lib/ecoli_tk_shlex.c

   1 /*
   2  * Copyright (c) 2016, Olivier MATZ <zer0@droids-corp.org>
   3  *
   4  * Redistribution and use in source and binary forms, with or without
   5  * modification, are permitted provided that the following conditions are met:
   6  *
   7  *     * Redistributions of source code must retain the above copyright
   8  *       notice, this list of conditions and the following disclaimer.
   9  *     * Redistributions in binary form must reproduce the above copyright
  10  *       notice, this list of conditions and the following disclaimer in the
  11  *       documentation and/or other materials provided with the distribution.
  12  *     * Neither the name of the University of California, Berkeley nor the
  13  *       names of its contributors may be used to endorse or promote products
  14  *       derived from this software without specific prior written permission.
  15  *
  16  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND ANY
  17  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  18  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
  19  * DISCLAIMED. IN NO EVENT SHALL THE REGENTS AND CONTRIBUTORS BE LIABLE FOR ANY
  20  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
  21  * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  22  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
  23  * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  24  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
  25  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  26  */
  27
  28 #include <stdio.h>
  29 #include <stdlib.h>
  30 #include <string.h>
  31 #include <assert.h>
  32 #include <stdarg.h>
  33 #include <ctype.h>
  34 #include <errno.h>
  35
  36 #include <ecoli_malloc.h>
  37 #include <ecoli_log.h>
  38 #include <ecoli_test.h>
  39 #include <ecoli_strvec.h>
  40 #include <ecoli_tk.h>
  41 #include <ecoli_tk_seq.h>
  42 #include <ecoli_tk_str.h>
  43 #include <ecoli_tk_option.h>
  44 #include <ecoli_tk_shlex.h>
  45
  46 struct ec_tk_shlex {
  47         struct ec_tk gen;
  48         struct ec_tk *child;
  49 };
  50
  51 static size_t eat_spaces(const char *str)
  52 {
  53         size_t i = 0;
  54
  55         /* skip spaces */
  56         while (isblank(str[i]))
  57                 i++;
  58
  59         return i;
  60 }
  61
  62 /*
  63  * Allocate a new string which is a copy of the input string with quotes
  64  * removed. If quotes are not closed properly, set missing_quote to the
  65  * missing quote char.
  66  */
  67 static char *unquote_str(const char *str, size_t n, int allow_missing_quote,
  68         char *missing_quote)
  69 {
  70         unsigned s = 1, d = 0;
  71         char quote = str[0];
  72         char *dst;
  73         int closed = 0;
  74
  75         dst = ec_malloc(n);
  76         if (dst == NULL) {
  77                 errno = ENOMEM;
  78                 return NULL;
  79         }
  80
  81         /* copy token and remove quotes */
  82         while (s < n && d < n && str[s] != '\0') {
  83                 if (str[s] == '\\' && str[s+1] == quote) {
  84                         dst[d++] = quote;
  85                         s += 2;
  86                         continue;
  87                 }
  88                 if (str[s] == '\\' && str[s+1] == '\\') {
  89                         dst[d++] = '\\';
  90                         s += 2;
  91                         continue;
  92                 }
  93                 if (str[s] == quote) {
  94                         s++;
  95                         closed = 1;
  96                         break;
  97                 }
  98                 dst[d++] = str[s++];
  99         }
 100
 101         /* not enough room in dst buffer (should not happen) */
 102         if (d >= n) {
 103                 ec_free(dst);
 104                 errno = EMSGSIZE;
 105                 return NULL;
 106         }
 107
 108         /* quote not closed */
 109         if (closed == 0) {
 110                 if (missing_quote != NULL)
 111                         *missing_quote = str[0];
 112                 if (allow_missing_quote == 0) {
 113                         ec_free(dst);
 114                         errno = EINVAL;
 115                         return NULL;
 116                 }
 117         }
 118         dst[d++] = '\0';
 119
 120         return dst;
 121 }
 122
 123 static size_t eat_quoted_str(const char *str)
 124 {
 125         size_t i = 0;
 126         char quote = str[0];
 127
 128         while (str[i] != '\0') {
 129                 if (str[i] != '\\' && str[i+1] == quote)
 130                         return i + 2;
 131                 i++;
 132         }
 133
 134         /* unclosed quote, will be detected later */
 135         return i;
 136 }
 137
 138 static size_t eat_str(const char *str)
 139 {
 140         size_t i = 0;
 141
 142         /* skip spaces */
 143         while (!isblank(str[i]) && str[i] != '\0')
 144                 i++;
 145
 146         return i;
 147 }
 148
 149 static struct ec_strvec *tokenize(const char *str, int completion,
 150         int allow_missing_quote, char *missing_quote)
 151 {
 152         struct ec_strvec *strvec = NULL;
 153         size_t off = 0, len, suboff, sublen;
 154         char *word = NULL, *concat = NULL, *tmp;
 155         int last_is_space = 1;
 156
 157 //      printf("str=%s\n", str);
 158
 159         strvec = ec_strvec_new();
 160         if (strvec == NULL)
 161                 goto fail;
 162
 163         while (str[off] != '\0') {
 164                 len = eat_spaces(&str[off]);
 165                 if (len > 0)
 166                         last_is_space = 1;
 167 //              printf("space=%zd\n", len);
 168                 off += len;
 169
 170                 len = 0;
 171                 suboff = off;
 172                 while (str[suboff] != '\0') {
 173                         last_is_space = 0;
 174                         if (str[suboff] == '"' || str[suboff] == '\'') {
 175                                 sublen = eat_quoted_str(&str[suboff]);
 176 //                              printf("sublen=%zd\n", sublen);
 177                                 word = unquote_str(&str[suboff], sublen,
 178                                         allow_missing_quote, missing_quote);
 179                         } else {
 180                                 sublen = eat_str(&str[suboff]);
 181 //                              printf("sublen=%zd\n", sublen);
 182                                 if (sublen == 0)
 183                                         break;
 184                                 word = ec_strndup(&str[suboff], sublen);
 185                         }
 186
 187                         if (word == NULL)
 188                                 goto fail;
 189 //                      printf("word=%s\n", word);
 190
 191                         len += sublen;
 192                         suboff += sublen;
 193
 194                         if (concat == NULL) {
 195                                 concat = word;
 196                                 word = NULL;
 197                         } else {
 198                                 tmp = ec_realloc(concat, len + 1);
 199                                 if (tmp == NULL)
 200                                         goto fail;
 201                                 concat = tmp;
 202                                 strcat(concat, word);
 203                                 ec_free(word);
 204                                 word = NULL;
 205                         }
 206                 }
 207
 208                 if (concat != NULL) {
 209                         if (ec_strvec_add(strvec, concat) < 0)
 210                                 goto fail;
 211                         ec_free(concat);
 212                         concat = NULL;
 213                 }
 214
 215                 /* XXX remove all printf comments */
 216 //              printf("str off=%zd len=%zd\n", off, len);
 217                 off += len;
 218         }
 219
 220         /* in completion mode, append an empty token if the string ends
 221          * with space */
 222         if (completion && last_is_space) {
 223                 if (ec_strvec_add(strvec, "") < 0)
 224                         goto fail;
 225         }
 226
 227         return strvec;
 228
 229  fail:
 230         ec_free(word);
 231         ec_free(concat);
 232         ec_strvec_free(strvec);
 233         return NULL;
 234 }
 235
 236 static struct ec_parsed_tk *ec_tk_shlex_parse(const struct ec_tk *gen_tk,
 237         const struct ec_strvec *strvec)
 238 {
 239         struct ec_tk_shlex *tk = (struct ec_tk_shlex *)gen_tk;
 240         struct ec_strvec *new_vec = NULL, *match_strvec;
 241         struct ec_parsed_tk *parsed_tk = NULL, *child_parsed_tk;
 242         const char *str;
 243
 244         parsed_tk = ec_parsed_tk_new();
 245         if (parsed_tk == NULL)
 246                 return NULL;
 247
 248         if (ec_strvec_len(strvec) == 0)
 249                 return parsed_tk;
 250
 251         str = ec_strvec_val(strvec, 0);
 252         new_vec = tokenize(str, 0, 0, NULL);
 253         if (new_vec == NULL)
 254                 goto fail;
 255
 256         child_parsed_tk = ec_tk_parse_tokens(tk->child, new_vec);
 257         if (child_parsed_tk == NULL)
 258                 goto fail;
 259
 260         if (!ec_parsed_tk_matches(child_parsed_tk) ||
 261                         ec_parsed_tk_len(child_parsed_tk) !=
 262                                 ec_strvec_len(new_vec)) {
 263                 ec_strvec_free(new_vec);
 264                 ec_parsed_tk_free(child_parsed_tk);
 265                 return parsed_tk;
 266         }
 267         ec_strvec_free(new_vec);
 268         new_vec = NULL;
 269
 270         ec_parsed_tk_add_child(parsed_tk, child_parsed_tk);
 271         match_strvec = ec_strvec_ndup(strvec, 0, 1);
 272         if (match_strvec == NULL)
 273                 goto fail;
 274         ec_parsed_tk_set_match(parsed_tk, gen_tk, match_strvec);
 275
 276         return parsed_tk;
 277
 278  fail:
 279         ec_strvec_free(new_vec);
 280         ec_parsed_tk_free(parsed_tk);
 281
 282         return NULL;
 283 }
 284
 285 static struct ec_completed_tk *ec_tk_shlex_complete(const struct ec_tk *gen_tk,
 286         const struct ec_strvec *strvec)
 287 {
 288         struct ec_tk_shlex *tk = (struct ec_tk_shlex *)gen_tk;
 289         struct ec_completed_tk *completed_tk, *child_completed_tk = NULL;
 290         struct ec_strvec *new_vec = NULL;
 291         const char *str;
 292         char missing_quote;
 293
 294 //      printf("==================\n");
 295         completed_tk = ec_completed_tk_new();
 296         if (completed_tk == NULL)
 297                 return NULL;
 298
 299         if (ec_strvec_len(strvec) != 1)
 300                 return completed_tk;
 301
 302         str = ec_strvec_val(strvec, 0);
 303         new_vec = tokenize(str, 1, 1, &missing_quote);
 304         if (new_vec == NULL)
 305                 goto fail;
 306
 307 //      ec_strvec_dump(new_vec, stdout);
 308
 309         child_completed_tk = ec_tk_complete_tokens(tk->child, new_vec);
 310         if (child_completed_tk == NULL)
 311                 goto fail;
 312
 313         ec_strvec_free(new_vec);
 314         new_vec = NULL;
 315         ec_completed_tk_merge(completed_tk, child_completed_tk);
 316
 317         return completed_tk;
 318
 319
 320 #if 0
 321         for (i = 0, t = &tokens[0]; i < tk->len; i++, t++) {
 322                 if (*(t + 1) != NULL) {
 323                         child_parsed_tk = ec_tk_parse(tk->table[i], *t);
 324                         if (child_parsed_tk == NULL)
 325                                 goto fail;
 326
 327                         if (strlen(child_parsed_tk->str) == 0)
 328                                 t--;
 329                         else if (strlen(child_parsed_tk->str) != strlen(*t)) {
 330                                 ec_parsed_tk_free(child_parsed_tk);
 331                                 goto fail;
 332                         }
 333
 334                         ec_parsed_tk_free(child_parsed_tk);
 335                 } else {
 336                         child_completed_tk = ec_tk_complete(tk->table[i], *t);
 337                         if (child_completed_tk == NULL) {
 338                                 ec_completed_tk_free(completed_tk);
 339                                 return NULL;
 340                         }
 341                         ec_completed_tk_merge(completed_tk, child_completed_tk);
 342
 343                         child_parsed_tk = ec_tk_parse(tk->table[i], "");
 344                         if (child_parsed_tk == NULL)
 345                                 break;
 346                         ec_parsed_tk_free(child_parsed_tk);
 347                         t--;
 348                 }
 349         }
 350
 351         if (tokens != NULL) {
 352                 for (t = &tokens[0]; *t != NULL; t++)
 353                         ec_free(*t);
 354                 ec_free(tokens);
 355                 tokens = NULL;
 356         }
 357
 358         ec_completed_tk_dump(stdout, completed_tk);
 359 #endif
 360
 361  fail:
 362         ec_strvec_free(new_vec);
 363         ec_completed_tk_free(completed_tk);
 364         return NULL;
 365 }
 366
 367 static void ec_tk_shlex_free_priv(struct ec_tk *gen_tk)
 368 {
 369         struct ec_tk_shlex *tk = (struct ec_tk_shlex *)gen_tk;
 370
 371         ec_tk_free(tk->child);
 372 }
 373
 374 static struct ec_tk_ops ec_tk_shlex_ops = {
 375         .typename = "shlex",
 376         .parse = ec_tk_shlex_parse,
 377         .complete = ec_tk_shlex_complete,
 378         .free_priv = ec_tk_shlex_free_priv,
 379 };
 380
 381 struct ec_tk *ec_tk_shlex_new(const char *id, struct ec_tk *child)
 382 {
 383         struct ec_tk_shlex *tk = NULL;
 384
 385         if (child == NULL)
 386                 return NULL;
 387
 388         tk = (struct ec_tk_shlex *)ec_tk_new(id, &ec_tk_shlex_ops,
 389                 sizeof(*tk));
 390         if (tk == NULL) {
 391                 ec_tk_free(child);
 392                 return NULL;
 393         }
 394
 395         tk->child = child;
 396
 397         return &tk->gen;
 398 }
 399
 400 static int ec_tk_shlex_testcase(void)
 401 {
 402         struct ec_tk *tk;
 403         int ret = 0;
 404
 405         tk = ec_tk_shlex_new(NULL,
 406                 EC_TK_SEQ(NULL,
 407                         ec_tk_str(NULL, "foo"),
 408                         ec_tk_option_new(NULL,
 409                                 ec_tk_str(NULL, "toto")
 410                         ),
 411                         ec_tk_str(NULL, "bar")
 412                 )
 413         );
 414         if (tk == NULL) {
 415                 ec_log(EC_LOG_ERR, "cannot create tk\n");
 416                 return -1;
 417         }
 418         ret |= EC_TEST_CHECK_TK_PARSE(tk, 1, "foo bar");
 419         ret |= EC_TEST_CHECK_TK_PARSE(tk, 1, "  foo   bar");
 420         ret |= EC_TEST_CHECK_TK_PARSE(tk, 1, "  'foo' \"bar\"");
 421         ret |= EC_TEST_CHECK_TK_PARSE(tk, 1, "  'f'oo 'toto' bar");
 422         ec_tk_free(tk);
 423
 424         /* test completion */
 425         tk = ec_tk_shlex_new(NULL,
 426                 EC_TK_SEQ(NULL,
 427                         ec_tk_str(NULL, "foo"),
 428                         ec_tk_option_new(NULL,
 429                                 ec_tk_str(NULL, "toto")
 430                         ),
 431                         ec_tk_str(NULL, "bar"),
 432                         ec_tk_str(NULL, "titi")
 433                 )
 434         );
 435         if (tk == NULL) {
 436                 ec_log(EC_LOG_ERR, "cannot create tk\n");
 437                 return -1;
 438         }
 439         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 440                 "", EC_TK_ENDLIST,
 441                 "foo", EC_TK_ENDLIST,
 442                 "foo");
 443         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 444                 " ", EC_TK_ENDLIST,
 445                 "foo", EC_TK_ENDLIST,
 446                 "foo");
 447         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 448                 "f", EC_TK_ENDLIST,
 449                 "oo", EC_TK_ENDLIST,
 450                 "oo");
 451         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 452                 "foo", EC_TK_ENDLIST,
 453                 "", EC_TK_ENDLIST,
 454                 "");
 455         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 456                 "foo ", EC_TK_ENDLIST,
 457                 "bar", "toto", EC_TK_ENDLIST,
 458                 "");
 459         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 460                 "foo t", EC_TK_ENDLIST,
 461                 "oto", EC_TK_ENDLIST,
 462                 "oto");
 463         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 464                 "foo b", EC_TK_ENDLIST,
 465                 "ar", EC_TK_ENDLIST,
 466                 "ar");
 467         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 468                 "foo bar", EC_TK_ENDLIST,
 469                 "", EC_TK_ENDLIST,
 470                 "");
 471         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 472                 "foo bar ", EC_TK_ENDLIST,
 473                 "titi", EC_TK_ENDLIST,
 474                 "titi");
 475         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 476                 "foo toto bar ", EC_TK_ENDLIST,
 477                 "titi", EC_TK_ENDLIST,
 478                 "titi");
 479         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 480                 "x", EC_TK_ENDLIST,
 481                 EC_TK_ENDLIST,
 482                 "");
 483         ret |= EC_TEST_CHECK_TK_COMPLETE(tk,
 484                 "foo barx", EC_TK_ENDLIST,
 485                 EC_TK_ENDLIST,
 486                 "");
 487
 488         ec_tk_free(tk);
 489         return ret;
 490 }
 491
 492 static struct ec_test ec_tk_shlex_test = {
 493         .name = "tk_shlex",
 494         .test = ec_tk_shlex_testcase,
 495 };
 496
 497 EC_REGISTER_TEST(ec_tk_shlex_test);