1

/*

     2

 *  matiec - a compiler for the programming languages defined in IEC 61131-3

     3

 *

     4

 *  Copyright (C) 2009-2011  Mario de Sousa (msousa@fe.up.pt)

     5

 *  Copyright (C) 2007-2011  Laurent Bessard and Edouard Tisserant

     6

 *

     7

 *  This program is free software: you can redistribute it and/or modify

     8

 *  it under the terms of the GNU General Public License as published by

     9

 *  the Free Software Foundation, either version 3 of the License, or

    10

 *  (at your option) any later version.

    11

 *

    12

 *  This program is distributed in the hope that it will be useful,

    13

 *  but WITHOUT ANY WARRANTY; without even the implied warranty of

    14

 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

    15

 *  GNU General Public License for more details.

    16

 *

    17

 *  You should have received a copy of the GNU General Public License

    18

 *  along with this program.  If not, see <http://www.gnu.org/licenses/>.

    19

 *

    20

 *

    21

 * This code is made available on the understanding that it will not be

    22

 * used in safety-critical situations without a full and competent review.

    23

 */

    24

    25

/*

    26

 * An IEC 61131-3 compiler.

    27

 *

    28

 * Based on the

    29

 * FINAL DRAFT - IEC 61131-3, 2nd Ed. (2001-12-10)

    30

 *

    31

 */

    32

    33

/* Determine the size, in bits, of the data type.

    34

 * 

    35

 * NOTE: Currently, only elementary data types with well defined sizes (in the standard) are supported.

    36

 *       - derived data types are not supported, and these will return 0

    37

 *       - TIME, DATE, TIME_OF_DAY, and DATE_AND_TIME are not supported, and will return 0

    38

 *       - STRING and WSTRING are not supported, and the standard merely defines bit per character,

    39

 *              and not the maximum number of characters, so these will return 0

    40

 *

    41

 *       We also support the 'Numeric Literals' Data types.

    42

 *       i.e., numeric literals are considerd basic data types

    43

 *       as their data type is undefined (e.g. the datat type of '30'

    44

 *       could be 'INT' or 'SINT' or 'LINT' or 'USINT' or ...

    45

 *       NOTE: for base 10 numeric literals, any number taking up more than 64 bits

    46

 *             will only return a bitsize of 1024!

    47

 *

    48

 *       NOTE: The code that does the following has been commented out, since we no longer need it!

    49

 *             It has been superceded by the constant_folding.cc class.

    50

 *       // For numeric literals, we return the minimum number of bits

    51

 *       // required to store the value.

    52

 *

    53

 * E.g. TYPE new_int_t : INT; END_TYPE;

    54

 *      TYPE new_int2_t : INT = 2; END_TYPE;

    55

 *      TYPE new_subr_t : INT (4..5); END_TYPE;

    56

 *

    57

 *    sizeof(SINT) ->  8

    58

 *    sizeof(INT)  -> 16

    59

 *    sizeof(DINT) -> 32

    60

 *    sizeof(LINT) -> 64

    61

 *

    62

 *       NOTE: The code that does the following has been commented out, since we no longer need it!

    63

 *             It has been superceded by the constant_folding.cc class.

    64

 *    // sizeof('1')       ->  1

    65

 *    // sizeof('015')     ->  4    # Leading zeros are ignored!

    66

 *    // sizeof('0')       ->  1    # This is a special case! Even the value 0 needs at least 1 bit to store!

    67

 *    // sizeof('16')      ->  5

    68

 *    // sizeof('2#00101') ->  3

    69

 *    // sizeof('8#334')   ->  9

    70

 *    // sizeof('16#2A')   ->  8

    71

 *

    72

 *    // sizeof('7.4')     ->  32   # all real literals return 32 bits, the size of a 'REAL'

    73

 *    //                            # TODO: study IEC 60559 for the range of values that may be

    74

 *    //                            #       stored in a REAL (basic single width floating point format)

    75

 *    //                            #       and in a LREAL (basic double width floating point format)

    76

 *    //                            #       and see if some real literals need to return 64 instead!

    77

 */

    78

    79

#include "get_sizeof_datatype.hh"

    80

    81

#include <stdlib.h>

    82

#include <string.h>

    83

#include <limits.h>  // get definition of ULLONG_MAX

    84

#include <errno.h>

    85

    86

#include "../main.hh" // required for ERROR() and ERROR_MSG() macros, and uint64_t and UINT64_MAX

    87

    88

    89

    90

/* This class is a singleton.

    91

 * So we need a pointer to the singe instance...

    92

 */

    93

get_sizeof_datatype_c *get_sizeof_datatype_c::singleton = NULL;

    94

    95

    96

#define _encode_int(value)   ((void *)(((char *)NULL) + value))

    97

#define _decode_int(ptr)     (((char *)ptr) - ((char *)NULL))

    98

    99

   100

#if 0   /* We no longer need the code for handling numeric literals. But keep it around for a little while longer... */

   101

/* divide a base 10 literal in a string by 2 */

   102

/* returns remainder of division (0 or 1)    */

   103

static int strdivby2(char **strptr) {

   104

  char *str = *strptr;

   105

  int carry = 0;

   106

   107

  while (*str != '\0') {

   108

    /* Assumes ASCII */

   109

    int newcarry;

   110

//     newcarry = ((*str-'0') mod 2);

   111

    newcarry = ((*str-'0') - ((*str-'0')/2)*2);

   112

    *str = (((*str-'0') + 10*carry)/2) + '0';

   113

    carry = newcarry;

   114

    str++;

   115

  }

   116

   117

  /* ignore leading zeros in result... */

   118

  while (**strptr == '0') 

   119

    (*strptr)++;

   120

   121

  return carry;

   122

}

   123

#endif

   124

   125

/* Constructor for the singleton class */

   126

int get_sizeof_datatype_c::getsize(symbol_c *data_type_symbol) {

   127

      if (NULL == singleton) {

   128

        singleton = new get_sizeof_datatype_c;

   129

        if (NULL == singleton)

   130

          ERROR;

   131

      }

   132

      return _decode_int(data_type_symbol->accept(*singleton));

   133

    }

   134

   135

/* Destructor for the singleton class */

   136

get_sizeof_datatype_c::~get_sizeof_datatype_c(void) {

   137

      if (NULL != singleton) delete singleton;

   138

      singleton = NULL;

   139

    }

   140

   141

#if 0   /* We no longer need the code for handling numeric literals. But keep it around for a little while longer... */

   142

/*********************/

   143

/* B 1.2 - Constants */

   144

/*********************/

   145

   146

/******************************/

   147

/* B 1.2.1 - Numeric Literals */

   148

/******************************/

   149

 /* Numeric literals without any explicit type cast have unknown data type, 

   150

  * so we continue considering them as their own basic data types until

   151

  * they can be resolved (for example, when using '30+x' where 'x' is a LINT variable, the

   152

  * numeric literal '30' must then be considered a LINT so the ADD function may be called

   153

  * with all inputs of the same data type.

   154

  * If 'x' were a SINT, then the '30' would have to be a SINT too!

   155

  */

   156

   157

/* NOTE: all integer_c and real_c tokens will always be positive (i.e. no leading '-')

   158

 * due to the way the source code is parsed by iec.flex.

   159

 */

   160

   161

/*

   162

 * IEC6113-3 and C++ use IEC 60559 to rappresent floating point data types

   163

 * REAL  => float       => single precision 	32 bit

   164

 * LREAL => double      => double precision 	64 bit

   165

 * ????? => long double => quadruple precision 128 bit

   166

 */

   167

void *get_sizeof_datatype_c::visit(real_c *symbol) {

   168

  char *endp;

   169

  long double ld_test;

   170

  double d_test;

   171

  float  f_test;

   172

   173

  /* copy the original string, but leave out any underscores... */

   174

  char *sval, *oval;

   175

  const char *pval;

   176

  oval = sval = (char *)malloc(strlen(symbol->value)+1);

   177

  if (NULL ==  sval) ERROR;

   178

   179

  for (pval = symbol->value, sval = oval; *pval != '\0'; pval++) {

   180

    if ('_' != *pval) {*sval = *pval; sval++;}

   181

  }  

   182

  *sval = '\0';  

   183

   184

  sval = oval;

   185

  if ('\0' == *sval) ERROR;

   186

   187

  /* now do the conversion using the new string... */

   188

  f_test = strtof(sval, &endp);

   189

  if (*endp != '\0') ERROR;

   190

  if (ERANGE != errno) {

   191

    /* No overflow/underflow! => It fits in a float! */

   192

    free(oval);

   193

    return _encode_int(32);

   194

  }

   195

   196

  d_test = strtod(sval, &endp);

   197

  if (*endp != '\0') ERROR;

   198

  if (ERANGE != errno) {

   199

    /* No overflow/underflow! => It fits in a double! */

   200

    free(oval);

   201

    return _encode_int(64);

   202

  }

   203

   204

  ld_test = strtold(sval, &endp);

   205

  if (*endp != '\0') ERROR;

   206

  if (ERANGE != errno) {

   207

    /* No overflow/underflow! => It fits in a long double! */

   208

    free(oval);

   209

    return _encode_int(128);

   210

  }

   211

   212

  free(oval);

   213

  return _encode_int(65535); /* a very large number!!! */

   214

}

   215

   216

void *get_sizeof_datatype_c::visit(neg_real_c *symbol) {

   217

  return symbol->exp->accept(*this);

   218

}

   219

   220

   221

/* NOTE: all integer_c and real_c literal tokens will always be positive (i.e. no leading '-')

   222

 * due to the way the source code is parsed by iec.flex.

   223

 */

   224

void *get_sizeof_datatype_c::visit(integer_c *symbol) {

   225

  int bitsize = 0;

   226

   227

  if (NULL ==   symbol->value ) ERROR;

   228

  if ('\0' == *(symbol->value)) ERROR;

   229

   230

#if 0

   231

  char *endptr;

   232

  /* Convert the string to an unsigned 64 bit integer */

   233

  /* We can use strtoull(), but we are not guaranteed that an unsigned long long int

   234

   * is 64 bits wide. First make sure that it is...

   235

   *

   236

   * We could also use the strtouq() instead, which converts

   237

   * to a quad word (64 bits). However, this requires either GCC or BSD extensions.

   238

   */

   239

  #ifdef strtoull // this ifdef does not work!!

   240

    /* we have long long int, use it... */

   241

    #define ival_MAX ULLONG_MAX

   242

    unsigned long long int ival = 0;

   243

    ival = strtoull (symbol->value, &endptr, 10 /* base */);

   244

  #else

   245

    /* use long int ... */

   246

    #define ival_MAX ULONG_MAX

   247

    unsigned long int ival = 0;

   248

    ival = strtoul (symbol->value, &endptr, 10 /* base */);

   249

  #endif

   250

   251

  #if (ival_MAX < UINT64_MAX)

   252

  #error Largest strtoint() conversion function converts to an int less than 64 bits wide!

   253

  #endif

   254

   255

  if (NULL ==  endptr) ERROR;

   256

  if ('\0' != *endptr) ERROR;

   257

  // TODO: return _encode_int(1024) if value takes up more than 64 bits! 

   258

   259

  /* determine the number of bits used... */

   260

  for (bitsize = 0; ival > 0; ival /= 2, bitsize++);

   261

   262

  /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */

   263

  if (0 == bitsize) bitsize = 1;

   264

   265

  return _encode_int(bitsize);

   266

#endif

   267

  /* We could first convert from string to some kind of integer, and then

   268

   * determine the the bitsize using integer aritmetic.

   269

   * However, we are then limited to the bit size of the widest available integer

   270

   * (usually 64 bits), which is not good at all!

   271

   */

   272

  /* Let's try to determine bitsize by converting directly from the string!! */

   273

  char *sval, *oval;

   274

  const char *pval;

   275

  oval = sval = (char *)malloc(strlen(symbol->value)+1);

   276

  if (NULL ==  sval) ERROR;

   277

   278

  /* copy the original string, but leave out any underscores... */

   279

  for (pval = symbol->value, sval = oval; *pval != '\0'; pval++) {

   280

    if ('_' != *pval) {*sval = *pval; sval++;}

   281

  }  

   282

  *sval = '\0';  

   283

   284

  sval = oval;

   285

  if ('\0' == *sval) ERROR;

   286

   287

  for (bitsize = 0; *sval != '\0'; strdivby2(&sval), bitsize ++);

   288

   289

/*

   290

  char *sval = strdup(symbol->value);

   291

  char *oval = sval;

   292

  if (NULL ==  sval) ERROR;

   293

  if ('\0' == *sval) ERROR;

   294

  for (bitsize = 0; *sval != '\0'; strdivby2(&sval), bitsize ++);

   295

*/

   296

   297

  /* Even for (value == 0), the above loop will return bitsize == 1!, 

   298

   * so we don't need to handle the special case...

   299

   */

   300

  /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */

   301

  // if (0 == bitsize) bitsize = 1;

   302

   303

  free(oval);

   304

  return _encode_int(bitsize);

   305

}

   306

   307

   308

void *get_sizeof_datatype_c::visit(neg_integer_c *symbol) {

   309

  return symbol->exp->accept(*this);

   310

}

   311

   312

   313

/* NOTE: all binary_integer_c tokens will always be positive (i.e. no leading '-')

   314

 * due to the syntax definition of IEC 61131-3.

   315

 */

   316

void *get_sizeof_datatype_c::visit(binary_integer_c *symbol) {

   317

  const char *sval = symbol->value;

   318

  int bitsize = 0;

   319

   320

  /* first 2 characters had better be "2#" ! */

   321

  if (NULL ==  sval) ERROR;

   322

  if ('\0' == *sval) ERROR;

   323

  if ( '2' != *sval) ERROR;

   324

  sval++;

   325

  if ('\0' == *sval) ERROR;

   326

  if ( '#' != *sval) ERROR;

   327

  sval++;

   328

  if ('\0' == *sval) ERROR;

   329

   330

  /* determine the number of bits used... */

   331

  for (bitsize = 0; '\0' != *sval; sval++) {

   332

    /* consistency check: make sure we only have binary digits! */

   333

    if (('0' != *sval) && ('1' != *sval) && ('_' != *sval))

   334

      ERROR;

   335

   336

    if ('_' != *sval) bitsize++; /* 1 bits per binary digit */

   337

  }

   338

   339

  /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */

   340

  if (0 == bitsize) bitsize = 1;

   341

   342

  return _encode_int(bitsize);

   343

}

   344

   345

   346

/* NOTE: all octal_integer_c tokens will always be positive (i.e. no leading '-')

   347

 * due to the syntax definition of IEC 61131-3.

   348

 */

   349

void *get_sizeof_datatype_c::visit(octal_integer_c *symbol) {

   350

  const char *sval = symbol->value;

   351

  int bitsize = 0;

   352

   353

  /* first 2 characters had better be "8#" ! */

   354

  if (NULL ==  sval) ERROR;

   355

  if ('\0' == *sval) ERROR;

   356

  if ( '8' != *sval) ERROR;

   357

  sval++;

   358

  if ('\0' == *sval) ERROR;

   359

  if ( '#' != *sval) ERROR;

   360

  sval++;

   361

  if ('\0' == *sval) ERROR;

   362

   363

  /* determine the number of bits used... */

   364

  for (bitsize = 0; '\0' != *sval; sval++) {

   365

    /* consistency check: make sure we only have octal digits! */

   366

    /* Assumes ASCII */

   367

    if ((('0' > *sval) || ('7' < *sval)) && ('_' != *sval))

   368

      ERROR;

   369

   370

    if ('_' != *sval) bitsize += 3; /* 3 bits per octal digit */

   371

  }

   372

   373

  /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */

   374

  if (0 == bitsize) bitsize = 1;

   375

   376

  return _encode_int(bitsize);

   377

}

   378

   379

   380

/* NOTE: all hex_integer_c tokens will always be positive (i.e. no leading '-')

   381

 * due to the syntax definition of IEC 61131-3.

   382

 */

   383

void *get_sizeof_datatype_c::visit(hex_integer_c *symbol) {

   384

  const char *sval = symbol->value;

   385

  int bitsize = 0;

   386

   387

  /* first 3 characters had better be "16#" ! */

   388

  if (NULL ==  sval) ERROR;

   389

  if ('\0' == *sval) ERROR;

   390

  if ( '1' != *sval) ERROR;

   391

  sval++;

   392

  if ('\0' == *sval) ERROR;

   393

  if ( '6' != *sval) ERROR;

   394

  sval++;

   395

  if ('\0' == *sval) ERROR;

   396

  if ( '#' != *sval) ERROR;

   397

  sval++;

   398

  if ('\0' == *sval) ERROR;

   399

   400

  /* determine the number of bits used... */

   401

  for (bitsize = 0; '\0' != *sval; sval++) {

   402

    /* consistency check: make sure we only have hex digits or underscores! */

   403

    /* Assumes ASCII */

   404

    if (!(('0' <= *sval) && ('9' >= *sval)) && 

   405

        !(('A' <= *sval) && ('F' >= *sval)) &&

   406

        !(('a' <= *sval) && ('f' >= *sval)) &&

   407

        ! ('_' == *sval))

   408

      ERROR;

   409

   410

    if ('_' != *sval) bitsize += 4; /* 4 bits per hex digit */

   411

  }

   412

   413

  /* special case... if (value == 0) <=> (bitsize == 0), return bit size of 1 ! */

   414

  if (0 == bitsize) bitsize = 1;

   415

   416

  return _encode_int(bitsize);

   417

}

   418

#endif

   419

   420

/***********************************/

   421

/* B 1.3.1 - Elementary Data Types */

   422

/***********************************/

   423

// void *get_sizeof_datatype_c::visit(time_type_name_c *symbol) {return _encode_int(0); }

   424

void *get_sizeof_datatype_c::visit(bool_type_name_c *symbol)    {return _encode_int(1); }

   425

void *get_sizeof_datatype_c::visit(sint_type_name_c *symbol)    {return _encode_int(8); }

   426

void *get_sizeof_datatype_c::visit(int_type_name_c *symbol)     {return _encode_int(16);}

   427

void *get_sizeof_datatype_c::visit(dint_type_name_c *symbol)    {return _encode_int(32);}

   428

void *get_sizeof_datatype_c::visit(lint_type_name_c *symbol)    {return _encode_int(64);}

   429

void *get_sizeof_datatype_c::visit(usint_type_name_c *symbol)   {return _encode_int(8); }

   430

void *get_sizeof_datatype_c::visit(uint_type_name_c *symbol)    {return _encode_int(16);}

   431

void *get_sizeof_datatype_c::visit(udint_type_name_c *symbol)   {return _encode_int(32);}

   432

void *get_sizeof_datatype_c::visit(ulint_type_name_c *symbol)   {return _encode_int(64);}

   433

void *get_sizeof_datatype_c::visit(real_type_name_c *symbol)    {return _encode_int(32);}

   434

void *get_sizeof_datatype_c::visit(lreal_type_name_c *symbol)   {return _encode_int(64);}

   435

// void *get_sizeof_datatype_c::visit(date_type_name_c *symbol) {return _encode_int(0); }

   436

// void *get_sizeof_datatype_c::visit(tod_type_name_c *symbol)  {return _encode_int(0); }

   437

// void *get_sizeof_datatype_c::visit(dt_type_name_c *symbol)   {return _encode_int(0); }

   438

void *get_sizeof_datatype_c::visit(byte_type_name_c *symbol)    {return _encode_int(8); }

   439

void *get_sizeof_datatype_c::visit(word_type_name_c *symbol)    {return _encode_int(16);}

   440

void *get_sizeof_datatype_c::visit(dword_type_name_c *symbol)   {return _encode_int(32);}

   441

void *get_sizeof_datatype_c::visit(lword_type_name_c *symbol)   {return _encode_int(64);}

   442

// void *get_sizeof_datatype_c::visit(string_type_name_c *symbol)    {return _encode_int(0); }

   443

// void *get_sizeof_datatype_c::visit(wstring_type_name_c *symbol)   {return _encode_int(0); }

   444

/******************************************************/

   445

/* Extensions to the base standard as defined in      */

   446

/* "Safety Software Technical Specification,          */

   447

/*  Part 1: Concepts and Function Blocks,             */

   448

/*  Version 1.0 – Official Release"                   */

   449

/* by PLCopen - Technical Committee 5 - 2006-01-31    */

   450

/******************************************************/

   451

//     void *get_sizeof_datatype_c::visit(safetime_type_name_c *symbol); {return _encode_int(0); }

   452

void *get_sizeof_datatype_c::visit(safebool_type_name_c *symbol)         {return _encode_int(1); }

   453

void *get_sizeof_datatype_c::visit(safesint_type_name_c *symbol)         {return _encode_int(8); }

   454

void *get_sizeof_datatype_c::visit(safeint_type_name_c *symbol)          {return _encode_int(16);}

   455

void *get_sizeof_datatype_c::visit(safedint_type_name_c *symbol)         {return _encode_int(32);}

   456

void *get_sizeof_datatype_c::visit(safelint_type_name_c *symbol)         {return _encode_int(64);}

   457

void *get_sizeof_datatype_c::visit(safeusint_type_name_c *symbol)        {return _encode_int(8); }

   458

void *get_sizeof_datatype_c::visit(safeuint_type_name_c *symbol)         {return _encode_int(16);}

   459

void *get_sizeof_datatype_c::visit(safeudint_type_name_c *symbol)        {return _encode_int(32);}

   460

void *get_sizeof_datatype_c::visit(safeulint_type_name_c *symbol)        {return _encode_int(64);}

   461

void *get_sizeof_datatype_c::visit(safereal_type_name_c *symbol)         {return _encode_int(32);}

   462

void *get_sizeof_datatype_c::visit(safelreal_type_name_c *symbol)        {return _encode_int(64);}

   463

//     void *get_sizeof_datatype_c::visit(safedate_type_name_c *symbol); {return _encode_int(0); }

   464

//     void *get_sizeof_datatype_c::visit(safetod_type_name_c *symbol);  {return _encode_int(0); }

   465

//     void *get_sizeof_datatype_c::visit(safedt_type_name_c *symbol);   {return _encode_int(0); }

   466

void *get_sizeof_datatype_c::visit(safebyte_type_name_c *symbol)         {return _encode_int(8); }

   467

void *get_sizeof_datatype_c::visit(safeword_type_name_c *symbol)         {return _encode_int(16);}

   468

void *get_sizeof_datatype_c::visit(safedword_type_name_c *symbol)        {return _encode_int(32);}

   469

void *get_sizeof_datatype_c::visit(safelword_type_name_c *symbol)        {return _encode_int(64);}

   470

//     void *get_sizeof_datatype_c::visit(safestring_type_name_c *symbol);  {return _encode_int(0); }

   471

//     void *get_sizeof_datatype_c::visit(safewstring_type_name_c *symbol); {return _encode_int(0); }

   472

   473

   474

/********************************/

   475

/* B 1.3.3 - Derived data types */

   476

/********************************/

   477

// Not yet supported...