lib/libedit/tokenizer.c

   1 /*-
   2  * Copyright (c) 1992, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  *
   5  * This code is derived from software contributed to Berkeley by
   6  * Christos Zoulas of Cornell University.
   7  *
   8  * Redistribution and use in source and binary forms, with or without
   9  * modification, are permitted provided that the following conditions
  10  * are met:
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in the
  15  *    documentation and/or other materials provided with the distribution.
  16  * 3. All advertising materials mentioning features or use of this software
  17  *    must display the following acknowledgement:
  18  *      This product includes software developed by the University of
  19  *      California, Berkeley and its contributors.
  20  * 4. Neither the name of the University nor the names of its contributors
  21  *    may be used to endorse or promote products derived from this software
  22  *    without specific prior written permission.
  23  *
  24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  34  * SUCH DAMAGE.
  35  *
  36  *      $NetBSD: tokenizer.c,v 1.6 2000/09/04 22:06:33 lukem Exp $
  37  */
  38
  39 #if !defined(lint) && !defined(SCCSID)
  40 static char sccsid[] = "@(#)tokenizer.c 8.1 (Berkeley) 6/4/93";
  41 #endif /* not lint && not SCCSID */
  42 #include <sys/cdefs.h>
  43 __FBSDID("$FreeBSD$");
  44
  45 /*
  46  * tokenize.c: Bourne shell like tokenizer
  47  */
  48 #include "sys.h"
  49 #include <string.h>
  50 #include <stdlib.h>
  51 #include "tokenizer.h"
  52
  53 typedef enum {
  54         Q_none, Q_single, Q_double, Q_one, Q_doubleone
  55 } quote_t;
  56
  57 #define IFS             "\t \n"
  58
  59 #define TOK_KEEP        1
  60 #define TOK_EAT         2
  61
  62 #define WINCR           20
  63 #define AINCR           10
  64
  65 #define tok_malloc(a)           malloc(a)
  66 #define tok_free(a)             free(a)
  67 #define tok_realloc(a, b)       realloc(a, b)
  68 #define tok_reallocf(a, b)      reallocf(a, b)
  69
  70
  71 struct tokenizer {
  72         char    *ifs;           /* In field separator                    */
  73         int      argc, amax;    /* Current and maximum number of args    */
  74         char   **argv;          /* Argument list                         */
  75         char    *wptr, *wmax;   /* Space and limit on the word buffer    */
  76         char    *wstart;        /* Beginning of next word                */
  77         char    *wspace;        /* Space of word buffer                  */
  78         quote_t  quote;         /* Quoting state                         */
  79         int      flags;         /* flags;                                */
  80 };
  81
  82
  83 private void tok_finish(Tokenizer *);
  84
  85
  86 /* tok_finish():
  87  *      Finish a word in the tokenizer.
  88  */
  89 private void
  90 tok_finish(Tokenizer *tok)
  91 {
  92
  93         *tok->wptr = '\0';
  94         if ((tok->flags & TOK_KEEP) || tok->wptr != tok->wstart) {
  95                 tok->argv[tok->argc++] = tok->wstart;
  96                 tok->argv[tok->argc] = NULL;
  97                 tok->wstart = ++tok->wptr;
  98         }
  99         tok->flags &= ~TOK_KEEP;
 100 }
 101
 102
 103 /* tok_init():
 104  *      Initialize the tokenizer
 105  */
 106 public Tokenizer *
 107 tok_init(const char *ifs)
 108 {
 109         Tokenizer *tok = (Tokenizer *) tok_malloc(sizeof(Tokenizer));
 110
 111         tok->ifs = strdup(ifs ? ifs : IFS);
 112         tok->argc = 0;
 113         tok->amax = AINCR;
 114         tok->argv = (char **) tok_malloc(sizeof(char *) * tok->amax);
 115         if (tok->argv == NULL)
 116                 return (NULL);
 117         tok->argv[0] = NULL;
 118         tok->wspace = (char *) tok_malloc(WINCR);
 119         if (tok->wspace == NULL)
 120                 return (NULL);
 121         tok->wmax = tok->wspace + WINCR;
 122         tok->wstart = tok->wspace;
 123         tok->wptr = tok->wspace;
 124         tok->flags = 0;
 125         tok->quote = Q_none;
 126
 127         return (tok);
 128 }
 129
 130
 131 /* tok_reset():
 132  *      Reset the tokenizer
 133  */
 134 public void
 135 tok_reset(Tokenizer *tok)
 136 {
 137
 138         tok->argc = 0;
 139         tok->wstart = tok->wspace;
 140         tok->wptr = tok->wspace;
 141         tok->flags = 0;
 142         tok->quote = Q_none;
 143 }
 144
 145
 146 /* tok_end():
 147  *      Clean up
 148  */
 149 public void
 150 tok_end(Tokenizer *tok)
 151 {
 152
 153         tok_free((ptr_t) tok->ifs);
 154         tok_free((ptr_t) tok->wspace);
 155         tok_free((ptr_t) tok->argv);
 156         tok_free((ptr_t) tok);
 157 }
 158
 159
 160
 161 /* tok_line():
 162  *      Bourne shell like tokenizing
 163  *      Return:
 164  *              -1: Internal error
 165  *               3: Quoted return
 166  *               2: Unmatched double quote
 167  *               1: Unmatched single quote
 168  *               0: Ok
 169  */
 170 public int
 171 tok_line(Tokenizer *tok, const char *line, int *argc, char ***argv)
 172 {
 173         const char *ptr;
 174
 175         for (;;) {
 176                 switch (*(ptr = line++)) {
 177                 case '\'':
 178                         tok->flags |= TOK_KEEP;
 179                         tok->flags &= ~TOK_EAT;
 180                         switch (tok->quote) {
 181                         case Q_none:
 182                                 tok->quote = Q_single;  /* Enter single quote
 183                                                          * mode */
 184                                 break;
 185
 186                         case Q_single:  /* Exit single quote mode */
 187                                 tok->quote = Q_none;
 188                                 break;
 189
 190                         case Q_one:     /* Quote this ' */
 191                                 tok->quote = Q_none;
 192                                 *tok->wptr++ = *ptr;
 193                                 break;
 194
 195                         case Q_double:  /* Stay in double quote mode */
 196                                 *tok->wptr++ = *ptr;
 197                                 break;
 198
 199                         case Q_doubleone:       /* Quote this ' */
 200                                 tok->quote = Q_double;
 201                                 *tok->wptr++ = *ptr;
 202                                 break;
 203
 204                         default:
 205                                 return (-1);
 206                         }
 207                         break;
 208
 209                 case '"':
 210                         tok->flags &= ~TOK_EAT;
 211                         tok->flags |= TOK_KEEP;
 212                         switch (tok->quote) {
 213                         case Q_none:    /* Enter double quote mode */
 214                                 tok->quote = Q_double;
 215                                 break;
 216
 217                         case Q_double:  /* Exit double quote mode */
 218                                 tok->quote = Q_none;
 219                                 break;
 220
 221                         case Q_one:     /* Quote this " */
 222                                 tok->quote = Q_none;
 223                                 *tok->wptr++ = *ptr;
 224                                 break;
 225
 226                         case Q_single:  /* Stay in single quote mode */
 227                                 *tok->wptr++ = *ptr;
 228                                 break;
 229
 230                         case Q_doubleone:       /* Quote this " */
 231                                 tok->quote = Q_double;
 232                                 *tok->wptr++ = *ptr;
 233                                 break;
 234
 235                         default:
 236                                 return (-1);
 237                         }
 238                         break;
 239
 240                 case '\\':
 241                         tok->flags |= TOK_KEEP;
 242                         tok->flags &= ~TOK_EAT;
 243                         switch (tok->quote) {
 244                         case Q_none:    /* Quote next character */
 245                                 tok->quote = Q_one;
 246                                 break;
 247
 248                         case Q_double:  /* Quote next character */
 249                                 tok->quote = Q_doubleone;
 250                                 break;
 251
 252                         case Q_one:     /* Quote this, restore state */
 253                                 *tok->wptr++ = *ptr;
 254                                 tok->quote = Q_none;
 255                                 break;
 256
 257                         case Q_single:  /* Stay in single quote mode */
 258                                 *tok->wptr++ = *ptr;
 259                                 break;
 260
 261                         case Q_doubleone:       /* Quote this \ */
 262                                 tok->quote = Q_double;
 263                                 *tok->wptr++ = *ptr;
 264                                 break;
 265
 266                         default:
 267                                 return (-1);
 268                         }
 269                         break;
 270
 271                 case '\n':
 272                         tok->flags &= ~TOK_EAT;
 273                         switch (tok->quote) {
 274                         case Q_none:
 275                                 tok_finish(tok);
 276                                 *argv = tok->argv;
 277                                 *argc = tok->argc;
 278                                 return (0);
 279
 280                         case Q_single:
 281                         case Q_double:
 282                                 *tok->wptr++ = *ptr;    /* Add the return */
 283                                 break;
 284
 285                         case Q_doubleone:   /* Back to double, eat the '\n' */
 286                                 tok->flags |= TOK_EAT;
 287                                 tok->quote = Q_double;
 288                                 break;
 289
 290                         case Q_one:     /* No quote, more eat the '\n' */
 291                                 tok->flags |= TOK_EAT;
 292                                 tok->quote = Q_none;
 293                                 break;
 294
 295                         default:
 296                                 return (0);
 297                         }
 298                         break;
 299
 300                 case '\0':
 301                         switch (tok->quote) {
 302                         case Q_none:
 303                                 /* Finish word and return */
 304                                 if (tok->flags & TOK_EAT) {
 305                                         tok->flags &= ~TOK_EAT;
 306                                         return (3);
 307                                 }
 308                                 tok_finish(tok);
 309                                 *argv = tok->argv;
 310                                 *argc = tok->argc;
 311                                 return (0);
 312
 313                         case Q_single:
 314                                 return (1);
 315
 316                         case Q_double:
 317                                 return (2);
 318
 319                         case Q_doubleone:
 320                                 tok->quote = Q_double;
 321                                 *tok->wptr++ = *ptr;
 322                                 break;
 323
 324                         case Q_one:
 325                                 tok->quote = Q_none;
 326                                 *tok->wptr++ = *ptr;
 327                                 break;
 328
 329                         default:
 330                                 return (-1);
 331                         }
 332                         break;
 333
 334                 default:
 335                         tok->flags &= ~TOK_EAT;
 336                         switch (tok->quote) {
 337                         case Q_none:
 338                                 if (strchr(tok->ifs, *ptr) != NULL)
 339                                         tok_finish(tok);
 340                                 else
 341                                         *tok->wptr++ = *ptr;
 342                                 break;
 343
 344                         case Q_single:
 345                         case Q_double:
 346                                 *tok->wptr++ = *ptr;
 347                                 break;
 348
 349
 350                         case Q_doubleone:
 351                                 *tok->wptr++ = '\\';
 352                                 tok->quote = Q_double;
 353                                 *tok->wptr++ = *ptr;
 354                                 break;
 355
 356                         case Q_one:
 357                                 tok->quote = Q_none;
 358                                 *tok->wptr++ = *ptr;
 359                                 break;
 360
 361                         default:
 362                                 return (-1);
 363
 364                         }
 365                         break;
 366                 }
 367
 368                 if (tok->wptr >= tok->wmax - 4) {
 369                         size_t size = tok->wmax - tok->wspace + WINCR;
 370                         char *s = (char *) tok_realloc(tok->wspace, size);
 371                         /* SUPPRESS 22 */
 372                         int offs = s - tok->wspace;
 373                         if (s == NULL)
 374                                 return (-1);
 375
 376                         if (offs != 0) {
 377                                 int i;
 378                                 for (i = 0; i < tok->argc; i++)
 379                                         tok->argv[i] = tok->argv[i] + offs;
 380                                 tok->wptr = tok->wptr + offs;
 381                                 tok->wstart = tok->wstart + offs;
 382                                 tok->wmax = s + size;
 383                                 tok->wspace = s;
 384                         }
 385                 }
 386                 if (tok->argc >= tok->amax - 4) {
 387                         char **p;
 388                         tok->amax += AINCR;
 389                         p = (char **) tok_reallocf(tok->argv,
 390                             tok->amax * sizeof(char *));
 391                         if (p == NULL)
 392                                 return (-1);
 393                         tok->argv = p;
 394                 }
 395         }
 396 }