
---|
| 139 | floating_fraction "."{decimal_digits}{exponent}?{floating_suffix_opt} |
---|
| 140 | floating_constant ({decimal_digits}{exponent}{floating_suffix_opt})|({decimal_digits}{floating_fraction}) |
---|
[51b7345] | 141 | |
---|
| 142 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits} |
---|
[dc7db63] | 143 | hex_floating_suffix_opt ("_"?({floating_suffix}))? |
---|
[930f69e] | 144 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".") |
---|
[f4b77f2] | 145 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt} |
---|
[51b7345] | 146 | |
---|
[b87a5ed] | 147 | // character escape sequence, GCC: \e => esc character |
---|
[51b7345] | 148 | simple_escape "\\"[abefnrtv'"?\\] |
---|
[930f69e] | 149 | // ' stop editor highlighting |
---|
[3848e0e] | 150 | octal_escape "\\"{octal}("_"?{octal}){0,2} |
---|
| 151 | hex_escape "\\""x""_"?{hex_digits} |
---|
[51b7345] | 152 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char} |
---|
[59db689] | 153 | cwide_prefix "L"|"U"|"u" |
---|
| 154 | swide_prefix {cwide_prefix}|"u8" |
---|
[51b7345] | 155 | |
---|
[b87a5ed] | 156 | // display/white-space characters |
---|
[51b7345] | 157 | h_tab [\011] |
---|
| 158 | form_feed [\014] |
---|
| 159 | v_tab [\013] |
---|
| 160 | c_return [\015] |
---|
| 161 | h_white [ ]|{h_tab} |
---|
| 162 | |
---|
[e7aed49] | 163 | // overloadable operators |
---|
[51b7345] | 164 | op_unary_only "~"|"!" |
---|
| 165 | op_unary_binary "+"|"-"|"*" |
---|
| 166 | op_unary_pre_post "++"|"--" |
---|
| 167 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post} |
---|
| 168 | |
---|
[994d080] | 169 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>=" |
---|
[51b7345] | 170 | op_binary_over {op_unary_binary}|{op_binary_only} |
---|
[e7aed49] | 171 | // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@=" |
---|
| 172 | // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over} |
---|
[51b7345] | 173 | |
---|
| 174 | %x COMMENT |
---|
[3848e0e] | 175 | %x BKQUOTE |
---|
| 176 | %x QUOTE |
---|
| 177 | %x STRING |
---|
[51b7345] | 178 | |
---|
| 179 | %% |
---|
[994d080] | 180 | /* line directives */ |
---|
[4040425] | 181 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" { |
---|
[930f69e] | 182 | /* " stop editor highlighting */ |
---|
[c1c1112] | 183 | static char filename[FILENAME_MAX]; // temporarily store current source-file name |
---|
[51b7345] | 184 | char *end_num; |
---|
| 185 | char *begin_string, *end_string; |
---|
| 186 | long lineno, length; |
---|
| 187 | lineno = strtol( yytext + 1, &end_num, 0 ); |
---|
| 188 | begin_string = strchr( end_num, '"' ); |
---|
[f487962] | 189 | if ( begin_string ) { // file name ? |
---|
| 190 | end_string = strchr( begin_string + 1, '"' ); // look for ending delimiter |
---|
| 191 | assert( end_string ); // closing quote ? |
---|
| 192 | length = end_string - begin_string - 1; // file-name length without quotes or sentinel |
---|
| 193 | assert( length < FILENAME_MAX ); // room for sentinel ? |
---|
| 194 | memcpy( &filename, begin_string + 1, length ); // copy file name from yytext |
---|
| 195 | filename[ length ] = '\0'; // terminate string with sentinel |
---|
[9ed4f94] | 196 | //cout << "file " << filename << " line " << lineno << endl; |
---|
[f487962] | 197 | yylineno = lineno; |
---|
[c1c1112] | 198 | yyfilename = filename; |
---|
[b87a5ed] | 199 | } // if |
---|
[51b7345] | 200 | } |
---|
| 201 | |
---|
[637dd9c] | 202 | /* preprocessor-style directives */ |
---|
[61fc4f6] | 203 | ^{h_white}*"#"[^\n]*"\n" { RETURN_VAL( DIRECTIVE ); } |
---|
[51b7345] | 204 | |
---|
[cd623a4] | 205 | /* ignore C style comments (ALSO HANDLED BY CPP) */ |
---|
[3848e0e] | 206 | "/*" { BEGIN COMMENT; } |
---|
[cd623a4] | 207 | <COMMENT>.|\n ; |
---|
| 208 | <COMMENT>"*/" { BEGIN 0; } |
---|
[51b7345] | 209 | |
---|
[cd623a4] | 210 | /* ignore C++ style comments (ALSO HANDLED BY CPP) */ |
---|
| 211 | "//"[^\n]*"\n" ; |
---|
[51b7345] | 212 | |
---|
[b87a5ed] | 213 | /* ignore whitespace */ |
---|
[3848e0e] | 214 | {h_white}+ { WHITE_RETURN(' '); } |
---|
| 215 | ({v_tab}|{c_return}|{form_feed})+ { WHITE_RETURN(' '); } |
---|
| 216 | ({h_white}|{v_tab}|{c_return}|{form_feed})*"\n" { NEWLINE_RETURN(); } |
---|
[51b7345] | 217 | |
---|
[b87a5ed] | 218 | /* keywords */ |
---|
| 219 | _Alignas { KEYWORD_RETURN(ALIGNAS); } // C11 |
---|
| 220 | _Alignof { KEYWORD_RETURN(ALIGNOF); } // C11 |
---|
| 221 | __alignof { KEYWORD_RETURN(ALIGNOF); } // GCC |
---|
| 222 | __alignof__ { KEYWORD_RETURN(ALIGNOF); } // GCC |
---|
[9fd9d015] | 223 | and { QKEYWORD_RETURN(WAND); } // CFA |
---|
[b87a5ed] | 224 | asm { KEYWORD_RETURN(ASM); } |
---|
| 225 | __asm { KEYWORD_RETURN(ASM); } // GCC |
---|
| 226 | __asm__ { KEYWORD_RETURN(ASM); } // GCC |
---|
| 227 | _Atomic { KEYWORD_RETURN(ATOMIC); } // C11 |
---|
| 228 | __attribute { KEYWORD_RETURN(ATTRIBUTE); } // GCC |
---|
| 229 | __attribute__ { KEYWORD_RETURN(ATTRIBUTE); } // GCC |
---|
[3848e0e] | 230 | auto { KEYWORD_RETURN(AUTO); } |
---|
[f673c13c] | 231 | __auto_type { KEYWORD_RETURN(AUTO_TYPE); } |
---|
[b6ad601] | 232 | basetypeof { KEYWORD_RETURN(BASETYPEOF); } // CFA |
---|
[b87a5ed] | 233 | _Bool { KEYWORD_RETURN(BOOL); } // C99 |
---|
[3848e0e] | 234 | break { KEYWORD_RETURN(BREAK); } |
---|
| 235 | case { KEYWORD_RETURN(CASE); } |
---|
[9fb1367] | 236 | catch { QKEYWORD_RETURN(CATCH); } // CFA |
---|
| 237 | catchResume { QKEYWORD_RETURN(CATCHRESUME); } // CFA |
---|
[3848e0e] | 238 | char { KEYWORD_RETURN(CHAR); } |
---|
[b87a5ed] | 239 | choose { KEYWORD_RETURN(CHOOSE); } // CFA |
---|
[1528a2c] | 240 | coerce { KEYWORD_RETURN(COERCE); } // CFA |
---|
[b87a5ed] | 241 | _Complex { KEYWORD_RETURN(COMPLEX); } // C99 |
---|
| 242 | __complex { KEYWORD_RETURN(COMPLEX); } // GCC |
---|
| 243 | __complex__ { KEYWORD_RETURN(COMPLEX); } // GCC |
---|
[3848e0e] | 244 | const { KEYWORD_RETURN(CONST); } |
---|
[b87a5ed] | 245 | __const { KEYWORD_RETURN(CONST); } // GCC |
---|
| 246 | __const__ { KEYWORD_RETURN(CONST); } // GCC |
---|
[3848e0e] | 247 | continue { KEYWORD_RETURN(CONTINUE); } |
---|
[e04b636] | 248 | coroutine { KEYWORD_RETURN(COROUTINE); } // CFA |
---|
[15f769c] | 249 | _Decimal32 { KEYWORD_RETURN(DECIMAL32); } // GCC |
---|
| 250 | _Decimal64 { KEYWORD_RETURN(DECIMAL64); } // GCC |
---|
| 251 | _Decimal128 { KEYWORD_RETURN(DECIMAL128); } // GCC |
---|
[3848e0e] | 252 | default { KEYWORD_RETURN(DEFAULT); } |
---|
[02e5ab6] | 253 | disable { KEYWORD_RETURN(DISABLE); } // CFA |
---|
[b87a5ed] | 254 | do { KEYWORD_RETURN(DO); } |
---|
[3848e0e] | 255 | double { KEYWORD_RETURN(DOUBLE); } |
---|
[b87a5ed] | 256 | dtype { KEYWORD_RETURN(DTYPE); } // CFA |
---|
[3848e0e] | 257 | else { KEYWORD_RETURN(ELSE); } |
---|
[02e5ab6] | 258 | enable { KEYWORD_RETURN(ENABLE); } // CFA |
---|
[3848e0e] | 259 | enum { KEYWORD_RETURN(ENUM); } |
---|
[c27fb59] | 260 | exception { KEYWORD_RETURN(EXCEPTION); } // CFA |
---|
[8c25760] | 261 | __extension__ { KEYWORD_RETURN(EXTENSION); } // GCC |
---|
[3848e0e] | 262 | extern { KEYWORD_RETURN(EXTERN); } |
---|
[bd7f401] | 263 | fallthrough { KEYWORD_RETURN(FALLTHROUGH); } // CFA |
---|
[35f730f] | 264 | fallthru { KEYWORD_RETURN(FALLTHRU); } // CFA |
---|
[9fb1367] | 265 | finally { QKEYWORD_RETURN(FINALLY); } // CFA |
---|
| 266 | fixup { QKEYWORD_RETURN(FIXUP); } // CFA |
---|
[3848e0e] | 267 | float { KEYWORD_RETURN(FLOAT); } |
---|
[e15853c] | 268 | __float80 { KEYWORD_RETURN(uuFLOAT80); } // GCC |
---|
| 269 | float80 { KEYWORD_RETURN(uuFLOAT80); } // GCC |
---|
| 270 | __float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC |
---|
| 271 | float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC |
---|
| 272 | _Float16 { FLOATXX(uFLOAT16); } // GCC |
---|
| 273 | _Float32 { FLOATXX(uFLOAT32); } // GCC |
---|
| 274 | _Float32x { FLOATXX(uFLOAT32X); } // GCC |
---|
| 275 | _Float64 { FLOATXX(uFLOAT64); } // GCC |
---|
| 276 | _Float64x { FLOATXX(uFLOAT64X); } // GCC |
---|
| 277 | _Float128 { FLOATXX(uFLOAT128); } // GCC |
---|
| 278 | _Float128x { FLOATXX(uFLOAT128); } // GCC |
---|
[b87a5ed] | 279 | for { KEYWORD_RETURN(FOR); } |
---|
| 280 | forall { KEYWORD_RETURN(FORALL); } // CFA |
---|
[3848e0e] | 281 | fortran { KEYWORD_RETURN(FORTRAN); } |
---|
[b87a5ed] | 282 | ftype { KEYWORD_RETURN(FTYPE); } // CFA |
---|
[553772b] | 283 | generator { KEYWORD_RETURN(GENERATOR); } // CFA |
---|
[b87a5ed] | 284 | _Generic { KEYWORD_RETURN(GENERIC); } // C11 |
---|
[3848e0e] | 285 | goto { KEYWORD_RETURN(GOTO); } |
---|
[b87a5ed] | 286 | if { KEYWORD_RETURN(IF); } |
---|
| 287 | _Imaginary { KEYWORD_RETURN(IMAGINARY); } // C99 |
---|
| 288 | __imag { KEYWORD_RETURN(IMAGINARY); } // GCC |
---|
| 289 | __imag__ { KEYWORD_RETURN(IMAGINARY); } // GCC |
---|
| 290 | inline { KEYWORD_RETURN(INLINE); } // C99 |
---|
| 291 | __inline { KEYWORD_RETURN(INLINE); } // GCC |
---|
| 292 | __inline__ { KEYWORD_RETURN(INLINE); } // GCC |
---|
| 293 | int { KEYWORD_RETURN(INT); } |
---|
[f1da02c] | 294 | int128 { KEYWORD_RETURN(INT128); } // CFA |
---|
[201aeb9] | 295 | __int128 { KEYWORD_RETURN(INT128); } // GCC |
---|
[f1da02c] | 296 | __int128_t { KEYWORD_RETURN(INT128); } // GCC |
---|
[b87a5ed] | 297 | __label__ { KEYWORD_RETURN(LABEL); } // GCC |
---|
[3848e0e] | 298 | long { KEYWORD_RETURN(LONG); } |
---|
[6016c87] | 299 | monitor { KEYWORD_RETURN(MONITOR); } // CFA |
---|
[a7c90d4] | 300 | mutex { KEYWORD_RETURN(MUTEX); } // CFA |
---|
[b87a5ed] | 301 | _Noreturn { KEYWORD_RETURN(NORETURN); } // C11 |
---|
[5721a6d] | 302 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); } // GCC |
---|
[3a2128f] | 303 | one_t { NUMERIC_RETURN(ONE_T); } // CFA |
---|
[ce28c7b] | 304 | or { QKEYWORD_RETURN(WOR); } // CFA |
---|
[4040425] | 305 | otype { KEYWORD_RETURN(OTYPE); } // CFA |
---|
[9fb1367] | 306 | recover { QKEYWORD_RETURN(RECOVER); } // CFA |
---|
[3848e0e] | 307 | register { KEYWORD_RETURN(REGISTER); } |
---|
[9fb1367] | 308 | report { KEYWORD_RETURN(THROWRESUME); } // CFA |
---|
[b87a5ed] | 309 | restrict { KEYWORD_RETURN(RESTRICT); } // C99 |
---|
| 310 | __restrict { KEYWORD_RETURN(RESTRICT); } // GCC |
---|
| 311 | __restrict__ { KEYWORD_RETURN(RESTRICT); } // GCC |
---|
[3848e0e] | 312 | return { KEYWORD_RETURN(RETURN); } |
---|
[25744d2] | 313 | /* resume { KEYWORD_RETURN(RESUME); } // CFA */ |
---|
[3848e0e] | 314 | short { KEYWORD_RETURN(SHORT); } |
---|
| 315 | signed { KEYWORD_RETURN(SIGNED); } |
---|
[b87a5ed] | 316 | __signed { KEYWORD_RETURN(SIGNED); } // GCC |
---|
| 317 | __signed__ { KEYWORD_RETURN(SIGNED); } // GCC |
---|
[3848e0e] | 318 | sizeof { KEYWORD_RETURN(SIZEOF); } |
---|
| 319 | static { KEYWORD_RETURN(STATIC); } |
---|
[b87a5ed] | 320 | _Static_assert { KEYWORD_RETURN(STATICASSERT); } // C11 |
---|
[01fac29] | 321 | _static_assert { KEYWORD_RETURN(STATICASSERT); } // C23 |
---|
[3848e0e] | 322 | struct { KEYWORD_RETURN(STRUCT); } |
---|
[37cdd97] | 323 | suspend { KEYWORD_RETURN(SUSPEND); } // CFA |
---|
[3848e0e] | 324 | switch { KEYWORD_RETURN(SWITCH); } |
---|
[bd4d011] | 325 | thread { KEYWORD_RETURN(THREAD); } // C11 |
---|
[ed9a1ae] | 326 | __thread { KEYWORD_RETURN(THREADLOCALGCC); } // GCC |
---|
| 327 | _Thread_local { KEYWORD_RETURN(THREADLOCALC11); } // C11 |
---|
[b87a5ed] | 328 | throw { KEYWORD_RETURN(THROW); } // CFA |
---|
[02e5ab6] | 329 | throwResume { KEYWORD_RETURN(THROWRESUME); } // CFA |
---|
[5b2edbc] | 330 | timeout { QKEYWORD_RETURN(TIMEOUT); } // CFA |
---|
[4040425] | 331 | trait { KEYWORD_RETURN(TRAIT); } // CFA |
---|
[b87a5ed] | 332 | try { KEYWORD_RETURN(TRY); } // CFA |
---|
[8f60f0b] | 333 | ttype { KEYWORD_RETURN(TTYPE); } // CFA |
---|
[3848e0e] | 334 | typedef { KEYWORD_RETURN(TYPEDEF); } |
---|
[b87a5ed] | 335 | typeof { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
| 336 | __typeof { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
| 337 | __typeof__ { KEYWORD_RETURN(TYPEOF); } // GCC |
---|
[1f652a7] | 338 | typeid { KEYWORD_RETURN(TYPEID); } // GCC |
---|
[3848e0e] | 339 | union { KEYWORD_RETURN(UNION); } |
---|
[f1da02c] | 340 | __uint128_t { KEYWORD_RETURN(UINT128); } // GCC |
---|
[3848e0e] | 341 | unsigned { KEYWORD_RETURN(UNSIGNED); } |
---|
[59c7e3e] | 342 | __builtin_va_arg { KEYWORD_RETURN(VA_ARG); } // GCC |
---|
| 343 | __builtin_va_list { KEYWORD_RETURN(VA_LIST); } // GCC |
---|
[72457b6] | 344 | virtual { KEYWORD_RETURN(VIRTUAL); } // CFA |
---|
[3848e0e] | 345 | void { KEYWORD_RETURN(VOID); } |
---|
| 346 | volatile { KEYWORD_RETURN(VOLATILE); } |
---|
[b87a5ed] | 347 | __volatile { KEYWORD_RETURN(VOLATILE); } // GCC |
---|
| 348 | __volatile__ { KEYWORD_RETURN(VOLATILE); } // GCC |
---|
[1f652a7] | 349 | vtable { KEYWORD_RETURN(VTABLE); } // CFA |
---|
[9fb1367] | 350 | waitfor { KEYWORD_RETURN(WAITFOR); } // CFA |
---|
[466787a] | 351 | waituntil { KEYWORD_RETURN(WAITUNTIL); } // CFA |
---|
[9fb1367] | 352 | when { KEYWORD_RETURN(WHEN); } // CFA |
---|
[3848e0e] | 353 | while { KEYWORD_RETURN(WHILE); } |
---|
[8b47e50] | 354 | with { KEYWORD_RETURN(WITH); } // CFA |
---|
[3a2128f] | 355 | zero_t { NUMERIC_RETURN(ZERO_T); } // CFA |
---|
[51b7345] | 356 | |
---|
[b87a5ed] | 357 | /* identifier */ |
---|
| 358 | {identifier} { IDENTIFIER_RETURN(); } |
---|
[095ac99] | 359 | "``"{identifier} { // CFA |
---|
| 360 | yytext[yyleng] = '\0'; yytext += 2; // SKULLDUGGERY: remove backquotes (ok to shorten?) |
---|
[dc7db63] | 361 | IDENTIFIER_RETURN(); |
---|
| 362 | } |
---|
[51b7345] | 363 | |
---|
[b87a5ed] | 364 | /* numeric constants */ |
---|
[0a2168f] | 365 | {binary_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[59db689] | 366 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[0a2168f] | 367 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[59db689] | 368 | {hex_constant} { NUMERIC_RETURN(INTEGERconstant); } |
---|
[930f69e] | 369 | {floating_decimal} { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant |
---|
| 370 | {floating_fraction} { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant |
---|
[3848e0e] | 371 | {floating_constant} { NUMERIC_RETURN(FLOATINGconstant); } |
---|
| 372 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); } |
---|
[51b7345] | 373 | |
---|
[b87a5ed] | 374 | /* character constant, allows empty value */ |
---|
[9ed4f94] | 375 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); } |
---|
[c1c1112] | 376 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); } |
---|
[dc7db63] | 377 | <QUOTE>['\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); } |
---|
[930f69e] | 378 | /* ' stop editor highlighting */ |
---|
[51b7345] | 379 | |
---|
[b87a5ed] | 380 | /* string constant */ |
---|
[9ed4f94] | 381 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); } |
---|
[c1c1112] | 382 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); } |
---|
[dc7db63] | 383 | <STRING>["\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); } |
---|
[930f69e] | 384 | /* " stop editor highlighting */ |
---|
[51b7345] | 385 | |
---|
[59db689] | 386 | /* common character/string constant */ |
---|
[c1c1112] | 387 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); } |
---|
[cd623a4] | 388 | <QUOTE,STRING>"\\"{h_white}*"\n" {} // continuation (ALSO HANDLED BY CPP) |
---|
[c1c1112] | 389 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character |
---|
[3848e0e] | 390 | |
---|
[b87a5ed] | 391 | /* punctuation */ |
---|
[615a096] | 392 | "@" { ASCIIOP_RETURN(); } |
---|
[dc7db63] | 393 | "`" { ASCIIOP_RETURN(); } |
---|
[b87a5ed] | 394 | "[" { ASCIIOP_RETURN(); } |
---|
| 395 | "]" { ASCIIOP_RETURN(); } |
---|
| 396 | "(" { ASCIIOP_RETURN(); } |
---|
| 397 | ")" { ASCIIOP_RETURN(); } |
---|
| 398 | "{" { ASCIIOP_RETURN(); } |
---|
| 399 | "}" { ASCIIOP_RETURN(); } |
---|
| 400 | "," { ASCIIOP_RETURN(); } // also operator |
---|
| 401 | ":" { ASCIIOP_RETURN(); } |
---|
| 402 | ";" { ASCIIOP_RETURN(); } |
---|
| 403 | "." { ASCIIOP_RETURN(); } // also operator |
---|
[3848e0e] | 404 | "..." { NAMEDOP_RETURN(ELLIPSIS); } |
---|
| 405 | |
---|
[b87a5ed] | 406 | /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */ |
---|
[3848e0e] | 407 | "<:" { RETURN_VAL('['); } |
---|
| 408 | ":>" { RETURN_VAL(']'); } |
---|
| 409 | "<%" { RETURN_VAL('{'); } |
---|
| 410 | "%>" { RETURN_VAL('}'); } |
---|
[51b7345] | 411 | |
---|
[b87a5ed] | 412 | /* operators */ |
---|
| 413 | "!" { ASCIIOP_RETURN(); } |
---|
| 414 | "+" { ASCIIOP_RETURN(); } |
---|
| 415 | "-" { ASCIIOP_RETURN(); } |
---|
| 416 | "*" { ASCIIOP_RETURN(); } |
---|
[e5f2a67] | 417 | "\\" { ASCIIOP_RETURN(); } // CFA, exponentiation |
---|
[b87a5ed] | 418 | "/" { ASCIIOP_RETURN(); } |
---|
| 419 | "%" { ASCIIOP_RETURN(); } |
---|
| 420 | "^" { ASCIIOP_RETURN(); } |
---|
| 421 | "~" { ASCIIOP_RETURN(); } |
---|
| 422 | "&" { ASCIIOP_RETURN(); } |
---|
| 423 | "|" { ASCIIOP_RETURN(); } |
---|
| 424 | "<" { ASCIIOP_RETURN(); } |
---|
| 425 | ">" { ASCIIOP_RETURN(); } |
---|
| 426 | "=" { ASCIIOP_RETURN(); } |
---|
| 427 | "?" { ASCIIOP_RETURN(); } |
---|
[3848e0e] | 428 | |
---|
| 429 | "++" { NAMEDOP_RETURN(ICR); } |
---|
| 430 | "--" { NAMEDOP_RETURN(DECR); } |
---|
| 431 | "==" { NAMEDOP_RETURN(EQ); } |
---|
| 432 | "!=" { NAMEDOP_RETURN(NE); } |
---|
| 433 | "<<" { NAMEDOP_RETURN(LS); } |
---|
| 434 | ">>" { NAMEDOP_RETURN(RS); } |
---|
| 435 | "<=" { NAMEDOP_RETURN(LE); } |
---|
| 436 | ">=" { NAMEDOP_RETURN(GE); } |
---|
| 437 | "&&" { NAMEDOP_RETURN(ANDAND); } |
---|
| 438 | "||" { NAMEDOP_RETURN(OROR); } |
---|
| 439 | "->" { NAMEDOP_RETURN(ARROW); } |
---|
| 440 | "+=" { NAMEDOP_RETURN(PLUSassign); } |
---|
| 441 | "-=" { NAMEDOP_RETURN(MINUSassign); } |
---|
[e5f2a67] | 442 | "\\=" { NAMEDOP_RETURN(EXPassign); } // CFA, exponentiation |
---|
[3848e0e] | 443 | "*=" { NAMEDOP_RETURN(MULTassign); } |
---|
| 444 | "/=" { NAMEDOP_RETURN(DIVassign); } |
---|
| 445 | "%=" { NAMEDOP_RETURN(MODassign); } |
---|
| 446 | "&=" { NAMEDOP_RETURN(ANDassign); } |
---|
| 447 | "|=" { NAMEDOP_RETURN(ORassign); } |
---|
| 448 | "^=" { NAMEDOP_RETURN(ERassign); } |
---|
| 449 | "<<=" { NAMEDOP_RETURN(LSassign); } |
---|
| 450 | ">>=" { NAMEDOP_RETURN(RSassign); } |
---|
[51b7345] | 451 | |
---|
[08061589] | 452 | "@=" { NAMEDOP_RETURN(ATassign); } // CFA |
---|
[d69f4bb4] | 453 | "~=" { NAMEDOP_RETURN(ErangeUpEq); } // CFA |
---|
| 454 | "-~" { NAMEDOP_RETURN(ErangeDown); } // CFA |
---|
| 455 | "-~=" { NAMEDOP_RETURN(ErangeDownEq); } // CFA |
---|
[097e2b0] | 456 | |
---|
[b87a5ed] | 457 | /* CFA, operator identifier */ |
---|
| 458 | {op_unary}"?" { IDENTIFIER_RETURN(); } // unary |
---|
[a61fea9a] | 459 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); } |
---|
[76c62b2] | 460 | "^?{}" { IDENTIFIER_RETURN(); } |
---|
[60a8062] | 461 | "?`"{identifier} { // postfix operator |
---|
| 462 | yylval.tok.str = new string( &yytext[2] ); // remove ?` |
---|
| 463 | yylval.tok.str = build_postfix_name( yylval.tok.str ); // add prefix |
---|
| 464 | RETURN_LOCN( typedefTable.isKind( *yylval.tok.str ) ); |
---|
| 465 | } |
---|
[b87a5ed] | 466 | "?"{op_binary_over}"?" { IDENTIFIER_RETURN(); } // binary |
---|
[51b7345] | 467 | /* |
---|
[daf9671] | 468 | This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?" can be |
---|
| 469 | lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier, |
---|
| 470 | e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is |
---|
| 471 | a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then |
---|
| 472 | reparsing the trailing operator identifier. Otherwise a space is needed between the unary operator and operator |
---|
| 473 | identifier to disambiguate this common case. |
---|
| 474 | |
---|
| 475 | A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers. The ambiguity |
---|
| 476 | occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace |
---|
| 477 | look-ahead for the routine-call parameter-list to disambiguate. However, the dereference operator must have a |
---|
| 478 | parameter/argument to dereference *?(...). Hence, always interpreting the string *?() as * ?() does not preclude |
---|
| 479 | any meaningful program. |
---|
| 480 | |
---|
| 481 | The remaining cases are with the increment/decrement operators and conditional expression: |
---|
| 482 | |
---|
| 483 | i++? ...(...); |
---|
| 484 | i?++ ...(...); |
---|
| 485 | |
---|
| 486 | requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an |
---|
| 487 | incorrect expression (juxtaposed identifiers). Therefore, it is necessary to disambiguate these cases with a |
---|
| 488 | space: |
---|
| 489 | |
---|
| 490 | i++ ? i : 0; |
---|
| 491 | i? ++i : 0; |
---|
[51b7345] | 492 | */ |
---|
[daf9671] | 493 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") { |
---|
[b87a5ed] | 494 | // 1 or 2 character unary operator ? |
---|
| 495 | int i = yytext[1] == '?' ? 1 : 2; |
---|
| 496 | yyless( i ); // put back characters up to first '?' |
---|
| 497 | if ( i > 1 ) { |
---|
| 498 | NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR ); |
---|
| 499 | } else { |
---|
| 500 | ASCIIOP_RETURN(); |
---|
| 501 | } // if |
---|
| 502 | } |
---|
| 503 | |
---|
[9ed4f94] | 504 | /* unknown character */ |
---|
| 505 | . { yyerror( "unknown character" ); } |
---|
[51b7345] | 506 | |
---|
| 507 | %% |
---|
[8551b859] | 508 | |
---|
[9ed4f94] | 509 | // ----end of lexer---- |
---|
| 510 | |
---|
| 511 | void yyerror( const char * errmsg ) { |
---|
[8551b859] | 512 | SemanticErrorThrow = true; |
---|
[a8615fd1] | 513 | cerr << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1 |
---|
[79a6b17] | 514 | << ": " << ErrorHelpers::error_str() << errmsg << " before token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl; |
---|
[9ed4f94] | 515 | } |
---|
[51b7345] | 516 | |
---|
[b87a5ed] | 517 | // Local Variables: // |
---|
| 518 | // mode: c++ // |
---|
[de62360d] | 519 | // tab-width: 4 // |
---|
[b87a5ed] | 520 | // compile-command: "make install" // |
---|
| 521 | // End: // |
---|