src/ecoli_node_re_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 <stdbool.h>
   8 #include <string.h>
   9 #include <regex.h>
  10 #include <errno.h>
  11
  12 #include <ecoli_malloc.h>
  13 #include <ecoli_log.h>
  14 #include <ecoli_test.h>
  15 #include <ecoli_strvec.h>
  16 #include <ecoli_dict.h>
  17 #include <ecoli_node.h>
  18 #include <ecoli_complete.h>
  19 #include <ecoli_parse.h>
  20 #include <ecoli_config.h>
  21 #include <ecoli_node_many.h>
  22 #include <ecoli_node_or.h>
  23 #include <ecoli_node_str.h>
  24 #include <ecoli_node_int.h>
  25 #include <ecoli_node_re_lex.h>
  26
  27 EC_LOG_TYPE_REGISTER(node_re_lex);
  28
  29 struct regexp_pattern {
  30         char *pattern;
  31         char *attr_name;
  32         regex_t r;
  33         bool keep;
  34 };
  35
  36 struct ec_node_re_lex {
  37         struct ec_node *child;
  38         struct regexp_pattern *table;
  39         size_t len;
  40 };
  41
  42 static struct ec_strvec *
  43 tokenize(struct regexp_pattern *table, size_t table_len, const char *str)
  44 {
  45         struct ec_strvec *strvec = NULL;
  46         struct ec_dict *attrs = NULL;
  47         char *dup = NULL;
  48         char c;
  49         size_t len, off = 0;
  50         size_t i;
  51         int ret;
  52         regmatch_t pos;
  53
  54         dup = ec_strdup(str);
  55         if (dup == NULL)
  56                 goto fail;
  57
  58         strvec = ec_strvec();
  59         if (strvec == NULL)
  60                 goto fail;
  61
  62         len = strlen(dup);
  63         while (off < len) {
  64                 for (i = 0; i < table_len; i++) {
  65                         ret = regexec(&table[i].r, &dup[off], 1, &pos, 0);
  66                         if (ret != 0)
  67                                 continue;
  68                         if (pos.rm_so != 0 || pos.rm_eo == 0) {
  69                                 ret = -1;
  70                                 continue;
  71                         }
  72
  73                         if (table[i].keep == 0)
  74                                 break;
  75
  76                         c = dup[pos.rm_eo + off];
  77                         dup[pos.rm_eo + off] = '\0';
  78                         EC_LOG(EC_LOG_DEBUG, "re_lex match <%s>\n", &dup[off]);
  79                         if (ec_strvec_add(strvec, &dup[off]) < 0)
  80                                 goto fail;
  81
  82                         if (table[i].attr_name != NULL) {
  83                                 attrs = ec_dict();
  84                                 if (attrs == NULL)
  85                                         goto fail;
  86                                 if (ec_dict_set(attrs, table[i].attr_name,
  87                                                 NULL, NULL) < 0)
  88                                         goto fail;
  89                                 if (ec_strvec_set_attrs(strvec,
  90                                                 ec_strvec_len(strvec) - 1,
  91                                                 attrs) < 0) {
  92                                         attrs = NULL;
  93                                         goto fail;
  94                                 }
  95                                 attrs = NULL;
  96                         }
  97
  98                         dup[pos.rm_eo + off] = c;
  99                         break;
 100                 }
 101
 102                 if (ret != 0)
 103                         goto fail;
 104
 105                 off += pos.rm_eo;
 106         }
 107
 108         ec_free(dup);
 109         return strvec;
 110
 111 fail:
 112         ec_free(dup);
 113         ec_strvec_free(strvec);
 114         return NULL;
 115 }
 116
 117 static int
 118 ec_node_re_lex_parse(const struct ec_node *node,
 119                 struct ec_pnode *pstate,
 120                 const struct ec_strvec *strvec)
 121 {
 122         struct ec_node_re_lex *priv = ec_node_priv(node);
 123         struct ec_strvec *new_vec = NULL;
 124         struct ec_pnode *child_parse;
 125         const char *str;
 126         int ret;
 127
 128         if (priv->child == NULL) {
 129                 errno = EINVAL;
 130                 goto fail;
 131         }
 132
 133         if (ec_strvec_len(strvec) == 0) {
 134                 new_vec = ec_strvec();
 135         } else {
 136                 str = ec_strvec_val(strvec, 0);
 137                 new_vec = tokenize(priv->table, priv->len, str);
 138         }
 139         if (new_vec == NULL)
 140                 goto fail;
 141
 142         ret = ec_parse_child(priv->child, pstate, new_vec);
 143         if (ret < 0)
 144                 goto fail;
 145
 146         if ((unsigned)ret == ec_strvec_len(new_vec)) {
 147                 ret = 1;
 148         } else if (ret != EC_PARSE_NOMATCH) {
 149                 child_parse = ec_pnode_get_last_child(pstate);
 150                 ec_pnode_unlink_child(pstate, child_parse);
 151                 ec_pnode_free(child_parse);
 152                 ret = EC_PARSE_NOMATCH;
 153         }
 154
 155         ec_strvec_free(new_vec);
 156         new_vec = NULL;
 157
 158         return ret;
 159
 160  fail:
 161         ec_strvec_free(new_vec);
 162         return -1;
 163 }
 164
 165 static void ec_node_re_lex_free_priv(struct ec_node *node)
 166 {
 167         struct ec_node_re_lex *priv = ec_node_priv(node);
 168         unsigned int i;
 169
 170         ec_node_free(priv->child);
 171         for (i = 0; i < priv->len; i++) {
 172                 ec_free(priv->table[i].pattern);
 173                 ec_free(priv->table[i].attr_name);
 174                 regfree(&priv->table[i].r);
 175         }
 176
 177         ec_free(priv->table);
 178 }
 179
 180 static size_t
 181 ec_node_re_lex_get_children_count(const struct ec_node *node)
 182 {
 183         struct ec_node_re_lex *priv = ec_node_priv(node);
 184
 185         if (priv->child)
 186                 return 1;
 187         return 0;
 188 }
 189
 190 static int
 191 ec_node_re_lex_get_child(const struct ec_node *node, size_t i,
 192                         struct ec_node **child, unsigned int *refs)
 193 {
 194         struct ec_node_re_lex *priv = ec_node_priv(node);
 195
 196         if (i >= 1)
 197                 return -1;
 198
 199         *child = priv->child;
 200         *refs = 2;
 201         return 0;
 202 }
 203
 204 static const struct ec_config_schema ec_node_re_lex_dict[] = {
 205         {
 206                 .key = "pattern",
 207                 .desc = "The pattern to match.",
 208                 .type = EC_CONFIG_TYPE_STRING,
 209         },
 210         {
 211                 .key = "keep",
 212                 .desc = "Whether to keep or drop the string matching "
 213                 "the regular expression.",
 214                 .type = EC_CONFIG_TYPE_BOOL,
 215         },
 216         {
 217                 .key = "attr",
 218                 .desc = "The optional attribute name to attach.",
 219                 .type = EC_CONFIG_TYPE_STRING,
 220         },
 221         {
 222                 .type = EC_CONFIG_TYPE_NONE,
 223         },
 224 };
 225
 226 static const struct ec_config_schema ec_node_re_lex_elt[] = {
 227         {
 228                 .desc = "A pattern element.",
 229                 .type = EC_CONFIG_TYPE_DICT,
 230                 .subschema = ec_node_re_lex_dict,
 231         },
 232         {
 233                 .type = EC_CONFIG_TYPE_NONE,
 234         },
 235 };
 236
 237 static const struct ec_config_schema ec_node_re_lex_schema[] = {
 238         {
 239                 .key = "patterns",
 240                 .desc = "The list of patterns elements.",
 241                 .type = EC_CONFIG_TYPE_LIST,
 242                 .subschema = ec_node_re_lex_elt,
 243         },
 244         {
 245                 .key = "child",
 246                 .desc = "The child node.",
 247                 .type = EC_CONFIG_TYPE_NODE,
 248         },
 249         {
 250                 .type = EC_CONFIG_TYPE_NONE,
 251         },
 252 };
 253
 254 static int ec_node_re_lex_set_config(struct ec_node *node,
 255                                 const struct ec_config *config)
 256 {
 257         struct ec_node_re_lex *priv = ec_node_priv(node);
 258         struct regexp_pattern *table = NULL;
 259         const struct ec_config *patterns, *child, *elt, *pattern, *keep, *attr;
 260         char *pattern_str = NULL, *attr_name = NULL;
 261         ssize_t i, n = 0;
 262         int ret;
 263
 264         child = ec_config_dict_get(config, "child");
 265         if (child == NULL)
 266                 goto fail;
 267         if (ec_config_get_type(child) != EC_CONFIG_TYPE_NODE) {
 268                 errno = EINVAL;
 269                 goto fail;
 270         }
 271
 272         patterns = ec_config_dict_get(config, "patterns");
 273         if (patterns != NULL) {
 274                 n = ec_config_count(patterns);
 275                 if (n < 0)
 276                         goto fail;
 277
 278                 table = ec_calloc(n, sizeof(*table));
 279                 if (table == NULL)
 280                         goto fail;
 281
 282                 n = 0;
 283                 TAILQ_FOREACH(elt, &patterns->list, next) {
 284                         if (ec_config_get_type(elt) != EC_CONFIG_TYPE_DICT) {
 285                                 errno = EINVAL;
 286                                 goto fail;
 287                         }
 288                         pattern = ec_config_dict_get(elt, "pattern");
 289                         if (pattern == NULL) {
 290                                 errno = EINVAL;
 291                                 goto fail;
 292                         }
 293                         if (ec_config_get_type(pattern) != EC_CONFIG_TYPE_STRING) {
 294                                 errno = EINVAL;
 295                                 goto fail;
 296                         }
 297                         keep = ec_config_dict_get(elt, "keep");
 298                         if (keep == NULL) {
 299                                 errno = EINVAL;
 300                                 goto fail;
 301                         }
 302                         if (ec_config_get_type(keep) != EC_CONFIG_TYPE_BOOL) {
 303                                 errno = EINVAL;
 304                                 goto fail;
 305                         }
 306                         attr = ec_config_dict_get(elt, "attr");
 307                         if (attr != NULL && ec_config_get_type(attr) !=
 308                                         EC_CONFIG_TYPE_STRING) {
 309                                 errno = EINVAL;
 310                                 goto fail;
 311                         }
 312                         pattern_str = ec_strdup(pattern->string);
 313                         if (pattern_str == NULL)
 314                                 goto fail;
 315                         if (attr != NULL && attr->string != NULL) {
 316                                 attr_name = ec_strdup(attr->string);
 317                                 if (attr_name == NULL)
 318                                         goto fail;
 319                         }
 320
 321                         ret = regcomp(&table[n].r, pattern_str, REG_EXTENDED);
 322                         if (ret != 0) {
 323                                 EC_LOG(EC_LOG_ERR,
 324                                         "Regular expression <%s> compilation failed: %d\n",
 325                                         pattern_str, ret);
 326                                 if (ret == REG_ESPACE)
 327                                         errno = ENOMEM;
 328                                 else
 329                                         errno = EINVAL;
 330                                 goto fail;
 331                         }
 332                         table[n].pattern = pattern_str;
 333                         table[n].keep = keep->boolean;
 334                         table[n].attr_name = attr_name;
 335                         pattern_str = NULL;
 336                         attr_name = NULL;
 337
 338                         n++;
 339                 }
 340         }
 341
 342         if (priv->child != NULL)
 343                 ec_node_free(priv->child);
 344         priv->child = ec_node_clone(child->node);
 345         for (i = 0; i < (ssize_t)priv->len; i++) {
 346                 ec_free(priv->table[i].pattern);
 347                 ec_free(priv->table[i].attr_name);
 348                 regfree(&priv->table[i].r);
 349         }
 350         ec_free(priv->table);
 351         priv->table = table;
 352         priv->len = n;
 353
 354         return 0;
 355
 356 fail:
 357         if (table != NULL) {
 358                 for (i = 0; i < n; i++) {
 359                         if (table[i].pattern != NULL) {
 360                                 ec_free(table[i].pattern);
 361                                 regfree(&table[i].r);
 362                         }
 363                 }
 364         }
 365         ec_free(table);
 366         ec_free(pattern_str);
 367         return -1;
 368 }
 369
 370 static struct ec_node_type ec_node_re_lex_type = {
 371         .name = "re_lex",
 372         .schema = ec_node_re_lex_schema,
 373         .set_config = ec_node_re_lex_set_config,
 374         .parse = ec_node_re_lex_parse,
 375         .size = sizeof(struct ec_node_re_lex),
 376         .free_priv = ec_node_re_lex_free_priv,
 377         .get_children_count = ec_node_re_lex_get_children_count,
 378         .get_child = ec_node_re_lex_get_child,
 379 };
 380
 381 EC_NODE_TYPE_REGISTER(ec_node_re_lex_type);
 382
 383 int ec_node_re_lex_add(struct ec_node *node, const char *pattern, int keep,
 384         const char *attr_name)
 385 {
 386         const struct ec_config *cur_config = NULL;
 387         struct ec_config *config = NULL, *patterns = NULL, *elt = NULL;
 388         int ret;
 389
 390         if (ec_node_check_type(node, &ec_node_re_lex_type) < 0)
 391                 goto fail;
 392
 393         elt = ec_config_dict();
 394         if (elt == NULL)
 395                 goto fail;
 396         if (ec_config_dict_set(elt, "pattern", ec_config_string(pattern)) < 0)
 397                 goto fail;
 398         if (ec_config_dict_set(elt, "keep", ec_config_bool(keep)) < 0)
 399                 goto fail;
 400         if (attr_name != NULL) {
 401                 if (ec_config_dict_set(elt, "attr",
 402                                         ec_config_string(attr_name)) < 0)
 403                         goto fail;
 404         }
 405
 406         cur_config = ec_node_get_config(node);
 407         if (cur_config == NULL)
 408                 config = ec_config_dict();
 409         else
 410                 config = ec_config_dup(cur_config);
 411         if (config == NULL)
 412                 goto fail;
 413
 414         patterns = ec_config_dict_get(config, "patterns");
 415         if (patterns == NULL) {
 416                 patterns = ec_config_list();
 417                 if (patterns == NULL)
 418                         goto fail;
 419
 420                 if (ec_config_dict_set(config, "patterns", patterns) < 0)
 421                         goto fail; /* patterns list is freed on error */
 422         }
 423
 424         if (ec_config_list_add(patterns, elt) < 0) {
 425                 elt = NULL;
 426                 goto fail;
 427         }
 428         elt = NULL;
 429
 430         ret = ec_node_set_config(node, config);
 431         config = NULL; /* freed */
 432         if (ret < 0)
 433                 goto fail;
 434
 435         return 0;
 436
 437 fail:
 438         ec_config_free(config);
 439         ec_config_free(elt);
 440         return -1;
 441 }
 442
 443 static int
 444 ec_node_re_lex_set_child(struct ec_node *node, struct ec_node *child)
 445 {
 446         const struct ec_config *cur_config = NULL;
 447         struct ec_config *config = NULL;
 448         int ret;
 449
 450         if (ec_node_check_type(node, &ec_node_re_lex_type) < 0)
 451                 goto fail;
 452
 453         cur_config = ec_node_get_config(node);
 454         if (cur_config == NULL)
 455                 config = ec_config_dict();
 456         else
 457                 config = ec_config_dup(cur_config);
 458         if (config == NULL)
 459                 goto fail;
 460
 461         if (ec_config_dict_set(config, "child", ec_config_node(child)) < 0) {
 462                 child = NULL; /* freed */
 463                 goto fail;
 464         }
 465         child = NULL; /* freed */
 466
 467         ret = ec_node_set_config(node, config);
 468         config = NULL; /* freed */
 469         if (ret < 0)
 470                 goto fail;
 471
 472         return 0;
 473
 474 fail:
 475         ec_config_free(config);
 476         ec_node_free(child);
 477         return -1;
 478 }
 479
 480 struct ec_node *ec_node_re_lex(const char *id, struct ec_node *child)
 481 {
 482         struct ec_node *node = NULL;
 483
 484         if (child == NULL)
 485                 return NULL;
 486
 487         node = ec_node_from_type(&ec_node_re_lex_type, id);
 488         if (node == NULL)
 489                 goto fail;
 490
 491         if (ec_node_re_lex_set_child(node, child) < 0) {
 492                 child = NULL; /* freed */
 493                 goto fail;
 494         }
 495
 496         return node;
 497
 498 fail:
 499         ec_node_free(node);
 500         ec_node_free(child);
 501         return NULL;
 502 }
 503
 504 /* LCOV_EXCL_START */
 505 static int ec_node_re_lex_testcase(void)
 506 {
 507         struct ec_node *node;
 508         int ret, testres = 0;
 509
 510         node = ec_node_re_lex(EC_NO_ID,
 511                 ec_node_many(EC_NO_ID,
 512                         EC_NODE_OR(EC_NO_ID,
 513                                 ec_node_str(EC_NO_ID, "foo"),
 514                                 ec_node_str(EC_NO_ID, "bar"),
 515                                 ec_node_int(EC_NO_ID, 0, 1000, 0)
 516                         ), 0, 0
 517                 )
 518         );
 519         if (node == NULL) {
 520                 EC_LOG(EC_LOG_ERR, "cannot create node\n");
 521                 return -1;
 522         }
 523
 524         ret = ec_node_re_lex_add(node, "[a-zA-Z]+", 1, NULL);
 525         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 526         ret = ec_node_re_lex_add(node, "[0-9]+", 1, NULL);
 527         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 528         ret = ec_node_re_lex_add(node, "=", 1, NULL);
 529         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 530         ret = ec_node_re_lex_add(node, "-", 1, NULL);
 531         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 532         ret = ec_node_re_lex_add(node, "\\+", 1, NULL);
 533         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 534         ret = ec_node_re_lex_add(node, "[       ]+", 0, NULL);
 535         testres |= EC_TEST_CHECK(ret == 0, "cannot add regexp");
 536         if (ret != 0) {
 537                 EC_LOG(EC_LOG_ERR, "cannot add regexp to node\n");
 538                 ec_node_free(node);
 539                 return -1;
 540         }
 541
 542         testres |= EC_TEST_CHECK_PARSE(node, 1, "  foo bar  324 bar234");
 543         testres |= EC_TEST_CHECK_PARSE(node, 1, "foo bar324");
 544         testres |= EC_TEST_CHECK_PARSE(node, 1, "");
 545         testres |= EC_TEST_CHECK_PARSE(node, -1, "foobar");
 546
 547         /* no completion */
 548         testres |= EC_TEST_CHECK_COMPLETE(node,
 549                 "", EC_VA_END,
 550                 EC_VA_END);
 551
 552         ec_node_free(node);
 553
 554         return testres;
 555 }
 556 /* LCOV_EXCL_STOP */
 557
 558 static struct ec_test ec_node_re_lex_test = {
 559         .name = "node_re_lex",
 560         .test = ec_node_re_lex_testcase,
 561 };
 562
 563 EC_TEST_REGISTER(ec_node_re_lex_test);