quote.c

   1 /* quote.c      library routines for the Netwide Assembler
   2  *
   3  * The Netwide Assembler is copyright (C) 1996 Simon Tatham and
   4  * Julian Hall. All rights reserved. The software is
   5  * redistributable under the license given in the file "LICENSE"
   6  * distributed in the NASM archive.
   7  */
   8
   9 #include "compiler.h"
  10
  11 #include <assert.h>
  12 #include <stdlib.h>
  13
  14 #include "nasmlib.h"
  15 #include "quote.h"
  16
  17 #define numvalue(c)  ((c)>='a' ? (c)-'a'+10 : (c)>='A' ? (c)-'A'+10 : (c)-'0')
  18
  19 char *nasm_quote(char *str, size_t len)
  20 {
  21     char c, c1, *p, *q, *nstr, *ep;
  22     bool sq_ok, dq_ok;
  23     size_t qlen;
  24
  25     sq_ok = dq_ok = true;
  26     ep = str+len;
  27     qlen = 0;                   /* Length if we need `...` quotes */
  28     for (p = str; p < ep; p++) {
  29         c = *p;
  30         switch (c) {
  31         case '\'':
  32             sq_ok = false;
  33             qlen++;
  34             break;
  35         case '\"':
  36             dq_ok = false;
  37             qlen++;
  38             break;
  39         case '`':
  40         case '\\':
  41             qlen += 2;
  42             break;
  43         default:
  44             if (c < ' ' || c > '~') {
  45                 sq_ok = dq_ok = false;
  46                 switch (c) {
  47                 case '\a':
  48                 case '\b':
  49                 case '\t':
  50                 case '\n':
  51                 case '\v':
  52                 case '\f':
  53                 case '\r':
  54                 case 27:
  55                     qlen += 2;
  56                     break;
  57                 default:
  58                     c1 = (p+1 < ep) ? p[1] : 0;
  59                     if (c > 077 || (c1 >= '0' && c1 <= '7'))
  60                         qlen += 4; /* Must use the full form */
  61                     else if (c > 07)
  62                         qlen += 3;
  63                     else
  64                         qlen += 2;
  65                     break;
  66                 }
  67             } else {
  68                 qlen++;
  69             }
  70             break;
  71         }
  72     }
  73
  74     if (sq_ok || dq_ok) {
  75         /* Use '...' or "..." */
  76         nstr = nasm_malloc(len+3);
  77         nstr[0] = nstr[len+1] = sq_ok ? '\'' : '\"';
  78         nstr[len+2] = '\0';
  79         memcpy(nstr+1, str, len);
  80     } else {
  81         /* Need to use `...` quoted syntax */
  82         nstr = nasm_malloc(qlen+3);
  83         q = nstr;
  84         *q++ = '`';
  85         for (p = str; p < ep; p++) {
  86             c = *p;
  87             switch (c) {
  88             case '`':
  89             case '\\':
  90                 *q++ = '\\';
  91                 *q++ = c;
  92                 break;
  93             case 7:
  94                 *q++ = '\\';
  95                 *q++ = 'a';
  96                 break;
  97             case 8:
  98                 *q++ = '\\';
  99                 *q++ = 'b';
 100                 break;
 101             case 9:
 102                 *q++ = '\\';
 103                 *q++ = 't';
 104                 break;
 105             case 10:
 106                 *q++ = '\\';
 107                 *q++ = 'n';
 108                 break;
 109             case 11:
 110                 *q++ = '\\';
 111                 *q++ = 'v';
 112                 break;
 113             case 12:
 114                 *q++ = '\\';
 115                 *q++ = 'f';
 116                 break;
 117             case 13:
 118                 *q++ = '\\';
 119                 *q++ = 'r';
 120                 break;
 121             case 27:
 122                 *q++ = '\\';
 123                 *q++ = 'e';
 124                 break;
 125             default:
 126                 if (c < ' ' || c > '~') {
 127                     c1 = (p+1 < ep) ? p[1] : 0;
 128                     if (c1 >= '0' && c1 <= '7')
 129                         q += sprintf(q, "\\%03o", (unsigned char)c);
 130                     else
 131                         q += sprintf(q, "\\%o", (unsigned char)c);
 132                 } else {
 133                     *q++ = c;
 134                 }
 135                 break;
 136             }
 137         }
 138         *q++ = '`';
 139         *q++ = '\0';
 140         assert((size_t)(q-nstr) == qlen+3);
 141     }
 142     return nstr;
 143 }
 144
 145 static char *emit_utf8(char *q, int32_t v)
 146 {
 147     if (v < 0) {
 148         /* Impossible - do nothing */
 149     } else if (v <= 0x7f) {
 150         *q++ = v;
 151     } else if (v <= 0x000007ff) {
 152         *q++ = 0xc0 | (v >> 6);
 153         *q++ = 0x80 | (v & 63);
 154     } else if (v <= 0x0000ffff) {
 155         *q++ = 0xe0 | (v >> 12);
 156         *q++ = 0x80 | ((v >> 6) & 63);
 157         *q++ = 0x80 | (v & 63);
 158     } else if (v <= 0x001fffff) {
 159         *q++ = 0xf0 | (v >> 18);
 160         *q++ = 0x80 | ((v >> 12) & 63);
 161         *q++ = 0x80 | ((v >> 6) & 63);
 162         *q++ = 0x80 | (v & 63);
 163     } else if (v <= 0x03ffffff) {
 164         *q++ = 0xf8 | (v >> 24);
 165         *q++ = 0x80 | ((v >> 18) & 63);
 166         *q++ = 0x80 | ((v >> 12) & 63);
 167         *q++ = 0x80 | ((v >> 6) & 63);
 168         *q++ = 0x80 | (v & 63);
 169     } else {
 170         *q++ = 0xfc | (v >> 30);
 171         *q++ = 0x80 | ((v >> 24) & 63);
 172         *q++ = 0x80 | ((v >> 18) & 63);
 173         *q++ = 0x80 | ((v >> 12) & 63);
 174         *q++ = 0x80 | ((v >> 6) & 63);
 175         *q++ = 0x80 | (v & 63);
 176     }
 177     return q;
 178 }
 179
 180 /*
 181  * Do an *in-place* dequoting of the specified string, returning the
 182  * resulting length (which may be containing embedded nulls.)
 183  *
 184  * In-place replacement is possible since the unquoted length is always
 185  * shorter than or equal to the quoted length.
 186  */
 187 size_t nasm_unquote(char *str)
 188 {
 189     size_t ln;
 190     char bq, eq;
 191     char *p, *q, *ep;
 192     char *escp = NULL;
 193     char c;
 194     enum unq_state {
 195         st_start,
 196         st_backslash,
 197         st_hex,
 198         st_oct,
 199         st_ucs,
 200     } state;
 201     int ndig = 0;
 202     int32_t nval = 0;
 203
 204     bq = str[0];
 205     if (!bq)
 206         return 0;
 207     ln = strlen(str);
 208     eq = str[ln-1];
 209
 210     if ((bq == '\'' || bq == '\"') && bq == eq) {
 211         /* '...' or "..." string */
 212         memmove(str, str+1, ln-2);
 213         str[ln-2] = '\0';
 214         return ln-2;
 215     }
 216     if (bq == '`' || eq == '`') {
 217         /* `...` string */
 218         q = str;
 219         p = str+1;
 220         ep = str+ln-1;
 221         state = st_start;
 222
 223         while (p < ep) {
 224             c = *p++;
 225             switch (state) {
 226             case st_start:
 227                 if (c == '\\')
 228                     state = st_backslash;
 229                 else
 230                     *q++ = c;
 231                 break;
 232
 233             case st_backslash:
 234                 state = st_start;
 235                 escp = p;       /* Beginning of argument sequence */
 236                 nval = 0;
 237                 switch (c) {
 238                 case 'a':
 239                     *q++ = 7;
 240                     break;
 241                 case 'b':
 242                     *q++ = 8;
 243                     break;
 244                 case 'e':
 245                     *q++ = 27;
 246                     break;
 247                 case 'f':
 248                     *q++ = 12;
 249                     break;
 250                 case 'n':
 251                     *q++ = 10;
 252                     break;
 253                 case 'r':
 254                     *q++ = 13;
 255                     break;
 256                 case 't':
 257                     *q++ = 9;
 258                     break;
 259                 case 'u':
 260                     state = st_ucs;
 261                     ndig = 4;
 262                     break;
 263                 case 'U':
 264                     state = st_ucs;
 265                     ndig = 8;
 266                     break;
 267                 case 'v':
 268                     *q++ = 11;
 269                     break;
 270                 case 'x':
 271                 case 'X':
 272                     state = st_hex;
 273                     ndig = 2;
 274                     break;
 275                 case '0':
 276                 case '1':
 277                 case '2':
 278                 case '3':
 279                 case '4':
 280                 case '5':
 281                 case '6':
 282                 case '7':
 283                     state = st_oct;
 284                     ndig = 2;   /* Up to two more digits */
 285                     nval = c - '0';
 286                     break;
 287                 default:
 288                     *q++ = c;
 289                     break;
 290                 }
 291                 break;
 292
 293             case st_oct:
 294                 if (c >= '0' && c <= '7') {
 295                     nval = (nval << 3) + (c - '0');
 296                     if (!--ndig) {
 297                         *q++ = nval;
 298                         state = st_start;
 299                     }
 300                 } else {
 301                     p--;        /* Process this character again */
 302                     *q++ = nval;
 303                     state = st_start;
 304                 }
 305                 break;
 306
 307             case st_hex:
 308                 if ((c >= '0' && c <= '9') ||
 309                     (c >= 'A' && c <= 'F') ||
 310                     (c >= 'a' && c <= 'f')) {
 311                     nval = (nval << 4) + numvalue(c);
 312                     if (--ndig) {
 313                         *q++ = nval;
 314                         state = st_start;
 315                     }
 316                 } else {
 317                     p--;        /* Process this character again */
 318                     *q++ = (p > escp) ? nval : escp[-1];
 319                     state = st_start;
 320                 }
 321                 break;
 322
 323             case st_ucs:
 324                 if ((c >= '0' && c <= '9') ||
 325                     (c >= 'A' && c <= 'F') ||
 326                     (c >= 'a' && c <= 'f')) {
 327                     nval = (nval << 4) + numvalue(c);
 328                     if (!--ndig) {
 329                         q = emit_utf8(q, nval);
 330                         state = st_start;
 331                     }
 332                 } else {
 333                     p--;        /* Process this character again */
 334                     if (p > escp)
 335                         q = emit_utf8(q, nval);
 336                     else
 337                         *q++ = escp[-1];
 338                     state = st_start;
 339                 }
 340                 break;
 341             }
 342         }
 343         switch (state) {
 344         case st_start:
 345         case st_backslash:
 346             break;
 347         case st_oct:
 348             *q++ = nval;
 349             break;
 350         case st_hex:
 351             *q++ = (p > escp) ? nval : escp[-1];
 352             break;
 353         case st_ucs:
 354             if (p > escp)
 355                 q = emit_utf8(q, nval);
 356             else
 357                 *q++ = escp[-1];
 358             break;
 359         }
 360         *q = '\0';
 361         return q-str;
 362     }
 363
 364     /* Otherwise, just return the input... */
 365     return ln;
 366 }
 367
 368 /*
 369  * Find the end of a quoted string; returns the pointer to the terminating
 370  * character (either the ending quote or the null character, if unterminated.)
 371  */
 372 char *nasm_skip_string(char *str)
 373 {
 374     char bq;
 375     char *p;
 376     char c;
 377     enum unq_state {
 378         st_start,
 379         st_backslash,
 380     } state;
 381
 382     bq = str[0];
 383     if (bq == '\'' || bq == '\"') {
 384         /* '...' or "..." string */
 385         for (p = str+1; *p && *p != bq; p++)
 386             ;
 387         return p;
 388     } else if (bq == '`') {
 389         /* `...` string */
 390         p = str+1;
 391         state = st_start;
 392
 393         while ((c = *p++)) {
 394             switch (state) {
 395             case st_start:
 396                 switch (c) {
 397                 case '\\':
 398                     state = st_backslash;
 399                     break;
 400                 case '`':
 401                     return p-1; /* Found the end */
 402                 default:
 403                     break;
 404                 }
 405                 break;
 406
 407             case st_backslash:
 408                 /*
 409                  * Note: for the purpose of finding the end of the string,
 410                  * all successor states to st_backslash are functionally
 411                  * equivalent to st_start, since either a backslash or
 412                  * a backquote will force a return to the st_start state.
 413                  */
 414                 state = st_start;
 415                 break;
 416             }
 417         }
 418         return p;               /* Unterminated string... */
 419     } else {
 420         return str;             /* Not a string... */
 421     }
 422 }