| 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.
|
|---|
| 6 | *
|
|---|
| 7 | * lex.ll --
|
|---|
| 8 | *
|
|---|
| 9 | * Author : Peter A. Buhr
|
|---|
| 10 | * Created On : Sat Sep 22 08:58:10 2001
|
|---|
| 11 | * Last Modified By : Peter A. Buhr
|
|---|
| 12 | * Last Modified On : Fri Apr 18 15:23:40 2025
|
|---|
| 13 | * Update Count : 884
|
|---|
| 14 | */
|
|---|
| 15 |
|
|---|
| 16 | %option yylineno
|
|---|
| 17 | %option noyywrap
|
|---|
| 18 | %option nounput
|
|---|
| 19 |
|
|---|
| 20 | %{
|
|---|
| 21 | // The lexer assumes the program has been preprocessed by cpp. Hence, all user level preprocessor directive have been
|
|---|
| 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.
|
|---|
| 24 |
|
|---|
| 25 | // *************************** Includes and Defines ****************************
|
|---|
| 26 |
|
|---|
| 27 | #ifdef __clang__
|
|---|
| 28 | #pragma GCC diagnostic ignored "-Wnull-conversion"
|
|---|
| 29 | #endif
|
|---|
| 30 |
|
|---|
| 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 : "";
|
|---|
| 39 | unsigned int column = 0; // position of the end of the last token parsed
|
|---|
| 40 |
|
|---|
| 41 | #include <string>
|
|---|
| 42 | #include <cstdio> // FILENAME_MAX
|
|---|
| 43 | using namespace std;
|
|---|
| 44 |
|
|---|
| 45 | #include "config.h" // configure info
|
|---|
| 46 | #include "DeclarationNode.hpp" // for DeclarationNode
|
|---|
| 47 | #include "ExpressionNode.hpp" // for LabelNode
|
|---|
| 48 | #include "InitializerNode.hpp" // for InitializerNode
|
|---|
| 49 | #include "ParseNode.hpp"
|
|---|
| 50 | #include "ParserTypes.hpp" // for Token
|
|---|
| 51 | #include "StatementNode.hpp" // for CondCtrl, ForCtrl
|
|---|
| 52 | #include "TypedefTable.hpp"
|
|---|
| 53 | // This (generated) header must come late as it is missing includes.
|
|---|
| 54 | #include "parser.hh" // generated info
|
|---|
| 55 |
|
|---|
| 56 | string * build_postfix_name( string * name );
|
|---|
| 57 |
|
|---|
| 58 | char *yyfilename;
|
|---|
| 59 | string *strtext; // accumulate parts of character and string constant value
|
|---|
| 60 |
|
|---|
| 61 | #define RETURN_LOCN(x) yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x )
|
|---|
| 62 | #define RETURN_VAL(x) yylval.tok.str = new string( yytext ); RETURN_LOCN( x )
|
|---|
| 63 | #define RETURN_CHAR(x) yylval.tok.str = nullptr; RETURN_LOCN( x )
|
|---|
| 64 | #define RETURN_STR(x) yylval.tok.str = strtext; RETURN_LOCN( x )
|
|---|
| 65 |
|
|---|
| 66 | #define WHITE_RETURN(x) // do nothing
|
|---|
| 67 | #define NEWLINE_RETURN() column = 0; WHITE_RETURN( '\n' )
|
|---|
| 68 | #define ASCIIOP_RETURN() RETURN_CHAR( (int)yytext[0] ) // single character operator
|
|---|
| 69 | #define NAMEDOP_RETURN(x) RETURN_CHAR( x ) // multichar operator, with a name
|
|---|
| 70 | #define NUMERIC_RETURN(x) rm_underscore(); RETURN_VAL( x ) // numeric constant
|
|---|
| 71 | #define KEYWORD_RETURN(x) RETURN_CHAR( x ) // keyword
|
|---|
| 72 | #define QKEYWORD_RETURN(x) RETURN_VAL(x); // quasi-keyword
|
|---|
| 73 | #define IDENTIFIER_RETURN() RETURN_VAL( typedefTable.isKind( yytext ) )
|
|---|
| 74 |
|
|---|
| 75 | #ifdef HAVE_KEYWORDS_FLOATXX // GCC >= 7 => keyword, otherwise typedef
|
|---|
| 76 | #define FLOATXX(v) KEYWORD_RETURN(v);
|
|---|
| 77 | #else
|
|---|
| 78 | #define FLOATXX(v) IDENTIFIER_RETURN();
|
|---|
| 79 | #endif // HAVE_KEYWORDS_FLOATXX
|
|---|
| 80 |
|
|---|
| 81 | void rm_underscore() {
|
|---|
| 82 | // SKULLDUGGERY: remove underscores (ok to shorten?)
|
|---|
| 83 | yyleng = 0;
|
|---|
| 84 | for ( int i = 0; yytext[i] != '\0'; i += 1 ) { // copying non-underscore characters to front of string
|
|---|
| 85 | if ( yytext[i] != '_' ) {
|
|---|
| 86 | yytext[yyleng] = yytext[i];
|
|---|
| 87 | yyleng += 1;
|
|---|
| 88 | } // if
|
|---|
| 89 | } // for
|
|---|
| 90 | yytext[yyleng] = '\0';
|
|---|
| 91 | } // rm_underscore
|
|---|
| 92 |
|
|---|
| 93 | // Stop warning due to incorrectly generated flex code.
|
|---|
| 94 | #pragma GCC diagnostic ignored "-Wsign-compare"
|
|---|
| 95 | %}
|
|---|
| 96 |
|
|---|
| 97 | binary [0-1]
|
|---|
| 98 | octal [0-7]
|
|---|
| 99 | nonzero [1-9]
|
|---|
| 100 | decimal [0-9]
|
|---|
| 101 | hex [0-9a-fA-F]
|
|---|
| 102 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2}))
|
|---|
| 103 |
|
|---|
| 104 | // identifier, GCC: $ in identifier
|
|---|
| 105 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})*
|
|---|
| 106 |
|
|---|
| 107 | // numeric constants, CFA: '_' in constant
|
|---|
| 108 | hex_quad {hex}("_"?{hex}){3}
|
|---|
| 109 | size_opt (8|16|32|64|128)?
|
|---|
| 110 | // CFA: explicit l8/l16/l32/l64/l128, char 'hh', short 'h', int 'n'
|
|---|
| 111 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hHnN])
|
|---|
| 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]))?
|
|---|
| 114 |
|
|---|
| 115 | octal_digits ({octal})|({octal}({octal}|"_")*{octal})
|
|---|
| 116 | octal_prefix "0""_"?
|
|---|
| 117 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt}
|
|---|
| 118 |
|
|---|
| 119 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal})
|
|---|
| 120 | decimal_constant {nonzero_digits}{integer_suffix_opt}
|
|---|
| 121 |
|
|---|
| 122 | binary_digits ({binary})|({binary}({binary}|"_")*{binary})
|
|---|
| 123 | binary_prefix "0"[bB]"_"?
|
|---|
| 124 | binary_constant {binary_prefix}{binary_digits}{integer_suffix_opt}
|
|---|
| 125 |
|
|---|
| 126 | hex_digits ({hex})|({hex}({hex}|"_")*{hex})
|
|---|
| 127 | hex_prefix "0"[xX]"_"?
|
|---|
| 128 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt}
|
|---|
| 129 |
|
|---|
| 130 | // GCC: floating D (double), imaginary iI, and decimal floating DF, DD, DL
|
|---|
| 131 | exponent "_"?[eE]"_"?[+-]?{decimal_digits}
|
|---|
| 132 | floating_size 16|32|32x|64|64x|80|128|128x
|
|---|
| 133 | floating_length ([fFdDlLwWqQ]|[fF]{floating_size})
|
|---|
| 134 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length})
|
|---|
| 135 | decimal_floating_suffix [dD][fFdDlL]
|
|---|
| 136 | floating_suffix_opt ("_"?({floating_suffix}|{decimal_floating_suffix}))?
|
|---|
| 137 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal})
|
|---|
| 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})
|
|---|
| 141 |
|
|---|
| 142 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits}
|
|---|
| 143 | hex_floating_suffix_opt ("_"?({floating_suffix}))?
|
|---|
| 144 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".")
|
|---|
| 145 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt}
|
|---|
| 146 |
|
|---|
| 147 | // character escape sequence, GCC: \e => esc character
|
|---|
| 148 | simple_escape "\\"[abefnrtv'"?\\]
|
|---|
| 149 | // " stop editor highlighting
|
|---|
| 150 | octal_escape "\\"{octal}("_"?{octal}){0,2}
|
|---|
| 151 | hex_escape "\\""x""_"?{hex_digits}
|
|---|
| 152 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char}
|
|---|
| 153 | cwide_prefix "L"|"U"|"u"
|
|---|
| 154 | swide_prefix {cwide_prefix}|"u8"
|
|---|
| 155 |
|
|---|
| 156 | // display/white-space characters
|
|---|
| 157 | h_tab "\t"
|
|---|
| 158 | form_feed "\f"
|
|---|
| 159 | v_tab "\v"
|
|---|
| 160 | new_line "\n"
|
|---|
| 161 | c_return "\r"
|
|---|
| 162 | h_white " "|{h_tab}
|
|---|
| 163 | v_white {v_tab}|{c_return}|{form_feed}
|
|---|
| 164 | hv_white {h_white}|{v_tab}|{new_line}|{c_return}|{form_feed}
|
|---|
| 165 |
|
|---|
| 166 | // overloadable operators
|
|---|
| 167 | op_unary_only "~"|"!"
|
|---|
| 168 | op_unary_binary "+"|"-"|"*"
|
|---|
| 169 | op_unary_pre_post "++"|"--"
|
|---|
| 170 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post}
|
|---|
| 171 |
|
|---|
| 172 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>="
|
|---|
| 173 | op_binary_over {op_unary_binary}|{op_binary_only}
|
|---|
| 174 | // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@="
|
|---|
| 175 | // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over}
|
|---|
| 176 |
|
|---|
| 177 | // C23 attributes, CPP also handles missing quote delimiter
|
|---|
| 178 | attr_string "\""([^\"\\\n]|{escape_seq})*["\n]
|
|---|
| 179 | attr_arg_opt ({hv_white}*"("({hv_white}*{attr_string}{hv_white}*)+")")?
|
|---|
| 180 | attributes "deprecated"{attr_arg_opt}|"fallthrough"|"nodiscard"{attr_arg_opt}|"maybe_unused"|"noreturn"|"_Noreturn"|"unsequenced"|"unused"|"reproducible"|{identifier}{hv_white}*"::"{hv_white}*{identifier}
|
|---|
| 181 |
|
|---|
| 182 | %x COMMENT
|
|---|
| 183 | %x QUOTE
|
|---|
| 184 | %x STRING
|
|---|
| 185 |
|
|---|
| 186 | %%
|
|---|
| 187 | /* line directives */
|
|---|
| 188 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" {
|
|---|
| 189 | /* " stop editor highlighting */
|
|---|
| 190 | static char filename[FILENAME_MAX]; // temporarily store current source-file name
|
|---|
| 191 | char *end_num;
|
|---|
| 192 | char *begin_string, *end_string;
|
|---|
| 193 | long lineno, length;
|
|---|
| 194 | lineno = strtol( yytext + 1, &end_num, 0 );
|
|---|
| 195 | begin_string = strchr( end_num, '"' );
|
|---|
| 196 | if ( begin_string ) { // file name ?
|
|---|
| 197 | end_string = strchr( begin_string + 1, '"' ); // look for ending delimiter
|
|---|
| 198 | assert( end_string ); // closing quote ?
|
|---|
| 199 | length = end_string - begin_string - 1; // file-name length without quotes or sentinel
|
|---|
| 200 | assert( length < FILENAME_MAX ); // room for sentinel ?
|
|---|
| 201 | memcpy( &filename, begin_string + 1, length ); // copy file name from yytext
|
|---|
| 202 | filename[ length ] = '\0'; // terminate string with sentinel
|
|---|
| 203 | //cout << "file " << filename << " line " << lineno << endl;
|
|---|
| 204 | yylineno = lineno;
|
|---|
| 205 | yyfilename = filename;
|
|---|
| 206 | } // if
|
|---|
| 207 | }
|
|---|
| 208 |
|
|---|
| 209 | /* preprocessor-style directives */
|
|---|
| 210 | ^{h_white}*"#"[^\n]*"\n" { RETURN_VAL( DIRECTIVE ); }
|
|---|
| 211 |
|
|---|
| 212 | /* ignore C style comments (ALSO HANDLED BY CPP) */
|
|---|
| 213 | "/*" { BEGIN COMMENT; }
|
|---|
| 214 | <COMMENT>.|\n ;
|
|---|
| 215 | <COMMENT>"*/" { BEGIN 0; }
|
|---|
| 216 |
|
|---|
| 217 | /* ignore C++ style comments (ALSO HANDLED BY CPP) */
|
|---|
| 218 | "//"[^\n]*"\n" ;
|
|---|
| 219 |
|
|---|
| 220 | /* ignore whitespace */
|
|---|
| 221 | ({h_white}|{v_white})+ { WHITE_RETURN(' '); } // do nothing
|
|---|
| 222 | "\n" { NEWLINE_RETURN(); } // reset column counter
|
|---|
| 223 |
|
|---|
| 224 | /* keywords */
|
|---|
| 225 | alignas { KEYWORD_RETURN(ALIGNAS); } // CFA
|
|---|
| 226 | _Alignas { KEYWORD_RETURN(ALIGNAS); } // C11
|
|---|
| 227 | alignof { KEYWORD_RETURN(ALIGNOF); } // CFA
|
|---|
| 228 | _Alignof { KEYWORD_RETURN(ALIGNOF); } // C11
|
|---|
| 229 | __alignof { KEYWORD_RETURN(ALIGNOF); } // GCC
|
|---|
| 230 | __alignof__ { KEYWORD_RETURN(ALIGNOF); } // GCC
|
|---|
| 231 | and { QKEYWORD_RETURN(WAND); } // CFA
|
|---|
| 232 | asm { KEYWORD_RETURN(ASM); }
|
|---|
| 233 | __asm { KEYWORD_RETURN(ASM); } // GCC
|
|---|
| 234 | __asm__ { KEYWORD_RETURN(ASM); } // GCC
|
|---|
| 235 | _Atomic { KEYWORD_RETURN(ATOMIC); } // C11
|
|---|
| 236 | __attribute { KEYWORD_RETURN(ATTRIBUTE); } // GCC
|
|---|
| 237 | __attribute__ { KEYWORD_RETURN(ATTRIBUTE); } // GCC
|
|---|
| 238 | auto { KEYWORD_RETURN(AUTO); }
|
|---|
| 239 | __auto_type { KEYWORD_RETURN(AUTO_TYPE); }
|
|---|
| 240 | basetypeof { KEYWORD_RETURN(BASETYPEOF); } // CFA
|
|---|
| 241 | _Bool { KEYWORD_RETURN(BOOL); } // C99
|
|---|
| 242 | __SVBool_t { KEYWORD_RETURN(BOOL); } // gcc (ARM)
|
|---|
| 243 | break { KEYWORD_RETURN(BREAK); }
|
|---|
| 244 | case { KEYWORD_RETURN(CASE); }
|
|---|
| 245 | catch { QKEYWORD_RETURN(CATCH); } // CFA
|
|---|
| 246 | catchResume { QKEYWORD_RETURN(CATCHRESUME); } // CFA
|
|---|
| 247 | char { KEYWORD_RETURN(CHAR); }
|
|---|
| 248 | choose { KEYWORD_RETURN(CHOOSE); } // CFA
|
|---|
| 249 | coerce { KEYWORD_RETURN(COERCE); } // CFA
|
|---|
| 250 | corun { KEYWORD_RETURN(CORUN); } // CFA
|
|---|
| 251 | cofor { KEYWORD_RETURN(COFOR); } // CFA
|
|---|
| 252 | _Complex { KEYWORD_RETURN(COMPLEX); } // C99
|
|---|
| 253 | __complex { KEYWORD_RETURN(COMPLEX); } // GCC
|
|---|
| 254 | __complex__ { KEYWORD_RETURN(COMPLEX); } // GCC
|
|---|
| 255 | const { KEYWORD_RETURN(CONST); }
|
|---|
| 256 | __const { KEYWORD_RETURN(CONST); } // GCC
|
|---|
| 257 | __const__ { KEYWORD_RETURN(CONST); } // GCC
|
|---|
| 258 | continue { KEYWORD_RETURN(CONTINUE); }
|
|---|
| 259 | coroutine { KEYWORD_RETURN(COROUTINE); } // CFA
|
|---|
| 260 | _Decimal32 { KEYWORD_RETURN(DECIMAL32); } // GCC
|
|---|
| 261 | _Decimal64 { KEYWORD_RETURN(DECIMAL64); } // GCC
|
|---|
| 262 | _Decimal128 { KEYWORD_RETURN(DECIMAL128); } // GCC
|
|---|
| 263 | default { KEYWORD_RETURN(DEFAULT); }
|
|---|
| 264 | disable { KEYWORD_RETURN(DISABLE); } // CFA
|
|---|
| 265 | do { KEYWORD_RETURN(DO); }
|
|---|
| 266 | double { KEYWORD_RETURN(DOUBLE); }
|
|---|
| 267 | dtype { KEYWORD_RETURN(DTYPE); } // CFA
|
|---|
| 268 | else { KEYWORD_RETURN(ELSE); }
|
|---|
| 269 | enable { KEYWORD_RETURN(ENABLE); } // CFA
|
|---|
| 270 | enum { KEYWORD_RETURN(ENUM); }
|
|---|
| 271 | exception { KEYWORD_RETURN(EXCEPTION); } // CFA
|
|---|
| 272 | __extension__ { KEYWORD_RETURN(EXTENSION); } // GCC
|
|---|
| 273 | extern { KEYWORD_RETURN(EXTERN); }
|
|---|
| 274 | fallthrough { KEYWORD_RETURN(FALLTHROUGH); } // CFA
|
|---|
| 275 | finally { QKEYWORD_RETURN(FINALLY); } // CFA
|
|---|
| 276 | fixup { QKEYWORD_RETURN(FIXUP); } // CFA
|
|---|
| 277 | float { KEYWORD_RETURN(FLOAT); }
|
|---|
| 278 | __float80 { KEYWORD_RETURN(FLOAT80); } // GCC
|
|---|
| 279 | float80 { KEYWORD_RETURN(FLOAT80); } // GCC
|
|---|
| 280 | __float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC
|
|---|
| 281 | float128 { KEYWORD_RETURN(uuFLOAT128); } // GCC
|
|---|
| 282 | _Float16 { FLOATXX(FLOAT16); } // GCC
|
|---|
| 283 | _Float32 { FLOATXX(FLOAT32); } // GCC
|
|---|
| 284 | _Float32x { FLOATXX(FLOAT32X); } // GCC
|
|---|
| 285 | _Float64 { FLOATXX(FLOAT64); } // GCC
|
|---|
| 286 | _Float64x { FLOATXX(FLOAT64X); } // GCC
|
|---|
| 287 | _Float128 { FLOATXX(FLOAT128); } // GCC
|
|---|
| 288 | _Float128x { FLOATXX(FLOAT128X); } // GCC
|
|---|
| 289 | __Float32x4_t { FLOATXX(FLOAT128); } // GCC (ARM)
|
|---|
| 290 | __Float64x2_t { FLOATXX(FLOAT128); } // GCC (ARM)
|
|---|
| 291 | __SVFloat32_t { FLOATXX(FLOAT128); } // GCC (ARM)
|
|---|
| 292 | __SVFloat64_t { FLOATXX(FLOAT128); } // GCC (ARM)
|
|---|
| 293 | for { KEYWORD_RETURN(FOR); }
|
|---|
| 294 | forall { KEYWORD_RETURN(FORALL); } // CFA
|
|---|
| 295 | fortran { KEYWORD_RETURN(FORTRAN); }
|
|---|
| 296 | ftype { KEYWORD_RETURN(FTYPE); } // CFA
|
|---|
| 297 | generator { KEYWORD_RETURN(GENERATOR); } // CFA
|
|---|
| 298 | _Generic { KEYWORD_RETURN(GENERIC); } // C11
|
|---|
| 299 | goto { KEYWORD_RETURN(GOTO); }
|
|---|
| 300 | if { KEYWORD_RETURN(IF); }
|
|---|
| 301 | _Imaginary { KEYWORD_RETURN(IMAGINARY); } // C99
|
|---|
| 302 | __imag { KEYWORD_RETURN(IMAGINARY); } // GCC
|
|---|
| 303 | __imag__ { KEYWORD_RETURN(IMAGINARY); } // GCC
|
|---|
| 304 | inline { KEYWORD_RETURN(INLINE); } // C99
|
|---|
| 305 | __inline { KEYWORD_RETURN(INLINE); } // GCC
|
|---|
| 306 | __inline__ { KEYWORD_RETURN(INLINE); } // GCC
|
|---|
| 307 | int { KEYWORD_RETURN(INT); }
|
|---|
| 308 | int128 { KEYWORD_RETURN(INT128); } // CFA
|
|---|
| 309 | __int128 { KEYWORD_RETURN(INT128); } // GCC
|
|---|
| 310 | __int128_t { KEYWORD_RETURN(INT128); } // GCC
|
|---|
| 311 | __label__ { KEYWORD_RETURN(LABEL); } // GCC
|
|---|
| 312 | long { KEYWORD_RETURN(LONG); }
|
|---|
| 313 | monitor { KEYWORD_RETURN(MONITOR); } // CFA
|
|---|
| 314 | mutex { KEYWORD_RETURN(MUTEX); } // CFA
|
|---|
| 315 | _Noreturn { KEYWORD_RETURN(NORETURN); } // C11
|
|---|
| 316 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); } // GCC
|
|---|
| 317 | one_t { NUMERIC_RETURN(ONE_T); } // CFA
|
|---|
| 318 | or { QKEYWORD_RETURN(WOR); } // CFA
|
|---|
| 319 | otype { KEYWORD_RETURN(OTYPE); } // CFA
|
|---|
| 320 | recover { QKEYWORD_RETURN(RECOVER); } // CFA
|
|---|
| 321 | register { KEYWORD_RETURN(REGISTER); }
|
|---|
| 322 | report { KEYWORD_RETURN(THROWRESUME); } // CFA
|
|---|
| 323 | restrict { KEYWORD_RETURN(RESTRICT); } // C99
|
|---|
| 324 | __restrict { KEYWORD_RETURN(RESTRICT); } // GCC
|
|---|
| 325 | __restrict__ { KEYWORD_RETURN(RESTRICT); } // GCC
|
|---|
| 326 | return { KEYWORD_RETURN(RETURN); }
|
|---|
| 327 | /* resume { KEYWORD_RETURN(RESUME); } // CFA */
|
|---|
| 328 | short { KEYWORD_RETURN(SHORT); }
|
|---|
| 329 | signed { KEYWORD_RETURN(SIGNED); }
|
|---|
| 330 | __signed { KEYWORD_RETURN(SIGNED); } // GCC
|
|---|
| 331 | __signed__ { KEYWORD_RETURN(SIGNED); } // GCC
|
|---|
| 332 | sizeof { KEYWORD_RETURN(SIZEOF); }
|
|---|
| 333 | countof { KEYWORD_RETURN(COUNTOF); } // GCC
|
|---|
| 334 | static { KEYWORD_RETURN(STATIC); }
|
|---|
| 335 | _Static_assert { KEYWORD_RETURN(STATICASSERT); } // C11
|
|---|
| 336 | static_assert { KEYWORD_RETURN(STATICASSERT); } // C23
|
|---|
| 337 | struct { KEYWORD_RETURN(STRUCT); }
|
|---|
| 338 | suspend { KEYWORD_RETURN(SUSPEND); } // CFA
|
|---|
| 339 | switch { KEYWORD_RETURN(SWITCH); }
|
|---|
| 340 | thread { KEYWORD_RETURN(THREAD); } // C11
|
|---|
| 341 | __thread { KEYWORD_RETURN(THREADLOCALGCC); } // GCC
|
|---|
| 342 | _Thread_local { KEYWORD_RETURN(THREADLOCALC11); } // C11
|
|---|
| 343 | thread_local { KEYWORD_RETURN(THREADLOCALC11); } // C23
|
|---|
| 344 | throw { KEYWORD_RETURN(THROW); } // CFA
|
|---|
| 345 | throwResume { KEYWORD_RETURN(THROWRESUME); } // CFA
|
|---|
| 346 | timeout { QKEYWORD_RETURN(TIMEOUT); } // CFA
|
|---|
| 347 | trait { KEYWORD_RETURN(TRAIT); } // CFA
|
|---|
| 348 | try { KEYWORD_RETURN(TRY); } // CFA
|
|---|
| 349 | ttype { KEYWORD_RETURN(TTYPE); } // CFA
|
|---|
| 350 | typedef { KEYWORD_RETURN(TYPEDEF); }
|
|---|
| 351 | typeof { KEYWORD_RETURN(TYPEOF); } // GCC
|
|---|
| 352 | __typeof { KEYWORD_RETURN(TYPEOF); } // GCC
|
|---|
| 353 | __typeof__ { KEYWORD_RETURN(TYPEOF); } // GCC
|
|---|
| 354 | typeid { KEYWORD_RETURN(TYPEID); } // GCC
|
|---|
| 355 | union { KEYWORD_RETURN(UNION); }
|
|---|
| 356 | __uint128_t { KEYWORD_RETURN(UINT128); } // GCC
|
|---|
| 357 | unsigned { KEYWORD_RETURN(UNSIGNED); }
|
|---|
| 358 | __builtin_va_arg { KEYWORD_RETURN(VA_ARG); } // GCC
|
|---|
| 359 | __builtin_va_list { KEYWORD_RETURN(VA_LIST); } // GCC
|
|---|
| 360 | virtual { KEYWORD_RETURN(VIRTUAL); } // CFA
|
|---|
| 361 | void { KEYWORD_RETURN(VOID); }
|
|---|
| 362 | volatile { KEYWORD_RETURN(VOLATILE); }
|
|---|
| 363 | __volatile { KEYWORD_RETURN(VOLATILE); } // GCC
|
|---|
| 364 | __volatile__ { KEYWORD_RETURN(VOLATILE); } // GCC
|
|---|
| 365 | vtable { KEYWORD_RETURN(VTABLE); } // CFA
|
|---|
| 366 | waitfor { KEYWORD_RETURN(WAITFOR); } // CFA
|
|---|
| 367 | waituntil { KEYWORD_RETURN(WAITUNTIL); } // CFA
|
|---|
| 368 | when { KEYWORD_RETURN(WHEN); } // CFA
|
|---|
| 369 | while { KEYWORD_RETURN(WHILE); }
|
|---|
| 370 | with { KEYWORD_RETURN(WITH); } // CFA
|
|---|
| 371 | zero_t { NUMERIC_RETURN(ZERO_T); } // CFA
|
|---|
| 372 |
|
|---|
| 373 | /* identifier */
|
|---|
| 374 | {identifier} { IDENTIFIER_RETURN(); }
|
|---|
| 375 | "``"{identifier} { // CFA
|
|---|
| 376 | yytext[yyleng] = '\0'; yytext += 2; // SKULLDUGGERY: remove backquotes (ok to shorten?)
|
|---|
| 377 | IDENTIFIER_RETURN();
|
|---|
| 378 | }
|
|---|
| 379 |
|
|---|
| 380 | /* C23 attributes */
|
|---|
| 381 | "[["{hv_white}*{attributes}({hv_white}*","{hv_white}*{attributes})*{hv_white}*"]]" {
|
|---|
| 382 | strtext = new string( &yytext[2], yyleng - 4 ); // remove delimiters "[[" and "]]"
|
|---|
| 383 | RETURN_STR(C23_ATTRIBUTE);
|
|---|
| 384 | }
|
|---|
| 385 |
|
|---|
| 386 | /* numeric constants */
|
|---|
| 387 | {binary_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
|---|
| 388 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
|---|
| 389 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
|---|
| 390 | {hex_constant} { NUMERIC_RETURN(INTEGERconstant); }
|
|---|
| 391 | {floating_decimal} { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant
|
|---|
| 392 | {floating_fraction} { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant
|
|---|
| 393 | {floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
|
|---|
| 394 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
|
|---|
| 395 |
|
|---|
| 396 | /* character constant, allows empty value, CPP also handles missing quote delimiter */
|
|---|
| 397 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); }
|
|---|
| 398 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); }
|
|---|
| 399 | <QUOTE>['\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); }
|
|---|
| 400 | /* ' stop editor highlighting */
|
|---|
| 401 |
|
|---|
| 402 | /* string constant, CPP also handles missing quote delimiter */
|
|---|
| 403 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); }
|
|---|
| 404 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); }
|
|---|
| 405 | <STRING>["\n] { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); }
|
|---|
| 406 | /* " stop editor highlighting */
|
|---|
| 407 |
|
|---|
| 408 | /* common character/string constant */
|
|---|
| 409 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); }
|
|---|
| 410 | <QUOTE,STRING>"\\"{h_white}*"\n" {} // continuation (ALSO HANDLED BY CPP)
|
|---|
| 411 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character
|
|---|
| 412 |
|
|---|
| 413 | /* punctuation */
|
|---|
| 414 | "@" { ASCIIOP_RETURN(); }
|
|---|
| 415 | "`" { ASCIIOP_RETURN(); }
|
|---|
| 416 | "[" { ASCIIOP_RETURN(); }
|
|---|
| 417 | "]" { ASCIIOP_RETURN(); }
|
|---|
| 418 | "(" { ASCIIOP_RETURN(); }
|
|---|
| 419 | ")" { ASCIIOP_RETURN(); }
|
|---|
| 420 | "{" { ASCIIOP_RETURN(); }
|
|---|
| 421 | "}" { ASCIIOP_RETURN(); }
|
|---|
| 422 | "," { ASCIIOP_RETURN(); } // also operator
|
|---|
| 423 | ":" { ASCIIOP_RETURN(); }
|
|---|
| 424 | ";" { ASCIIOP_RETURN(); }
|
|---|
| 425 | "." { ASCIIOP_RETURN(); } // also operator
|
|---|
| 426 | "@[" { NAMEDOP_RETURN(ATTR); } // CFA, attribute shorthand
|
|---|
| 427 | "..." { NAMEDOP_RETURN(ELLIPSIS); }
|
|---|
| 428 |
|
|---|
| 429 | /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */
|
|---|
| 430 | "<:" { RETURN_VAL('['); }
|
|---|
| 431 | ":>" { RETURN_VAL(']'); }
|
|---|
| 432 | "<%" { RETURN_VAL('{'); }
|
|---|
| 433 | "%>" { RETURN_VAL('}'); }
|
|---|
| 434 |
|
|---|
| 435 | /* operators */
|
|---|
| 436 | "!" { ASCIIOP_RETURN(); }
|
|---|
| 437 | "+" { ASCIIOP_RETURN(); }
|
|---|
| 438 | "-" { ASCIIOP_RETURN(); }
|
|---|
| 439 | "*" { ASCIIOP_RETURN(); }
|
|---|
| 440 | "\\" { ASCIIOP_RETURN(); } // CFA, exponentiation
|
|---|
| 441 | "/" { ASCIIOP_RETURN(); }
|
|---|
| 442 | "%" { ASCIIOP_RETURN(); }
|
|---|
| 443 | "^" { ASCIIOP_RETURN(); }
|
|---|
| 444 | "~" { ASCIIOP_RETURN(); }
|
|---|
| 445 | "&" { ASCIIOP_RETURN(); }
|
|---|
| 446 | "|" { ASCIIOP_RETURN(); }
|
|---|
| 447 | "<" { ASCIIOP_RETURN(); }
|
|---|
| 448 | ">" { ASCIIOP_RETURN(); }
|
|---|
| 449 | "=" { ASCIIOP_RETURN(); }
|
|---|
| 450 | "?" { ASCIIOP_RETURN(); }
|
|---|
| 451 |
|
|---|
| 452 | "++" { NAMEDOP_RETURN(ICR); }
|
|---|
| 453 | "--" { NAMEDOP_RETURN(DECR); }
|
|---|
| 454 | "==" { NAMEDOP_RETURN(EQ); }
|
|---|
| 455 | "!=" { NAMEDOP_RETURN(NE); }
|
|---|
| 456 | "<<" { NAMEDOP_RETURN(LS); }
|
|---|
| 457 | ">>" { NAMEDOP_RETURN(RS); }
|
|---|
| 458 | "<=" { NAMEDOP_RETURN(LE); }
|
|---|
| 459 | ">=" { NAMEDOP_RETURN(GE); }
|
|---|
| 460 | "&&" { NAMEDOP_RETURN(ANDAND); }
|
|---|
| 461 | "||" { NAMEDOP_RETURN(OROR); }
|
|---|
| 462 | "->" { NAMEDOP_RETURN(ARROW); }
|
|---|
| 463 | "+=" { NAMEDOP_RETURN(PLUSassign); }
|
|---|
| 464 | "-=" { NAMEDOP_RETURN(MINUSassign); }
|
|---|
| 465 | "\\=" { NAMEDOP_RETURN(EXPassign); } // CFA, exponentiation
|
|---|
| 466 | "*=" { NAMEDOP_RETURN(MULTassign); }
|
|---|
| 467 | "/=" { NAMEDOP_RETURN(DIVassign); }
|
|---|
| 468 | "%=" { NAMEDOP_RETURN(MODassign); }
|
|---|
| 469 | "&=" { NAMEDOP_RETURN(ANDassign); }
|
|---|
| 470 | "|=" { NAMEDOP_RETURN(ORassign); }
|
|---|
| 471 | "^=" { NAMEDOP_RETURN(ERassign); }
|
|---|
| 472 | "<<=" { NAMEDOP_RETURN(LSassign); }
|
|---|
| 473 | ">>=" { NAMEDOP_RETURN(RSassign); }
|
|---|
| 474 |
|
|---|
| 475 | "@=" { NAMEDOP_RETURN(ATassign); } // CFA
|
|---|
| 476 | "+~" { NAMEDOP_RETURN(ErangeUpLt); } // CFA
|
|---|
| 477 | "~=" { NAMEDOP_RETURN(ErangeUpLe); } // CFA
|
|---|
| 478 | "+~=" { NAMEDOP_RETURN(ErangeUpLe); } // CFA
|
|---|
| 479 | "-~" { NAMEDOP_RETURN(ErangeDownGt); } // CFA
|
|---|
| 480 | "-~=" { NAMEDOP_RETURN(ErangeDownGe); } // CFA
|
|---|
| 481 | "~==" { NAMEDOP_RETURN(ErangeEq); } // CFA
|
|---|
| 482 | "~!=" { NAMEDOP_RETURN(ErangeNe); } // CFA
|
|---|
| 483 | "-~==" { NAMEDOP_RETURN(ErangeDownEq); } // CFA
|
|---|
| 484 | "-~!=" { NAMEDOP_RETURN(ErangeDownNe); } // CFA
|
|---|
| 485 |
|
|---|
| 486 | /* CFA, operator identifier */
|
|---|
| 487 | {op_unary}"?" { IDENTIFIER_RETURN(); } // unary
|
|---|
| 488 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); }
|
|---|
| 489 | "^?{}" { IDENTIFIER_RETURN(); }
|
|---|
| 490 | "?`"{identifier} { // postfix operator
|
|---|
| 491 | yylval.tok.str = new string( &yytext[2] ); // remove ?`
|
|---|
| 492 | yylval.tok.str = build_postfix_name( yylval.tok.str ); // add prefix
|
|---|
| 493 | RETURN_LOCN( typedefTable.isKind( *yylval.tok.str ) );
|
|---|
| 494 | }
|
|---|
| 495 | "?"{op_binary_over}"?" { IDENTIFIER_RETURN(); } // binary
|
|---|
| 496 | /*
|
|---|
| 497 | This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?" can be
|
|---|
| 498 | lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier,
|
|---|
| 499 | e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is
|
|---|
| 500 | a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then
|
|---|
| 501 | reparsing the trailing operator identifier. Otherwise a space is needed between the unary operator and operator
|
|---|
| 502 | identifier to disambiguate this common case.
|
|---|
| 503 |
|
|---|
| 504 | A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers. The ambiguity
|
|---|
| 505 | occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace
|
|---|
| 506 | look-ahead for the routine-call parameter-list to disambiguate. However, the dereference operator must have a
|
|---|
| 507 | parameter/argument to dereference *?(...). Hence, always interpreting the string *?() as * ?() does not preclude
|
|---|
| 508 | any meaningful program.
|
|---|
| 509 |
|
|---|
| 510 | The remaining cases are with the increment/decrement operators and conditional expression:
|
|---|
| 511 |
|
|---|
| 512 | i++? ...(...);
|
|---|
| 513 | i?++ ...(...);
|
|---|
| 514 |
|
|---|
| 515 | requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an
|
|---|
| 516 | incorrect expression (juxtaposed identifiers). Therefore, it is necessary to disambiguate these cases with a
|
|---|
| 517 | space:
|
|---|
| 518 |
|
|---|
| 519 | i++ ? i : 0;
|
|---|
| 520 | i? ++i : 0;
|
|---|
| 521 | */
|
|---|
| 522 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") {
|
|---|
| 523 | // 1 or 2 character unary operator ?
|
|---|
| 524 | int i = yytext[1] == '?' ? 1 : 2;
|
|---|
| 525 | yyless( i ); // put back characters up to first '?'
|
|---|
| 526 | if ( i > 1 ) {
|
|---|
| 527 | NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR );
|
|---|
| 528 | } else {
|
|---|
| 529 | ASCIIOP_RETURN();
|
|---|
| 530 | } // if
|
|---|
| 531 | }
|
|---|
| 532 |
|
|---|
| 533 | /* unknown character */
|
|---|
| 534 | . { yyerror( "unknown character" ); }
|
|---|
| 535 |
|
|---|
| 536 | %%
|
|---|
| 537 |
|
|---|
| 538 | // ----end of lexer----
|
|---|
| 539 |
|
|---|
| 540 | void yyerror( const char * errmsg ) {
|
|---|
| 541 | SemanticErrorThrow = true;
|
|---|
| 542 | cerr << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1
|
|---|
| 543 | << ": " << ErrorHelpers::error_str() << errmsg << " before token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl;
|
|---|
| 544 | }
|
|---|
| 545 |
|
|---|
| 546 | // Local Variables: //
|
|---|
| 547 | // mode: c++ //
|
|---|
| 548 | // tab-width: 4 //
|
|---|
| 549 | // compile-command: "make install" //
|
|---|
| 550 | // End: //
|
|---|