| 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  6 15:16:15 2018
 | 
|---|
| 13 |  * Update Count     : 670
 | 
|---|
| 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 | unsigned int column = 0;                                                                // position of the end of the last token parsed
 | 
|---|
| 28 | #define YY_USER_ACTION yylloc.first_line = yylineno; yylloc.first_column = column; column += yyleng; yylloc.last_column = column; yylloc.last_line = yylineno; yylloc.filename = yyfilename ? yyfilename : "";                          // trigger before each matching rule's action
 | 
|---|
| 29 | 
 | 
|---|
| 30 | #include <string>
 | 
|---|
| 31 | #include <cstdio>                                                                               // FILENAME_MAX
 | 
|---|
| 32 | using namespace std;
 | 
|---|
| 33 | 
 | 
|---|
| 34 | #include "ParseNode.h"
 | 
|---|
| 35 | #include "TypedefTable.h"
 | 
|---|
| 36 | 
 | 
|---|
| 37 | char *yyfilename;
 | 
|---|
| 38 | string *strtext;                                                                                // accumulate parts of character and string constant value
 | 
|---|
| 39 | 
 | 
|---|
| 40 | #define RETURN_LOCN(x)          yylval.tok.loc.file = yyfilename; yylval.tok.loc.line = yylineno; return( x )
 | 
|---|
| 41 | #define RETURN_VAL(x)           yylval.tok.str = new string( yytext ); RETURN_LOCN( x )
 | 
|---|
| 42 | #define RETURN_CHAR(x)          yylval.tok.str = nullptr; RETURN_LOCN( x )
 | 
|---|
| 43 | #define RETURN_STR(x)           yylval.tok.str = strtext; RETURN_LOCN( x )
 | 
|---|
| 44 | 
 | 
|---|
| 45 | #define WHITE_RETURN(x)         // do nothing
 | 
|---|
| 46 | #define NEWLINE_RETURN()        column = 0; WHITE_RETURN( '\n' )
 | 
|---|
| 47 | #define ASCIIOP_RETURN()        RETURN_CHAR( (int)yytext[0] ) // single character operator
 | 
|---|
| 48 | #define NAMEDOP_RETURN(x)       RETURN_CHAR( x )                        // multichar operator, with a name
 | 
|---|
| 49 | #define NUMERIC_RETURN(x)       rm_underscore(); RETURN_VAL( x ) // numeric constant
 | 
|---|
| 50 | #define KEYWORD_RETURN(x)       RETURN_CHAR( x )                        // keyword
 | 
|---|
| 51 | #define QKEYWORD_RETURN(x)      typedefTable.isKind( yytext ); RETURN_VAL(x); // quasi-keyword
 | 
|---|
| 52 | #define IDENTIFIER_RETURN()     RETURN_VAL( typedefTable.isKind( yytext ) )
 | 
|---|
| 53 | #define ATTRIBUTE_RETURN()      RETURN_VAL( ATTR_IDENTIFIER )
 | 
|---|
| 54 | 
 | 
|---|
| 55 | void rm_underscore() {
 | 
|---|
| 56 |         // SKULLDUGGERY: remove underscores (ok to shorten?)
 | 
|---|
| 57 |         yyleng = 0;
 | 
|---|
| 58 |         for ( int i = 0; yytext[i] != '\0'; i += 1 ) {          // copying non-underscore characters to front of string
 | 
|---|
| 59 |                 if ( yytext[i] != '_' ) {
 | 
|---|
| 60 |                         yytext[yyleng] = yytext[i];
 | 
|---|
| 61 |                         yyleng += 1;
 | 
|---|
| 62 |                 } // if
 | 
|---|
| 63 |         } // for
 | 
|---|
| 64 |         yytext[yyleng] = '\0';
 | 
|---|
| 65 | } // rm_underscore
 | 
|---|
| 66 | 
 | 
|---|
| 67 | // Stop warning due to incorrectly generated flex code.
 | 
|---|
| 68 | #pragma GCC diagnostic ignored "-Wsign-compare"
 | 
|---|
| 69 | %}
 | 
|---|
| 70 | 
 | 
|---|
| 71 | binary [0-1]
 | 
|---|
| 72 | octal [0-7]
 | 
|---|
| 73 | nonzero [1-9]
 | 
|---|
| 74 | decimal [0-9]
 | 
|---|
| 75 | hex [0-9a-fA-F]
 | 
|---|
| 76 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2}))
 | 
|---|
| 77 | 
 | 
|---|
| 78 |                                 // identifier, GCC: $ in identifier
 | 
|---|
| 79 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})*
 | 
|---|
| 80 | 
 | 
|---|
| 81 |                                 // attribute identifier, GCC: $ in identifier
 | 
|---|
| 82 | attr_identifier "@"{identifier}
 | 
|---|
| 83 | 
 | 
|---|
| 84 |                                 // numeric constants, CFA: '_' in constant
 | 
|---|
| 85 | hex_quad {hex}("_"?{hex}){3}
 | 
|---|
| 86 | size_opt (8|16|32|64|128)?
 | 
|---|
| 87 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hH])
 | 
|---|
| 88 | integer_suffix_opt ("_"?(([uU]({length}?[iI]?)|([iI]{length}))|([iI]({length}?[uU]?)|([uU]{length}))|({length}([iI]?[uU]?)|([uU][iI]))|[zZ]))?
 | 
|---|
| 89 | 
 | 
|---|
| 90 | octal_digits ({octal})|({octal}({octal}|"_")*{octal})
 | 
|---|
| 91 | octal_prefix "0""_"?
 | 
|---|
| 92 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt}
 | 
|---|
| 93 | 
 | 
|---|
| 94 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal})
 | 
|---|
| 95 | decimal_constant {nonzero_digits}{integer_suffix_opt}
 | 
|---|
| 96 | 
 | 
|---|
| 97 | binary_digits ({binary})|({binary}({binary}|"_")*{binary})
 | 
|---|
| 98 | binary_prefix "0"[bB]"_"?
 | 
|---|
| 99 | binary_constant {binary_prefix}{binary_digits}{integer_suffix_opt}
 | 
|---|
| 100 | 
 | 
|---|
| 101 | hex_digits ({hex})|({hex}({hex}|"_")*{hex})
 | 
|---|
| 102 | hex_prefix "0"[xX]"_"?
 | 
|---|
| 103 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt}
 | 
|---|
| 104 | 
 | 
|---|
| 105 |                                 // GCC: D (double) and iI (imaginary) suffixes, and DL (long double)
 | 
|---|
| 106 | exponent "_"?[eE]"_"?[+-]?{decimal_digits}
 | 
|---|
| 107 | floating_size 32|64|80|128
 | 
|---|
| 108 | floating_length ([fFdDlL]|[lL]{floating_size})
 | 
|---|
| 109 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length})
 | 
|---|
| 110 | floating_suffix_opt ("_"?({floating_suffix}|"DL"))?
 | 
|---|
| 111 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal})
 | 
|---|
| 112 | floating_decimal {decimal_digits}"."{exponent}?{floating_suffix_opt}
 | 
|---|
| 113 | floating_fraction "."{decimal_digits}{exponent}?{floating_suffix_opt}
 | 
|---|
| 114 | floating_constant ({decimal_digits}{exponent}{floating_suffix_opt})|({decimal_digits}{floating_fraction})
 | 
|---|
| 115 | 
 | 
|---|
| 116 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits}
 | 
|---|
| 117 | hex_floating_suffix_opt ("_"?({floating_suffix}))?
 | 
|---|
| 118 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".")
 | 
|---|
| 119 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt}
 | 
|---|
| 120 | 
 | 
|---|
| 121 |                                 // character escape sequence, GCC: \e => esc character
 | 
|---|
| 122 | simple_escape "\\"[abefnrtv'"?\\]
 | 
|---|
| 123 |                                 // ' stop editor highlighting
 | 
|---|
| 124 | octal_escape "\\"{octal}("_"?{octal}){0,2}
 | 
|---|
| 125 | hex_escape "\\""x""_"?{hex_digits}
 | 
|---|
| 126 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char}
 | 
|---|
| 127 | cwide_prefix "L"|"U"|"u"
 | 
|---|
| 128 | swide_prefix {cwide_prefix}|"u8"
 | 
|---|
| 129 | 
 | 
|---|
| 130 |                                 // display/white-space characters
 | 
|---|
| 131 | h_tab [\011]
 | 
|---|
| 132 | form_feed [\014]
 | 
|---|
| 133 | v_tab [\013]
 | 
|---|
| 134 | c_return [\015]
 | 
|---|
| 135 | h_white [ ]|{h_tab}
 | 
|---|
| 136 | 
 | 
|---|
| 137 |                                 // overloadable operators
 | 
|---|
| 138 | op_unary_only "~"|"!"
 | 
|---|
| 139 | op_unary_binary "+"|"-"|"*"
 | 
|---|
| 140 | op_unary_pre_post "++"|"--"
 | 
|---|
| 141 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post}
 | 
|---|
| 142 | 
 | 
|---|
| 143 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>="
 | 
|---|
| 144 | op_binary_over {op_unary_binary}|{op_binary_only}
 | 
|---|
| 145 |                                 // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@="
 | 
|---|
| 146 |                                 // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over}
 | 
|---|
| 147 | 
 | 
|---|
| 148 | %x COMMENT
 | 
|---|
| 149 | %x BKQUOTE
 | 
|---|
| 150 | %x QUOTE
 | 
|---|
| 151 | %x STRING
 | 
|---|
| 152 | 
 | 
|---|
| 153 | %%
 | 
|---|
| 154 |                                 /* line directives */
 | 
|---|
| 155 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" {
 | 
|---|
| 156 |         /* " stop editor highlighting */
 | 
|---|
| 157 |         static char filename[FILENAME_MAX];                                     // temporarily store current source-file name
 | 
|---|
| 158 |         char *end_num;
 | 
|---|
| 159 |         char *begin_string, *end_string;
 | 
|---|
| 160 |         long lineno, length;
 | 
|---|
| 161 |         lineno = strtol( yytext + 1, &end_num, 0 );
 | 
|---|
| 162 |         begin_string = strchr( end_num, '"' );
 | 
|---|
| 163 |         if ( begin_string ) {                                                           // file name ?
 | 
|---|
| 164 |                 end_string = strchr( begin_string + 1, '"' );   // look for ending delimiter
 | 
|---|
| 165 |                 assert( end_string );                                                   // closing quote ?
 | 
|---|
| 166 |                 length = end_string - begin_string - 1;                 // file-name length without quotes or sentinel
 | 
|---|
| 167 |                 assert( length < FILENAME_MAX );                                // room for sentinel ?
 | 
|---|
| 168 |                 memcpy( &filename, begin_string + 1, length );  // copy file name from yytext
 | 
|---|
| 169 |                 filename[ length ] = '\0';                                              // terminate string with sentinel
 | 
|---|
| 170 |                 //cout << "file " << filename << " line " << lineno << endl;
 | 
|---|
| 171 |                 yylineno = lineno;
 | 
|---|
| 172 |                 yyfilename = filename;
 | 
|---|
| 173 |         } // if
 | 
|---|
| 174 | }
 | 
|---|
| 175 | 
 | 
|---|
| 176 |                                 /* ignore preprocessor directives (for now) */
 | 
|---|
| 177 | ^{h_white}*"#"[^\n]*"\n" ;
 | 
|---|
| 178 | 
 | 
|---|
| 179 |                                 /* ignore C style comments (ALSO HANDLED BY CPP) */
 | 
|---|
| 180 | "/*"                    { BEGIN COMMENT; }
 | 
|---|
| 181 | <COMMENT>.|\n   ;
 | 
|---|
| 182 | <COMMENT>"*/"   { BEGIN 0; }
 | 
|---|
| 183 | 
 | 
|---|
| 184 |                                 /* ignore C++ style comments (ALSO HANDLED BY CPP) */
 | 
|---|
| 185 | "//"[^\n]*"\n"  ;
 | 
|---|
| 186 | 
 | 
|---|
| 187 |                                 /* ignore whitespace */
 | 
|---|
| 188 | {h_white}+              { WHITE_RETURN(' '); }
 | 
|---|
| 189 | ({v_tab}|{c_return}|{form_feed})+ { WHITE_RETURN(' '); }
 | 
|---|
| 190 | ({h_white}|{v_tab}|{c_return}|{form_feed})*"\n" { NEWLINE_RETURN(); }
 | 
|---|
| 191 | 
 | 
|---|
| 192 |                                 /* keywords */
 | 
|---|
| 193 | _Alignas                { KEYWORD_RETURN(ALIGNAS); }                    // C11
 | 
|---|
| 194 | _Alignof                { KEYWORD_RETURN(ALIGNOF); }                    // C11
 | 
|---|
| 195 | __alignof               { KEYWORD_RETURN(ALIGNOF); }                    // GCC
 | 
|---|
| 196 | __alignof__             { KEYWORD_RETURN(ALIGNOF); }                    // GCC
 | 
|---|
| 197 | asm                             { KEYWORD_RETURN(ASM); }
 | 
|---|
| 198 | __asm                   { KEYWORD_RETURN(ASM); }                                // GCC
 | 
|---|
| 199 | __asm__                 { KEYWORD_RETURN(ASM); }                                // GCC
 | 
|---|
| 200 | _Atomic                 { KEYWORD_RETURN(ATOMIC); }                             // C11
 | 
|---|
| 201 | __attribute             { KEYWORD_RETURN(ATTRIBUTE); }                  // GCC
 | 
|---|
| 202 | __attribute__   { KEYWORD_RETURN(ATTRIBUTE); }                  // GCC
 | 
|---|
| 203 | auto                    { KEYWORD_RETURN(AUTO); }
 | 
|---|
| 204 | _Bool                   { KEYWORD_RETURN(BOOL); }                               // C99
 | 
|---|
| 205 | break                   { KEYWORD_RETURN(BREAK); }
 | 
|---|
| 206 | case                    { KEYWORD_RETURN(CASE); }
 | 
|---|
| 207 | catch                   { KEYWORD_RETURN(CATCH); }                              // CFA
 | 
|---|
| 208 | catchResume             { KEYWORD_RETURN(CATCHRESUME); }                // CFA
 | 
|---|
| 209 | char                    { KEYWORD_RETURN(CHAR); }
 | 
|---|
| 210 | choose                  { KEYWORD_RETURN(CHOOSE); }                             // CFA
 | 
|---|
| 211 | _Complex                { KEYWORD_RETURN(COMPLEX); }                    // C99
 | 
|---|
| 212 | __complex               { KEYWORD_RETURN(COMPLEX); }                    // GCC
 | 
|---|
| 213 | __complex__             { KEYWORD_RETURN(COMPLEX); }                    // GCC
 | 
|---|
| 214 | const                   { KEYWORD_RETURN(CONST); }
 | 
|---|
| 215 | __const                 { KEYWORD_RETURN(CONST); }                              // GCC
 | 
|---|
| 216 | __const__               { KEYWORD_RETURN(CONST); }                              // GCC
 | 
|---|
| 217 | continue                { KEYWORD_RETURN(CONTINUE); }
 | 
|---|
| 218 | coroutine               { KEYWORD_RETURN(COROUTINE); }                  // CFA
 | 
|---|
| 219 | default                 { KEYWORD_RETURN(DEFAULT); }
 | 
|---|
| 220 | disable                 { KEYWORD_RETURN(DISABLE); }                    // CFA
 | 
|---|
| 221 | do                              { KEYWORD_RETURN(DO); }
 | 
|---|
| 222 | double                  { KEYWORD_RETURN(DOUBLE); }
 | 
|---|
| 223 | dtype                   { KEYWORD_RETURN(DTYPE); }                              // CFA
 | 
|---|
| 224 | else                    { KEYWORD_RETURN(ELSE); }
 | 
|---|
| 225 | enable                  { KEYWORD_RETURN(ENABLE); }                             // CFA
 | 
|---|
| 226 | enum                    { KEYWORD_RETURN(ENUM); }
 | 
|---|
| 227 | __extension__   { KEYWORD_RETURN(EXTENSION); }                  // GCC
 | 
|---|
| 228 | exception               { KEYWORD_RETURN(EXCEPTION); }                  // CFA
 | 
|---|
| 229 | extern                  { KEYWORD_RETURN(EXTERN); }
 | 
|---|
| 230 | fallthrough             { KEYWORD_RETURN(FALLTHROUGH); }                // CFA
 | 
|---|
| 231 | fallthru                { KEYWORD_RETURN(FALLTHRU); }                   // CFA
 | 
|---|
| 232 | finally                 { KEYWORD_RETURN(FINALLY); }                    // CFA
 | 
|---|
| 233 | float                   { KEYWORD_RETURN(FLOAT); }
 | 
|---|
| 234 | __float80               { KEYWORD_RETURN(FLOAT80); }                    // GCC
 | 
|---|
| 235 | float80                 { KEYWORD_RETURN(FLOAT80); }                    // GCC
 | 
|---|
| 236 | __float128              { KEYWORD_RETURN(FLOAT128); }                   // GCC
 | 
|---|
| 237 | float128                { KEYWORD_RETURN(FLOAT128); }                   // GCC
 | 
|---|
| 238 | for                             { KEYWORD_RETURN(FOR); }
 | 
|---|
| 239 | forall                  { KEYWORD_RETURN(FORALL); }                             // CFA
 | 
|---|
| 240 | fortran                 { KEYWORD_RETURN(FORTRAN); }
 | 
|---|
| 241 | ftype                   { KEYWORD_RETURN(FTYPE); }                              // CFA
 | 
|---|
| 242 | _Generic                { KEYWORD_RETURN(GENERIC); }                    // C11
 | 
|---|
| 243 | goto                    { KEYWORD_RETURN(GOTO); }
 | 
|---|
| 244 | if                              { KEYWORD_RETURN(IF); }
 | 
|---|
| 245 | _Imaginary              { KEYWORD_RETURN(IMAGINARY); }                  // C99
 | 
|---|
| 246 | __imag                  { KEYWORD_RETURN(IMAGINARY); }                  // GCC
 | 
|---|
| 247 | __imag__                { KEYWORD_RETURN(IMAGINARY); }                  // GCC
 | 
|---|
| 248 | inline                  { KEYWORD_RETURN(INLINE); }                             // C99
 | 
|---|
| 249 | __inline                { KEYWORD_RETURN(INLINE); }                             // GCC
 | 
|---|
| 250 | __inline__              { KEYWORD_RETURN(INLINE); }                             // GCC
 | 
|---|
| 251 | int                             { KEYWORD_RETURN(INT); }
 | 
|---|
| 252 | __int128                { KEYWORD_RETURN(INT128); }                             // GCC
 | 
|---|
| 253 | int128                  { KEYWORD_RETURN(INT128); }                             // GCC
 | 
|---|
| 254 | __label__               { KEYWORD_RETURN(LABEL); }                              // GCC
 | 
|---|
| 255 | long                    { KEYWORD_RETURN(LONG); }
 | 
|---|
| 256 | monitor                 { KEYWORD_RETURN(MONITOR); }                    // CFA
 | 
|---|
| 257 | mutex                   { KEYWORD_RETURN(MUTEX); }                              // CFA
 | 
|---|
| 258 | _Noreturn               { KEYWORD_RETURN(NORETURN); }                   // C11
 | 
|---|
| 259 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); }                // GCC
 | 
|---|
| 260 | one_t                   { NUMERIC_RETURN(ONE_T); }                              // CFA
 | 
|---|
| 261 | or                              { QKEYWORD_RETURN(WOR); }                               // CFA
 | 
|---|
| 262 | otype                   { KEYWORD_RETURN(OTYPE); }                              // CFA
 | 
|---|
| 263 | register                { KEYWORD_RETURN(REGISTER); }
 | 
|---|
| 264 | restrict                { KEYWORD_RETURN(RESTRICT); }                   // C99
 | 
|---|
| 265 | __restrict              { KEYWORD_RETURN(RESTRICT); }                   // GCC
 | 
|---|
| 266 | __restrict__    { KEYWORD_RETURN(RESTRICT); }                   // GCC
 | 
|---|
| 267 | return                  { KEYWORD_RETURN(RETURN); }
 | 
|---|
| 268 | short                   { KEYWORD_RETURN(SHORT); }
 | 
|---|
| 269 | signed                  { KEYWORD_RETURN(SIGNED); }
 | 
|---|
| 270 | __signed                { KEYWORD_RETURN(SIGNED); }                             // GCC
 | 
|---|
| 271 | __signed__              { KEYWORD_RETURN(SIGNED); }                             // GCC
 | 
|---|
| 272 | sizeof                  { KEYWORD_RETURN(SIZEOF); }
 | 
|---|
| 273 | static                  { KEYWORD_RETURN(STATIC); }
 | 
|---|
| 274 | _Static_assert  { KEYWORD_RETURN(STATICASSERT); }               // C11
 | 
|---|
| 275 | struct                  { KEYWORD_RETURN(STRUCT); }
 | 
|---|
| 276 | switch                  { KEYWORD_RETURN(SWITCH); }
 | 
|---|
| 277 | thread                  { KEYWORD_RETURN(THREAD); }                             // C11
 | 
|---|
| 278 | _Thread_local   { KEYWORD_RETURN(THREADLOCAL); }                // C11
 | 
|---|
| 279 | throw                   { KEYWORD_RETURN(THROW); }                              // CFA
 | 
|---|
| 280 | throwResume             { KEYWORD_RETURN(THROWRESUME); }                // CFA
 | 
|---|
| 281 | timeout                 { QKEYWORD_RETURN(TIMEOUT); }                   // CFA
 | 
|---|
| 282 | trait                   { KEYWORD_RETURN(TRAIT); }                              // CFA
 | 
|---|
| 283 | try                             { KEYWORD_RETURN(TRY); }                                // CFA
 | 
|---|
| 284 | ttype                   { KEYWORD_RETURN(TTYPE); }                              // CFA
 | 
|---|
| 285 | typedef                 { KEYWORD_RETURN(TYPEDEF); }
 | 
|---|
| 286 | typeof                  { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 287 | __typeof                { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 288 | __typeof__              { KEYWORD_RETURN(TYPEOF); }                             // GCC
 | 
|---|
| 289 | union                   { KEYWORD_RETURN(UNION); }
 | 
|---|
| 290 | unsigned                { KEYWORD_RETURN(UNSIGNED); }
 | 
|---|
| 291 | __builtin_va_list { KEYWORD_RETURN(VALIST); }                   // GCC
 | 
|---|
| 292 | virtual                 { KEYWORD_RETURN(VIRTUAL); }                    // CFA
 | 
|---|
| 293 | void                    { KEYWORD_RETURN(VOID); }
 | 
|---|
| 294 | volatile                { KEYWORD_RETURN(VOLATILE); }
 | 
|---|
| 295 | __volatile              { KEYWORD_RETURN(VOLATILE); }                   // GCC
 | 
|---|
| 296 | __volatile__    { KEYWORD_RETURN(VOLATILE); }                   // GCC
 | 
|---|
| 297 | waitfor                 { KEYWORD_RETURN(WAITFOR); }
 | 
|---|
| 298 | when                    { KEYWORD_RETURN(WHEN); }
 | 
|---|
| 299 | while                   { KEYWORD_RETURN(WHILE); }
 | 
|---|
| 300 | with                    { KEYWORD_RETURN(WITH); }                               // CFA
 | 
|---|
| 301 | zero_t                  { NUMERIC_RETURN(ZERO_T); }                             // CFA
 | 
|---|
| 302 | 
 | 
|---|
| 303 |                                 /* identifier */
 | 
|---|
| 304 | {identifier}    { IDENTIFIER_RETURN(); }
 | 
|---|
| 305 | "`"{identifier}"`" {                                                                    // CFA
 | 
|---|
| 306 |         yytext[yyleng - 1] = '\0'; yytext += 1;                         // SKULLDUGGERY: remove backquotes (ok to shorten?)
 | 
|---|
| 307 |         IDENTIFIER_RETURN();
 | 
|---|
| 308 | }
 | 
|---|
| 309 | {attr_identifier} { ATTRIBUTE_RETURN(); }
 | 
|---|
| 310 | 
 | 
|---|
| 311 |                                 /* numeric constants */
 | 
|---|
| 312 | {binary_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 313 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 314 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 315 | {hex_constant}  { NUMERIC_RETURN(INTEGERconstant); }
 | 
|---|
| 316 | {floating_decimal}      { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant
 | 
|---|
| 317 | {floating_fraction}     { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant
 | 
|---|
| 318 | {floating_constant}     { NUMERIC_RETURN(FLOATINGconstant); }
 | 
|---|
| 319 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); }
 | 
|---|
| 320 | 
 | 
|---|
| 321 |                                 /* character constant, allows empty value */
 | 
|---|
| 322 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); }
 | 
|---|
| 323 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); }
 | 
|---|
| 324 | <QUOTE>['\n]    { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); }
 | 
|---|
| 325 |                                 /* ' stop editor highlighting */
 | 
|---|
| 326 | 
 | 
|---|
| 327 |                                 /* string constant */
 | 
|---|
| 328 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); }
 | 
|---|
| 329 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); }
 | 
|---|
| 330 | <STRING>["\n]   { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); }
 | 
|---|
| 331 |                                 /* " stop editor highlighting */
 | 
|---|
| 332 | 
 | 
|---|
| 333 |                                 /* common character/string constant */
 | 
|---|
| 334 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); }
 | 
|---|
| 335 | <QUOTE,STRING>"\\"{h_white}*"\n" {}                                             // continuation (ALSO HANDLED BY CPP)
 | 
|---|
| 336 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character
 | 
|---|
| 337 | 
 | 
|---|
| 338 |                                 /* punctuation */
 | 
|---|
| 339 | "@"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 340 | "`"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 341 | "["                             { ASCIIOP_RETURN(); }
 | 
|---|
| 342 | "]"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 343 | "("                             { ASCIIOP_RETURN(); }
 | 
|---|
| 344 | ")"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 345 | "{"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 346 | "}"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 347 | ","                             { ASCIIOP_RETURN(); }                                   // also operator
 | 
|---|
| 348 | ":"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 349 | ";"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 350 | "."                             { ASCIIOP_RETURN(); }                                   // also operator
 | 
|---|
| 351 | "..."                   { NAMEDOP_RETURN(ELLIPSIS); }
 | 
|---|
| 352 | 
 | 
|---|
| 353 |                                 /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */
 | 
|---|
| 354 | "<:"                    { RETURN_VAL('['); }
 | 
|---|
| 355 | ":>"                    { RETURN_VAL(']'); }
 | 
|---|
| 356 | "<%"                    { RETURN_VAL('{'); }
 | 
|---|
| 357 | "%>"                    { RETURN_VAL('}'); }
 | 
|---|
| 358 | 
 | 
|---|
| 359 |                                 /* operators */
 | 
|---|
| 360 | "!"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 361 | "+"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 362 | "-"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 363 | "*"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 364 | "\\"                    { ASCIIOP_RETURN(); }                                   // CFA, exponentiation
 | 
|---|
| 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(); }
 | 
|---|
| 374 | "?"                             { ASCIIOP_RETURN(); }
 | 
|---|
| 375 | 
 | 
|---|
| 376 | "++"                    { NAMEDOP_RETURN(ICR); }
 | 
|---|
| 377 | "--"                    { NAMEDOP_RETURN(DECR); }
 | 
|---|
| 378 | "=="                    { NAMEDOP_RETURN(EQ); }
 | 
|---|
| 379 | "!="                    { NAMEDOP_RETURN(NE); }
 | 
|---|
| 380 | "<<"                    { NAMEDOP_RETURN(LS); }
 | 
|---|
| 381 | ">>"                    { NAMEDOP_RETURN(RS); }
 | 
|---|
| 382 | "<="                    { NAMEDOP_RETURN(LE); }
 | 
|---|
| 383 | ">="                    { NAMEDOP_RETURN(GE); }
 | 
|---|
| 384 | "&&"                    { NAMEDOP_RETURN(ANDAND); }
 | 
|---|
| 385 | "||"                    { NAMEDOP_RETURN(OROR); }
 | 
|---|
| 386 | "->"                    { NAMEDOP_RETURN(ARROW); }
 | 
|---|
| 387 | "+="                    { NAMEDOP_RETURN(PLUSassign); }
 | 
|---|
| 388 | "-="                    { NAMEDOP_RETURN(MINUSassign); }
 | 
|---|
| 389 | "\\="                   { NAMEDOP_RETURN(EXPassign); }                  // CFA, exponentiation
 | 
|---|
| 390 | "*="                    { NAMEDOP_RETURN(MULTassign); }
 | 
|---|
| 391 | "/="                    { NAMEDOP_RETURN(DIVassign); }
 | 
|---|
| 392 | "%="                    { NAMEDOP_RETURN(MODassign); }
 | 
|---|
| 393 | "&="                    { NAMEDOP_RETURN(ANDassign); }
 | 
|---|
| 394 | "|="                    { NAMEDOP_RETURN(ORassign); }
 | 
|---|
| 395 | "^="                    { NAMEDOP_RETURN(ERassign); }
 | 
|---|
| 396 | "<<="                   { NAMEDOP_RETURN(LSassign); }
 | 
|---|
| 397 | ">>="                   { NAMEDOP_RETURN(RSassign); }
 | 
|---|
| 398 | 
 | 
|---|
| 399 | "@="                    { NAMEDOP_RETURN(ATassign); }                   // CFA
 | 
|---|
| 400 | 
 | 
|---|
| 401 |                                 /* CFA, operator identifier */
 | 
|---|
| 402 | {op_unary}"?"   { IDENTIFIER_RETURN(); }                                // unary
 | 
|---|
| 403 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); }
 | 
|---|
| 404 | "^?{}"                  { IDENTIFIER_RETURN(); }
 | 
|---|
| 405 | "?`"{identifier} { IDENTIFIER_RETURN(); }                               // postfix operator
 | 
|---|
| 406 | "?"{op_binary_over}"?"  { IDENTIFIER_RETURN(); }                // binary
 | 
|---|
| 407 |         /*
 | 
|---|
| 408 |           This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?"  can be
 | 
|---|
| 409 |           lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier,
 | 
|---|
| 410 |           e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is
 | 
|---|
| 411 |           a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then
 | 
|---|
| 412 |           reparsing the trailing operator identifier.  Otherwise a space is needed between the unary operator and operator
 | 
|---|
| 413 |           identifier to disambiguate this common case.
 | 
|---|
| 414 | 
 | 
|---|
| 415 |           A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers.  The ambiguity
 | 
|---|
| 416 |           occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace
 | 
|---|
| 417 |           look-ahead for the routine-call parameter-list to disambiguate.  However, the dereference operator must have a
 | 
|---|
| 418 |           parameter/argument to dereference *?(...).  Hence, always interpreting the string *?() as * ?() does not preclude
 | 
|---|
| 419 |           any meaningful program.
 | 
|---|
| 420 | 
 | 
|---|
| 421 |           The remaining cases are with the increment/decrement operators and conditional expression:
 | 
|---|
| 422 | 
 | 
|---|
| 423 |           i++? ...(...);
 | 
|---|
| 424 |           i?++ ...(...);
 | 
|---|
| 425 | 
 | 
|---|
| 426 |           requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an
 | 
|---|
| 427 |       incorrect expression (juxtaposed identifiers).  Therefore, it is necessary to disambiguate these cases with a
 | 
|---|
| 428 |       space:
 | 
|---|
| 429 | 
 | 
|---|
| 430 |           i++ ? i : 0;
 | 
|---|
| 431 |           i? ++i : 0;
 | 
|---|
| 432 |         */
 | 
|---|
| 433 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") {
 | 
|---|
| 434 |         // 1 or 2 character unary operator ?
 | 
|---|
| 435 |         int i = yytext[1] == '?' ? 1 : 2;
 | 
|---|
| 436 |         yyless( i );            // put back characters up to first '?'
 | 
|---|
| 437 |         if ( i > 1 ) {
 | 
|---|
| 438 |                 NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR );
 | 
|---|
| 439 |         } else {
 | 
|---|
| 440 |                 ASCIIOP_RETURN();
 | 
|---|
| 441 |         } // if
 | 
|---|
| 442 | }
 | 
|---|
| 443 | 
 | 
|---|
| 444 |                                 /* unknown character */
 | 
|---|
| 445 | .                               { yyerror( "unknown character" ); }
 | 
|---|
| 446 | 
 | 
|---|
| 447 | %%
 | 
|---|
| 448 | // ----end of lexer----
 | 
|---|
| 449 | 
 | 
|---|
| 450 | void yyerror( const char * errmsg ) {
 | 
|---|
| 451 |         cout << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1
 | 
|---|
| 452 |                  << ": " << ErrorHelpers::error_str() << errmsg << " at token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl;
 | 
|---|
| 453 | }
 | 
|---|
| 454 | 
 | 
|---|
| 455 | // Local Variables: //
 | 
|---|
| 456 | // mode: c++ //
 | 
|---|
| 457 | // tab-width: 4 //
 | 
|---|
| 458 | // compile-command: "make install" //
 | 
|---|
| 459 | // End: //
 | 
|---|