git.ozlabs.org Git - ccan/blob - ccan/cdump/cdump.c

   1 /* MIT (BSD) license - see LICENSE file for details */
   2 #include "cdump.h"
   3 #include <ccan/tal/str/str.h>
   4 #include <assert.h>
   5
   6 struct token {
   7         const char *p;
   8         size_t len;
   9 };
  10
  11 static void add_token(struct token **toks, const char *p, size_t len)
  12 {
  13         size_t n = tal_count(*toks);
  14         tal_resize(toks, n+1);
  15         (*toks)[n].p = p;
  16         (*toks)[n].len = len;
  17 }
  18
  19 /* Simplified tokenizer: comments and preproc directives removed,
  20    identifiers are a token, others are single char tokens. */
  21 static struct token *tokenize(const void *ctx, const char *code)
  22 {
  23         unsigned int i, len, tok_start = -1;
  24         bool start_of_line = true;
  25         struct token *toks = tal_arr(ctx, struct token, 0);
  26
  27         for (i = 0; code[i]; i += len) {
  28                 if (code[i] == '#' && start_of_line) {
  29                         /* Preprocessor line. */
  30                         len = strcspn(code+i, "\n");
  31                 } else if (code[i] == '/' && code[i+1] == '/') {
  32                         /* One line comment. */
  33                         len = strcspn(code+i, "\n");
  34                         if (tok_start != -1U) {
  35                                 add_token(&toks, code+tok_start, i - tok_start);
  36                                 tok_start = -1U;
  37                         }
  38                 } else if (code[i] == '/' && code[i+1] == '*') {
  39                         /* Multi-line comment. */
  40                         const char *end = strstr(code+i+2, "*/");
  41                         len = (end + 2) - (code + i);
  42                         if (!end)
  43                                 len = strlen(code + i);
  44                         if (tok_start != -1U) {
  45                                 add_token(&toks, code+tok_start, i - tok_start);
  46                                 tok_start = -1U;
  47                         }
  48                 } else if (cisalnum(code[i]) || code[i] == '_') {
  49                         /* Identifier or part thereof */
  50                         if (tok_start == -1U)
  51                                 tok_start = i;
  52                         len = 1;
  53                 } else if (!cisspace(code[i])) {
  54                         /* Punctuation: treat as single char token. */
  55                         if (tok_start != -1U) {
  56                                 add_token(&toks, code+tok_start, i - tok_start);
  57                                 tok_start = -1U;
  58                         }
  59                         add_token(&toks, code+i, 1);
  60                         len = 1;
  61                 } else {
  62                         /* Whitespace. */
  63                         if (tok_start != -1U) {
  64                                 add_token(&toks, code+tok_start, i - tok_start);
  65                                 tok_start = -1U;
  66                         }
  67                         len = 1;
  68                 }
  69                 if (code[i] == '\n')
  70                         start_of_line = true;
  71                 else if (!cisspace(code[i]))
  72                         start_of_line = false;
  73         }
  74
  75         /* Add terminating NULL. */
  76         tal_resizez(&toks, tal_count(toks) + 1);
  77         return toks;
  78 }
  79
  80 struct parse_state {
  81         const char *code;
  82         const struct token *toks;
  83         struct cdump_definitions *defs;
  84         char *complaints;
  85 };
  86
  87 static const struct token *tok_peek(const struct token **toks)
  88 {
  89         /* Ignore removed tokens (eg. comments) */
  90         while (toks[0]->len == 0) {
  91                 if (!toks[0]->p)
  92                         return NULL;
  93                 (*toks)++;
  94         }
  95         return toks[0];
  96 }
  97
  98 static bool tok_is(const struct token **toks, const char *target)
  99 {
 100         const struct token *t = tok_peek(toks);
 101         return (t && t->len == strlen(target)
 102                 && memcmp(t->p, target, t->len) == 0);
 103 }
 104
 105 static const struct token *tok_take(const struct token **toks)
 106 {
 107         const struct token *t = tok_peek(toks);
 108         if (t)
 109                 (*toks)++;
 110
 111         return t;
 112 }
 113
 114 static const struct token *tok_take_if(const struct token **toks,
 115                                        const char *target)
 116 {
 117         if (tok_is(toks, target))
 118                 return tok_take(toks);
 119         return NULL;
 120 }
 121
 122 static const char *tok_take_ident(const tal_t *ctx, const struct token **toks)
 123 {
 124         const struct token *t = tok_peek(toks);
 125
 126         if (!t)
 127                 return NULL;
 128
 129         if (strspn(t->p, "_0123456789"
 130                    "abcdefghijklmnopqrstuvwxyz"
 131                    "ABCDEFGHIJKLMNOPQRSTUVWXYZ") < t->len)
 132                 return NULL;
 133
 134         t = tok_take(toks);
 135         return tal_strndup(ctx, t->p, t->len);
 136 }
 137
 138 static char *string_of_toks(const tal_t *ctx,
 139                             const struct token *first,
 140                             const struct token *until)
 141 {
 142         char *str, *p;
 143
 144         /* Careful to skip erased tokens (eg. comments) */
 145         str = p = tal_arr(ctx, char, until->p - first->p + 1);
 146         while (first != until) {
 147                 const struct token *next = first + 1;
 148
 149                 if (first->len) {
 150                         memcpy(p, first->p, first->len);
 151                         p += first->len;
 152                         /* Insert space if they weren't adjacent, unless last */
 153                         if (next != until) {
 154                                 if (first->p + first->len != next->p)
 155                                         *(p++) = ' ';
 156                         }
 157                 }
 158                 first = next;
 159         }
 160         *p = '\0';
 161
 162         return str;
 163 }
 164
 165 static char *tok_take_until(const tal_t *ctx,
 166                             const struct token **toks,
 167                             const char *delims)
 168 {
 169         const struct token *t, *start;
 170
 171         start = tok_peek(toks);
 172         while ((t = tok_peek(toks)) != NULL) {
 173                 /* If this contains a delimiter, copy up to prev token. */
 174                 if (strcspn(t->p, delims) < t->len)
 175                         return string_of_toks(ctx, start, t);
 176                 tok_take(toks);
 177         };
 178
 179         /* EOF without finding delimiter */
 180         return NULL;
 181 }
 182
 183 static bool type_defined(const struct cdump_type *t)
 184 {
 185         switch (t->kind) {
 186         case CDUMP_STRUCT:
 187         case CDUMP_UNION:
 188                 return (t->u.members != NULL);
 189         case CDUMP_ENUM:
 190                 return (t->u.enum_vals != NULL);
 191
 192         /* These shouldn't happen; we don't try to define them. */
 193         case CDUMP_UNKNOWN:
 194         case CDUMP_ARRAY:
 195         case CDUMP_POINTER:
 196                 break;
 197         }
 198         abort();
 199 }
 200
 201 /* May allocate a new type if not already found (steals @name) */
 202 static struct cdump_type *get_type(struct cdump_definitions *defs,
 203                                    enum cdump_type_kind kind,
 204                                    const char *name)
 205 {
 206         struct cdump_map *m;
 207         struct cdump_type *t;
 208
 209         switch (kind) {
 210         case CDUMP_STRUCT:
 211                 m = &defs->structs;
 212                 break;
 213         case CDUMP_UNION:
 214                 m = &defs->unions;
 215                 break;
 216         case CDUMP_ENUM:
 217                 m = &defs->enums;
 218                 break;
 219         case CDUMP_UNKNOWN:
 220         case CDUMP_ARRAY:
 221         case CDUMP_POINTER:
 222                 m = NULL;
 223         }
 224
 225         /* Do we already have it? */
 226         if (m) {
 227                 t = strmap_get(m, name);
 228                 if (t)
 229                         return t;
 230         }
 231
 232         t = tal(defs, struct cdump_type);
 233         t->kind = kind;
 234         t->name = name ? tal_steal(t, name) : NULL;
 235         /* These are actually the same, but be thorough */
 236         t->u.members = NULL;
 237         t->u.enum_vals = NULL;
 238         if (m)
 239                 strmap_add(m, t->name, t);
 240
 241         return t;
 242 }
 243
 244 static void complain(struct parse_state *ps, const char *complaint)
 245 {
 246         unsigned int linenum;
 247         const char *p = ps->code;
 248
 249         for (linenum = 1; p < ps->toks[0].p; linenum++) {
 250                 p = strchr(p+1, '\n');
 251                 if (!p)
 252                         break;
 253         }
 254
 255         tal_append_fmt(&ps->complaints,
 256                        "Line %u: '%.*s': %s\n",
 257                        linenum, (int)ps->toks[0].len,
 258                        ps->toks[0].p, complaint);
 259 }
 260
 261 static void tok_take_unknown_statement(struct parse_state *ps)
 262 {
 263         complain(ps, "Ignoring unknown statement until next semicolon");
 264         tal_free(tok_take_until(NULL, &ps->toks, ";"));
 265         tok_take_if(&ps->toks, ";");
 266 }
 267
 268 static bool tok_take_expr(struct parse_state *ps, const char *term)
 269 {
 270         while (!tok_is(&ps->toks, term)) {
 271                 if (tok_take_if(&ps->toks, "(")) {
 272                         if (!tok_take_expr(ps, ")"))
 273                                 return false;
 274                 } else if (tok_take_if(&ps->toks, "[")) {
 275                         if (!tok_take_expr(ps, "]"))
 276                                 return false;
 277                 } else if (!tok_take(&ps->toks))
 278                         return false;
 279         }
 280         return tok_take(&ps->toks);
 281 }
 282
 283 /* [ ... */
 284 static bool tok_take_array(struct parse_state *ps, struct cdump_type **type)
 285 {
 286         /* This will be some arbitrary expression! */
 287         struct cdump_type *arr = get_type(ps->defs, CDUMP_ARRAY, NULL);
 288         const struct token *start = tok_peek(&ps->toks);
 289
 290         if (!tok_take_expr(ps, "]")) {
 291                 complain(ps, "Could not find closing array size ]");
 292                 return false;
 293         }
 294
 295         arr->u.arr.size = string_of_toks(arr, start, ps->toks - 1);
 296         arr->u.arr.type = *type;
 297         *type = arr;
 298
 299         return true;
 300 }
 301
 302 static struct cdump_type *ptr_of(struct parse_state *ps,
 303                                  const struct cdump_type *ptr_to)
 304 {
 305         struct cdump_type *ptr = get_type(ps->defs, CDUMP_POINTER, NULL);
 306         ptr->u.ptr = ptr_to;
 307         return ptr;
 308 }
 309
 310 static bool tok_take_type(struct parse_state *ps, struct cdump_type **type)
 311 {
 312         const char *name;
 313         const struct token *types;
 314         enum cdump_type_kind kind;
 315
 316         /* Ignoring weird typedefs, only these can be combined. */
 317         types = ps->toks;
 318         while (tok_take_if(&ps->toks, "int")
 319                || tok_take_if(&ps->toks, "long")
 320                || tok_take_if(&ps->toks, "short")
 321                || tok_take_if(&ps->toks, "double")
 322                || tok_take_if(&ps->toks, "float")
 323                || tok_take_if(&ps->toks, "char")
 324                || tok_take_if(&ps->toks, "signed")
 325                || tok_take_if(&ps->toks, "unsigned"));
 326
 327         /* Did we get some? */
 328         if (ps->toks != types) {
 329                 name = string_of_toks(NULL, types, tok_peek(&ps->toks));
 330                 kind = CDUMP_UNKNOWN;
 331         } else {
 332                 /* Try normal types (or simple typedefs, etc). */
 333                 if (tok_take_if(&ps->toks, "struct")) {
 334                         kind = CDUMP_STRUCT;
 335                 } else if (tok_take_if(&ps->toks, "union")) {
 336                         kind = CDUMP_UNION;
 337                 } else if (tok_take_if(&ps->toks, "enum")) {
 338                         kind = CDUMP_ENUM;
 339                 } else
 340                         kind = CDUMP_UNKNOWN;
 341
 342                 name = tok_take_ident(ps->defs, &ps->toks);
 343                 if (!name) {
 344                         complain(ps, "Invalid typename");
 345                         return false;
 346                 }
 347         }
 348
 349         *type = get_type(ps->defs, kind, name);
 350         return true;
 351 }
 352
 353 /* struct|union ... */
 354 static bool tok_take_conglom(struct parse_state *ps,
 355                              enum cdump_type_kind conglom_kind)
 356 {
 357         struct cdump_type *e;
 358         const char *name;
 359         size_t n;
 360
 361         assert(conglom_kind == CDUMP_STRUCT || conglom_kind == CDUMP_UNION);
 362
 363         name = tok_take_ident(ps->defs, &ps->toks);
 364         if (!name) {
 365                 complain(ps, "Invalid struct/union name");
 366                 return false;
 367         }
 368
 369         e = get_type(ps->defs, conglom_kind, name);
 370         if (type_defined(e)) {
 371                 complain(ps, "Type already defined");
 372                 return false;
 373         }
 374
 375         if (!tok_take_if(&ps->toks, "{")) {
 376                 complain(ps, "Expected { for struct/union");
 377                 return false;
 378         }
 379
 380         e->u.members = tal_arr(e, struct cdump_member, n = 0);
 381         while (!tok_is(&ps->toks, "}")) {
 382                 struct cdump_type *basetype;
 383                 const struct token *quals;
 384                 unsigned int num_quals = 0;
 385
 386                 /* Anything can have these prepended. */
 387                 quals = ps->toks;
 388                 while (tok_take_if(&ps->toks, "const")
 389                        || tok_take_if(&ps->toks, "volatile"))
 390                         num_quals++;
 391
 392                 /* eg. "struct foo" or "varint_t" */
 393                 if (!tok_take_type(ps, &basetype)) {
 394                         complain(ps, "Expected typename inside struct/union");
 395                         return false;
 396                 }
 397
 398                 do {
 399                         struct cdump_member *m;
 400
 401                         tal_resize(&e->u.members, n+1);
 402                         m = &e->u.members[n++];
 403                         m->type = basetype;
 404                         if (num_quals) {
 405                                 m->qualifiers
 406                                         = string_of_toks(e, quals,
 407                                                          quals + num_quals);
 408                         } else
 409                                 m->qualifiers = NULL;
 410
 411                         /* May have multiple asterisks. */
 412                         while (tok_take_if(&ps->toks, "*"))
 413                                 m->type = ptr_of(ps, m->type);
 414
 415                         m->name = tok_take_ident(e, &ps->toks);
 416                         if (!m->name) {
 417                                 complain(ps, "Expected name for member");
 418                                 return false;
 419                         }
 420
 421                         /* May be an array. */
 422                         while (tok_take_if(&ps->toks, "[")) {
 423                                 if (!tok_take_array(ps, &m->type))
 424                                         return false;
 425                         }
 426                 } while (tok_take_if(&ps->toks, ","));
 427
 428                 if (!tok_take_if(&ps->toks, ";")) {
 429                         complain(ps, "Expected ; at end of member");
 430                         return false;
 431                 }
 432         }
 433
 434         if (tok_take_if(&ps->toks, "}") && tok_take_if(&ps->toks, ";"))
 435                 return true;
 436         complain(ps, "Expected }; at end of struct/union");
 437         return false;
 438 }
 439
 440 /* enum ... */
 441 static bool tok_take_enum(struct parse_state *ps)
 442 {
 443         size_t n = 0;
 444         struct cdump_type *e;
 445         const char *name;
 446
 447         name = tok_take_ident(ps->defs, &ps->toks);
 448         if (!name) {
 449                 complain(ps, "Expected enum name");
 450                 return false;
 451         }
 452
 453         e = get_type(ps->defs, CDUMP_ENUM, name);
 454
 455         /* Duplicate name? */
 456         if (type_defined(e)) {
 457                 complain(ps, "enum already defined");
 458                 return false;
 459         }
 460
 461         if (!tok_take_if(&ps->toks, "{")) {
 462                 complain(ps, "Expected { after enum name");
 463                 return false;
 464         }
 465
 466         e->u.enum_vals = tal_arr(e, struct cdump_enum_val, n);
 467         do {
 468                 struct cdump_enum_val *v;
 469
 470                 /* GCC extension: comma and end of enum */
 471                 if (tok_is(&ps->toks, "}"))
 472                         break;
 473
 474                 tal_resize(&e->u.enum_vals, n+1);
 475                 v = &e->u.enum_vals[n++];
 476
 477                 v->name = tok_take_ident(e, &ps->toks);
 478                 if (!v->name) {
 479                         complain(ps, "Expected enum value name");
 480                         return false;
 481                 }
 482                 if (tok_take_if(&ps->toks, "=")) {
 483                         v->value = tok_take_until(e, &ps->toks, ",}");
 484                         if (!v->value) {
 485                                 complain(ps, "Expected , or } to end value");
 486                                 return false;
 487                         }
 488                 } else
 489                         v->value = NULL;
 490         } while (tok_take_if(&ps->toks, ","));
 491
 492         if (tok_take_if(&ps->toks, "}") && tok_take_if(&ps->toks, ";"))
 493                 return true;
 494
 495         complain(ps, "Expected }; at end of enum");
 496         return false;
 497 }
 498
 499 static bool gather_undefines(const char *name,
 500                              struct cdump_type *t,
 501                              struct cdump_map *undefs)
 502 {
 503         if (!type_defined(t))
 504                 strmap_add(undefs, name, t);
 505         return true;
 506 }
 507
 508 static bool remove_from_map(const char *name,
 509                             struct cdump_type *t,
 510                             struct cdump_map *map)
 511 {
 512         strmap_del(map, name, NULL);
 513         return true;
 514 }
 515
 516 static void remove_undefined(struct cdump_map *map)
 517 {
 518         struct cdump_map undefs;
 519
 520         /* We can't delete inside iterator, so gather all the undefs
 521          * then remove them. */
 522         strmap_init(&undefs);
 523
 524         strmap_iterate(map, gather_undefines, &undefs);
 525         strmap_iterate(&undefs, remove_from_map, map);
 526         strmap_clear(&undefs);
 527 }
 528
 529 static void destroy_definitions(struct cdump_definitions *defs)
 530 {
 531         strmap_clear(&defs->enums);
 532         strmap_clear(&defs->structs);
 533         strmap_clear(&defs->unions);
 534 }
 535
 536 /* Simple LL(1) parser, inspired by Tridge's genstruct.pl. */
 537 struct cdump_definitions *cdump_extract(const tal_t *ctx, const char *code,
 538                                         char **complaints)
 539 {
 540         struct parse_state ps;
 541         const struct token *toks;
 542
 543         ps.defs = tal(ctx, struct cdump_definitions);
 544         ps.complaints = tal_strdup(ctx, "");
 545         ps.code = code;
 546
 547         strmap_init(&ps.defs->enums);
 548         strmap_init(&ps.defs->structs);
 549         strmap_init(&ps.defs->unions);
 550         tal_add_destructor(ps.defs, destroy_definitions);
 551
 552         toks = ps.toks = tokenize(ps.defs, code);
 553         while (tok_peek(&ps.toks)) {
 554                 if (tok_take_if(&ps.toks, "struct")) {
 555                         if (!tok_take_conglom(&ps, CDUMP_STRUCT))
 556                                 goto fail;
 557                 } else if (tok_take_if(&ps.toks, "union")) {
 558                         if (!tok_take_conglom(&ps, CDUMP_UNION))
 559                                 goto fail;
 560                 } else if (tok_take_if(&ps.toks, "enum")) {
 561                         if (!tok_take_enum(&ps))
 562                                 goto fail;
 563                 } else
 564                         tok_take_unknown_statement(&ps);
 565         }
 566
 567         /* Now, remove any undefined types! */
 568         remove_undefined(&ps.defs->enums);
 569         remove_undefined(&ps.defs->structs);
 570         remove_undefined(&ps.defs->unions);
 571         tal_free(toks);
 572
 573 out:
 574         if (streq(ps.complaints, ""))
 575                 ps.complaints = tal_free(ps.complaints);
 576
 577         if (complaints)
 578                 *complaints = ps.complaints;
 579         else
 580                 tal_free(ps.complaints);
 581         return ps.defs;
 582
 583 fail:
 584         ps.defs = tal_free(ps.defs);
 585         goto out;
 586 }