absyntax_utils/get_sizeof_datatype.cc
changeset 202 da1a8186f86f
child 257 90782e241346
equal deleted inserted replaced
201:e657008f43d0 202:da1a8186f86f
       
     1 /*
       
     2  * (c) 2003 Mario de Sousa
       
     3  *
       
     4  * Offered to the public under the terms of the GNU General Public License
       
     5  * as published by the Free Software Foundation; either version 2 of the
       
     6  * License, or (at your option) any later version.
       
     7  *
       
     8  * This program is distributed in the hope that it will be useful, but
       
     9  * WITHOUT ANY WARRANTY; without even the implied warranty of
       
    10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
       
    11  * Public License for more details.
       
    12  *
       
    13  * This code is made available on the understanding that it will not be
       
    14  * used in safety-critical situations without a full and competent review.
       
    15  */
       
    16 
       
    17 /*
       
    18  * An IEC 61131-3 IL and ST compiler.
       
    19  *
       
    20  * Based on the
       
    21  * FINAL DRAFT - IEC 61131-3, 2nd Ed. (2001-12-10)
       
    22  *
       
    23  */
       
    24 
       
    25 /* Determine the size, in bits, of the data type.
       
    26  * 
       
    27  * NOTE: Currently, only elementary data types with well defined sizes (in the standard) are supported.
       
    28  *       - derived data types are not supported, and these will return 0
       
    29  *       - TIME, DATE, TIME_OF_DAY, and DATE_AND_TIME are not supported, and will return 0
       
    30  *       - STRING and WSTRING are not supported, and the standard merely defines bit per character,
       
    31  *              and not the maximum number of characters, so these will return 0
       
    32  *
       
    33  *       We also support the 'Numeric Literals' Data types.
       
    34  *       i.e., numeric literals are considerd basic data types
       
    35  *       as their data type is undefined (e.g. the datat type of '30'
       
    36  *       could be 'INT' or 'SINT' or 'LINT' or 'USINT' or ...
       
    37  *       NOTE: for base 10 numeric literals, any number taking up more than 64 bits
       
    38  *             will only return a bitsize of 1024!
       
    39  *
       
    40  *       For numeric literals, we return the minimum number of bits
       
    41  *       required to store the value.
       
    42  *
       
    43  * E.g. TYPE new_int_t : INT; END_TYPE;
       
    44  *      TYPE new_int2_t : INT = 2; END_TYPE;
       
    45  *      TYPE new_subr_t : INT (4..5); END_TYPE;
       
    46  *
       
    47  *    sizeof(SINT) ->  8
       
    48  *    sizeof(INT)  -> 16
       
    49  *    sizeof(DINT) -> 32
       
    50  *    sizeof(LINT) -> 64
       
    51  *
       
    52  *    sizeof('1')       ->  1
       
    53  *    sizeof('015')     ->  4    # Leading zeros are ignored!
       
    54  *    sizeof('0')       ->  1    # This is a special case! Even the value 0 needs at least 1 bit to store!
       
    55  *    sizeof('16')      ->  5
       
    56  *    sizeof('2#00101') ->  3
       
    57  *    sizeof('8#334')   ->  9
       
    58  *    sizeof('16#2A')   ->  8
       
    59  *
       
    60  *    sizeof('7.4')     ->  32   # all real literals return 32 bits, the size of a 'REAL'
       
    61  *                               # TODO: study IEC 60559 for the range of values that may be
       
    62  *                               #       stored in a REAL (basic single width floating point format)
       
    63  *                               #       and in a LREAL (basic double width floating point format)
       
    64  *                               #       and see if some real literals need to return 64 instead!
       
    65  */
       
    66 
       
    67 #include "get_sizeof_datatype.hh"
       
    68 
       
    69 #include <stdlib.h>
       
    70 #include <string.h>
       
    71 #include <limits.h>  // get definition of ULLONG_MAX
       
    72 /* tell stdint.h we want the definition of UINT64_MAX */
       
    73 #define __STDC_LIMIT_MACROS
       
    74 #include <stdint.h>  // get definition of uint64_t and UINT64_MAX
       
    75 
       
    76 
       
    77 #define ERROR error_exit(__FILE__,__LINE__)
       
    78 /* function defined in main.cc */
       
    79 extern void error_exit(const char *file_name, int line_no);
       
    80 
       
    81 
       
    82 /* This class is a singleton.
       
    83  * So we need a pointer to the singe instance...
       
    84  */
       
    85 get_sizeof_datatype_c *get_sizeof_datatype_c::singleton = NULL;
       
    86 
       
    87 
       
    88 #define _encode_int(value)   ((void *)(((char *)NULL) + value))
       
    89 #define _decode_int(ptr)     (((char *)ptr) - ((char *)NULL))
       
    90 
       
    91 
       
    92 
       
    93 
       
    94 /* divide a base 10 literal in a string by 2 */
       
    95 /* returns remainder of division (0 or 1)    */
       
    96 static int strdivby2(char **strptr) {
       
    97   char *str = *strptr;
       
    98   int carry = 0;
       
    99 
       
   100   while (*str != '\0') {
       
   101     /* Assumes ASCII */
       
   102     int newcarry;
       
   103 //     newcarry = ((*str-'0') mod 2);
       
   104     newcarry = ((*str-'0') - ((*str-'0')/2)*2);
       
   105     *str = (((*str-'0') + 10*carry)/2) + '0';
       
   106     carry = newcarry;
       
   107     str++;
       
   108   }
       
   109 
       
   110   /* ignore leading zeros in result... */
       
   111   while (**strptr == '0') 
       
   112     (*strptr)++;
       
   113 
       
   114   return carry;
       
   115 }
       
   116 
       
   117 
       
   118 /* Constructor for the singleton class */
       
   119 int get_sizeof_datatype_c::getsize(symbol_c *data_type_symbol) {
       
   120       if (NULL == singleton) {
       
   121         singleton = new get_sizeof_datatype_c;
       
   122         if (NULL == singleton)
       
   123           ERROR;
       
   124       }
       
   125       return _decode_int(data_type_symbol->accept(*singleton));
       
   126     }
       
   127 
       
   128 /* Destructor for the singleton class */
       
   129 get_sizeof_datatype_c::~get_sizeof_datatype_c(void) {
       
   130       if (NULL != singleton) delete singleton;
       
   131       singleton = NULL;
       
   132     }
       
   133 
       
   134 
       
   135 /*********************/
       
   136 /* B 1.2 - Constants */
       
   137 /*********************/
       
   138 
       
   139 /******************************/
       
   140 /* B 1.2.1 - Numeric Literals */
       
   141 /******************************/
       
   142  /* Numeric literals without any explicit type cast have unknown data type, 
       
   143   * so we continue considering them as their own basic data types until
       
   144   * they can be resolved (for example, when using '30+x' where 'x' is a LINT variable, the
       
   145   * numeric literal '30' must then be considered a LINT so the ADD function may be called
       
   146   * with all inputs of the same data type.
       
   147   * If 'x' were a SINT, then the '30' would have to be a SINT too!
       
   148   */
       
   149 
       
   150 /* NOTE: all integer and real literal tokens will always be positive (i.e. no leading '-')
       
   151  * due to the way the source code is parsed by iec.flex.
       
   152  */
       
   153 void *get_sizeof_datatype_c::visit(real_c *symbol) {
       
   154   return _encode_int(32);
       
   155 }
       
   156 
       
   157 /* NOTE: all integer and real literal tokens will always be positive (i.e. no leading '-')
       
   158  * due to the way the source code is parsed by iec.flex.
       
   159  */
       
   160 void *get_sizeof_datatype_c::visit(integer_c *symbol) {
       
   161   int bitsize = 0;
       
   162 
       
   163   if (NULL ==   symbol->value ) ERROR;
       
   164   if ('\0' == *(symbol->value)) ERROR;
       
   165 
       
   166 #if 0
       
   167   char *endptr;
       
   168   /* Convert the string to an unsigned 64 bit integer */
       
   169   /* We can use strtoull(), but we are not guaranteed that an unsigned long long int
       
   170    * is 64 bits wide. First make sure that it is...
       
   171    *
       
   172    * We could also use the strtouq() instead, which converts
       
   173    * to a quad word (64 bits). However, this requires either GCC or BSD extensions.
       
   174    */
       
   175   #ifdef strtoull // this ifdef does not work!!
       
   176     /* we have long long int, use it... */
       
   177     #define ival_MAX ULLONG_MAX
       
   178     unsigned long long int ival = 0;
       
   179     ival = strtoull (symbol->value, &endptr, 10 /* base */);
       
   180   #else
       
   181     /* use long int ... */
       
   182     #define ival_MAX ULONG_MAX
       
   183     unsigned long int ival = 0;
       
   184     ival = strtoul (symbol->value, &endptr, 10 /* base */);
       
   185   #endif
       
   186 
       
   187   #if (ival_MAX < UINT64_MAX)
       
   188   #error Largest strtoint() conversion function converts to an int less than 64 bits wide!
       
   189   #endif
       
   190 
       
   191   if (NULL ==  endptr) ERROR;
       
   192   if ('\0' != *endptr) ERROR;
       
   193   // TODO: return _encode_int(1024) if value takes up more than 64 bits! 
       
   194 
       
   195   /* determine the number of bits used... */
       
   196   for (bitsize = 0; ival > 0; ival /= 2, bitsize++);
       
   197 
       
   198   /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */
       
   199   if (0 == bitsize) bitsize = 1;
       
   200 
       
   201   return _encode_int(bitsize);
       
   202 #endif
       
   203   /* We could first convert from string to some kind of integer, and then
       
   204    * determine the the bitsize using integer aritmetic.
       
   205    * However, we are then limited to the bit size of the widest available integer
       
   206    * (usually 64 bits), which is not good at all!
       
   207    */
       
   208   /* Let's try to determine bitsize by converting directly from the string!! */
       
   209   char *sval = strdup(symbol->value);
       
   210   char *oval = sval;
       
   211   if (NULL ==  sval) ERROR;
       
   212   if ('\0' == *sval) ERROR;
       
   213 
       
   214   for (bitsize = 0; *sval != '\0'; strdivby2(&sval), bitsize ++);
       
   215 
       
   216   /* Even for (value == 0), the above loop will return bitsize == 1!, 
       
   217    * so we don't need to handle the special case...
       
   218    */
       
   219   /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */
       
   220   // if (0 == bitsize) bitsize = 1;
       
   221 
       
   222   free(oval);
       
   223   return _encode_int(bitsize);
       
   224 }
       
   225 
       
   226 
       
   227 /* NOTE: all integer and real literal tokens will always be positive (i.e. no leading '-')
       
   228  * due to the way the source code is parsed by iec.flex.
       
   229  */
       
   230 void *get_sizeof_datatype_c::visit(binary_integer_c *symbol) {
       
   231   const char *sval = symbol->value;
       
   232   int bitsize = 0;
       
   233 
       
   234   /* first 2 characters had better be "2#" ! */
       
   235   if (NULL ==  sval) ERROR;
       
   236   if ('\0' == *sval) ERROR;
       
   237   if ( '2' != *sval) ERROR;
       
   238   sval++;
       
   239   if ('\0' == *sval) ERROR;
       
   240   if ( '#' != *sval) ERROR;
       
   241   sval++;
       
   242   if ('\0' == *sval) ERROR;
       
   243 
       
   244   /* determine the number of bits used... */
       
   245   for (bitsize = 0; '\0' != *sval; sval++, bitsize++) {
       
   246     /* consistency check: make sure we only have binary digits! */
       
   247     if (('0' != *sval) && ('1' != *sval))
       
   248       ERROR;
       
   249   }
       
   250 
       
   251   /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */
       
   252   if (0 == bitsize) bitsize = 1;
       
   253 
       
   254   return _encode_int(bitsize);
       
   255 }
       
   256 
       
   257 
       
   258 /* NOTE: all integer and real literal tokens will always be positive (i.e. no leading '-')
       
   259  * due to the way the source code is parsed by iec.flex.
       
   260  */
       
   261 void *get_sizeof_datatype_c::visit(octal_integer_c *symbol) {
       
   262   const char *sval = symbol->value;
       
   263   int bitsize = 0;
       
   264 
       
   265   /* first 2 characters had better be "8#" ! */
       
   266   if (NULL ==  sval) ERROR;
       
   267   if ('\0' == *sval) ERROR;
       
   268   if ( '8' != *sval) ERROR;
       
   269   sval++;
       
   270   if ('\0' == *sval) ERROR;
       
   271   if ( '#' != *sval) ERROR;
       
   272   sval++;
       
   273   if ('\0' == *sval) ERROR;
       
   274 
       
   275   /* determine the number of bits used... */
       
   276   for (bitsize = 0; '\0' != *sval; sval++, bitsize += 3 /* 3 bits per octal digit */) {
       
   277     /* consistency check: make sure we only have octal digits! */
       
   278     /* Assumes ASCII */
       
   279     if (('0' > *sval) || ('7' < *sval))
       
   280       ERROR;
       
   281   }
       
   282 
       
   283   /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */
       
   284   if (0 == bitsize) bitsize = 1;
       
   285 
       
   286   return _encode_int(bitsize);
       
   287 }
       
   288 
       
   289 
       
   290 /* NOTE: all integer and real literal tokens will always be positive (i.e. no leading '-')
       
   291  * due to the way the source code is parsed by iec.flex.
       
   292  */
       
   293 void *get_sizeof_datatype_c::visit(hex_integer_c *symbol) {
       
   294   const char *sval = symbol->value;
       
   295   int bitsize = 0;
       
   296 
       
   297   /* first 3 characters had better be "16#" ! */
       
   298   if (NULL ==  sval) ERROR;
       
   299   if ('\0' == *sval) ERROR;
       
   300   if ( '1' != *sval) ERROR;
       
   301   sval++;
       
   302   if ('\0' == *sval) ERROR;
       
   303   if ( '6' != *sval) ERROR;
       
   304   sval++;
       
   305   if ('\0' == *sval) ERROR;
       
   306   if ( '#' != *sval) ERROR;
       
   307   sval++;
       
   308   if ('\0' == *sval) ERROR;
       
   309 
       
   310   /* determine the number of bits used... */
       
   311   for (bitsize = 0; '\0' != *sval; sval++, bitsize += 4 /* 4 bits per hex digit */) {
       
   312     /* consistency check: make sure we only have hex digits! */
       
   313     /* Assumes ASCII */
       
   314     if (!(('0' <= *sval) && ('9' >= *sval)) && 
       
   315         !(('A' <= *sval) && ('F' >= *sval)) &&
       
   316         !(('a' <= *sval) && ('b' >= *sval)))
       
   317       ERROR;
       
   318   }
       
   319 
       
   320   /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */
       
   321   if (0 == bitsize) bitsize = 1;
       
   322 
       
   323   return _encode_int(bitsize);
       
   324 }
       
   325 
       
   326 
       
   327 /***********************************/
       
   328 /* B 1.3.1 - Elementary Data Types */
       
   329 /***********************************/
       
   330 // void *get_sizeof_datatype_c::visit(time_type_name_c *symbol)	{return _encode_int(0); }
       
   331 void *get_sizeof_datatype_c::visit(bool_type_name_c *symbol)	{return _encode_int(1); }
       
   332 void *get_sizeof_datatype_c::visit(sint_type_name_c *symbol)	{return _encode_int(8); }
       
   333 void *get_sizeof_datatype_c::visit(int_type_name_c *symbol)	{return _encode_int(16);}
       
   334 void *get_sizeof_datatype_c::visit(dint_type_name_c *symbol)	{return _encode_int(32);}
       
   335 void *get_sizeof_datatype_c::visit(lint_type_name_c *symbol)	{return _encode_int(64);}
       
   336 void *get_sizeof_datatype_c::visit(usint_type_name_c *symbol)	{return _encode_int(8); }
       
   337 void *get_sizeof_datatype_c::visit(uint_type_name_c *symbol)	{return _encode_int(16);}
       
   338 void *get_sizeof_datatype_c::visit(udint_type_name_c *symbol)	{return _encode_int(32);}
       
   339 void *get_sizeof_datatype_c::visit(ulint_type_name_c *symbol)	{return _encode_int(64);}
       
   340 void *get_sizeof_datatype_c::visit(real_type_name_c *symbol)	{return _encode_int(32);}
       
   341 void *get_sizeof_datatype_c::visit(lreal_type_name_c *symbol)	{return _encode_int(64);}
       
   342 // void *get_sizeof_datatype_c::visit(date_type_name_c *symbol)	{return _encode_int(0); }
       
   343 // void *get_sizeof_datatype_c::visit(tod_type_name_c *symbol)	{return _encode_int(0); }
       
   344 // void *get_sizeof_datatype_c::visit(dt_type_name_c *symbol)	{return _encode_int(0); }
       
   345 void *get_sizeof_datatype_c::visit(byte_type_name_c *symbol)	{return _encode_int(8); }
       
   346 void *get_sizeof_datatype_c::visit(word_type_name_c *symbol)	{return _encode_int(16);}
       
   347 void *get_sizeof_datatype_c::visit(dword_type_name_c *symbol)	{return _encode_int(32);}
       
   348 void *get_sizeof_datatype_c::visit(lword_type_name_c *symbol)	{return _encode_int(64);}
       
   349 // void *get_sizeof_datatype_c::visit(string_type_name_c *symbol)	{return _encode_int(0); }
       
   350 // void *get_sizeof_datatype_c::visit(wstring_type_name_c *symbol)	{return _encode_int(0); }
       
   351 /******************************************************/
       
   352 /* Extensions to the base standard as defined in      */
       
   353 /* "Safety Software Technical Specification,          */
       
   354 /*  Part 1: Concepts and Function Blocks,             */
       
   355 /*  Version 1.0 – Official Release"                   */
       
   356 /* by PLCopen - Technical Committee 5 - 2006-01-31    */
       
   357 /******************************************************/
       
   358 void *get_sizeof_datatype_c::visit(safebool_type_name_c *symbol)	{return _encode_int(1);}
       
   359 
       
   360 /********************************/
       
   361 /* B 1.3.3 - Derived data types */
       
   362 /********************************/
       
   363 // Not yet supported...