| 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 : Sun Aug  4 20:53:47 2019
 | 
|---|
| 13 |  * Update Count     : 719
 | 
|---|
| 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 | // 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 : "";
 | 
|---|
| 35 | unsigned int column = 0;                                                                // position of the end of the last token parsed
 | 
|---|
| 36 | 
 | 
|---|
| 37 | #include <string>
 | 
|---|
| 38 | #include <cstdio>                                                                               // FILENAME_MAX
 | 
|---|
| 39 | using namespace std;
 | 
|---|
| 40 | 
 | 
|---|
| 41 | #include "config.h"                                                                             // configure info
 | 
|---|
| 42 | #include "ParseNode.h"
 | 
|---|
| 43 | #include "TypedefTable.h"
 | 
|---|
| 44 | 
 | 
|---|
| 45 | char *yyfilename;
 | 
|---|
| 46 | string *strtext;                                                                                // accumulate parts of character and string constant value
 | 
|---|
| 47 | 
 | 
|---|
| 48 | #define RETURN_LOCN(x)          yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x )
 | 
|---|
| 49 | #define RETURN_VAL(x)           yylval.tok.str = new string( yytext ); RETURN_LOCN( x )
 | 
|---|
| 50 | #define RETURN_CHAR(x)          yylval.tok.str = nullptr; RETURN_LOCN( x )
 | 
|---|
| 51 | #define RETURN_STR(x)           yylval.tok.str = strtext; RETURN_LOCN( x )
 | 
|---|
| 52 | 
 | 
|---|
| 53 | #define WHITE_RETURN(x)         // do nothing
 | 
|---|
| 54 | #define NEWLINE_RETURN()        column = 0; WHITE_RETURN( '\n' )
 | 
|---|
| 55 | #define ASCIIOP_RETURN()        RETURN_CHAR( (int)yytext[0] ) // single character operator
 | 
|---|
| 56 | #define NAMEDOP_RETURN(x)       RETURN_CHAR( x )                        // multichar operator, with a name
 | 
|---|
| 57 | #define NUMERIC_RETURN(x)       rm_underscore(); RETURN_VAL( x ) // numeric constant
 | 
|---|
| 58 | #define KEYWORD_RETURN(x)       RETURN_CHAR( x )                        // keyword
 | 
|---|
| 59 | #define QKEYWORD_RETURN(x)      RETURN_VAL(x);                          // quasi-keyword
 | 
|---|
| 60 | #define IDENTIFIER_RETURN()     RETURN_VAL( typedefTable.isKind( yytext ) )
 | 
|---|
| 61 | 
 | 
|---|
| 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 | 
 | 
|---|
| 68 | void rm_underscore() {
 | 
|---|
| 69 |         // SKULLDUGGERY: remove underscores (ok to shorten?)
 | 
|---|
| 70 |         yyleng = 0;
 | 
|---|
| 71 |         for ( int i = 0; yytext[i] != '\0'; i += 1 ) {          // copying non-underscore characters to front of string
 | 
|---|
| 72 |                 if ( yytext[i] != '_' ) {
 | 
|---|
| 73 |                         yytext[yyleng] = yytext[i];
 | 
|---|
| 74 |                         yyleng += 1;
 | 
|---|
| 75 |                 } // if
 | 
|---|
| 76 |         } // for
 | 
|---|
| 77 |         yytext[yyleng] = '\0';
 | 
|---|
| 78 | } // rm_underscore
 | 
|---|
| 79 | 
 | 
|---|
| 80 | // Stop warning due to incorrectly generated flex code.
 | 
|---|
| 81 | #pragma GCC diagnostic ignored "-Wsign-compare"
 | 
|---|
| 82 | %}
 | 
|---|
| 83 | 
 | 
|---|
| 84 | binary [0-1]
 | 
|---|
| 85 | octal [0-7]
 | 
|---|
| 86 | nonzero [1-9]
 | 
|---|
| 87 | decimal [0-9]
 | 
|---|
| 88 | hex [0-9a-fA-F]
 | 
|---|
| 89 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2}))
 | 
|---|
| 90 | 
 | 
|---|
| 91 |                                 // identifier, GCC: $ in identifier
 | 
|---|
| 92 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})*
 | 
|---|
| 93 | 
 | 
|---|
| 94 |                                 // numeric constants, CFA: '_' in constant
 | 
|---|
| 95 | hex_quad {hex}("_"?{hex}){3}
 | 
|---|
| 96 | size_opt (8|16|32|64|128)?
 | 
|---|
| 97 |                                 // CFA: explicit l8/l16/l32/l64/l128, char 'hh', short 'h', int 'n'
 | 
|---|
| 98 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hHnN])
 | 
|---|
| 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]))?
 | 
|---|
| 101 | 
 | 
|---|
| 102 | octal_digits ({octal})|({octal}({octal}|"_")*{octal})
 | 
|---|
| 103 | octal_prefix "0""_"?
 | 
|---|
| 104 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt}
 | 
|---|
| 105 | 
 | 
|---|
| 106 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal})
 | 
|---|
| 107 | decimal_constant {nonzero_digits}{integer_suffix_opt}
 | 
|---|
| 108 | 
 | 
|---|
| 109 | binary_digits ({binary})|({binary}({binary}|"_")*{binary})
 | 
|---|
| 110 | binary_prefix "0"[bB]"_"?
 | 
|---|
| 111 | binary_constant {binary_prefix}{binary_digits}{integer_suffix_opt}
 | 
|---|
| 112 | 
 | 
|---|
| 113 | hex_digits ({hex})|({hex}({hex}|"_")*{hex})
 | 
|---|
| 114 | hex_prefix "0"[xX]"_"?
 | 
|---|
| 115 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt}
 | 
|---|
| 116 | 
 | 
|---|
| 117 |                                 // GCC: D (double) and iI (imaginary) suffixes, and DL (long double)
 | 
|---|
| 118 | exponent "_"?[eE]"_"?[+-]?{decimal_digits}
 | 
|---|
| 119 | floating_size 16|32|32x|64|64x|80|128|128x
 | 
|---|
| 120 | floating_length ([fFdDlLwWqQ]|[fF]{floating_size})
 | 
|---|
| 121 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length})
 | 
|---|
| 122 | floating_suffix_opt ("_"?({floating_suffix}|"DL"))?
 | 
|---|
| 123 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal})
 | 
|---|
| 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})
 | 
|---|
| 127 | 
 | 
|---|
| 128 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits}
 | 
|---|
| 129 | hex_floating_suffix_opt ("_"?({floating_suffix}))?
 | 
|---|
| 130 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".")
 | 
|---|
| 131 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt}
 | 
|---|
| 132 | 
 | 
|---|
| 133 |                                 // character escape sequence, GCC: \e => esc character
 | 
|---|
| 134 | simple_escape "\\"[abefnrtv'"?\\]
 | 
|---|
| 135 |                                 // ' stop editor highlighting
 | 
|---|
| 136 | octal_escape "\\"{octal}("_"?{octal}){0,2}
 | 
|---|
| 137 | hex_escape "\\""x""_"?{hex_digits}
 | 
|---|
| 138 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char}
 | 
|---|
| 139 | cwide_prefix "L"|"U"|"u"
 | 
|---|
| 140 | swide_prefix {cwide_prefix}|"u8"
 | 
|---|
| 141 | 
 | 
|---|
| 142 |                                 // display/white-space characters
 | 
|---|
| 143 | h_tab [\011]
 | 
|---|
| 144 | form_feed [\014]
 | 
|---|
| 145 | v_tab [\013]
 | 
|---|
| 146 | c_return [\015]
 | 
|---|
| 147 | h_white [ ]|{h_tab}
 | 
|---|
| 148 | 
 | 
|---|
| 149 |                                 // overloadable operators
 | 
|---|
| 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 | 
 | 
|---|
| 155 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>="
 | 
|---|
| 156 | op_binary_over {op_unary_binary}|{op_binary_only}
 | 
|---|
| 157 |                                 // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@="
 | 
|---|
| 158 |                                 // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over}
 | 
|---|
| 159 | 
 | 
|---|
| 160 | %x COMMENT
 | 
|---|
| 161 | %x BKQUOTE
 | 
|---|
| 162 | %x QUOTE
 | 
|---|
| 163 | %x STRING
 | 
|---|
| 164 | 
 | 
|---|
| 165 | %%
 | 
|---|
| 166 |                                 /* line directives */
 | 
|---|
| 167 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" {
 | 
|---|
| 168 |         /* " stop editor highlighting */
 | 
|---|
| 169 |         static char filename[FILENAME_MAX];                                     // temporarily store current source-file name
 | 
|---|
| 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, '"' );
 | 
|---|
| 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
 | 
|---|
| 182 |                 //cout << "file " << filename << " line " << lineno << endl;
 | 
|---|
| 183 |                 yylineno = lineno;
 | 
|---|
| 184 |                 yyfilename = filename;
 | 
|---|
| 185 |         } // if
 | 
|---|
| 186 | }
 | 
|---|
| 187 | 
 | 
|---|
| 188 |                                 /* preprocessor-style directives */
 | 
|---|
| 189 | ^{h_white}*"#"[^\n]*"\n" { RETURN_VAL( DIRECTIVE ); }
 | 
|---|
| 190 | 
 | 
|---|
| 191 |                                 /* ignore C style comments (ALSO HANDLED BY CPP) */
 | 
|---|
| 192 | "/*"                    { BEGIN COMMENT; }
 | 
|---|
| 193 | <COMMENT>.|\n   ;
 | 
|---|
| 194 | <COMMENT>"*/"   { BEGIN 0; }
 | 
|---|
| 195 | 
 | 
|---|
| 196 |                                 /* ignore C++ style comments (ALSO HANDLED BY CPP) */
 | 
|---|
| 197 | "//"[^\n]*"\n"  ;
 | 
|---|
| 198 | 
 | 
|---|
| 199 |                                 /* ignore whitespace */
 | 
|---|
| 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(); }
 | 
|---|
| 203 | 
 | 
|---|
| 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
 | 
|---|
| 215 | auto                    { KEYWORD_RETURN(AUTO); }
 | 
|---|
| 216 | __auto_type             { KEYWORD_RETURN(AUTO_TYPE); }
 | 
|---|
| 217 | basetypeof              { KEYWORD_RETURN(BASETYPEOF); }                 // CFA
 | 
|---|
| 218 | _Bool                   { KEYWORD_RETURN(BOOL); }                               // C99
 | 
|---|
| 219 | break                   { KEYWORD_RETURN(BREAK); }
 | 
|---|
| 220 | case                    { KEYWORD_RETURN(CASE); }
 | 
|---|
| 221 | catch                   { KEYWORD_RETURN(CATCH); }                              // CFA
 | 
|---|
| 222 | catchResume             { KEYWORD_RETURN(CATCHRESUME); }                // CFA
 | 
|---|
| 223 | char                    { KEYWORD_RETURN(CHAR); }
 | 
|---|
| 224 | choose                  { KEYWORD_RETURN(CHOOSE); }                             // CFA
 | 
|---|
| 225 | coerce                  { KEYWORD_RETURN(COERCE); }                             // CFA
 | 
|---|
| 226 | _Complex                { KEYWORD_RETURN(COMPLEX); }                    // C99
 | 
|---|
| 227 | __complex               { KEYWORD_RETURN(COMPLEX); }                    // GCC
 | 
|---|
| 228 | __complex__             { KEYWORD_RETURN(COMPLEX); }                    // GCC
 | 
|---|
| 229 | const                   { KEYWORD_RETURN(CONST); }
 | 
|---|
| 230 | __const                 { KEYWORD_RETURN(CONST); }                              // GCC
 | 
|---|
| 231 | __const__               { KEYWORD_RETURN(CONST); }                              // GCC
 | 
|---|
| 232 | continue                { KEYWORD_RETURN(CONTINUE); }
 | 
|---|
| 233 | coroutine               { KEYWORD_RETURN(COROUTINE); }                  // CFA
 | 
|---|
| 234 | default                 { KEYWORD_RETURN(DEFAULT); }
 | 
|---|
| 235 | disable                 { KEYWORD_RETURN(DISABLE); }                    // CFA
 | 
|---|
| 236 | do                              { KEYWORD_RETURN(DO); }
 | 
|---|
| 237 | double                  { KEYWORD_RETURN(DOUBLE); }
 | 
|---|
| 238 | dtype                   { KEYWORD_RETURN(DTYPE); }                              // CFA
 | 
|---|
| 239 | else                    { KEYWORD_RETURN(ELSE); }
 | 
|---|
| 240 | enable                  { KEYWORD_RETURN(ENABLE); }                             // CFA
 | 
|---|
| 241 | enum                    { KEYWORD_RETURN(ENUM); }
 | 
|---|
| 242 | __extension__   { KEYWORD_RETURN(EXTENSION); }                  // GCC
 | 
|---|
| 243 | exception               { KEYWORD_RETURN(EXCEPTION); }                  // CFA
 | 
|---|
| 244 | extern                  { KEYWORD_RETURN(EXTERN); }
 | 
|---|
| 245 | fallthrough             { KEYWORD_RETURN(FALLTHROUGH); }                // CFA
 | 
|---|
| 246 | fallthru                { KEYWORD_RETURN(FALLTHRU); }                   // CFA
 | 
|---|
| 247 | finally                 { KEYWORD_RETURN(FINALLY); }                    // CFA
 | 
|---|
| 248 | float                   { KEYWORD_RETURN(FLOAT); }
 | 
|---|
| 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
 | 
|---|
| 260 | for                             { KEYWORD_RETURN(FOR); }
 | 
|---|
| 261 | forall                  { KEYWORD_RETURN(FORALL); }                             // CFA
 | 
|---|
| 262 | fortran                 { KEYWORD_RETURN(FORTRAN); }
 | 
|---|
| 263 | ftype                   { KEYWORD_RETURN(FTYPE); }                              // CFA
 | 
|---|
| 264 | generator               { KEYWORD_RETURN(GENERATOR); }                  // CFA
 | 
|---|
| 265 | _Generic                { KEYWORD_RETURN(GENERIC); }                    // C11
 | 
|---|
| 266 | goto                    { KEYWORD_RETURN(GOTO); }
 | 
|---|
| 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); }
 | 
|---|
| 275 | int128                  { KEYWORD_RETURN(INT128); }                             // CFA
 | 
|---|
| 276 | __int128                { KEYWORD_RETURN(INT128); }                             // GCC
 | 
|---|
| 277 | __int128_t              { KEYWORD_RETURN(INT128); }                             // GCC
 | 
|---|
| 278 | __label__               { KEYWORD_RETURN(LABEL); }                              // GCC
 | 
|---|
| 279 | long                    { KEYWORD_RETURN(LONG); }
 | 
|---|
| 280 | monitor                 { KEYWORD_RETURN(MONITOR); }                    // CFA
 | 
|---|
| 281 | mutex                   { KEYWORD_RETURN(MUTEX); }                              // CFA
 | 
|---|
| 282 | _Noreturn               { KEYWORD_RETURN(NORETURN); }                   // C11
 | 
|---|
| 283 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); }                // GCC
 | 
|---|
| 284 | one_t                   { NUMERIC_RETURN(ONE_T); }                              // CFA
 | 
|---|
| 285 | or                              { QKEYWORD_RETURN(WOR); }                               // CFA
 | 
|---|
| 286 | otype                   { KEYWORD_RETURN(OTYPE); }                              // CFA
 | 
|---|
| 287 | register                { KEYWORD_RETURN(REGISTER); }
 | 
|---|
| 288 | restrict                { KEYWORD_RETURN(RESTRICT); }                   // C99
 | 
|---|
| 289 | __restrict              { KEYWORD_RETURN(RESTRICT); }                   // GCC
 | 
|---|
| 290 | __restrict__    { KEYWORD_RETURN(RESTRICT); }                   // GCC
 | 
|---|
| 291 | return                  { KEYWORD_RETURN(RETURN); }
 | 
|---|
| 292 |         /* resume                       { KEYWORD_RETURN(RESUME); }                             // CFA */
 | 
|---|
| 293 | short                   { KEYWORD_RETURN(SHORT); }
 | 
|---|
| 294 | signed                  { KEYWORD_RETURN(SIGNED); }
 | 
|---|
| 295 | __signed                { KEYWORD_RETURN(SIGNED); }                             // GCC
 | 
|---|
| 296 | __signed__              { KEYWORD_RETURN(SIGNED); }                             // GCC
 | 
|---|
| 297 | sizeof                  { KEYWORD_RETURN(SIZEOF); }
 | 
|---|
| 298 | static                  { KEYWORD_RETURN(STATIC); }
 | 
|---|
| 299 | _Static_assert  { KEYWORD_RETURN(STATICASSERT); }               // C11
 | 
|---|
| 300 | struct                  { KEYWORD_RETURN(STRUCT); }
 | 
|---|
| 301 |         /* suspend                      { KEYWORD_RETURN(SUSPEND); }                    // CFA */
 | 
|---|
| 302 | switch                  { KEYWORD_RETURN(SWITCH); }
 | 
|---|
| 303 | thread                  { KEYWORD_RETURN(THREAD); }                             // C11
 | 
|---|
| 304 | _Thread_local   { KEYWORD_RETURN(THREADLOCAL); }                // C11
 | 
|---|
| 305 | throw                   { KEYWORD_RETURN(THROW); }                              // CFA
 | 
|---|
| 306 | throwResume             { KEYWORD_RETURN(THROWRESUME); }                // CFA
 | 
|---|
| 307 | timeout                 { QKEYWORD_RETURN(TIMEOUT); }                   // CFA
 | 
|---|
| 308 | trait                   { KEYWORD_RETURN(TRAIT); }                              // CFA
 | 
|---|
| 309 | try                             { KEYWORD_RETURN(TRY); }                                // CFA
 | 
|---|
| 310 | ttype                   { KEYWORD_RETURN(TTYPE); }                              // CFA
 | 
|---|
| 311 | typedef                 { KEYWORD_RETURN(TYPEDEF); }
 | 
|---|
| 312 | typeof                  { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 313 | __typeof                { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 314 | __typeof__              { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 315 | union                   { KEYWORD_RETURN(UNION); }
 | 
|---|
| 316 | __uint128_t             { KEYWORD_RETURN(UINT128); }                    // GCC
 | 
|---|
| 317 | unsigned                { KEYWORD_RETURN(UNSIGNED); }
 | 
|---|
| 318 | __builtin_va_list { KEYWORD_RETURN(VALIST); }                   // GCC
 | 
|---|
| 319 | virtual                 { KEYWORD_RETURN(VIRTUAL); }                    // CFA
 | 
|---|
| 320 | void                    { KEYWORD_RETURN(VOID); }
 | 
|---|
| 321 | volatile                { KEYWORD_RETURN(VOLATILE); }
 | 
|---|
| 322 | __volatile              { KEYWORD_RETURN(VOLATILE); }                   // GCC
 | 
|---|
| 323 | __volatile__    { KEYWORD_RETURN(VOLATILE); }                   // GCC
 | 
|---|
| 324 | waitfor                 { KEYWORD_RETURN(WAITFOR); }
 | 
|---|
| 325 | when                    { KEYWORD_RETURN(WHEN); }
 | 
|---|
| 326 | while                   { KEYWORD_RETURN(WHILE); }
 | 
|---|
| 327 | with                    { KEYWORD_RETURN(WITH); }                               // CFA
 | 
|---|
| 328 | zero_t                  { NUMERIC_RETURN(ZERO_T); }                             // CFA
 | 
|---|
| 329 | 
 | 
|---|
| 330 |                                 /* identifier */
 | 
|---|
| 331 | {identifier}    { IDENTIFIER_RETURN(); }
 | 
|---|
| 332 | "`"{identifier}"`" {                                                                    // CFA
 | 
|---|
| 333 |         yytext[yyleng - 1] = '\0'; yytext += 1;                         // SKULLDUGGERY: remove backquotes (ok to shorten?)
 | 
|---|
| 334 |         IDENTIFIER_RETURN();
 | 
|---|
| 335 | }
 | 
|---|
| 336 | 
 | 
|---|
| 337 |                                 /* numeric constants */
 | 
|---|
| 338 | {binary_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 339 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 340 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 341 | {hex_constant}  { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 342 | {floating_decimal}      { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant
 | 
|---|
| 343 | {floating_fraction}     { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant
 | 
|---|
| 344 | {floating_constant}     { NUMERIC_RETURN(FLOATINGconstant); }
 | 
|---|
| 345 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
 | 
|---|
| 346 | 
 | 
|---|
| 347 |                                 /* character constant, allows empty value */
 | 
|---|
| 348 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); }
 | 
|---|
| 349 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); }
 | 
|---|
| 350 | <QUOTE>['\n]    { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); }
 | 
|---|
| 351 |                                 /* ' stop editor highlighting */
 | 
|---|
| 352 | 
 | 
|---|
| 353 |                                 /* string constant */
 | 
|---|
| 354 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); }
 | 
|---|
| 355 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); }
 | 
|---|
| 356 | <STRING>["\n]   { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); }
 | 
|---|
| 357 |                                 /* " stop editor highlighting */
 | 
|---|
| 358 | 
 | 
|---|
| 359 |                                 /* common character/string constant */
 | 
|---|
| 360 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); }
 | 
|---|
| 361 | <QUOTE,STRING>"\\"{h_white}*"\n" {}                                             // continuation (ALSO HANDLED BY CPP)
 | 
|---|
| 362 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character
 | 
|---|
| 363 | 
 | 
|---|
| 364 |                                 /* punctuation */
 | 
|---|
| 365 | "@"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 366 | "`"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 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
 | 
|---|
| 377 | "..."                   { NAMEDOP_RETURN(ELLIPSIS); }
 | 
|---|
| 378 | 
 | 
|---|
| 379 |                                 /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */
 | 
|---|
| 380 | "<:"                    { RETURN_VAL('['); }
 | 
|---|
| 381 | ":>"                    { RETURN_VAL(']'); }
 | 
|---|
| 382 | "<%"                    { RETURN_VAL('{'); }
 | 
|---|
| 383 | "%>"                    { RETURN_VAL('}'); }
 | 
|---|
| 384 | 
 | 
|---|
| 385 |                                 /* operators */
 | 
|---|
| 386 | "!"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 387 | "+"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 388 | "-"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 389 | "*"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 390 | "\\"                    { ASCIIOP_RETURN(); }                                   // CFA, exponentiation
 | 
|---|
| 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(); }
 | 
|---|
| 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); }
 | 
|---|
| 415 | "\\="                   { NAMEDOP_RETURN(EXPassign); }                  // CFA, exponentiation
 | 
|---|
| 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); }
 | 
|---|
| 424 | 
 | 
|---|
| 425 | "@="                    { NAMEDOP_RETURN(ATassign); }                   // CFA
 | 
|---|
| 426 | "~="                    { NAMEDOP_RETURN(ErangeUpEq); }                 // CFA
 | 
|---|
| 427 | "-~"                    { NAMEDOP_RETURN(ErangeDown); }                 // CFA
 | 
|---|
| 428 | "-~="                   { NAMEDOP_RETURN(ErangeDownEq); }               // CFA
 | 
|---|
| 429 | 
 | 
|---|
| 430 |                                 /* CFA, operator identifier */
 | 
|---|
| 431 | {op_unary}"?"   { IDENTIFIER_RETURN(); }                                // unary
 | 
|---|
| 432 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); }
 | 
|---|
| 433 | "^?{}"                  { IDENTIFIER_RETURN(); }
 | 
|---|
| 434 | "?`"{identifier} { IDENTIFIER_RETURN(); }                               // postfix operator
 | 
|---|
| 435 | "?"{op_binary_over}"?"  { IDENTIFIER_RETURN(); }                // binary
 | 
|---|
| 436 |         /*
 | 
|---|
| 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;
 | 
|---|
| 461 |         */
 | 
|---|
| 462 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") {
 | 
|---|
| 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 | 
 | 
|---|
| 473 |                                 /* unknown character */
 | 
|---|
| 474 | .                               { yyerror( "unknown character" ); }
 | 
|---|
| 475 | 
 | 
|---|
| 476 | %%
 | 
|---|
| 477 | 
 | 
|---|
| 478 | // ----end of lexer----
 | 
|---|
| 479 | 
 | 
|---|
| 480 | void yyerror( const char * errmsg ) {
 | 
|---|
| 481 |         SemanticErrorThrow = true;
 | 
|---|
| 482 |         cerr << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1
 | 
|---|
| 483 |                  << ": " << ErrorHelpers::error_str() << errmsg << " at token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl;
 | 
|---|
| 484 | }
 | 
|---|
| 485 | 
 | 
|---|
| 486 | // Local Variables: //
 | 
|---|
| 487 | // mode: c++ //
 | 
|---|
| 488 | // tab-width: 4 //
 | 
|---|
| 489 | // compile-command: "make install" //
 | 
|---|
| 490 | // End: //
 | 
|---|