| 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 : Thu Feb 22 18:11:27 2018 | 
|---|
| 13 | * Update Count     : 637 | 
|---|
| 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 | // Remove underscores in numeric constant by copying the non-underscore characters to the front of the string. | 
|---|
| 57 | yyleng = 0; | 
|---|
| 58 | for ( int i = 0; yytext[i] != '\0'; i += 1 ) { | 
|---|
| 59 | if ( yytext[i] == '`' ) { | 
|---|
| 60 | // copy user suffix | 
|---|
| 61 | for ( ; yytext[i] != '\0'; i += 1 ) { | 
|---|
| 62 | yytext[yyleng] = yytext[i]; | 
|---|
| 63 | yyleng += 1; | 
|---|
| 64 | } // for | 
|---|
| 65 | break; | 
|---|
| 66 | } // if | 
|---|
| 67 | if ( yytext[i] != '_' ) { | 
|---|
| 68 | yytext[yyleng] = yytext[i]; | 
|---|
| 69 | yyleng += 1; | 
|---|
| 70 | } // if | 
|---|
| 71 | } // for | 
|---|
| 72 | yytext[yyleng] = '\0'; | 
|---|
| 73 | } | 
|---|
| 74 |  | 
|---|
| 75 | // Stop warning due to incorrectly generated flex code. | 
|---|
| 76 | #pragma GCC diagnostic ignored "-Wsign-compare" | 
|---|
| 77 | %} | 
|---|
| 78 |  | 
|---|
| 79 | octal [0-7] | 
|---|
| 80 | nonzero [1-9] | 
|---|
| 81 | decimal [0-9] | 
|---|
| 82 | hex [0-9a-fA-F] | 
|---|
| 83 | universal_char "\\"((u"_"?{hex_quad})|(U"_"?{hex_quad}{2})) | 
|---|
| 84 |  | 
|---|
| 85 | // identifier, GCC: $ in identifier | 
|---|
| 86 | identifier ([a-zA-Z_$]|{universal_char})([0-9a-zA-Z_$]|{universal_char})* | 
|---|
| 87 |  | 
|---|
| 88 | // attribute identifier, GCC: $ in identifier | 
|---|
| 89 | attr_identifier "@"{identifier} | 
|---|
| 90 |  | 
|---|
| 91 | user_suffix_opt ("`"{identifier})? | 
|---|
| 92 |  | 
|---|
| 93 | // numeric constants, CFA: '_' in constant | 
|---|
| 94 | hex_quad {hex}("_"?{hex}){3} | 
|---|
| 95 | size_opt (8|16|32|64|128)? | 
|---|
| 96 | length ("ll"|"LL"|[lL]{size_opt})|("hh"|"HH"|[hH]) | 
|---|
| 97 | integer_suffix_opt ("_"?(([uU]({length}?[iI]?)|([iI]{length}))|([iI]({length}?[uU]?)|([uU]{length}))|({length}([iI]?[uU]?)|([uU][iI]))|[zZ]))?{user_suffix_opt} | 
|---|
| 98 |  | 
|---|
| 99 | octal_digits ({octal})|({octal}({octal}|"_")*{octal}) | 
|---|
| 100 | octal_prefix "0""_"? | 
|---|
| 101 | octal_constant (("0")|({octal_prefix}{octal_digits})){integer_suffix_opt} | 
|---|
| 102 |  | 
|---|
| 103 | nonzero_digits ({nonzero})|({nonzero}({decimal}|"_")*{decimal}) | 
|---|
| 104 | decimal_constant {nonzero_digits}{integer_suffix_opt} | 
|---|
| 105 |  | 
|---|
| 106 | hex_digits ({hex})|({hex}({hex}|"_")*{hex}) | 
|---|
| 107 | hex_prefix "0"[xX]"_"? | 
|---|
| 108 | hex_constant {hex_prefix}{hex_digits}{integer_suffix_opt} | 
|---|
| 109 |  | 
|---|
| 110 | // GCC: D (double) and iI (imaginary) suffixes, and DL (long double) | 
|---|
| 111 | exponent "_"?[eE]"_"?[+-]?{decimal_digits} | 
|---|
| 112 | floating_size 32|64|80|128 | 
|---|
| 113 | floating_length ([fFdDlL]|[lL]{floating_size}) | 
|---|
| 114 | floating_suffix ({floating_length}?[iI]?)|([iI]{floating_length}) | 
|---|
| 115 | floating_suffix_opt ("_"?({floating_suffix}|"DL"))?{user_suffix_opt} | 
|---|
| 116 | decimal_digits ({decimal})|({decimal}({decimal}|"_")*{decimal}) | 
|---|
| 117 | floating_decimal {decimal_digits}"."{exponent}?{floating_suffix_opt} | 
|---|
| 118 | floating_fraction "."{decimal_digits}{exponent}?{floating_suffix_opt} | 
|---|
| 119 | floating_constant ({decimal_digits}{exponent}{floating_suffix_opt})|({decimal_digits}{floating_fraction}) | 
|---|
| 120 |  | 
|---|
| 121 | binary_exponent "_"?[pP]"_"?[+-]?{decimal_digits} | 
|---|
| 122 | hex_floating_suffix_opt ("_"?({floating_suffix}))?{user_suffix_opt} | 
|---|
| 123 | hex_floating_fraction ({hex_digits}?"."{hex_digits})|({hex_digits}".") | 
|---|
| 124 | hex_floating_constant {hex_prefix}(({hex_floating_fraction}{binary_exponent})|({hex_digits}{binary_exponent})){hex_floating_suffix_opt} | 
|---|
| 125 |  | 
|---|
| 126 | // character escape sequence, GCC: \e => esc character | 
|---|
| 127 | simple_escape "\\"[abefnrtv'"?\\] | 
|---|
| 128 | // ' stop editor highlighting | 
|---|
| 129 | octal_escape "\\"{octal}("_"?{octal}){0,2} | 
|---|
| 130 | hex_escape "\\""x""_"?{hex_digits} | 
|---|
| 131 | escape_seq {simple_escape}|{octal_escape}|{hex_escape}|{universal_char} | 
|---|
| 132 | cwide_prefix "L"|"U"|"u" | 
|---|
| 133 | swide_prefix {cwide_prefix}|"u8" | 
|---|
| 134 |  | 
|---|
| 135 | // display/white-space characters | 
|---|
| 136 | h_tab [\011] | 
|---|
| 137 | form_feed [\014] | 
|---|
| 138 | v_tab [\013] | 
|---|
| 139 | c_return [\015] | 
|---|
| 140 | h_white [ ]|{h_tab} | 
|---|
| 141 |  | 
|---|
| 142 | // overloadable operators | 
|---|
| 143 | op_unary_only "~"|"!" | 
|---|
| 144 | op_unary_binary "+"|"-"|"*" | 
|---|
| 145 | op_unary_pre_post "++"|"--" | 
|---|
| 146 | op_unary {op_unary_only}|{op_unary_binary}|{op_unary_pre_post} | 
|---|
| 147 |  | 
|---|
| 148 | op_binary_only "/"|"%"|"\\"|"^"|"&"|"|"|"<"|">"|"="|"=="|"!="|"<<"|">>"|"<="|">="|"+="|"-="|"*="|"/="|"%="|"\\="|"&="|"|="|"^="|"<<="|">>=" | 
|---|
| 149 | op_binary_over {op_unary_binary}|{op_binary_only} | 
|---|
| 150 | // op_binary_not_over "?"|"->"|"."|"&&"|"||"|"@=" | 
|---|
| 151 | // operator {op_unary_pre_post}|{op_binary_over}|{op_binary_not_over} | 
|---|
| 152 |  | 
|---|
| 153 | %x COMMENT | 
|---|
| 154 | %x BKQUOTE | 
|---|
| 155 | %x QUOTE | 
|---|
| 156 | %x STRING | 
|---|
| 157 |  | 
|---|
| 158 | %% | 
|---|
| 159 | /* line directives */ | 
|---|
| 160 | ^{h_white}*"#"{h_white}*[0-9]+{h_white}*["][^"\n]+["].*"\n" { | 
|---|
| 161 | /* " stop editor highlighting */ | 
|---|
| 162 | static char filename[FILENAME_MAX];                                     // temporarily store current source-file name | 
|---|
| 163 | char *end_num; | 
|---|
| 164 | char *begin_string, *end_string; | 
|---|
| 165 | long lineno, length; | 
|---|
| 166 | lineno = strtol( yytext + 1, &end_num, 0 ); | 
|---|
| 167 | begin_string = strchr( end_num, '"' ); | 
|---|
| 168 | if ( begin_string ) {                                                           // file name ? | 
|---|
| 169 | end_string = strchr( begin_string + 1, '"' );   // look for ending delimiter | 
|---|
| 170 | assert( end_string );                                                   // closing quote ? | 
|---|
| 171 | length = end_string - begin_string - 1;                 // file-name length without quotes or sentinel | 
|---|
| 172 | assert( length < FILENAME_MAX );                                // room for sentinel ? | 
|---|
| 173 | memcpy( &filename, begin_string + 1, length );  // copy file name from yytext | 
|---|
| 174 | filename[ length ] = '\0';                                              // terminate string with sentinel | 
|---|
| 175 | //cout << "file " << filename << " line " << lineno << endl; | 
|---|
| 176 | yylineno = lineno; | 
|---|
| 177 | yyfilename = filename; | 
|---|
| 178 | } // if | 
|---|
| 179 | } | 
|---|
| 180 |  | 
|---|
| 181 | /* ignore preprocessor directives (for now) */ | 
|---|
| 182 | ^{h_white}*"#"[^\n]*"\n" ; | 
|---|
| 183 |  | 
|---|
| 184 | /* ignore C style comments (ALSO HANDLED BY CPP) */ | 
|---|
| 185 | "/*"                    { BEGIN COMMENT; } | 
|---|
| 186 | <COMMENT>.|\n   ; | 
|---|
| 187 | <COMMENT>"*/"   { BEGIN 0; } | 
|---|
| 188 |  | 
|---|
| 189 | /* ignore C++ style comments (ALSO HANDLED BY CPP) */ | 
|---|
| 190 | "//"[^\n]*"\n"  ; | 
|---|
| 191 |  | 
|---|
| 192 | /* ignore whitespace */ | 
|---|
| 193 | {h_white}+              { WHITE_RETURN(' '); } | 
|---|
| 194 | ({v_tab}|{c_return}|{form_feed})+ { WHITE_RETURN(' '); } | 
|---|
| 195 | ({h_white}|{v_tab}|{c_return}|{form_feed})*"\n" { NEWLINE_RETURN(); } | 
|---|
| 196 |  | 
|---|
| 197 | /* keywords */ | 
|---|
| 198 | _Alignas                { KEYWORD_RETURN(ALIGNAS); }                    // C11 | 
|---|
| 199 | _Alignof                { KEYWORD_RETURN(ALIGNOF); }                    // C11 | 
|---|
| 200 | __alignof               { KEYWORD_RETURN(ALIGNOF); }                    // GCC | 
|---|
| 201 | __alignof__             { KEYWORD_RETURN(ALIGNOF); }                    // GCC | 
|---|
| 202 | asm                             { KEYWORD_RETURN(ASM); } | 
|---|
| 203 | __asm                   { KEYWORD_RETURN(ASM); }                                // GCC | 
|---|
| 204 | __asm__                 { KEYWORD_RETURN(ASM); }                                // GCC | 
|---|
| 205 | _At                             { KEYWORD_RETURN(AT); }                                 // CFA | 
|---|
| 206 | _Atomic                 { KEYWORD_RETURN(ATOMIC); }                             // C11 | 
|---|
| 207 | __attribute             { KEYWORD_RETURN(ATTRIBUTE); }                  // GCC | 
|---|
| 208 | __attribute__   { KEYWORD_RETURN(ATTRIBUTE); }                  // GCC | 
|---|
| 209 | auto                    { KEYWORD_RETURN(AUTO); } | 
|---|
| 210 | _Bool                   { KEYWORD_RETURN(BOOL); }                               // C99 | 
|---|
| 211 | break                   { KEYWORD_RETURN(BREAK); } | 
|---|
| 212 | case                    { KEYWORD_RETURN(CASE); } | 
|---|
| 213 | catch                   { KEYWORD_RETURN(CATCH); }                              // CFA | 
|---|
| 214 | catchResume             { KEYWORD_RETURN(CATCHRESUME); }                // CFA | 
|---|
| 215 | char                    { KEYWORD_RETURN(CHAR); } | 
|---|
| 216 | choose                  { KEYWORD_RETURN(CHOOSE); }                             // CFA | 
|---|
| 217 | _Complex                { KEYWORD_RETURN(COMPLEX); }                    // C99 | 
|---|
| 218 | __complex               { KEYWORD_RETURN(COMPLEX); }                    // GCC | 
|---|
| 219 | __complex__             { KEYWORD_RETURN(COMPLEX); }                    // GCC | 
|---|
| 220 | const                   { KEYWORD_RETURN(CONST); } | 
|---|
| 221 | __const                 { KEYWORD_RETURN(CONST); }                              // GCC | 
|---|
| 222 | __const__               { KEYWORD_RETURN(CONST); }                              // GCC | 
|---|
| 223 | continue                { KEYWORD_RETURN(CONTINUE); } | 
|---|
| 224 | coroutine               { KEYWORD_RETURN(COROUTINE); }                  // CFA | 
|---|
| 225 | default                 { KEYWORD_RETURN(DEFAULT); } | 
|---|
| 226 | disable                 { KEYWORD_RETURN(DISABLE); }                    // CFA | 
|---|
| 227 | do                              { KEYWORD_RETURN(DO); } | 
|---|
| 228 | double                  { KEYWORD_RETURN(DOUBLE); } | 
|---|
| 229 | dtype                   { KEYWORD_RETURN(DTYPE); }                              // CFA | 
|---|
| 230 | else                    { KEYWORD_RETURN(ELSE); } | 
|---|
| 231 | enable                  { KEYWORD_RETURN(ENABLE); }                             // CFA | 
|---|
| 232 | enum                    { KEYWORD_RETURN(ENUM); } | 
|---|
| 233 | __extension__   { KEYWORD_RETURN(EXTENSION); }                  // GCC | 
|---|
| 234 | exception               { KEYWORD_RETURN(EXCEPTION); }                  // CFA | 
|---|
| 235 | extern                  { KEYWORD_RETURN(EXTERN); } | 
|---|
| 236 | fallthru                { KEYWORD_RETURN(FALLTHRU); }                   // CFA | 
|---|
| 237 | fallthrough             { KEYWORD_RETURN(FALLTHROUGH); }                // CFA | 
|---|
| 238 | finally                 { KEYWORD_RETURN(FINALLY); }                    // CFA | 
|---|
| 239 | float                   { KEYWORD_RETURN(FLOAT); } | 
|---|
| 240 | __float80               { KEYWORD_RETURN(FLOAT80); }                    // GCC | 
|---|
| 241 | float80                 { KEYWORD_RETURN(FLOAT80); }                    // GCC | 
|---|
| 242 | __float128              { KEYWORD_RETURN(FLOAT128); }                   // GCC | 
|---|
| 243 | float128                { KEYWORD_RETURN(FLOAT128); }                   // GCC | 
|---|
| 244 | for                             { KEYWORD_RETURN(FOR); } | 
|---|
| 245 | forall                  { KEYWORD_RETURN(FORALL); }                             // CFA | 
|---|
| 246 | fortran                 { KEYWORD_RETURN(FORTRAN); } | 
|---|
| 247 | ftype                   { KEYWORD_RETURN(FTYPE); }                              // CFA | 
|---|
| 248 | _Generic                { KEYWORD_RETURN(GENERIC); }                    // C11 | 
|---|
| 249 | goto                    { KEYWORD_RETURN(GOTO); } | 
|---|
| 250 | if                              { KEYWORD_RETURN(IF); } | 
|---|
| 251 | _Imaginary              { KEYWORD_RETURN(IMAGINARY); }                  // C99 | 
|---|
| 252 | __imag                  { KEYWORD_RETURN(IMAGINARY); }                  // GCC | 
|---|
| 253 | __imag__                { KEYWORD_RETURN(IMAGINARY); }                  // GCC | 
|---|
| 254 | inline                  { KEYWORD_RETURN(INLINE); }                             // C99 | 
|---|
| 255 | __inline                { KEYWORD_RETURN(INLINE); }                             // GCC | 
|---|
| 256 | __inline__              { KEYWORD_RETURN(INLINE); }                             // GCC | 
|---|
| 257 | int                             { KEYWORD_RETURN(INT); } | 
|---|
| 258 | __int128                { KEYWORD_RETURN(INT128); }                             // GCC | 
|---|
| 259 | int128                  { KEYWORD_RETURN(INT128); }                             // GCC | 
|---|
| 260 | __label__               { KEYWORD_RETURN(LABEL); }                              // GCC | 
|---|
| 261 | long                    { KEYWORD_RETURN(LONG); } | 
|---|
| 262 | monitor                 { KEYWORD_RETURN(MONITOR); }                    // CFA | 
|---|
| 263 | mutex                   { KEYWORD_RETURN(MUTEX); }                              // CFA | 
|---|
| 264 | _Noreturn               { KEYWORD_RETURN(NORETURN); }                   // C11 | 
|---|
| 265 | __builtin_offsetof { KEYWORD_RETURN(OFFSETOF); }                // GCC | 
|---|
| 266 | one_t                   { NUMERIC_RETURN(ONE_T); }                              // CFA | 
|---|
| 267 | otype                   { KEYWORD_RETURN(OTYPE); }                              // CFA | 
|---|
| 268 | register                { KEYWORD_RETURN(REGISTER); } | 
|---|
| 269 | restrict                { KEYWORD_RETURN(RESTRICT); }                   // C99 | 
|---|
| 270 | __restrict              { KEYWORD_RETURN(RESTRICT); }                   // GCC | 
|---|
| 271 | __restrict__    { KEYWORD_RETURN(RESTRICT); }                   // GCC | 
|---|
| 272 | return                  { KEYWORD_RETURN(RETURN); } | 
|---|
| 273 | short                   { KEYWORD_RETURN(SHORT); } | 
|---|
| 274 | signed                  { KEYWORD_RETURN(SIGNED); } | 
|---|
| 275 | __signed                { KEYWORD_RETURN(SIGNED); }                             // GCC | 
|---|
| 276 | __signed__              { KEYWORD_RETURN(SIGNED); }                             // GCC | 
|---|
| 277 | sizeof                  { KEYWORD_RETURN(SIZEOF); } | 
|---|
| 278 | static                  { KEYWORD_RETURN(STATIC); } | 
|---|
| 279 | _Static_assert  { KEYWORD_RETURN(STATICASSERT); }               // C11 | 
|---|
| 280 | struct                  { KEYWORD_RETURN(STRUCT); } | 
|---|
| 281 | switch                  { KEYWORD_RETURN(SWITCH); } | 
|---|
| 282 | thread                  { KEYWORD_RETURN(THREAD); }                             // C11 | 
|---|
| 283 | _Thread_local   { KEYWORD_RETURN(THREADLOCAL); }                // C11 | 
|---|
| 284 | throw                   { KEYWORD_RETURN(THROW); }                              // CFA | 
|---|
| 285 | throwResume             { KEYWORD_RETURN(THROWRESUME); }                // CFA | 
|---|
| 286 | timeout                 { QKEYWORD_RETURN(TIMEOUT); }                   // CFA | 
|---|
| 287 | trait                   { KEYWORD_RETURN(TRAIT); }                              // CFA | 
|---|
| 288 | try                             { KEYWORD_RETURN(TRY); }                                // CFA | 
|---|
| 289 | ttype                   { KEYWORD_RETURN(TTYPE); }                              // CFA | 
|---|
| 290 | typedef                 { KEYWORD_RETURN(TYPEDEF); } | 
|---|
| 291 | typeof                  { KEYWORD_RETURN(TYPEOF); }                             // GCC | 
|---|
| 292 | __typeof                { KEYWORD_RETURN(TYPEOF); }                             // GCC | 
|---|
| 293 | __typeof__              { KEYWORD_RETURN(TYPEOF); }                             // GCC | 
|---|
| 294 | union                   { KEYWORD_RETURN(UNION); } | 
|---|
| 295 | unsigned                { KEYWORD_RETURN(UNSIGNED); } | 
|---|
| 296 | __builtin_va_list { KEYWORD_RETURN(VALIST); }                   // GCC | 
|---|
| 297 | virtual                 { KEYWORD_RETURN(VIRTUAL); }                    // CFA | 
|---|
| 298 | void                    { KEYWORD_RETURN(VOID); } | 
|---|
| 299 | volatile                { KEYWORD_RETURN(VOLATILE); } | 
|---|
| 300 | __volatile              { KEYWORD_RETURN(VOLATILE); }                   // GCC | 
|---|
| 301 | __volatile__    { KEYWORD_RETURN(VOLATILE); }                   // GCC | 
|---|
| 302 | waitfor                 { KEYWORD_RETURN(WAITFOR); } | 
|---|
| 303 | or                              { QKEYWORD_RETURN(WOR); }                               // CFA | 
|---|
| 304 | when                    { KEYWORD_RETURN(WHEN); } | 
|---|
| 305 | while                   { KEYWORD_RETURN(WHILE); } | 
|---|
| 306 | with                    { KEYWORD_RETURN(WITH); }                               // CFA | 
|---|
| 307 | zero_t                  { NUMERIC_RETURN(ZERO_T); }                             // CFA | 
|---|
| 308 |  | 
|---|
| 309 | /* identifier */ | 
|---|
| 310 | {identifier}    { IDENTIFIER_RETURN(); } | 
|---|
| 311 | {attr_identifier} { ATTRIBUTE_RETURN(); } | 
|---|
| 312 | "`"                             { BEGIN BKQUOTE; } | 
|---|
| 313 | <BKQUOTE>{identifier} { IDENTIFIER_RETURN(); } | 
|---|
| 314 | <BKQUOTE>"`"    { BEGIN 0; } | 
|---|
| 315 |  | 
|---|
| 316 | /* numeric constants */ | 
|---|
| 317 | {decimal_constant} { NUMERIC_RETURN(INTEGERconstant); } | 
|---|
| 318 | {octal_constant} { NUMERIC_RETURN(INTEGERconstant); } | 
|---|
| 319 | {hex_constant}  { NUMERIC_RETURN(INTEGERconstant); } | 
|---|
| 320 | {floating_decimal}      { NUMERIC_RETURN(FLOATING_DECIMALconstant); } // must appear before floating_constant | 
|---|
| 321 | {floating_fraction}     { NUMERIC_RETURN(FLOATING_FRACTIONconstant); } // must appear before floating_constant | 
|---|
| 322 | {floating_constant}     { NUMERIC_RETURN(FLOATINGconstant); } | 
|---|
| 323 | {hex_floating_constant} { NUMERIC_RETURN(FLOATINGconstant); } | 
|---|
| 324 |  | 
|---|
| 325 | /* character constant, allows empty value */ | 
|---|
| 326 | ({cwide_prefix}[_]?)?['] { BEGIN QUOTE; rm_underscore(); strtext = new string( yytext, yyleng ); } | 
|---|
| 327 | <QUOTE>[^'\\\n]* { strtext->append( yytext, yyleng ); } | 
|---|
| 328 | <QUOTE>['\n]{user_suffix_opt}   { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(CHARACTERconstant); } | 
|---|
| 329 | /* ' stop editor highlighting */ | 
|---|
| 330 |  | 
|---|
| 331 | /* string constant */ | 
|---|
| 332 | ({swide_prefix}[_]?)?["] { BEGIN STRING; rm_underscore(); strtext = new string( yytext, yyleng ); } | 
|---|
| 333 | <STRING>[^"\\\n]* { strtext->append( yytext, yyleng ); } | 
|---|
| 334 | <STRING>["\n]{user_suffix_opt}  { BEGIN 0; strtext->append( yytext, yyleng ); RETURN_STR(STRINGliteral); } | 
|---|
| 335 | /* " stop editor highlighting */ | 
|---|
| 336 |  | 
|---|
| 337 | /* common character/string constant */ | 
|---|
| 338 | <QUOTE,STRING>{escape_seq} { rm_underscore(); strtext->append( yytext, yyleng ); } | 
|---|
| 339 | <QUOTE,STRING>"\\"{h_white}*"\n" {}                                             // continuation (ALSO HANDLED BY CPP) | 
|---|
| 340 | <QUOTE,STRING>"\\" { strtext->append( yytext, yyleng ); } // unknown escape character | 
|---|
| 341 |  | 
|---|
| 342 | /* punctuation */ | 
|---|
| 343 | "@"                             { ASCIIOP_RETURN(); } | 
|---|
| 344 | "["                             { ASCIIOP_RETURN(); } | 
|---|
| 345 | "]"                             { ASCIIOP_RETURN(); } | 
|---|
| 346 | "("                             { ASCIIOP_RETURN(); } | 
|---|
| 347 | ")"                             { ASCIIOP_RETURN(); } | 
|---|
| 348 | "{"                             { ASCIIOP_RETURN(); } | 
|---|
| 349 | "}"                             { ASCIIOP_RETURN(); } | 
|---|
| 350 | ","                             { ASCIIOP_RETURN(); }                                   // also operator | 
|---|
| 351 | ":"                             { ASCIIOP_RETURN(); } | 
|---|
| 352 | ";"                             { ASCIIOP_RETURN(); } | 
|---|
| 353 | "."                             { ASCIIOP_RETURN(); }                                   // also operator | 
|---|
| 354 | "..."                   { NAMEDOP_RETURN(ELLIPSIS); } | 
|---|
| 355 |  | 
|---|
| 356 | /* alternative C99 brackets, "<:" & "<:<:" handled by preprocessor */ | 
|---|
| 357 | "<:"                    { RETURN_VAL('['); } | 
|---|
| 358 | ":>"                    { RETURN_VAL(']'); } | 
|---|
| 359 | "<%"                    { RETURN_VAL('{'); } | 
|---|
| 360 | "%>"                    { RETURN_VAL('}'); } | 
|---|
| 361 |  | 
|---|
| 362 | /* operators */ | 
|---|
| 363 | "!"                             { ASCIIOP_RETURN(); } | 
|---|
| 364 | "+"                             { ASCIIOP_RETURN(); } | 
|---|
| 365 | "-"                             { ASCIIOP_RETURN(); } | 
|---|
| 366 | "*"                             { ASCIIOP_RETURN(); } | 
|---|
| 367 | "\\"                    { ASCIIOP_RETURN(); }                                   // CFA, exponentiation | 
|---|
| 368 | "/"                             { ASCIIOP_RETURN(); } | 
|---|
| 369 | "%"                             { ASCIIOP_RETURN(); } | 
|---|
| 370 | "^"                             { ASCIIOP_RETURN(); } | 
|---|
| 371 | "~"                             { ASCIIOP_RETURN(); } | 
|---|
| 372 | "&"                             { ASCIIOP_RETURN(); } | 
|---|
| 373 | "|"                             { ASCIIOP_RETURN(); } | 
|---|
| 374 | "<"                             { ASCIIOP_RETURN(); } | 
|---|
| 375 | ">"                             { ASCIIOP_RETURN(); } | 
|---|
| 376 | "="                             { ASCIIOP_RETURN(); } | 
|---|
| 377 | "?"                             { ASCIIOP_RETURN(); } | 
|---|
| 378 |  | 
|---|
| 379 | "++"                    { NAMEDOP_RETURN(ICR); } | 
|---|
| 380 | "--"                    { NAMEDOP_RETURN(DECR); } | 
|---|
| 381 | "=="                    { NAMEDOP_RETURN(EQ); } | 
|---|
| 382 | "!="                    { NAMEDOP_RETURN(NE); } | 
|---|
| 383 | "<<"                    { NAMEDOP_RETURN(LS); } | 
|---|
| 384 | ">>"                    { NAMEDOP_RETURN(RS); } | 
|---|
| 385 | "<="                    { NAMEDOP_RETURN(LE); } | 
|---|
| 386 | ">="                    { NAMEDOP_RETURN(GE); } | 
|---|
| 387 | "&&"                    { NAMEDOP_RETURN(ANDAND); } | 
|---|
| 388 | "||"                    { NAMEDOP_RETURN(OROR); } | 
|---|
| 389 | "->"                    { NAMEDOP_RETURN(ARROW); } | 
|---|
| 390 | "+="                    { NAMEDOP_RETURN(PLUSassign); } | 
|---|
| 391 | "-="                    { NAMEDOP_RETURN(MINUSassign); } | 
|---|
| 392 | "\\="                   { NAMEDOP_RETURN(EXPassign); }                  // CFA, exponentiation | 
|---|
| 393 | "*="                    { NAMEDOP_RETURN(MULTassign); } | 
|---|
| 394 | "/="                    { NAMEDOP_RETURN(DIVassign); } | 
|---|
| 395 | "%="                    { NAMEDOP_RETURN(MODassign); } | 
|---|
| 396 | "&="                    { NAMEDOP_RETURN(ANDassign); } | 
|---|
| 397 | "|="                    { NAMEDOP_RETURN(ORassign); } | 
|---|
| 398 | "^="                    { NAMEDOP_RETURN(ERassign); } | 
|---|
| 399 | "<<="                   { NAMEDOP_RETURN(LSassign); } | 
|---|
| 400 | ">>="                   { NAMEDOP_RETURN(RSassign); } | 
|---|
| 401 |  | 
|---|
| 402 | "@="                    { NAMEDOP_RETURN(ATassign); }                   // CFA | 
|---|
| 403 |  | 
|---|
| 404 | /* CFA, operator identifier */ | 
|---|
| 405 | {op_unary}"?"   { IDENTIFIER_RETURN(); }                                // unary | 
|---|
| 406 | "?"({op_unary_pre_post}|"()"|"[?]"|"{}") { IDENTIFIER_RETURN(); } | 
|---|
| 407 | "^?{}"                  { IDENTIFIER_RETURN(); } | 
|---|
| 408 | "?`"{identifier} { IDENTIFIER_RETURN(); }                               // unit operator | 
|---|
| 409 | "?"{op_binary_over}"?"  { IDENTIFIER_RETURN(); }                // binary | 
|---|
| 410 | /* | 
|---|
| 411 | This rule handles ambiguous cases with operator identifiers, e.g., "int *?*?()", where the string "*?*?"  can be | 
|---|
| 412 | lexed as "*?"/"*?" or "*"/"?*?". Since it is common practise to put a unary operator juxtaposed to an identifier, | 
|---|
| 413 | e.g., "*i", users will be annoyed if they cannot do this with respect to operator identifiers. Therefore, there is | 
|---|
| 414 | a lexical look-ahead for the second case, with backtracking to return the leading unary operator and then | 
|---|
| 415 | reparsing the trailing operator identifier.  Otherwise a space is needed between the unary operator and operator | 
|---|
| 416 | identifier to disambiguate this common case. | 
|---|
| 417 |  | 
|---|
| 418 | A similar issue occurs with the dereference, *?(...), and routine-call, ?()(...) identifiers.  The ambiguity | 
|---|
| 419 | occurs when the deference operator has no parameters, *?() and *?()(...), requiring arbitrary whitespace | 
|---|
| 420 | look-ahead for the routine-call parameter-list to disambiguate.  However, the dereference operator must have a | 
|---|
| 421 | parameter/argument to dereference *?(...).  Hence, always interpreting the string *?() as * ?() does not preclude | 
|---|
| 422 | any meaningful program. | 
|---|
| 423 |  | 
|---|
| 424 | The remaining cases are with the increment/decrement operators and conditional expression: | 
|---|
| 425 |  | 
|---|
| 426 | i++? ...(...); | 
|---|
| 427 | i?++ ...(...); | 
|---|
| 428 |  | 
|---|
| 429 | requiring arbitrary whitespace look-ahead for the operator parameter-list, even though that interpretation is an | 
|---|
| 430 | incorrect expression (juxtaposed identifiers).  Therefore, it is necessary to disambiguate these cases with a | 
|---|
| 431 | space: | 
|---|
| 432 |  | 
|---|
| 433 | i++ ? i : 0; | 
|---|
| 434 | i? ++i : 0; | 
|---|
| 435 | */ | 
|---|
| 436 | {op_unary}"?"({op_unary_pre_post}|"()"|"[?]"|{op_binary_over}"?") { | 
|---|
| 437 | // 1 or 2 character unary operator ? | 
|---|
| 438 | int i = yytext[1] == '?' ? 1 : 2; | 
|---|
| 439 | yyless( i );            // put back characters up to first '?' | 
|---|
| 440 | if ( i > 1 ) { | 
|---|
| 441 | NAMEDOP_RETURN( yytext[0] == '+' ? ICR : DECR ); | 
|---|
| 442 | } else { | 
|---|
| 443 | ASCIIOP_RETURN(); | 
|---|
| 444 | } // if | 
|---|
| 445 | } | 
|---|
| 446 |  | 
|---|
| 447 | /* unknown character */ | 
|---|
| 448 | .                               { yyerror( "unknown character" ); } | 
|---|
| 449 |  | 
|---|
| 450 | %% | 
|---|
| 451 | // ----end of lexer---- | 
|---|
| 452 |  | 
|---|
| 453 | void yyerror( const char * errmsg ) { | 
|---|
| 454 | cout << (yyfilename ? yyfilename : "*unknown file*") << ':' << yylineno << ':' << column - yyleng + 1 | 
|---|
| 455 | << ": " << ErrorHelpers::error_str() << errmsg << " at token \"" << (yytext[0] == '\0' ? "EOF" : yytext) << '"' << endl; | 
|---|
| 456 | } | 
|---|
| 457 |  | 
|---|
| 458 | // Local Variables: // | 
|---|
| 459 | // mode: c++ // | 
|---|
| 460 | // tab-width: 4 // | 
|---|
| 461 | // compile-command: "make install" // | 
|---|
| 462 | // End: // | 
|---|