lib/ecoli_node_sh_lex.c

   1 /* SPDX-License-Identifier: BSD-3-Clause
   2  * Copyright 2016, Olivier MATZ <zer0@droids-corp.org>
   3  */
   4
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <string.h>
   8 #include <assert.h>
   9 #include <stdarg.h>
  10 #include <ctype.h>
  11 #include <errno.h>
  12
  13 #include <ecoli_malloc.h>
  14 #include <ecoli_log.h>
  15 #include <ecoli_string.h>
  16 #include <ecoli_test.h>
  17 #include <ecoli_strvec.h>
  18 #include <ecoli_node.h>
  19 #include <ecoli_parsed.h>
  20 #include <ecoli_completed.h>
  21 #include <ecoli_node_seq.h>
  22 #include <ecoli_node_str.h>
  23 #include <ecoli_node_option.h>
  24 #include <ecoli_node_sh_lex.h>
  25
  26 EC_LOG_TYPE_REGISTER(node_sh_lex);
  27
  28 struct ec_node_sh_lex {
  29         struct ec_node gen;
  30         struct ec_node *child;
  31 };
  32
  33 static size_t eat_spaces(const char *str)
  34 {
  35         size_t i = 0;
  36
  37         /* skip spaces */
  38         while (isblank(str[i]))
  39                 i++;
  40
  41         return i;
  42 }
  43
  44 /*
  45  * Allocate a new string which is a copy of the input string with quotes
  46  * removed. If quotes are not closed properly, set missing_quote to the
  47  * missing quote char.
  48  */
  49 static char *unquote_str(const char *str, size_t n, int allow_missing_quote,
  50         char *missing_quote)
  51 {
  52         unsigned s = 1, d = 0;
  53         char quote = str[0];
  54         char *dst;
  55         int closed = 0;
  56
  57         dst = ec_malloc(n);
  58         if (dst == NULL) {
  59                 errno = ENOMEM;
  60                 return NULL;
  61         }
  62
  63         /* copy string and remove quotes */
  64         while (s < n && d < n && str[s] != '\0') {
  65                 if (str[s] == '\\' && str[s+1] == quote) {
  66                         dst[d++] = quote;
  67                         s += 2;
  68                         continue;
  69                 }
  70                 if (str[s] == '\\' && str[s+1] == '\\') {
  71                         dst[d++] = '\\';
  72                         s += 2;
  73                         continue;
  74                 }
  75                 if (str[s] == quote) {
  76                         s++;
  77                         closed = 1;
  78                         break;
  79                 }
  80                 dst[d++] = str[s++];
  81         }
  82
  83         /* not enough room in dst buffer (should not happen) */
  84         if (d >= n) {
  85                 ec_free(dst);
  86                 errno = EMSGSIZE;
  87                 return NULL;
  88         }
  89
  90         /* quote not closed */
  91         if (closed == 0) {
  92                 if (missing_quote != NULL)
  93                         *missing_quote = str[0];
  94                 if (allow_missing_quote == 0) {
  95                         ec_free(dst);
  96                         errno = EINVAL;
  97                         return NULL;
  98                 }
  99         }
 100         dst[d++] = '\0';
 101
 102         return dst;
 103 }
 104
 105 static size_t eat_quoted_str(const char *str)
 106 {
 107         size_t i = 0;
 108         char quote = str[0];
 109
 110         while (str[i] != '\0') {
 111                 if (str[i] != '\\' && str[i+1] == quote)
 112                         return i + 2;
 113                 i++;
 114         }
 115
 116         /* unclosed quote, will be detected later */
 117         return i;
 118 }
 119
 120 static size_t eat_str(const char *str)
 121 {
 122         size_t i = 0;
 123
 124         /* eat chars until we find a quote, space, or end of string  */
 125         while (!isblank(str[i]) && str[i] != '\0' &&
 126                         str[i] != '"' && str[i] != '\'')
 127                 i++;
 128
 129         return i;
 130 }
 131
 132 static struct ec_strvec *tokenize(const char *str, int completion,
 133         int allow_missing_quote, char *missing_quote)
 134 {
 135         struct ec_strvec *strvec = NULL;
 136         size_t off = 0, len, suboff, sublen;
 137         char *word = NULL, *concat = NULL, *tmp;
 138         int last_is_space = 1;
 139
 140         strvec = ec_strvec();
 141         if (strvec == NULL)
 142                 goto fail;
 143
 144         while (str[off] != '\0') {
 145                 if (missing_quote != NULL)
 146                         *missing_quote = '\0';
 147                 len = eat_spaces(&str[off]);
 148                 if (len > 0)
 149                         last_is_space = 1;
 150                 off += len;
 151
 152                 len = 0;
 153                 suboff = off;
 154                 while (str[suboff] != '\0') {
 155                         if (missing_quote != NULL)
 156                                 *missing_quote = '\0';
 157                         last_is_space = 0;
 158                         if (str[suboff] == '"' || str[suboff] == '\'') {
 159                                 sublen = eat_quoted_str(&str[suboff]);
 160                                 word = unquote_str(&str[suboff], sublen,
 161                                         allow_missing_quote, missing_quote);
 162                         } else {
 163                                 sublen = eat_str(&str[suboff]);
 164                                 if (sublen == 0)
 165                                         break;
 166                                 word = ec_strndup(&str[suboff], sublen);
 167                         }
 168
 169                         if (word == NULL)
 170                                 goto fail;
 171
 172                         len += sublen;
 173                         suboff += sublen;
 174
 175                         if (concat == NULL) {
 176                                 concat = word;
 177                                 word = NULL;
 178                         } else {
 179                                 tmp = ec_realloc(concat, len + 1);
 180                                 if (tmp == NULL)
 181                                         goto fail;
 182                                 concat = tmp;
 183                                 strcat(concat, word);
 184                                 ec_free(word);
 185                                 word = NULL;
 186                         }
 187                 }
 188
 189                 if (concat != NULL) {
 190                         if (ec_strvec_add(strvec, concat) < 0)
 191                                 goto fail;
 192                         ec_free(concat);
 193                         concat = NULL;
 194                 }
 195
 196                 off += len;
 197         }
 198
 199         /* in completion mode, append an empty string in the vector if
 200          * the input string ends with space */
 201         if (completion && last_is_space) {
 202                 if (ec_strvec_add(strvec, "") < 0)
 203                         goto fail;
 204         }
 205
 206         return strvec;
 207
 208  fail:
 209         ec_free(word);
 210         ec_free(concat);
 211         ec_strvec_free(strvec);
 212         return NULL;
 213 }
 214
 215 static int
 216 ec_node_sh_lex_parse(const struct ec_node *gen_node,
 217                 struct ec_parsed *state,
 218                 const struct ec_strvec *strvec)
 219 {
 220         struct ec_node_sh_lex *node = (struct ec_node_sh_lex *)gen_node;
 221         struct ec_strvec *new_vec = NULL;
 222         struct ec_parsed *child_parsed;
 223         const char *str;
 224         int ret;
 225
 226         if (ec_strvec_len(strvec) == 0) {
 227                 new_vec = ec_strvec();
 228         } else {
 229                 str = ec_strvec_val(strvec, 0);
 230                 new_vec = tokenize(str, 0, 0, NULL);
 231         }
 232         if (new_vec == NULL) {
 233                 if (errno == EINVAL)
 234                         ret = EC_PARSED_NOMATCH;
 235                 else
 236                         ret = -ENOMEM;
 237                 goto fail;
 238         }
 239
 240         ret = ec_node_parse_child(node->child, state, new_vec);
 241         if (ret < 0)
 242                 goto fail;
 243
 244         if ((unsigned)ret == ec_strvec_len(new_vec)) {
 245                 ret = 1;
 246         } else if (ret != EC_PARSED_NOMATCH) {
 247                 child_parsed = ec_parsed_get_last_child(state);
 248                 ec_parsed_unlink_child(state, child_parsed);
 249                 ec_parsed_free(child_parsed);
 250                 ret = EC_PARSED_NOMATCH;
 251         }
 252
 253         ec_strvec_free(new_vec);
 254         new_vec = NULL;
 255
 256         return ret;
 257
 258  fail:
 259         ec_strvec_free(new_vec);
 260         return ret;
 261 }
 262
 263 static int
 264 ec_node_sh_lex_complete(const struct ec_node *gen_node,
 265                         struct ec_completed *completed,
 266                         const struct ec_strvec *strvec)
 267 {
 268         struct ec_node_sh_lex *node = (struct ec_node_sh_lex *)gen_node;
 269         struct ec_completed *tmp_completed = NULL;
 270         struct ec_strvec *new_vec = NULL;
 271         struct ec_completed_iter *iter = NULL;
 272         struct ec_completed_item *item = NULL;
 273         char *new_str = NULL;
 274         const char *str;
 275         char missing_quote = '\0';
 276         int ret;
 277
 278         if (ec_strvec_len(strvec) != 1)
 279                 return 0;
 280
 281         str = ec_strvec_val(strvec, 0);
 282         new_vec = tokenize(str, 1, 1, &missing_quote);
 283         if (new_vec == NULL)
 284                 goto fail;
 285
 286         /* we will store the completions in a temporary struct, because
 287          * we want to update them (ex: add missing quotes) */
 288         tmp_completed = ec_completed(ec_completed_get_state(completed));
 289         if (tmp_completed == NULL)
 290                 goto fail;
 291
 292         ret = ec_node_complete_child(node->child, tmp_completed, new_vec);
 293         if (ret < 0)
 294                 goto fail;
 295
 296         /* add missing quote for full completions  */
 297         if (missing_quote != '\0') {
 298                 iter = ec_completed_iter(tmp_completed, EC_COMP_FULL);
 299                 if (iter == NULL)
 300                         goto fail;
 301                 while ((item = ec_completed_iter_next(iter)) != NULL) {
 302                         str = ec_completed_item_get_str(item);
 303                         if (ec_asprintf(&new_str, "%c%s%c", missing_quote, str,
 304                                         missing_quote) < 0) {
 305                                 new_str = NULL;
 306                                 goto fail;
 307                         }
 308                         if (ec_completed_item_set_str(item, new_str) < 0)
 309                                 goto fail;
 310                         ec_free(new_str);
 311                         new_str = NULL;
 312
 313                         str = ec_completed_item_get_completion(item);
 314                         if (ec_asprintf(&new_str, "%s%c", str,
 315                                         missing_quote) < 0) {
 316                                 new_str = NULL;
 317                                 goto fail;
 318                         }
 319                         if (ec_completed_item_set_completion(item, new_str) < 0)
 320                                 goto fail;
 321                         ec_free(new_str);
 322                         new_str = NULL;
 323                 }
 324         }
 325
 326         ec_completed_iter_free(iter);
 327         ec_strvec_free(new_vec);
 328
 329         ec_completed_merge(completed, tmp_completed);
 330
 331         return 0;
 332
 333  fail:
 334         ec_completed_free(tmp_completed);
 335         ec_completed_iter_free(iter);
 336         ec_strvec_free(new_vec);
 337         ec_free(new_str);
 338
 339         return -1;
 340 }
 341
 342 static void ec_node_sh_lex_free_priv(struct ec_node *gen_node)
 343 {
 344         struct ec_node_sh_lex *node = (struct ec_node_sh_lex *)gen_node;
 345
 346         ec_node_free(node->child);
 347 }
 348
 349 static struct ec_node_type ec_node_sh_lex_type = {
 350         .name = "sh_lex",
 351         .parse = ec_node_sh_lex_parse,
 352         .complete = ec_node_sh_lex_complete,
 353         .size = sizeof(struct ec_node_sh_lex),
 354         .free_priv = ec_node_sh_lex_free_priv,
 355 };
 356
 357 EC_NODE_TYPE_REGISTER(ec_node_sh_lex_type);
 358
 359 struct ec_node *ec_node_sh_lex(const char *id, struct ec_node *child)
 360 {
 361         struct ec_node_sh_lex *node = NULL;
 362
 363         if (child == NULL)
 364                 return NULL;
 365
 366         node = (struct ec_node_sh_lex *)__ec_node(&ec_node_sh_lex_type, id);
 367         if (node == NULL) {
 368                 ec_node_free(child);
 369                 return NULL;
 370         }
 371
 372         node->child = child;
 373
 374         return &node->gen;
 375 }
 376
 377 /* LCOV_EXCL_START */
 378 static int ec_node_sh_lex_testcase(void)
 379 {
 380         struct ec_node *node;
 381         int testres = 0;
 382
 383         node = ec_node_sh_lex(EC_NO_ID,
 384                 EC_NODE_SEQ(EC_NO_ID,
 385                         ec_node_str(EC_NO_ID, "foo"),
 386                         ec_node_option(EC_NO_ID,
 387                                 ec_node_str(EC_NO_ID, "toto")
 388                         ),
 389                         ec_node_str(EC_NO_ID, "bar")
 390                 )
 391         );
 392         if (node == NULL) {
 393                 EC_LOG(EC_LOG_ERR, "cannot create node\n");
 394                 return -1;
 395         }
 396         testres |= EC_TEST_CHECK_PARSE(node, 1, "foo bar");
 397         testres |= EC_TEST_CHECK_PARSE(node, 1, "  foo   bar");
 398         testres |= EC_TEST_CHECK_PARSE(node, 1, "  'foo' \"bar\"");
 399         testres |= EC_TEST_CHECK_PARSE(node, 1, "  'f'oo 'toto' bar");
 400         testres |= EC_TEST_CHECK_PARSE(node, -1, "  foo toto bar'");
 401         ec_node_free(node);
 402
 403         /* test completion */
 404         node = ec_node_sh_lex(EC_NO_ID,
 405                 EC_NODE_SEQ(EC_NO_ID,
 406                         ec_node_str(EC_NO_ID, "foo"),
 407                         ec_node_option(EC_NO_ID,
 408                                 ec_node_str(EC_NO_ID, "toto")
 409                         ),
 410                         ec_node_str(EC_NO_ID, "bar"),
 411                         ec_node_str(EC_NO_ID, "titi")
 412                 )
 413         );
 414         if (node == NULL) {
 415                 EC_LOG(EC_LOG_ERR, "cannot create node\n");
 416                 return -1;
 417         }
 418         testres |= EC_TEST_CHECK_COMPLETE(node,
 419                 "", EC_NODE_ENDLIST,
 420                 "foo", EC_NODE_ENDLIST);
 421         testres |= EC_TEST_CHECK_COMPLETE(node,
 422                 " ", EC_NODE_ENDLIST,
 423                 "foo", EC_NODE_ENDLIST);
 424         testres |= EC_TEST_CHECK_COMPLETE(node,
 425                 "f", EC_NODE_ENDLIST,
 426                 "foo", EC_NODE_ENDLIST);
 427         testres |= EC_TEST_CHECK_COMPLETE(node,
 428                 "foo", EC_NODE_ENDLIST,
 429                 "foo", EC_NODE_ENDLIST);
 430         testres |= EC_TEST_CHECK_COMPLETE(node,
 431                 "foo ", EC_NODE_ENDLIST,
 432                 "bar", "toto", EC_NODE_ENDLIST);
 433         testres |= EC_TEST_CHECK_COMPLETE(node,
 434                 "foo t", EC_NODE_ENDLIST,
 435                 "toto", EC_NODE_ENDLIST);
 436         testres |= EC_TEST_CHECK_COMPLETE(node,
 437                 "foo b", EC_NODE_ENDLIST,
 438                 "bar", EC_NODE_ENDLIST);
 439         testres |= EC_TEST_CHECK_COMPLETE(node,
 440                 "foo bar", EC_NODE_ENDLIST,
 441                 "bar", EC_NODE_ENDLIST);
 442         testres |= EC_TEST_CHECK_COMPLETE(node,
 443                 "foo bar ", EC_NODE_ENDLIST,
 444                 "titi", EC_NODE_ENDLIST);
 445         testres |= EC_TEST_CHECK_COMPLETE(node,
 446                 "foo toto bar ", EC_NODE_ENDLIST,
 447                 "titi", EC_NODE_ENDLIST);
 448         testres |= EC_TEST_CHECK_COMPLETE(node,
 449                 "x", EC_NODE_ENDLIST,
 450                 EC_NODE_ENDLIST);
 451         testres |= EC_TEST_CHECK_COMPLETE(node,
 452                 "foo barx", EC_NODE_ENDLIST,
 453                 EC_NODE_ENDLIST);
 454         testres |= EC_TEST_CHECK_COMPLETE(node,
 455                 "foo 'b", EC_NODE_ENDLIST,
 456                 "'bar'", EC_NODE_ENDLIST);
 457
 458         ec_node_free(node);
 459         return testres;
 460 }
 461 /* LCOV_EXCL_STOP */
 462
 463 static struct ec_test ec_node_sh_lex_test = {
 464         .name = "node_sh_lex",
 465         .test = ec_node_sh_lex_testcase,
 466 };
 467
 468 EC_TEST_REGISTER(ec_node_sh_lex_test);