/* lexer.ll - Lexical analyser for GNU Flex */ %{ #include #include #include "DeclarationSequence.h" #include "CompoundStmt.h" #include "BinaryOpExpr.h" #include "UnaryOpExpr.h" #include "TranslationUnit.h" #include "FunctionDef.h" #include "ClassDef.h" #include "FormalParameter.h" #include "FormalParamList.h" #include "DeclarationSpecifierList.h" #include "parser.hh" /* generated by bison -d */ /* Note for Automake users: version 1.12 or later is required to generate parser.hh. Previous versions generate parser.h. */ using namespace std; string sourceFilename; int lineno = 1; // line number in the input text; first line is numbered 1 static bool doublePrecisionLiteralWarningIssued = false; static void warnIfDoublePrecisionLiteral(const char *realLiteralText) { if (!doublePrecisionLiteralWarningIssued && tolower(realLiteralText[strlen(realLiteralText) - 1]) != 'f') { warnmsg("double-precision numeric literals have same precision as `float' for this compiler"); doublePrecisionLiteralWarningIssued = true; } } extern "C" int yywrap(void) { return 1; } // yylex() returns 0 when EOF reached bool isOctal(uint64_t &value, const char *str); %} letter [A-Za-z_] digit [0-9] ws ([ \t\f\v\r\xA0]|\xC2\xA0) /* A0 = Latin-1 unbreakable space; C2 A0 = same in UTF-8 */ %% /* WARNING: The opening brace for code that follows a pattern must be on the same line as the pattern. C-style comments must be indented so as not to be taken as lexical rules. */ /* #include markers generated by cpp. Format: # NUM "FILENAME" [NUM ...] The 1st integer is the line number in FILENAME. One or more integers can appear after the filename. They are all ignored. */ ^\#\ [0-9]+\ \".*\"[ 0-9]*$ { sscanf(yytext + 2, "%d", &lineno); lineno--; char *fn = strchr(yytext, '\"') + 1; char *end = strchr(fn, '\"'); *end = '\0'; sourceFilename = fn; } /* Pragma directive. */ ^{ws}*\#{ws}*pragma{ws}+.*$ { // Pragma directive. // const char *src = strstr(yytext, "pragma") + 6; for (++src; isspace(*src); ++src) ; yylval.str = strdup(src); return PRAGMA; } /* Inline assembly: asm { ... }. Accept balanced braces inside the outer ones. */ asm({ws}|\n)*\{([^\{\}]|(\{[^\{\}]*\}))*\} { // Verbatim assembler text. // Copy contents of braces into yytext, and count newlines. // Update 'lineno' according to number of source lines taken. // char *p; for (p = yytext + 3; *p != '{'; ++p) if (*p == '\n') ++lineno; char *begin = p + 1; char *end = strrchr(begin, '}'); if (begin != end) // only return something if text not empty { char *writer = yytext; for (char *reader = begin; reader != end; ) { char c = *reader++; *writer++ = c; if (c == '\n') ++lineno; } *writer = '\0'; return VERBATIM_ASM; } } /* White space: ignored. */ {ws}+ ; /* Newline increments the line counter. */ \n { lineno++; } -- { return MINUS_MINUS; } \+\+ { return PLUS_PLUS; } == { return EQUALS_EQUALS; } \!= { return BANG_EQUALS; } \<= { return LOWER_EQUALS; } \>= { return GREATER_EQUALS; } \+= { return PLUS_EQUALS; } -= { return MINUS_EQUALS; } \*= { return ASTERISK_EQUALS; } \/= { return SLASH_EQUALS; } %= { return PERCENT_EQUALS; } \^= { return CARET_EQUALS; } &= { return AMP_EQUALS; } \|= { return PIPE_EQUALS; } && { return AMP_AMP; } \|\| { return PIPE_PIPE; } \<\< { return LT_LT; } \>\> { return GT_GT; } \<\<= { return LT_LT_EQUALS; } \>\>= { return GT_GT_EQUALS; } -\> { return RIGHT_ARROW; } \.\.\. { return ELLIPSIS; } /* Hexadecimal constant. */ 0x([0-9A-Za-z]+) { unsigned int n; sscanf(yytext + 2, "%x", &n); yylval.real = n; return REAL; } /* Decimal float constant. No optional minus sign at the beginning of this rule, because it creates an ambiguity where 'n-1' is seen as tokens 'n' and '-1'. It should be seen as 'n', '-' and '1', and the add_expr rule in parser.yy will recognize these 3 tokens as a subtraction. */ (([0-9]*\.[0-9]+))([eE][-+]?[0-9]+)?[fF]? { yylval.real = atof(yytext); warnIfDoublePrecisionLiteral(yytext); return REAL; } /* If there are no digits after the decimal point, there must be at least one digit before the point, e.g., "1.", "1.e6", "1.f", "1.e6f". */ ([0-9]+\.)([eE][-+]?[0-9]+)?[fF]? { yylval.real = atof(yytext); warnIfDoublePrecisionLiteral(yytext); return REAL; } /* Same as previous rule, but no decimal point, and required 'e'. */ ([0-9]+)([eE][-+]?[0-9]+)[fF]? { yylval.real = atof(yytext); warnIfDoublePrecisionLiteral(yytext); return REAL; } /* Decimal or octal integer, with optional "unsigned" or "long" suffix character. */ ([0-9]+)([uUlL]*) { uint64_t octalConversion = 0; if (yytext[0] == '0' && isOctal(octalConversion, yytext + 1)) yylval.real = double(octalConversion); else yylval.real = atof(yytext); return REAL; } /* Keyword, typedef name or identifier. */ {letter}({letter}|{digit})* { if (strcmp(yytext, "int") == 0) return INT; if (strcmp(yytext, "char") == 0) return CHAR; if (strcmp(yytext, "short") == 0) return SHORT; if (strcmp(yytext, "long") == 0) return LONG; if (strcmp(yytext, "float") == 0) return FLOAT; if (strcmp(yytext, "double") == 0) return DOUBLE; if (strcmp(yytext, "signed") == 0) return SIGNED; if (strcmp(yytext, "unsigned") == 0) return UNSIGNED; if (strcmp(yytext, "void") == 0) return VOID; if (strcmp(yytext, "if") == 0) return IF; if (strcmp(yytext, "else") == 0) return ELSE; if (strcmp(yytext, "while") == 0) return WHILE; if (strcmp(yytext, "do") == 0) return DO; if (strcmp(yytext, "for") == 0) return FOR; if (strcmp(yytext, "break") == 0) return BREAK; if (strcmp(yytext, "continue") == 0) return CONTINUE; if (strcmp(yytext, "return") == 0) return RETURN; if (strcmp(yytext, "asm") == 0) return ASM; if (strcmp(yytext, "struct") == 0) return STRUCT; if (strcmp(yytext, "union") == 0) return UNION; if (strcmp(yytext, "this") == 0) return THIS; if (strcmp(yytext, "interrupt") == 0) return INTERRUPT; if (strcmp(yytext, "sizeof") == 0) return SIZEOF; if (strcmp(yytext, "typedef") == 0) return TYPEDEF; if (strcmp(yytext, "switch") == 0) return SWITCH; if (strcmp(yytext, "case") == 0) return CASE; if (strcmp(yytext, "default") == 0) return DEFAULT; if (strcmp(yytext, "asm") == 0) return ASM; if (strcmp(yytext, "register") == 0) return REGISTER; if (strcmp(yytext, "goto") == 0) return GOTO; if (strcmp(yytext, "extern") == 0) return EXTERN; if (strcmp(yytext, "static") == 0) return STATIC; if (strcmp(yytext, "enum") == 0) return ENUM; if (strcmp(yytext, "const") == 0) return CONST; if (strcmp(yytext, "volatile") == 0) return VOLATILE; if (strcmp(yytext, "auto") == 0) return AUTO; const TypeDesc *td = TranslationUnit::getTypeManager().getTypeDef(yytext); if (td) // if yytext is the name of a typedef: { yylval.typeDesc = td; return TYPE_NAME; } yylval.str = strdup(yytext); return ID; } /* Double-quoted string constant. No double-quote or newline allowed inside. */ \"([^\"\n]|\\\")*\" { yylval.str = strdup(yytext + 1); yylval.str[strlen(yylval.str) - 1] = '\0'; return STRLIT; } /* Control code character constants. */ '' { yylval.character = '\0'; return CHARLIT; } '\\0' { yylval.character = '\0'; return CHARLIT; } '\\a' { yylval.character = '\a'; return CHARLIT; } '\\b' { yylval.character = '\b'; return CHARLIT; } '\\t' { yylval.character = '\t'; return CHARLIT; } '\\n' { yylval.character = '\n'; return CHARLIT; } '\\v' { yylval.character = '\v'; return CHARLIT; } '\\f' { yylval.character = '\f'; return CHARLIT; } '\\r' { yylval.character = '\r'; return CHARLIT; } '\\'' { yylval.character = '\''; return CHARLIT; } '\\\"' { yylval.character = '\"'; return CHARLIT; } '\\\\' { yylval.character = '\\'; return CHARLIT; } /* Character constant expressed as a hexadecimal character code (e.g., '\xFF'). */ '\\x[0-9A-Fa-f][0-9A-Fa-f]?' { unsigned int n; sscanf(yytext + 3, "%x", &n); yylval.character = (signed char) n; return CHARLIT; } /* Character constant expressed as an octal character code (e.g., '\0377'). */ '\\0[0-7][0-7]?[0-7]?' { unsigned int n; sscanf(yytext + 2, "%o", &n); yylval.character = (signed char) n; return CHARLIT;} /* Other character constant. */ '.' { yylval.character = yytext[1]; return CHARLIT; } /* Any non-blank single-character token. */ . { return yytext[0]; } %% /* Avoids warning about `yyunput' being defined but not used. */ void BlackHole(void) { yyunput('\0', NULL); } /* Returns true if 'str' contains only octal digits and puts the value in 'value'. Returns false otherwise; 'value' is then undefined. */ bool isOctal(uint64_t &value, const char *str) { value = 0; for ( ; *str; ++str) { if (*str < '0' || *str > '7') // if not octal digit return false; value = (value << 3) | (*str - '0'); } return true; }