diff --git a/vhdlpp/Makefile.in b/vhdlpp/Makefile.in index 7717473fb..4e51a4d1b 100644 --- a/vhdlpp/Makefile.in +++ b/vhdlpp/Makefile.in @@ -57,10 +57,10 @@ CXXFLAGS = @WARNING_FLAGS@ @WARNING_FLAGS_CXX@ @CXXFLAGS@ LDFLAGS = @LDFLAGS@ LIBS = @LIBS@ @EXTRALIBS@ -M = StringHeap.o LineInfo.o +M = StringHeap.o LineInfo.o O = main.o compiler.o entity.o entity_elaborate.o \ - lexor.o lexor_keyword.o parse.o $M + lexor.o lexor_keyword.o parse.o vhdlreal.o vhdlint.o $M all: dep vhdlpp@EXEEXT@ diff --git a/vhdlpp/lexor.lex b/vhdlpp/lexor.lex index dd59ad95b..f00addcb9 100644 --- a/vhdlpp/lexor.lex +++ b/vhdlpp/lexor.lex @@ -19,17 +19,20 @@ * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License - * along with this program; if not, write to the Free Software + * aint64_t with this program; if not, write to the Free Software * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA */ # include "parse_api.h" # include "lexor_keyword.h" -# include "vhdlnum.h" +# include "vhdlint.h" # include "vhdlreal.h" # include "parse_wrap.h" +# include +# include # include +# include //class vhdlnum; //class vhdlreal; @@ -45,10 +48,16 @@ extern int lexor_keyword_code (const char*str, unsigned len); */ extern YYLTYPE yylloc; -static int check_underscores(char* text); +static bool are_underscores_correct(char* text); +static bool is_based_correct(char* text); +static char* escape_quot_and_dup(char* text); +static char* escape_apostrophe_and_dup(char* text); -vhdlnum* make_unisized_dec(char* text); -vhdlnum* make_unisized_based(char* text); +static double make_double_from_based(char* text); +static int64_t make_long_from_based(char* text); + +static int64_t lpow(int64_t left, int64_t right); +static unsigned short short_from_hex_char(char ch); static char* strdupnew(char const *str) { @@ -90,13 +99,26 @@ based_integer [0-9a-fA-F](_?[0-9a-fA-F])* \n { yylloc.first_line += 1; } "*/" { BEGIN(comment_enter); } +\'.\' { + yylval.text = escape_apostrophe_and_dup(yytext); + return CHARACTER_LITERAL; +} + +(\"([^"]|(\"\"))*?\")|(\"[^\"]*\") { +/* first pattern: string literals with doubled quotation mark */ +/* second pattern: string literals without doubled quotation */ + yylval.text = escape_quot_and_dup(yytext); + assert(yylval.text); + return STRING_LITERAL; +} [a-zA-Z][a-zA-Z0-9_]* { int rc = lexor_keyword_code(yytext, yyleng); switch (rc) { case IDENTIFIER: - if(check_underscores(yytext)) - std::cerr << "An invalid underscore in the identifier" << std::endl; + if(!are_underscores_correct(yytext)) + std::cerr << "An invalid underscore in the identifier:" + << yytext << std::endl; //yywarn(yylloc, "An invalid underscore in the identifier"); yylval.text = strdupnew(yytext); break; @@ -106,26 +128,43 @@ based_integer [0-9a-fA-F](_?[0-9a-fA-F])* return rc; } -\\(.|\\\\)*\\ { /* extended identifiers */ +\\([^\\]|\\\\)*\\ { /* extended identifiers */ yylval.text = strdupnew(yytext); - printf("special %s\n", yytext); return IDENTIFIER; } {decimal_literal} { + if(!are_underscores_correct(yytext)) + std::cerr << "An invalid underscore in the decimal literal:" + << yytext << std::endl; + if(strchr(yytext, '.')) { yylval.real = new vhdlreal(yytext); return REAL_LITERAL; } else { - yylval.integer = new vhdlnum(yytext); + yylval.integer = new vhdlint(yytext); return INT_LITERAL; } } {based_literal} { - yylval.integer = new vhdlnum(yytext); - return INT_LITERAL; + if(!are_underscores_correct(yytext) || !is_based_correct(yytext)) + std::cerr << "An invalid form of based literal:" + << yytext << std::endl; + + if(strchr(yytext, '.')) + { + double val = make_double_from_based(yytext); + yylval.real = new vhdlreal(val); + return REAL_LITERAL; + } + else + { + int64_t val = make_long_from_based(yytext); + yylval.integer = new vhdlint(val); + return INT_LITERAL; + } } @@ -158,7 +197,14 @@ based_integer [0-9a-fA-F](_?[0-9a-fA-F])* extern void yyparse_set_filepath(const char*path); -static int check_underscores(char* text) +/** +* This function checks if underscores in an identifier +* or in a number are correct. +* +* \return true is returned if underscores are placed +* correctly according to specification +*/ +static bool are_underscores_correct(char* text) { unsigned char underscore_allowed = 0; const char* cp; @@ -167,13 +213,285 @@ static int check_underscores(char* text) if (*cp == '_') { if (!underscore_allowed || *(cp+1) == '\0') - return 1; + return 0; underscore_allowed = 0; } else underscore_allowed = 1; } - return 0; + return 1; +} + +/** +* This function checks if the format of a based number +* is correct according to the VHDL standard +* +* \return true is returned if a based number +* is formed well according to specification +*/ +static bool is_based_correct(char* text) +{ + char* ptr; + //BASE examination + char clean_base[4]; + clean_base[3] = '\0'; + char* clean_base_ptr = clean_base; + for(ptr = text; ptr != strchr(text, '#'); ++ptr) + { + if(*ptr == '_') + ++ptr; + if(!(*ptr >= '0' && *ptr <= '9')) //the base uses chars other than digits + return 0; + if(*clean_base_ptr == '\0') + break; + *clean_base_ptr = *ptr; + ++clean_base_ptr; + } + unsigned length = clean_base_ptr - clean_base; + unsigned base; + if(length > 2 || length == 0) + return 0; //the base is too big or too small + if(length == 2) + { + base = 10*(clean_base[0] - '0') + (clean_base[1] - '0'); + //the base exceeds 16 or equals 0 + if(base > 16 || base == 0) + return 0; + } + else + { //the base consists of one char and is equal to zero + base = clean_base[0] - '0'; + if(base == 0) + return 0; + } + bool point = 0; + set allowed_chars; + + unsigned c; + if(base <= 10) { + for(c = 0; c < base; ++c) + allowed_chars.insert(c + '0'); + } + else + { + for(c = 0; c < 10; ++c) + allowed_chars.insert(c + '0'); + for(c = 0; c < base - 10; ++c) + allowed_chars.insert(c + 'a'); + } + //MANTISSA examination + for(ptr = strchr(text, '#') + 1, length = 0; ptr != strrchr(text, '#'); ++ptr) + { + if(*ptr == '.') + { + //we found a dot and another one was already found + if(point == 1) + return 0; + else + { + //notice the fact of finding a point and continue, without increasing the length + point = 1; + continue; + } + } + //the number consists of other chars than allowed + if(allowed_chars.find(*ptr) == allowed_chars.end()) + return 0; + ++length; + } + if(length == 0) + return 0; + + //EXPONENT examination + if(strchr(text, '\0') - strrchr(text, '#') > 1) { //the number contains an exponent + if(*(strrchr(text, '#') + 2) == '-') + return 0; + length = 0; + for(ptr = strrchr(text, '#')+2; *ptr != '\0'; ++ptr) + { + //the exponent consists of other chars than {'0'.,'9','a'..'f'} + if(!((*ptr >= '0' && *ptr <= '9') || (*ptr >= 'a' && *ptr <= 'f'))) + return 0; + } + } + return 1; +} + +/** +* This function takes a string literal, gets rid of +* quotation marks and copies the remaining characters +* to a new persistent C-string +* +* \return pointer to the new string is returned +*/ +static char* escape_quot_and_dup(char* text) +{ + char* newstr = new char[strlen(text)+1]; + + unsigned old_idx, new_idx; + for(new_idx = 0, old_idx = 0; old_idx < strlen(text); ) + { + if(text[old_idx] == '"' && old_idx == 0) + { //the beginning of the literal + ++old_idx; + continue; + } + else + if(text[old_idx] == '"' && text[old_idx+1] == '\0') + { //the end + newstr[new_idx] = '\0'; + return newstr; + } + else + if(text[old_idx] == '"' && text[old_idx+1] == '"') + { + newstr[new_idx++] = '"'; + old_idx += 2; //jump across two chars + } + else + { + newstr[new_idx] = text[old_idx]; + ++old_idx; + ++new_idx; + } + } + //the function should never reach this point + return 0; +} + +/** +* This function takes a character literal, gets rid +* of the apostrophes and returns new C-string +* +* \return pointer to the new string is returned +*/ +static char* escape_apostrophe_and_dup(char* text) +{ + char* newstr = new char[2]; + newstr[0] = text[1]; + newstr[1] = '\0'; + return newstr; +} + +/** +* This function takes a floating point based number +* in form of a C-strings and converts it to a double. +* +* \return new double is returned +*/ +static double make_double_from_based(char* text) +{ + char* first_hash_ptr = strchr(text, '#'); + char* second_hash_ptr = strrchr(text, '#'); + char* last_char_ptr = strchr(text, '\0') - 1; + //put null byte in lieu of hashes + *first_hash_ptr = '\0'; + *second_hash_ptr = '\0'; + + //now lets deduce the base + unsigned base = (unsigned)strtol(text, 0, 10) ; + + double mantissa = 0.0; + char*ptr = first_hash_ptr + 1; + for( ; ptr != second_hash_ptr ; ++ptr) + { + if(*ptr == '.') + break; + if(*ptr != '_') + { + mantissa = mantissa*base + short_from_hex_char(*ptr); + } + } + double fraction = 0.0; + double factor = 1.0/base; + for(++ptr ; ptr != second_hash_ptr; ++ptr) + { + if(*ptr != '_') + { + fraction = fraction + short_from_hex_char(*ptr)*factor; + factor = factor / base; + } + } + if(last_char_ptr == second_hash_ptr) //there is no exponent + { + return mantissa + fraction; + } + + //now calculate the value of the exponent + double exponent = 0.0; + //leave 'e'/'E' and '+' + ptr = *(second_hash_ptr + 2) == '+' ? second_hash_ptr + 3 : second_hash_ptr + 2; + + for( ; *ptr != '\0'; ++ptr) + { + if(*ptr != '_') + { + exponent = exponent*base + short_from_hex_char(*ptr); + } + } + return pow(mantissa + fraction, exponent); +} + +/** +* This function takes a hexadecimal digit in form of +* a char and returns its litteral value as short +*/ +static unsigned short short_from_hex_char(char ch) +{ + if(ch >= '0' && ch <= '9') + return ch - '0'; + else + return ch - 'a' + 10; +} + +/** +* This function takes a based number in form of +* a C-strings and converts it to a int64_t. +* +* \return new double is returned +*/ +static int64_t make_long_from_based(char* text) { + char* first_hash_ptr = strchr(text, '#'); + char* second_hash_ptr = strrchr(text, '#'); + char* end_ptr = strrchr(text, '\0'); + //now lets deduce the base + *first_hash_ptr = '\0'; + unsigned base = (unsigned)strtol(text, 0, 10) ; + + char *ptr = first_hash_ptr + 1; + int64_t mantissa = 0; + for( ; ptr != second_hash_ptr ; ++ptr) + { + if(*ptr != '_') + { + mantissa = mantissa * base + short_from_hex_char(*ptr); + } + } + //if there is an exponent + if(end_ptr - second_hash_ptr > 1) + { + int64_t exponent = 0L; + + ptr = *(second_hash_ptr + 2) == '+' ? second_hash_ptr + 3 : second_hash_ptr + 2; + for( ; *ptr != '\0'; ++ptr) + { + if(*ptr != '_') + exponent = base*exponent + short_from_hex_char(*ptr); + } + return lpow(mantissa, exponent); + } + else + return mantissa; +} + +/** +* Recursive power function for int64_t +*/ +static int64_t lpow(int64_t left, int64_t right) { + if(right == 0) + return 1; + else + return left*lpow(left, right - 1); } void reset_lexor(FILE*fd, const char*path) diff --git a/vhdlpp/parse.y b/vhdlpp/parse.y index 9c24a65a1..d5f8de0e9 100644 --- a/vhdlpp/parse.y +++ b/vhdlpp/parse.y @@ -20,7 +20,7 @@ */ # include "vhdlpp_config.h" -# include "vhdlnum.h" +# include "vhdlint.h" # include "vhdlreal.h" # include "compiler.h" # include "parse_api.h" @@ -45,7 +45,7 @@ int parse_errors = 0; %union { port_mode_t port_mode; char*text; - vhdlnum* integer; + vhdlint* integer; vhdlreal* real; InterfacePort*interface_element; diff --git a/vhdlpp/parse_wrap.h b/vhdlpp/parse_wrap.h index 2d09ecd62..c5a30e353 100644 --- a/vhdlpp/parse_wrap.h +++ b/vhdlpp/parse_wrap.h @@ -26,7 +26,7 @@ */ # include -# include "vhdlnum.h" +# include "vhdlint.h" # include "vhdlreal.h" # include "parse.h" diff --git a/vhdlpp/vhdlint.cc b/vhdlpp/vhdlint.cc new file mode 100644 index 000000000..e2486b040 --- /dev/null +++ b/vhdlpp/vhdlint.cc @@ -0,0 +1,68 @@ +#include "config.h" +#include "vhdlint.h" +#include +#include +#include + +bool vhdlint::is_negative() const +{ + return value_ < 0L; +} + +bool vhdlint::is_positive() const +{ + return value_ > 0L; +} + +bool vhdlint::is_zero() const +{ + return value_ == 0L; +} + +vhdlint::vhdlint(const char* text) +{ + unsigned text_length = strlen(text); + if(text_length == 0) + { + value_ = 0L; + return; + } + + char* new_text = new char[text_length + 1]; + + const char* ptr; + char* new_ptr; + for(ptr = text, new_ptr = new_text; *ptr != '\0'; ++ptr) + { + if(*ptr == '_') + continue; + else + { + *new_ptr = *ptr; + ++new_ptr; + ++ptr; + } + } + *new_ptr = '\0'; + + istringstream str(new_text); + delete[] new_text; + + //TODO: check if numbers greater than MAX_INT are handled correctly + str >> value_; +} + +vhdlint::vhdlint(const int64_t& val) +{ + value_ = val; +} + +vhdlint::vhdlint(const vhdlint& val) +{ + value_ = val.as_long(); +} + +int64_t vhdlint::as_long() const +{ + return value_; +} \ No newline at end of file diff --git a/vhdlpp/vhdlint.h b/vhdlpp/vhdlint.h new file mode 100644 index 000000000..0ef5ef54a --- /dev/null +++ b/vhdlpp/vhdlint.h @@ -0,0 +1,28 @@ +#ifndef __vhdlint_H +#define __vhdlint_H + +#include "config.h" +#include + +using namespace std; + +class vhdlint +{ + public: + explicit vhdlint(const char* text); + explicit vhdlint(const int64_t& val); + explicit vhdlint(const vhdlint& val); + + bool is_negative() const; + bool is_positive() const; + bool is_zero() const; + + int64_t as_long() const; + //vhdlv get(const unsigned index) const; + //void set(const unsigned index, const unsigned val); + // unsigned short operator[](const unsigned index); + private: + int64_t value_; +}; + +#endif diff --git a/vhdlpp/vhdlreal.cc b/vhdlpp/vhdlreal.cc new file mode 100644 index 000000000..fd903d544 --- /dev/null +++ b/vhdlpp/vhdlreal.cc @@ -0,0 +1,59 @@ +#include "config.h" +#include "compiler.h" +#include "vhdlreal.h" +#include +#include +#include + +vhdlreal::vhdlreal() { + value_ = 0.0; +} + +vhdlreal::vhdlreal(const double& r) { + value_ = r; +} + +vhdlreal::vhdlreal(const vhdlreal& val) { + value_ = val.as_double(); +} + +vhdlreal::vhdlreal(const char* text) { + assert(strlen(text) != 0); + char* buffer = new char[strlen(text)+1]; + + char* buf_ptr; + for(buf_ptr = buffer; *text != 0; ++buf_ptr, ++text) { + if(*text == '_') + continue; + *buf_ptr = *text; + } + *buf_ptr = '\0'; + + value_ = strtod(buffer, NULL); + delete[] buffer; +} + +ostream& operator<< (ostream& str, const vhdlreal& r) { + return (str << r.as_double()); +} +vhdlreal operator+ (const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(r1.as_double() + r2.as_double()); +} +vhdlreal operator- (const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(r1.as_double() - r2.as_double()); +} +vhdlreal operator* (const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(r1.as_double() * r2.as_double()); +} +vhdlreal operator/ (const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(r1.as_double() / r2.as_double()); +} +vhdlreal operator% (const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(fmod(r1.as_double(), r2.as_double())); +} +vhdlreal pow(const vhdlreal& r1, const vhdlreal& r2) { + return vhdlreal(pow(r1.as_double(), r2.as_double())); +} +vhdlreal operator- (const vhdlreal& r) { + return vhdlreal(-r.as_double()); +} \ No newline at end of file diff --git a/vhdlpp/vhdlreal.h b/vhdlpp/vhdlreal.h index c74263fdd..72c41f101 100644 --- a/vhdlpp/vhdlreal.h +++ b/vhdlpp/vhdlreal.h @@ -2,10 +2,39 @@ #define __vhdlreal_h #include "config.h" +#include +#include -class vhdlreal { - public: - vhdlreal(char* text) {} +using namespace std; +/* +* This class holds a floating point decimal number. The number is +* stored as double. All based numbers are converted by an external +* function to a double and then stored as class instance. +*/ +class vhdlreal +{ +public: + friend ostream& operator<< (ostream&, const vhdlreal&); + friend vhdlreal operator+ (const vhdlreal&, const vhdlreal&); + friend vhdlreal operator- (const vhdlreal&, const vhdlreal&); + friend vhdlreal operator* (const vhdlreal&, const vhdlreal&); + friend vhdlreal operator/ (const vhdlreal&, const vhdlreal&); + friend vhdlreal operator% (const vhdlreal&, const vhdlreal&); + friend vhdlreal pow(const vhdlreal&, const vhdlreal&); + // Unary minus. + friend vhdlreal operator- (const vhdlreal&); + + explicit vhdlreal(); + explicit vhdlreal(const char*text); + explicit vhdlreal(const double& val); + vhdlreal(const vhdlreal& val); + virtual ~vhdlreal() {}; + + double as_double() const + { + return value_; + } +protected: + double value_; }; - -#endif +#endif \ No newline at end of file