%{ /***************************************/ /* * 'D' Tokeniser module * Version 0.0.0.0 GPL (alpha) * Copyright C.R.Chafer 2002/6/11 * Author C.R.Chafer * Contact c.chafer@herts.ac.uk * License GPL v2 (www.fsf.org) * (contact me for versions * under other licenses) * */ /***************************************/ /* * Notes: This is not to be assumed * definitive - see * http://www.digitalmars.com * for definitive D spec */ /***************************************/ #include "d.h" #include "y.tab.h" /***************************************/ #define YY_NO_UNPUT /***************************************/ static U32 lineNumberIncrement = 1 ; static U32 lineNumber = 0 ; /***************************************/ void lineInc() { ++lineNumberIncrement; return; } void lineUpdate() { lineNumber+=lineNumberIncrement; lineNumberIncrement=0; return; } U32 getLine() { return lineNumber; } U32 getLineInc() { return lineNumberIncrement; } void setInput( s ) CH*s; { if( !(yyin=fopen(s,"rb")) ) printError( "Unable to open input", s, 0 ); return; } static char* convertString( char*s ) { return s; /* later convert unprintable letters to escapes */ } static U32 getInt( char*s, int l ) { return strtol( s, 0, l ); } static F64 getFloat( char*s ) { return atof( s ); } static F64 hexFloat( char*s ) { return 1.0; /* later convert */ } static int suffix( char*s ) { return _I;/* have to extend these later (uUlL/fFiIlL) */ } static int floatSuffix( char*s ) { while( *s++ ) if( *s=='i' || *s=='I' ) return _J; return _F; } int yywrap( void ) { return 1; } int yyerror( char*s ) { printError( s, convertString( yytext ), 0 ); return 0; } /***************************************/ %} %% <> { return _EOF ;} [ \t\r]+ { /* whitespace */ ;} \n { lineInc() ;} "//"[^\n]* { /* line comment */ ;} "/*" { /* multi line comment #1 */ S32 c1 = 0, c2 = 0 ; while( !( c1 == '*' && c2 == '/' ) ) { if( c2 == EOF ) printError( "Unexpected end of file within comment", 0, 0 ) ; if( c1 == '[' && c2 == '`' ) printError( "Found comment header '/*' within comment", 0, 0 ) ; if( c1 == '\n' ) lineInc() ; c1 = c2 ; c2 = input() ;} } "/+" { /* multi line comment #2 (nestable) */ S32 c1 = 0, c2 = 0, c3 = 2 ; while( -- c3 ) ; while( !( c1 == '+' && c2 == '/' ) ) { if( c2 == EOF ) printError( "Unexpected end of file within comment", 0, 0 ) ; if( c1 == '/' && c2 == '+' ) ++ c3 ; if( c1 == '\n' ) lineInc() ; c1 = c2 ; c2 = input() ;} } \"([^"\n]|\\\")*\" { /* string constant - double quoted */ CH*s=yytext ; while( *s++ ) ; *(s-2)=0 ; yylval.s=newString( yytext+1 ) ; return _S ;} \'[^'\n]*\' { /* string constant - single quoted */ CH*s=yytext ; while( *s++ ) ; *(s-2)=0 ; yylval.s=newStringNoEsc( yytext+1 ) ; return _S ;} \"([^"\n]|\\["\n])*\n { printError( "Unterminated string constant",0,0 ); } \'[^\n]*\n { printError( "Unterminated character constant",0,0 ); } \\x[0-9a-fA-F][0-9a-fA-F] { /* escape sequence */ yylval.s=newString( yytext ) ; return _S ;} \\u[0-9a-fA-F][0-9a-fA-F][0-9a-fA-F][0-9a-fA-F] { /* escape sequence */ yylval.s=newString( yytext ) ; return _S ;} \\[0-7]([0-7][0-7]?)? { /* escape sequence */ yylval.s=newString( yytext ) ; return _S ;} \\[abfnrtv'"?] { /* escape sequence */ yylval.s=newString( yytext ) ; return _S ;} 0[bB][01]+[lLuU]* { /* number/binary */ yylval.i = getInt( yytext+2, 2 ) ; return suffix( yytext ) ;} 0[0-7]+[lLuU]* { /* number/octal */ yylval.i = getInt( yytext, 8 ) ; return suffix( yytext ) ;} [0-9]+[lLuU]* { /* number/decimal */ yylval.i = getInt( yytext, 10 ) ; return suffix( yytext ) ;} 0[xX][0-9a-fA-F]+[lLuU]* { /* number/hexadecimal */ yylval.i = getInt( yytext+2, 16 ) ; return suffix( yytext ) ;} [0-9][0-9a-fA-F]*[hH][lLuU]* { /* number/hexadecimal */ yylval.i = getInt( yytext, 16 ) ; return suffix( yytext ) ;} ([0-9][0-9]*\.[0-9]+)([eE][+-]?[0-9]+)?[iIfFlL]* { /* number/float */ yylval.f = getFloat( yytext ) ; return floatSuffix( yytext ) ;} 0[xX][0-9a-fA-F]*(\.[0-9a-fA-F]+)?[pP][+-]?[0-9]+[iIfFlL]* { /* number/hex_float */ yylval.f = hexFloat( yytext ) ; return floatSuffix( yytext ) ;} "this" { return _this ;} "super" { return _super ;} "assert" { return _assert ;} "null" { return _null ;} "true" { return _true ;} "false" { return _false ;} "cast" { return _cast ;} "new" { return _new ;} "delete" { return _delete ;} "throw" { return _throw ;} "void" { return _void ;} "byte" { return _byte ;} "ubyte" { return _ubyte ;} "short" { return _short ;} "ushort" { return _ushort ;} "int" { return _int ;} "uint" { return _uint ;} "long" { return _long ;} "ulong" { return _ulong ;} "cent" { return _cent ;} "ucent" { return _ucent ;} "float" { return _float ;} "double" { return _double ;} "extended" { return _extended ;} "bit" { return _bit ;} "char" { return _char ;} "wchar" { return _wchar ;} "imaginary" { return _imaginary ;} "complex" { return _complex ;} "if" { return _if ;} "else" { return _else ;} "while" { return _while ;} "for" { return _for ;} "do" { return _do ;} "switch" { return _switch ;} "case" { return _case ;} "default" { return _default ;} "break" { return _break ;} "continue" { return _continue ;} "synchronized" { return _synchronized ;} "return" { return _return ;} "goto" { return _goto ;} "try" { return _try ;} "catch" { return _catch ;} "finally" { return _finally ;} "with" { return _with ;} "struct" { return _struct ;} "class" { return _class ;} "interface" { return _interface ;} "union" { return _union ;} "enum" { return _enum ;} "import" { return _import ;} "static" { return _static ;} "final" { return _final ;} "const" { return _const ;} "typedef" { return _typedef ;} "alias" { return _alias ;} "override" { return _override ;} "abstract" { return _abstract ;} "debug" { return _debug ;} "version" { return _version ;} "deprecated" { return _deprecated ;} "in" { return _in ;} "out" { return _out ;} "inout" { return _inout ;} "align" { return _align ;} "extern" { return _extern ;} "private" { return _private ;} "protected" { return _protected ;} "public" { return _public ;} "export" { return _export ;} "body" { return _body ;} "invariant" { return _invariant ;} "asm" { return _asm ;} "delegate" { return _delegate ;} "/" { return '/' ;} "/=" { return _DIVE ;} "." { return '.' ;} ".." { return _DOT2 ;} "..." { return _DOT3 ;} "&" { return '&' ;} "&=" { return _ANDE ;} "&&" { return _ANDB ;} "|" { return '|' ;} "|=" { return _IORE ;} "||" { return _IORB ;} "-" { return '-' ;} "-=" { return _SUBE ;} "--" { return _DEC ;} "+" { return '+' ;} "+=" { return _ADDE ;} "++" { return _INC ;} "<" { return '<' ;} "<=" { return _LEQ ;} "<<" { return _SHL ;} "<<=" { return _SHLE ;} "<>" { return _GLS ;} "<>=" { return _GLSE ;} ">" { return '>' ;} ">=" { return _GEQ ;} ">>=" { return _SHRE ;} ">>>=" { return _SARE ;} ">>" { return _SHR ;} ">>>" { return _SAR ;} "!" { return '!' ;} "!=" { return _NEQU ;} "!==" { return _NEQUR ;} "!===" { return _NEQUV ;} "!<>" { return _NGLS ;} "!<>=" { return _NGLSE ;} "!<" { return _NLES ;} "!<=" { return _NLEQ ;} "!>" { return _NGRT ;} "!>=" { return _NGEQ ;} "(" { return '(' ;} ")" { return ')' ;} "[" { return '[' ;} "]" { return ']' ;} "{" { return '{' ;} "}" { return '}' ;} "?" { return '?' ;} "," { return ',' ;} ";" { return ';' ;} ":" { return ':' ;} "$" { return '$' ;} "=" { return '=' ;} "==" { return _EQU ;} "===" { return _EQUR ;} "*" { return '*' ;} "*=" { return _MULE ;} "%" { return '%' ;} "%=" { return _MODE ;} "^" { return '^' ;} "^=" { return _XORE ;} "~" { return '~' ;} "~=" { return _CAT ;} [a-zA-Z_][a-zA-Z0-9_]* { yylval.s=newSymbol( yytext ); return _V; /* identifier */ } . { printError( "Unexpected symbol", convertString( yytext ), 0 );} %%