| 1 |
/* |
| 2 |
* The Regina Rexx Interpreter |
| 3 |
* Copyright (C) 1992-1994 Anders Christensen <[email protected]> |
| 4 |
* |
| 5 |
* This library is free software; you can redistribute it and/or |
| 6 |
* modify it under the terms of the GNU Library General Public |
| 7 |
* License as published by the Free Software Foundation; either |
| 8 |
* version 2 of the License, or (at your option) any later version. |
| 9 |
* |
| 10 |
* This library is distributed in the hope that it will be useful, |
| 11 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
| 13 |
* Library General Public License for more details. |
| 14 |
* |
| 15 |
* You should have received a copy of the GNU Library General Public |
| 16 |
* License along with this library; if not, write to the Free |
| 17 |
* Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
| 18 |
*/ |
| 19 |
|
| 20 |
/* |
| 21 |
* The functions in this file converts between the different |
| 22 |
* 'datatypes' that REXX uses, such as decimal number, character |
| 23 |
* string and hex string. Some of the functions are meant for the |
| 24 |
* standard library, like c2x, x2c, d2x, x2d, c2d and d2c, and also |
| 25 |
* the new functions in language level 4.00: b2x and x2b. The routines |
| 26 |
* in the file might also be used elsewhere for conversion between |
| 27 |
* char, decimal and hex. |
| 28 |
* |
| 29 |
* In general, there are two levels, the interface for the functions |
| 30 |
* in the builtin library (std_c2x, std_c2d, etc) and the low-level |
| 31 |
* routines that do the actual job. The interface routines 'only' call |
| 32 |
* the low-level routines in the right order, to preprocess parameters |
| 33 |
* and to postprocess the result. However, this implementation is not |
| 34 |
* 100% clean in this respect, since std_b2x() and std_x2b() are |
| 35 |
* selfcontained. |
| 36 |
* |
| 37 |
* When converting to or from decimal numbers, the decimal number will |
| 38 |
* trigger an overflow condition if more bits are pushed into the |
| 39 |
* number than the is allowed by the current setting of NUMERIC |
| 40 |
* DIGITS. |
| 41 |
* |
| 42 |
* The low-level routines are: |
| 43 |
* |
| 44 |
* pack_hex() --- packs a hex string into a char string |
| 45 |
* unpack_hex() --- unpacks a char string into a hex string. |
| 46 |
* numberize() --- converts a char string to a whole number |
| 47 |
* |
| 48 |
* In addition, low level routine that operates on whole numbers are |
| 49 |
* placed in strmath.c. The relevant routines there which are called |
| 50 |
* from this file are (numberize() is only a frontend to the function |
| 51 |
* str_digitize()): |
| 52 |
* |
| 53 |
* str_binerize() --- converts a whole number into a char string |
| 54 |
* str_digitize() --- converts a char string into a whole number |
| 55 |
* |
| 56 |
* The high-level routines in this file are: |
| 57 |
* |
| 58 |
* std_b2x() --- converts bin string to hex string |
| 59 |
* std_c2d() --- converts char string to whole number |
| 60 |
* std_c2x() --- converts char string to hex string |
| 61 |
* std_d2c() --- converts whole number to char string |
| 62 |
* std_d2x() --- converts whole number to hex string |
| 63 |
* std_x2b() --- converts hex string to bin string |
| 64 |
* std_x2c() --- converts hex string to char string |
| 65 |
* std_x2d() --- converts hex string to whole number |
| 66 |
* |
| 67 |
* |
| 68 |
* There are four 'generic' dataformats used in this implementation: |
| 69 |
* bin string, hex string, char string and whole number. |
| 70 |
* |
| 71 |
* * CHAR STRING consists of chars (any of the 256, including 0). It |
| 72 |
* can have any length (including 0). This is the 'normal' strings |
| 73 |
* in Rexx, but note that in the context of this module, char strings |
| 74 |
* will often contain unprintable characters, while in normal use |
| 75 |
* char strings will often just contain printable characters. |
| 76 |
* To emphsize the difference, it will sometimes be refered to as |
| 77 |
* packed char string. |
| 78 |
* |
| 79 |
* * BIN STRING consists of the binary digits 0 and 1, and will be |
| 80 |
* implicitly padded out to a nibble (4 bit) boundary. They may have |
| 81 |
* any length, including 0. Optional spaces might be added between |
| 82 |
* nibble boundaries, but not at the start or end of the string. The |
| 83 |
* first group of bin digits may have any number of digits, and will |
| 84 |
* be padded with leading zeros at the left to make it a multiple |
| 85 |
* of four digits. |
| 86 |
* |
| 87 |
* * HEX STRING consists of the hex digits 0-9 and A-F (in upper or |
| 88 |
* lower case). It may have any length (including 0). Optional |
| 89 |
* spaces might be added between pairs of digits to form groups of |
| 90 |
* hex digits, but not at the start or end. If the first group does |
| 91 |
* not contain an even number of hex digits, it is padded to the |
| 92 |
* left with an zero. |
| 93 |
* |
| 94 |
* * WHOLE NUMBER must be a string which is a valid Rexx whole number, |
| 95 |
* which is a subset of Rexx numbers. If it has a decimal part, that |
| 96 |
* must be zero (i.e. 13.0 is a whole number). In addition, it must |
| 97 |
* have a precition which is sufficient high to identify the number |
| 98 |
* as a particular integer (i.e. 13E1 is not a whole number, since |
| 99 |
* it really is 130 plus/minus 5). |
| 100 |
* |
| 101 |
* All these, except whole numbers, can have zero length, in which |
| 102 |
* case they are considered the nullstring. When generating these |
| 103 |
* datatype for output, they will be normalized, which means: |
| 104 |
* |
| 105 |
* * For whole numbers, bin and hex string, no additional space is |
| 106 |
* added; neither internal, leading nor trailing. |
| 107 |
* |
| 108 |
* * Whole numbers will not be in exponential form, and the decimal |
| 109 |
* part of the number (if any) is truncated. |
| 110 |
* |
| 111 |
* * Hex string will use upper case alphanumberic characters in stead |
| 112 |
* of lower case. Bin string will be padded with zeros at the left |
| 113 |
* to a multiple of four. |
| 114 |
*/ |
| 115 |
|
| 116 |
|
| 117 |
#include "rexx.h" |
| 118 |
#include <string.h> |
| 119 |
#include <stdio.h> |
| 120 |
#include <assert.h> |
| 121 |
|
| 122 |
|
| 123 |
/* |
| 124 |
* The following string is used to convert a nibble value |
| 125 |
* into its hexadecimal character set representation. Perhaps this |
| 126 |
* should be placed somewhere else, e.g. in misc.c |
| 127 |
*/ |
| 128 |
static const char hexnum[] = "0123456789ABCDEF" ; |
| 129 |
|
| 130 |
|
| 131 |
|
| 132 |
/* --------------------------------------------------------------- |
| 133 |
* Takes a bin string as input, and returns a normalized hex |
| 134 |
* string representing the same data. The output will have an even |
| 135 |
* number of digits. |
| 136 |
* |
| 137 |
* Since we are reading byte values in the range 0-255, we must take |
| 138 |
* care to use unsigned chars, else we might get some surprises. |
| 139 |
* Therefore there are some casting in this routine. |
| 140 |
*/ |
| 141 |
static streng *unpack_hex( const tsd_t *TSD, const streng *string ) |
| 142 |
{ |
| 143 |
streng *result=NULL ; /* the output string */ |
| 144 |
char *res_ptr=NULL ; /* ptr to current char in output string */ |
| 145 |
unsigned char *ptr=NULL ; /* ptr to input string */ |
| 146 |
unsigned char *end_ptr=NULL ; /* ptr to end+1 of input string */ |
| 147 |
|
| 148 |
/* |
| 149 |
* Allocate the needed space, which is very easy to calculate |
| 150 |
*/ |
| 151 |
result = Str_makeTSD( Str_len( string )*2 ) ; |
| 152 |
res_ptr = result->value ; |
| 153 |
|
| 154 |
/* |
| 155 |
* Initialize loop ptr and the end ptr, and loop through each |
| 156 |
* character in the input string. Write two output hex digits |
| 157 |
* for each char in the input string |
| 158 |
*/ |
| 159 |
end_ptr = (unsigned char *)(Str_end(string)) ; |
| 160 |
ptr = (unsigned char *)(string->value) ; |
| 161 |
for (; ptr<end_ptr; ptr++ ) |
| 162 |
{ |
| 163 |
*res_ptr++ = hexnum[ (*ptr>>4) ] ; |
| 164 |
*res_ptr++ = hexnum[ (*ptr & 0x0f) ] ; |
| 165 |
} |
| 166 |
|
| 167 |
/* |
| 168 |
* That's it, set the length and return to caller |
| 169 |
*/ |
| 170 |
result->len = ( res_ptr - result->value ) ; |
| 171 |
assert( result->len <= result->max ) ; |
| 172 |
return result ; |
| 173 |
} |
| 174 |
|
| 175 |
|
| 176 |
|
| 177 |
/* ----------------------------------------------------------------- |
| 178 |
* Input is a hex string, which is converted to a char string |
| 179 |
* representing the same information and returned. |
| 180 |
* |
| 181 |
* We have to concider the optional grouping of hex digits by spaces at |
| 182 |
* byte boundaries, and the possibility of having to pad first group |
| 183 |
* with a zero. |
| 184 |
* |
| 185 |
* There is one performance problem with this. If the hex string is |
| 186 |
* normalized and long, we have to loop through the string twice, while |
| 187 |
* once would suffice. To determine whether or not to pad first group |
| 188 |
* with a zero, all of first group must be scanned, which is identical |
| 189 |
* to the whole string if it is normalized. |
| 190 |
*/ |
| 191 |
static streng *pack_hex( tsd_t *TSD, const char *bif, const streng *string ) |
| 192 |
{ |
| 193 |
streng *result=NULL ; /* output char string */ |
| 194 |
const char *ptr=NULL ; /* current digit in input hex string */ |
| 195 |
const char *end_ptr=NULL ; /* ptr to end+1 in input hex string */ |
| 196 |
char *res_ptr=NULL ; /* ptr to current char in output string */ |
| 197 |
int byte_boundary=0 ; /* boolean, are we at at byte bounary? */ |
| 198 |
int count; /* used to count positions */ |
| 199 |
|
| 200 |
/* |
| 201 |
* Allow one extra char for padding, ignore that allocated string |
| 202 |
* might be too long if there is spacing at byte boundaries. |
| 203 |
*/ |
| 204 |
result = Str_makeTSD( (Str_len(string)+1)/2 +1 ) ; |
| 205 |
res_ptr = result->value ; |
| 206 |
|
| 207 |
/* |
| 208 |
* Initiate pointers to current char in intput string, and to |
| 209 |
* end+1 in input string. |
| 210 |
*/ |
| 211 |
ptr = string->value ; |
| 212 |
end_ptr = Str_end(string) ; |
| 213 |
|
| 214 |
/* |
| 215 |
* Explicitly check for space at start or end. Illegal space within |
| 216 |
* the hex string is checked for during the loop. |
| 217 |
*/ |
| 218 |
if ((ptr<end_ptr) && ((rx_isspace(*ptr)) || (rx_isspace(*(end_ptr-1))))) |
| 219 |
{ |
| 220 |
goto invalid; |
| 221 |
} |
| 222 |
|
| 223 |
/* |
| 224 |
* Find the number of hex digits in the first group of hex digits. |
| 225 |
* Let the variable 'byte_boundary' be a boolean, indicating if |
| 226 |
* current char might be a byte boundary. I.e if byte_boundary is |
| 227 |
* set, spaces are legal. |
| 228 |
* |
| 229 |
* Also, set the first byte in the output string. That is not |
| 230 |
* necessary if the first group of hex digits has an even number of |
| 231 |
* digits, but it is cheaper to do it always that check for it. |
| 232 |
*/ |
| 233 |
for (; (ptr<end_ptr) && (rx_isxdigit(*ptr)); ptr++ ) ; |
| 234 |
byte_boundary = !((ptr-string->value)%2) ; |
| 235 |
|
| 236 |
/* Does this statement do anything useful? (Proabably, things crash if |
| 237 |
I remove it ... |
| 238 |
*/ |
| 239 |
*res_ptr = 0x00 ; |
| 240 |
|
| 241 |
/* |
| 242 |
* Loop through the elements of the input string. Skip over spaces. |
| 243 |
* Stuff hex digits into the output string, and report error |
| 244 |
* for any other type of data. |
| 245 |
*/ |
| 246 |
for (count=1,ptr=string->value; ptr<end_ptr; ptr++, count++) |
| 247 |
{ |
| 248 |
if (rx_isspace(*ptr)) |
| 249 |
{ |
| 250 |
/* |
| 251 |
* Just make sure that this space occurs at a byte boundary, |
| 252 |
* except from that, ignore it. |
| 253 |
*/ |
| 254 |
if (!byte_boundary) |
| 255 |
{ |
| 256 |
goto invalid; |
| 257 |
} |
| 258 |
} |
| 259 |
else if (rx_isxdigit(*ptr)) |
| 260 |
{ |
| 261 |
/* |
| 262 |
* Stuff it into the output array, either as upper or lower |
| 263 |
* part of a byte, depending on the value of 'byte_boundary'. |
| 264 |
* Then toggle the value of 'byte_boundary'. |
| 265 |
*/ |
| 266 |
if (byte_boundary) |
| 267 |
*res_ptr = (char)( HEXVAL(*ptr) << 4 ) ; |
| 268 |
else |
| 269 |
{ |
| 270 |
/* Damn'ed MSVC: */ |
| 271 |
*res_ptr = (char) (*res_ptr + (char) (HEXVAL(*ptr))) ; |
| 272 |
res_ptr++; |
| 273 |
} |
| 274 |
|
| 275 |
byte_boundary = !byte_boundary ; |
| 276 |
} |
| 277 |
else |
| 278 |
{ |
| 279 |
goto invalid; |
| 280 |
} |
| 281 |
} |
| 282 |
|
| 283 |
/* |
| 284 |
* Set the length and do 'redundant' check for problems. In |
| 285 |
* particular, check 'byte_boundary' to verify that the last group |
| 286 |
* of hex digits ended at a byte boundary; report error if not. |
| 287 |
*/ |
| 288 |
if (!byte_boundary) |
| 289 |
{ |
| 290 |
goto invalid; |
| 291 |
} |
| 292 |
|
| 293 |
result->len = res_ptr - result->value ; |
| 294 |
assert( result->len <= result->max ) ; |
| 295 |
|
| 296 |
return result ; |
| 297 |
|
| 298 |
invalid: |
| 299 |
Free_stringTSD( result ); |
| 300 |
exiterror( ERR_INCORRECT_CALL, 25, bif, tmpstr_of( TSD, string ) ); |
| 301 |
return NULL; /* not reached */ |
| 302 |
} |
| 303 |
|
| 304 |
|
| 305 |
|
| 306 |
/* ------------------------------------------------------------------ |
| 307 |
* Takes a char string input and concerts it into a whole number of |
| 308 |
* base 10, which can be signed. Actually, the real work is done in |
| 309 |
* str_digitize(). |
| 310 |
* |
| 311 |
* If length is -1 (i.e. unspecified) it will be interpreted as an |
| 312 |
* unsigned integer. |
| 313 |
* |
| 314 |
* If length is specified, the input string will be interpreted as an |
| 315 |
* two's complement number having that length. If parameter 'length' |
| 316 |
* is bigger than the length of parameter string, 'string' is |
| 317 |
* logically extended with '0' at the left. If parameter length is |
| 318 |
* smaller than the length of parameter string, only the rightmost |
| 319 |
* characters of string is significant. |
| 320 |
* |
| 321 |
* The output string will not contain any leading zeros (unless the |
| 322 |
* value of the number is zero, in which case '0' is returned). A '0' |
| 323 |
* will also be returned if 'length' is zero, or if 'string' is the |
| 324 |
* nullstring. |
| 325 |
*/ |
| 326 |
static streng *numerize( tsd_t *TSD, streng *string, int length, |
| 327 |
const char *bif, int removeStringOnError ) |
| 328 |
{ |
| 329 |
int start=0 ; /* character to start reading at */ |
| 330 |
int sign=0 ; /* is this to be interpreted as signed? */ |
| 331 |
|
| 332 |
/* The trivial case, either the nullstring, or length=0 */ |
| 333 |
if ((length==0) || (Str_len(string)==0)) |
| 334 |
return int_to_streng( TSD, 0 ) ; |
| 335 |
|
| 336 |
/* |
| 337 |
* Set the variable 'start' to the most significant byte in 'string'. |
| 338 |
* That is the first byte if 'length' is either unspecified or |
| 339 |
* bigger than (or equal to) the length of 'string'. |
| 340 |
* |
| 341 |
* If 'length' is specified and is less than the length of 'string', |
| 342 |
* then set 'start' to the the 'length'th byte, counted backward. |
| 343 |
*/ |
| 344 |
if ((length==-1) || (length>Str_len(string))) |
| 345 |
start = sign = 0 ; |
| 346 |
else |
| 347 |
{ |
| 348 |
assert((length>0) && (length<=Str_len(string))) ; |
| 349 |
start = Str_len(string) - length ; |
| 350 |
sign = 1 ; |
| 351 |
assert((start>=0) && (start<Str_len(string))) ; |
| 352 |
} |
| 353 |
|
| 354 |
/* |
| 355 |
* Call the correct routine in the string module. The number will |
| 356 |
* always be signed if length is specified. |
| 357 |
*/ |
| 358 |
return str_digitize( TSD, string, start, sign, bif, removeStringOnError ) ; |
| 359 |
} |
| 360 |
|
| 361 |
|
| 362 |
|
| 363 |
/* ------------------------------------------------------------------ |
| 364 |
* Converts a hex string to a decimal number using two's complement. |
| 365 |
* This is a high level routine, which just calls the above low level |
| 366 |
* routines to do the job. First the hex string is converted into |
| 367 |
* a char string, and then the char string is converted into a whole |
| 368 |
* number. |
| 369 |
*/ |
| 370 |
streng *std_x2d( tsd_t *TSD, cparamboxptr parms ) |
| 371 |
{ |
| 372 |
int length=0 ; /* the length of the input hex string */ |
| 373 |
streng *result=NULL ; /* the output string */ |
| 374 |
streng *packed=NULL ; /* tmp variable holding the char string */ |
| 375 |
|
| 376 |
/* |
| 377 |
* First read the parameters, and set length to -1 (meaning |
| 378 |
* unspecified) if the second parameter was not specified. |
| 379 |
*/ |
| 380 |
checkparam( parms, 1, 2 , "X2D" ) ; |
| 381 |
if ((parms->next)&&(parms->next->value)) |
| 382 |
length = atozpos( TSD, parms->next->value, "X2D", 2 ) ; |
| 383 |
else |
| 384 |
length = (-1) ; |
| 385 |
|
| 386 |
/* |
| 387 |
* Convert the hex string into a whole number in two steps |
| 388 |
* |
| 389 |
* Note that the 'length' variable is the length in hex digits, and |
| 390 |
* that numerize receives its input as a char string, so we have to |
| 391 |
* convert the 'length' a bit. Also, that means that we have to |
| 392 |
* sign extend the number at the left to a byte boundary. |
| 393 |
*/ |
| 394 |
packed = pack_hex( TSD, "X2D", parms->value ) ; |
| 395 |
if ((length>0) && (length%2)) |
| 396 |
{ |
| 397 |
/* |
| 398 |
* The char string was padded with an extra zero nibble in pack_hex() |
| 399 |
* so we must signextend that nibble. 'msb' is a tmp variable |
| 400 |
* that points to the most significant byte in packed. Hmmm this |
| 401 |
* is a kludge ... |
| 402 |
*/ |
| 403 |
int msb = Str_len(packed)-(length/2)-1 ; |
| 404 |
if (msb >= 0) /* only if length <= hexchars supplied */ |
| 405 |
{ |
| 406 |
if (packed->value[msb] & 0x08) |
| 407 |
packed->value[msb] |= 0xf0 ; |
| 408 |
else |
| 409 |
packed->value[msb] &= 0x0f ; |
| 410 |
} |
| 411 |
} |
| 412 |
result = numerize( TSD, |
| 413 |
packed, |
| 414 |
((length!=-1) ? ((length+1)/2) : -1), |
| 415 |
"X2D", |
| 416 |
1 ) ; |
| 417 |
|
| 418 |
/* |
| 419 |
* Clean up and return to caller |
| 420 |
*/ |
| 421 |
Free_stringTSD( packed ) ; |
| 422 |
|
| 423 |
return result ; |
| 424 |
} |
| 425 |
|
| 426 |
|
| 427 |
|
| 428 |
/* ------------------------------------------------------------------ |
| 429 |
* Converts a char string to a hex string. This a just a box around |
| 430 |
* one of the low level routines, that processes the parameters. |
| 431 |
*/ |
| 432 |
streng *std_x2c( tsd_t *TSD, cparamboxptr parms ) |
| 433 |
{ |
| 434 |
checkparam( parms, 1, 1 , "X2C" ) ; |
| 435 |
return pack_hex( TSD, "X2C", parms->value ) ; |
| 436 |
} |
| 437 |
|
| 438 |
|
| 439 |
|
| 440 |
/* ------------------------------------------------------------------ |
| 441 |
* Builtin function that converts a binary string to a normalized |
| 442 |
* hexstring. The hexstring will be padded to *nibble* boundary, |
| 443 |
* (note: not byte boundary). |
| 444 |
* |
| 445 |
* This function should have been implemented as a convertion from |
| 446 |
* bin string to packed binary string, and then a call to unpack |
| 447 |
* the binary string to a hex string. Unfortunately, accuracy would |
| 448 |
* be lost, since packed binary strings are padded to byte boundary. |
| 449 |
* |
| 450 |
* Therefore, this function converts directly into a hex string. If |
| 451 |
* more functions are added to Rexx (like b2d() and b2c()) in the |
| 452 |
* future, it might be more effective to create a nibble format (or |
| 453 |
* change the packed binary string to nibble array), in order to |
| 454 |
* decrease the number of lines of code. |
| 455 |
* |
| 456 |
* The code for this is mostly taken from lexsrc.l, these two pieces |
| 457 |
* should perhaps be tuned to use the same function. |
| 458 |
*/ |
| 459 |
|
| 460 |
streng *std_b2x( tsd_t *TSD, cparamboxptr parms ) |
| 461 |
{ |
| 462 |
char *ptr=NULL ; /* loop variable */ |
| 463 |
char *endptr=NULL ; /* pointer to end+1 of input string */ |
| 464 |
char *res_ptr=NULL ; /* pointer to result string */ |
| 465 |
streng *result=NULL ; /* result string */ |
| 466 |
streng *string=NULL ; /* tmp variable to avoid pointer chasing */ |
| 467 |
int first_group=0 ; /* number of bin digits in first bin group */ |
| 468 |
int cur_bit=0 ; /* current bit in input string */ |
| 469 |
int nibble=0 ; /* collects bin digits to nibbles */ |
| 470 |
|
| 471 |
/* |
| 472 |
* Have we been called correctly? |
| 473 |
*/ |
| 474 |
checkparam( parms, 1, 1 , "B2X" ) ; |
| 475 |
string = parms->value ; |
| 476 |
|
| 477 |
/* |
| 478 |
* Since a bin string can have any number of digits in its first |
| 479 |
* group, we have to find the number of digits, in order to be able |
| 480 |
* to pad that group with leading zeros. The number can also be |
| 481 |
* zero (i.e. the bin string is empty.) The number first_group |
| 482 |
* contains the number of binary digits in the first group. |
| 483 |
*/ |
| 484 |
endptr = Str_end(string) ; |
| 485 |
ptr = string->value ; |
| 486 |
for (; (ptr<endptr) && (((*ptr)=='0') || ((*ptr)=='1')); ptr++ ) ; |
| 487 |
first_group = ptr - string->value ; |
| 488 |
|
| 489 |
/* |
| 490 |
* If the first group contained zero binary digits, then either does |
| 491 |
* it contain leading space, or it is the nullstring. The former is |
| 492 |
* an error, so report it if that is the case. |
| 493 |
*/ |
| 494 |
if (Str_len(string) && ((first_group==0) || (rx_isspace(*(endptr-1))))) |
| 495 |
{ |
| 496 |
/* fixes 1107969 */ |
| 497 |
exiterror( ERR_INCORRECT_CALL, 24, "B2X", tmpstr_of( TSD, string ) ); |
| 498 |
} |
| 499 |
|
| 500 |
/* |
| 501 |
* If the string is a proper bin string, we need one hex digit for |
| 502 |
* each fourth bin digit (after having taken into account that the |
| 503 |
* first group might have a maximum of three implied zeros.) If it |
| 504 |
* is not a proper bin string, the error will be caught later. |
| 505 |
* |
| 506 |
* Actually, this might be more than we need, since there might be |
| 507 |
* spaces embedded within the bin string, but don't bother about |
| 508 |
* that since it just takes CPU time to figure out exactly how many |
| 509 |
* bytes we need. Just allocate enough. |
| 510 |
*/ |
| 511 |
result = Str_makeTSD( (Str_len(string)+3)/4 ) ; |
| 512 |
res_ptr = result->value ; |
| 513 |
|
| 514 |
/* |
| 515 |
* Initialize some variables, 'cur_bit' is the current bit within |
| 516 |
* the current nibble. It must be initialized to the number of |
| 517 |
* implied leading zeros in the bin string. The following |
| 518 |
* transformation from 'frist_group%4' to 'cur_bit' has the following |
| 519 |
* mapping {0,1,2,3}->{0,3,2,1}. |
| 520 |
*/ |
| 521 |
cur_bit = (4 - first_group%4)%4 ; |
| 522 |
nibble = 0 ; |
| 523 |
|
| 524 |
/* |
| 525 |
* Then, loop for each character in the input string, and perform |
| 526 |
* some action, based on whether it is a space or a binary digit. |
| 527 |
* If it is neither, report an error. |
| 528 |
*/ |
| 529 |
for (ptr=string->value; ptr<endptr; ptr++) |
| 530 |
{ |
| 531 |
if (rx_isspace(*ptr)) |
| 532 |
{ |
| 533 |
/* |
| 534 |
* The variable 'cur_bit' is a number containing the relative |
| 535 |
* position of the current bit within the current group of |
| 536 |
* binary digits. After reading a complete nibble, it reaches |
| 537 |
* 4, and is reset to 0. So if this is a nibble boundary, it |
| 538 |
* better be 0, or else there is space within a nibble. |
| 539 |
*/ |
| 540 |
if (cur_bit!=0) |
| 541 |
{ |
| 542 |
Free_stringTSD( result ); |
| 543 |
exiterror( ERR_INCORRECT_CALL, 24, "B2X", tmpstr_of( TSD, string ) ); |
| 544 |
} |
| 545 |
} |
| 546 |
|
| 547 |
else if (((*ptr)=='0')||((*ptr)=='1')) |
| 548 |
{ |
| 549 |
/* |
| 550 |
* If it is a binary digit, shift 'nibble' and add the digit. |
| 551 |
* If 'cur_bit' (after being incremented) shows 4 (i.e we have |
| 552 |
* completed processing a nibble), reset it to 0, and and |
| 553 |
* flush 'nibble' to the result string and reset 'nibble' too. |
| 554 |
*/ |
| 555 |
nibble = nibble * 2 + ((*ptr)-'0') ; |
| 556 |
if ((++cur_bit)==4) |
| 557 |
{ |
| 558 |
*(res_ptr++) = hexnum[nibble] ; |
| 559 |
nibble = 0 ; |
| 560 |
cur_bit = 0 ; |
| 561 |
} |
| 562 |
} |
| 563 |
else |
| 564 |
{ |
| 565 |
exiterror( ERR_INCORRECT_CALL, 24, "B2X", tmpstr_of( TSD, string ) ); |
| 566 |
} |
| 567 |
} |
| 568 |
if (cur_bit!=0) |
| 569 |
{ |
| 570 |
Free_stringTSD( result ); |
| 571 |
exiterror( ERR_INCORRECT_CALL, 24, "B2X", tmpstr_of( TSD, string ) ); |
| 572 |
} |
| 573 |
/* |
| 574 |
* Wow, we're finished, we just have to set the length of 'result' |
| 575 |
*/ |
| 576 |
result->len = (res_ptr - result->value ) ; |
| 577 |
assert( result->len <= result->max ) ; |
| 578 |
return( result ) ; |
| 579 |
} |
| 580 |
|
| 581 |
|
| 582 |
|
| 583 |
|
| 584 |
/* ------------------------------------------------------------------ |
| 585 |
* Function that converts a hex string to a binary string. The hex |
| 586 |
* string may have spaces at byte boundaries as usual, but the first |
| 587 |
* group of hex digits may have an odd number of digits, in which |
| 588 |
* case it is *not* padded with an zero hex digit. |
| 589 |
* |
| 590 |
* Just like std_b2x(), this function should really use a common data |
| 591 |
* format, but it does not matter before more rexx gets more functions |
| 592 |
* that converts to/from bin strings. |
| 593 |
*/ |
| 594 |
|
| 595 |
streng *std_x2b( tsd_t *TSD, cparamboxptr parms ) |
| 596 |
{ |
| 597 |
int space_stat=0 ; /* state machine: nibble or byte boundary? */ |
| 598 |
char *ptr=NULL ; /* loop control variable */ |
| 599 |
char *end_ptr=NULL ; /* points to end+1 of input, endcondition in loop */ |
| 600 |
streng *result=NULL ; /* the output streng */ |
| 601 |
char *res_ptr=NULL ; /* ptr to contents of 'result' */ |
| 602 |
int nibble=0 ; /* holds a nibble while extracting bin digits */ |
| 603 |
int count=0 ; /* loop control variable */ |
| 604 |
int pos=0 ; /* position in string for error reporting */ |
| 605 |
|
| 606 |
/* |
| 607 |
* Check that we got the parameters that we needed. Then initialize |
| 608 |
* some of the variables. |
| 609 |
*/ |
| 610 |
checkparam( parms, 1, 1 ,"X2B" ) ; |
| 611 |
ptr = parms->value->value ; |
| 612 |
end_ptr = Str_end( parms->value ) ; |
| 613 |
space_stat = 0 ; |
| 614 |
|
| 615 |
/* |
| 616 |
* Let us allocate enough space, we could tune some space here |
| 617 |
* if we actually checked how many hex digits that the input |
| 618 |
* string contiained. |
| 619 |
*/ |
| 620 |
result = Str_makeTSD( (end_ptr-ptr) * 4 ) ; |
| 621 |
res_ptr = result->value ; |
| 622 |
|
| 623 |
/* |
| 624 |
* Check for leading or trailing space in the hex string. |
| 625 |
*/ |
| 626 |
if (end_ptr>ptr) |
| 627 |
{ |
| 628 |
if (rx_isspace(*ptr) || rx_isspace(*(end_ptr-1))) |
| 629 |
{ |
| 630 |
goto invalid; |
| 631 |
} |
| 632 |
} |
| 633 |
|
| 634 |
/* |
| 635 |
* The main loop. For each hex digit, output four bin digits to |
| 636 |
* the output string, and check for illegal spaces within bytes. |
| 637 |
* If anything other than spaces or hex digits are found, report |
| 638 |
* an error. |
| 639 |
*/ |
| 640 |
for (pos=1; ptr<end_ptr; ptr++,pos++) |
| 641 |
{ |
| 642 |
if (rx_isspace(*ptr)) |
| 643 |
{ |
| 644 |
/* |
| 645 |
* We have found space in the hex string, eat it up, and keep |
| 646 |
* the statemachine 'space_state' going. If state is 0 (end |
| 647 |
* of first group) go to state 2 (at byte boundary). If state |
| 648 |
* is 1 (inside a byte) report an error, since space may not |
| 649 |
* occur there. |
| 650 |
*/ |
| 651 |
if (space_stat==0) |
| 652 |
{ |
| 653 |
space_stat = 2 ; |
| 654 |
} |
| 655 |
else if (space_stat==1) |
| 656 |
{ |
| 657 |
goto invalid; |
| 658 |
} |
| 659 |
} |
| 660 |
else if (rx_isxdigit(*ptr)) |
| 661 |
{ |
| 662 |
/* |
| 663 |
* We have found a hex digit, chop it into four parts, and |
| 664 |
* stuff them into 'result'. Requires that the character set |
| 665 |
* value of '1' is one higher than that of '0'. |
| 666 |
*/ |
| 667 |
nibble = HEXVAL( *ptr ) ; |
| 668 |
for (count=0; count<4; count++) |
| 669 |
{ |
| 670 |
*(res_ptr++) = (char)(( (nibble & 0x08) != 0 ) + '0') ; |
| 671 |
nibble <<= 1 ; |
| 672 |
} |
| 673 |
|
| 674 |
/* |
| 675 |
* Remember to toggle the statemachine between states 1 and 2 |
| 676 |
* so we can keep track of byte and nibble boundaries. If in |
| 677 |
* state 0 (whitin first group), stay there. |
| 678 |
*/ |
| 679 |
if (space_stat) |
| 680 |
space_stat = ((space_stat==1) ? 2 : 1) ; |
| 681 |
} |
| 682 |
else |
| 683 |
{ |
| 684 |
goto invalid; |
| 685 |
} |
| 686 |
} |
| 687 |
|
| 688 |
/* |
| 689 |
* Set the length, and get out of here. |
| 690 |
*/ |
| 691 |
result->len = res_ptr - result->value ; |
| 692 |
return result ; |
| 693 |
|
| 694 |
invalid: |
| 695 |
Free_stringTSD( result ); |
| 696 |
exiterror( ERR_INCORRECT_CALL, 25, "X2B", tmpstr_of( TSD, parms->value ) ); |
| 697 |
return NULL; /* not reached */ |
| 698 |
} |
| 699 |
|
| 700 |
|
| 701 |
|
| 702 |
|
| 703 |
/* -------------------------------------------------------------------- |
| 704 |
* Converts a char string to a decimal string. Really just a box around |
| 705 |
* numerize, that only preprocesses the parameters. |
| 706 |
*/ |
| 707 |
streng *std_c2d( tsd_t *TSD, cparamboxptr parms ) |
| 708 |
{ |
| 709 |
int length ; /* The length of the input char string */ |
| 710 |
|
| 711 |
checkparam( parms, 1, 2 , "C2D" ) ; |
| 712 |
if ((parms->next)&&(parms->next->value)) |
| 713 |
length = atozpos( TSD, parms->next->value, "C2D", 2 ) ; |
| 714 |
else |
| 715 |
length = -1 ; |
| 716 |
|
| 717 |
return numerize( TSD, parms->value, length, "C2D", 0 ) ; |
| 718 |
} |
| 719 |
|
| 720 |
|
| 721 |
|
| 722 |
|
| 723 |
/* --------------------------------------------------------------------- |
| 724 |
* Converts a packed binary string to a hexadecimal string |
| 725 |
*/ |
| 726 |
streng *std_c2x( tsd_t *TSD, cparamboxptr parms ) |
| 727 |
{ |
| 728 |
checkparam( parms, 1, 1 , "C2X" ) ; |
| 729 |
return unpack_hex( TSD, parms->value ) ; |
| 730 |
} |
| 731 |
|
| 732 |
static void check_wholenum( tsd_t *TSD, const char *bif, const streng *arg, |
| 733 |
num_descr **num ) |
| 734 |
{ |
| 735 |
if ( !myiswnumber( TSD, arg, num, |
| 736 |
!get_options_flag( TSD->currlevel, EXT_STRICT_ANSI ) ) ) |
| 737 |
exiterror( ERR_INCORRECT_CALL, 12, bif, 1, tmpstr_of( TSD, arg ) ); |
| 738 |
} |
| 739 |
|
| 740 |
/* --------------------------------------------------------------------- |
| 741 |
* Converts a whole number into char string. This is just a wrapper |
| 742 |
* around str_binerize(), which preprocesses the parameters. |
| 743 |
*/ |
| 744 |
streng *std_d2c( tsd_t *TSD, cparamboxptr parms ) |
| 745 |
{ |
| 746 |
int length; /* the length of the output string */ |
| 747 |
num_descr *num; |
| 748 |
|
| 749 |
checkparam( parms, 1, 2 , "D2C" ); |
| 750 |
|
| 751 |
check_wholenum( TSD, "D2C", parms->value, &num ); |
| 752 |
if ( parms->next && parms->next->value ) |
| 753 |
length = atozpos( TSD, parms->next->value, "D2C", 2 ); |
| 754 |
else |
| 755 |
{ |
| 756 |
/* |
| 757 |
* The strange syntax forces a check for non-negative first arg if the |
| 758 |
* second doesn't exist. |
| 759 |
*/ |
| 760 |
if ( num->negative ) |
| 761 |
exiterror( ERR_INCORRECT_CALL, 13, "D2C", 1, tmpstr_of( TSD, parms->value ) ); |
| 762 |
|
| 763 |
length = -1; |
| 764 |
} |
| 765 |
|
| 766 |
return str_binerize( TSD, num, length, ERR_INCORRECT_CALL, 12, "D2C", 1 ); |
| 767 |
} |
| 768 |
|
| 769 |
|
| 770 |
/* ------------------------------------------------------------------ |
| 771 |
* Converts a decimal string into a hexadecimal string, using char |
| 772 |
* string as an intermediate format. Due to the use of intermediate |
| 773 |
* format, an extra hex digit might slip into the left end of the |
| 774 |
* answer, and code is added to remove it. |
| 775 |
*/ |
| 776 |
streng *std_d2x( tsd_t *TSD, cparamboxptr parms ) |
| 777 |
{ |
| 778 |
int length; /* holds the requested langth of the result */ |
| 779 |
streng *result; /* the output streng */ |
| 780 |
streng *packed; /* tmp variable, holds the packed string */ |
| 781 |
num_descr *num; |
| 782 |
|
| 783 |
/* |
| 784 |
* Check the parameters, and set 'length' to the specified length, or |
| 785 |
* to -1 if the second parameter was not specified. |
| 786 |
*/ |
| 787 |
checkparam( parms, 1, 2 , "D2X" ); |
| 788 |
|
| 789 |
check_wholenum( TSD, "D2X", parms->value, &num ); |
| 790 |
if ( parms->next && parms->next->value ) |
| 791 |
length = atozpos( TSD, parms->next->value, "D2X", 2 ); |
| 792 |
else |
| 793 |
{ |
| 794 |
/* |
| 795 |
* The strange syntax forces a check for non-negative first arg if the |
| 796 |
* second doesn't exist. |
| 797 |
*/ |
| 798 |
if ( num->negative ) |
| 799 |
exiterror( ERR_INCORRECT_CALL, 13, "D2X", 1, tmpstr_of( TSD, parms->value ) ); |
| 800 |
|
| 801 |
length = -1; |
| 802 |
} |
| 803 |
|
| 804 |
/* |
| 805 |
* Convert the whole number into a hex string in a two step operation. |
| 806 |
* First it is converted into a char string, and then that char string |
| 807 |
* is converted into a hexstring. |
| 808 |
*/ |
| 809 |
packed = str_binerize( TSD, num, ( length == -1 ) ? -1 : ( length+1 ) / 2, ERR_INCORRECT_CALL, 12, "D2X", 1 ); |
| 810 |
result = unpack_hex( TSD, packed ); |
| 811 |
Free_stringTSD( packed ); |
| 812 |
|
| 813 |
/* |
| 814 |
* Since we used char string as a temporary format, the hex string |
| 815 |
* will now be padded with one extra zero at the left. If we specified |
| 816 |
* length, and that length does not match the actual length, we must |
| 817 |
* strip away the first zero in 'result'. |
| 818 |
* |
| 819 |
* Here we check for length>0, since we want to catch it if length was |
| 820 |
* specified, but not if it was 0. In the latter case, the string will |
| 821 |
* be the nullstring, and we don't need to do anything anyway. |
| 822 |
*/ |
| 823 |
if ( ( length > 0 ) && ( Str_len( result ) != length ) && Str_len( result ) ) |
| 824 |
{ |
| 825 |
assert( Str_len( result ) == length + 1 ); |
| 826 |
memmove( result->value, &result->value[1], --result->len ); |
| 827 |
} |
| 828 |
|
| 829 |
/* |
| 830 |
* Just to be safe, check that we did get the nullstring if length |
| 831 |
* was specified to 0 |
| 832 |
*/ |
| 833 |
assert( ( length != 0 ) || ( Str_len(result) == 0 ) ); |
| 834 |
|
| 835 |
/* |
| 836 |
* If length was not specified, there might be leading zeros in the |
| 837 |
* answer that we might want to get rid of. However, the algoritm |
| 838 |
* that is used in str_binerize() should ensure that there will not |
| 839 |
* be more than one leading zero in this case. This part of the code |
| 840 |
* could easily be merged with the other call to memmove() above, |
| 841 |
* but have been placed here to improve readability. |
| 842 |
* |
| 843 |
* If there are more than one leading zero, this will not work. The |
| 844 |
* situation where the result only consists of one zero, should not |
| 845 |
* occur, but it will be handled, since the result should then be |
| 846 |
* the nullstring. |
| 847 |
*/ |
| 848 |
if ( ( length == -1 ) && ( result->value[0] == '0' ) ) |
| 849 |
{ |
| 850 |
assert( Str_len( result ) > 1 ); |
| 851 |
assert( ( result->value[0] == '0') && ( ( result->value[1] != '0' ) || ( Str_len( result ) == 2 ) ) ); |
| 852 |
|
| 853 |
memmove( result->value, &result->value[1], --result->len ); |
| 854 |
} |
| 855 |
|
| 856 |
/* |
| 857 |
* That's it, now we just have to get out of here |
| 858 |
*/ |
| 859 |
return result; |
| 860 |
} |