git.ozlabs.org Git - ccan/blob - tools/ccanlint/file_analysis.c

   1 #include "ccanlint.h"
   2 #include <ccan/talloc/talloc.h>
   3 #include <ccan/str/str.h>
   4 #include <ccan/str_talloc/str_talloc.h>
   5 #include <ccan/talloc_link/talloc_link.h>
   6 #include <ccan/hash/hash.h>
   7 #include <ccan/htable/htable_type.h>
   8 #include <ccan/grab_file/grab_file.h>
   9 #include <ccan/noerr/noerr.h>
  10 #include <ccan/foreach/foreach.h>
  11 #include <ccan/asort/asort.h>
  12 #include "../tools.h"
  13 #include <unistd.h>
  14 #include <sys/types.h>
  15 #include <sys/stat.h>
  16 #include <fcntl.h>
  17 #include <err.h>
  18 #include <errno.h>
  19 #include <dirent.h>
  20 #include <ctype.h>
  21 #include <stdarg.h>
  22 #include <assert.h>
  23
  24 const char *ccan_dir;
  25
  26 static size_t dir_hash(const char *name)
  27 {
  28         return hash(name, strlen(name), 0);
  29 }
  30
  31 static const char *manifest_name(const struct manifest *m)
  32 {
  33         return m->dir;
  34 }
  35
  36 static bool dir_cmp(const struct manifest *m, const char *dir)
  37 {
  38         return strcmp(m->dir, dir) == 0;
  39 }
  40
  41 HTABLE_DEFINE_TYPE(struct manifest, manifest_name, dir_hash, dir_cmp, manifest);
  42 static struct htable_manifest *manifests;
  43
  44 const char *get_ccan_file_contents(struct ccan_file *f)
  45 {
  46         if (!f->contents) {
  47                 f->contents = grab_file(f, f->fullname, &f->contents_size);
  48                 if (!f->contents)
  49                         err(1, "Reading file %s", f->fullname);
  50         }
  51         return f->contents;
  52 }
  53
  54 char **get_ccan_file_lines(struct ccan_file *f)
  55 {
  56         if (!f->lines)
  57                 f->lines = strsplit(f, get_ccan_file_contents(f), "\n");
  58
  59         /* FIXME: is f->num_lines necessary? */
  60         f->num_lines = talloc_array_length(f->lines) - 1;
  61         return f->lines;
  62 }
  63
  64 struct list_head *get_ccan_file_docs(struct ccan_file *f)
  65 {
  66         if (!f->doc_sections) {
  67                 get_ccan_file_lines(f);
  68                 f->doc_sections = extract_doc_sections(f->lines);
  69         }
  70         return f->doc_sections;
  71 }
  72
  73 struct ccan_file *new_ccan_file(const void *ctx, const char *dir, char *name)
  74 {
  75         struct ccan_file *f;
  76
  77         assert(dir[0] == '/');
  78
  79         f = talloc(ctx, struct ccan_file);
  80         f->lines = NULL;
  81         f->line_info = NULL;
  82         f->doc_sections = NULL;
  83         f->compiled = NULL;
  84         f->name = talloc_steal(f, name);
  85         f->fullname = talloc_asprintf(f, "%s/%s", dir, f->name);
  86         f->contents = NULL;
  87         f->cov_compiled = NULL;
  88         return f;
  89 }
  90
  91 static void add_files(struct manifest *m, const char *dir)
  92 {
  93         DIR *d;
  94         struct dirent *ent;
  95
  96         if (dir[0])
  97                 d = opendir(dir);
  98         else
  99                 d = opendir(".");
 100         if (!d)
 101                 err(1, "Opening directory %s", dir[0] ? dir : ".");
 102
 103         while ((ent = readdir(d)) != NULL) {
 104                 struct stat st;
 105                 struct ccan_file *f;
 106                 struct list_head *dest;
 107                 bool is_c_src;
 108
 109                 if (ent->d_name[0] == '.')
 110                         continue;
 111
 112                 f = new_ccan_file(m, m->dir,
 113                                   talloc_asprintf(m, "%s%s",
 114                                                   dir, ent->d_name));
 115                 if (lstat(f->name, &st) != 0)
 116                         err(1, "lstat %s", f->name);
 117
 118                 if (S_ISDIR(st.st_mode)) {
 119                         f->name = talloc_append_string(f->name, "/");
 120                         add_files(m, f->name);
 121                         continue;
 122                 }
 123                 if (!S_ISREG(st.st_mode)) {
 124                         talloc_free(f);
 125                         continue;
 126                 }
 127
 128                 if (streq(f->name, "_info")) {
 129                         m->info_file = f;
 130                         continue;
 131                 }
 132
 133                 is_c_src = strends(f->name, ".c");
 134                 if (!is_c_src && !strends(f->name, ".h")) {
 135                         dest = &m->other_files;
 136                         continue;
 137                 }
 138
 139                 if (!strchr(f->name, '/')) {
 140                         if (is_c_src)
 141                                 dest = &m->c_files;
 142                         else
 143                                 dest = &m->h_files;
 144                 } else if (strstarts(f->name, "test/")) {
 145                         if (is_c_src) {
 146                                 if (strstarts(f->name, "test/api"))
 147                                         dest = &m->api_tests;
 148                                 else if (strstarts(f->name, "test/run"))
 149                                         dest = &m->run_tests;
 150                                 else if (strstarts(f->name, "test/compile_ok"))
 151                                         dest = &m->compile_ok_tests;
 152                                 else if (strstarts(f->name, "test/compile_fail"))
 153                                         dest = &m->compile_fail_tests;
 154                                 else
 155                                         dest = &m->other_test_c_files;
 156                         } else
 157                                 dest = &m->other_test_files;
 158                 } else
 159                         dest = &m->other_files;
 160
 161                 list_add(dest, &f->list);
 162         }
 163         closedir(d);
 164 }
 165
 166 static int cmp_names(struct ccan_file *const *a, struct ccan_file *const *b,
 167                      void *unused)
 168 {
 169         return strcmp((*a)->name, (*b)->name);
 170 }
 171
 172 static void sort_files(struct list_head *list)
 173 {
 174         struct ccan_file **files = NULL, *f;
 175         unsigned int i, num;
 176
 177         num = 0;
 178         while ((f = list_top(list, struct ccan_file, list)) != NULL) {
 179                 files = talloc_realloc(NULL, files, struct ccan_file *, num+1);
 180                 files[num++] = f;
 181                 list_del(&f->list);
 182         }
 183         asort(files, num, cmp_names, NULL);
 184
 185         for (i = 0; i < num; i++)
 186                 list_add_tail(list, &files[i]->list);
 187         talloc_free(files);
 188 }
 189
 190 struct manifest *get_manifest(const void *ctx, const char *dir)
 191 {
 192         struct manifest *m;
 193         char *olddir, *canon_dir;
 194         unsigned int len;
 195         struct list_head *list;
 196
 197         if (!manifests)
 198                 manifests = htable_manifest_new();
 199
 200         olddir = talloc_getcwd(NULL);
 201         if (!olddir)
 202                 err(1, "Getting current directory");
 203
 204         if (chdir(dir) != 0)
 205                 err(1, "Failed to chdir to %s", dir);
 206
 207         canon_dir = talloc_getcwd(olddir);
 208         if (!canon_dir)
 209                 err(1, "Getting current directory");
 210
 211         m = htable_manifest_get(manifests, canon_dir);
 212         if (m)
 213                 goto done;
 214
 215         m = talloc_linked(ctx, talloc(NULL, struct manifest));
 216         m->info_file = NULL;
 217         m->compiled = NULL;
 218         m->dir = talloc_steal(m, canon_dir);
 219         list_head_init(&m->c_files);
 220         list_head_init(&m->h_files);
 221         list_head_init(&m->api_tests);
 222         list_head_init(&m->run_tests);
 223         list_head_init(&m->compile_ok_tests);
 224         list_head_init(&m->compile_fail_tests);
 225         list_head_init(&m->other_test_c_files);
 226         list_head_init(&m->other_test_files);
 227         list_head_init(&m->other_files);
 228         list_head_init(&m->examples);
 229         list_head_init(&m->mangled_examples);
 230         list_head_init(&m->deps);
 231
 232         len = strlen(m->dir);
 233         while (len && m->dir[len-1] == '/')
 234                 m->dir[--len] = '\0';
 235
 236         m->basename = strrchr(m->dir, '/');
 237         if (!m->basename)
 238                 errx(1, "I don't expect to be run from the root directory");
 239         m->basename++;
 240
 241         /* We expect the ccan dir to be two levels above module dir. */
 242         if (!ccan_dir) {
 243                 char *p;
 244                 ccan_dir = talloc_strdup(NULL, m->dir);
 245                 p = strrchr(ccan_dir, '/');
 246                 *p = '\0';
 247                 p = strrchr(ccan_dir, '/');
 248                 *p = '\0';
 249         }
 250
 251         add_files(m, "");
 252
 253         /* Nicer to run tests in a predictable order. */
 254         foreach_ptr(list, &m->api_tests, &m->run_tests, &m->compile_ok_tests,
 255                     &m->compile_fail_tests)
 256                 sort_files(list);
 257
 258         htable_manifest_add(manifests, m);
 259
 260 done:
 261         if (chdir(olddir) != 0)
 262                 err(1, "Returning to original directory '%s'", olddir);
 263         talloc_free(olddir);
 264
 265         return m;
 266 }
 267
 268
 269 /**
 270  * remove_comments - strip comments from a line, return copy.
 271  * @line: line to copy
 272  * @in_comment: are we already within a comment (from prev line).
 273  * @unterminated: are we still in a comment for next line.
 274  */
 275 static char *remove_comments(const char *line, bool in_comment,
 276                              bool *unterminated)
 277 {
 278         char *p, *ret = talloc_array(line, char, strlen(line) + 1);
 279
 280         p = ret;
 281         for (;;) {
 282                 if (!in_comment) {
 283                         /* Find first comment. */
 284                         const char *old_comment = strstr(line, "/*");
 285                         const char *new_comment = strstr(line, "//");
 286                         const char *comment;
 287
 288                         if (new_comment && old_comment)
 289                                 comment = new_comment < old_comment
 290                                         ? new_comment : old_comment;
 291                         else if (old_comment)
 292                                 comment = old_comment;
 293                         else if (new_comment)
 294                                 comment = new_comment;
 295                         else {
 296                                 /* Nothing more. */
 297                                 strcpy(p, line);
 298                                 *unterminated = false;
 299                                 break;
 300                         }
 301
 302                         /* Copy up to comment. */
 303                         memcpy(p, line, comment - line);
 304                         p += comment - line;
 305                         line += comment - line + 2;
 306
 307                         if (comment == new_comment) {
 308                                 /* We're done: goes to EOL. */
 309                                 p[0] = '\0';
 310                                 *unterminated = false;
 311                                 break;
 312                         }
 313                         in_comment = true;
 314                 }
 315
 316                 if (in_comment) {
 317                         const char *end = strstr(line, "*/");
 318                         if (!end) {
 319                                 *unterminated = true;
 320                                 p[0] = '\0';
 321                                 break;
 322                         }
 323                         line = end+2;
 324                         in_comment = false;
 325                 }
 326         }
 327         return ret;
 328 }
 329
 330 static bool is_empty(const char *line)
 331 {
 332         return strspn(line, " \t") == strlen(line);
 333 }
 334
 335 static bool continues(const char *line)
 336 {
 337         /* Technically, any odd number of these.  But who cares? */
 338         return strends(line, "\\");
 339 }
 340
 341 /* Get token if it's equal to token. */
 342 bool get_token(const char **line, const char *token)
 343 {
 344         unsigned int toklen;
 345
 346         *line += strspn(*line, " \t");
 347         if (isalnum(token[0]) || token[0] == '_')
 348                 toklen = strspn(*line, IDENT_CHARS);
 349         else {
 350                 /* FIXME: real tokenizer handles ++ and other multi-chars.  */
 351                 toklen = strlen(token);
 352         }
 353
 354         if (toklen == strlen(token) && !strncmp(*line, token, toklen)) {
 355                 *line += toklen;
 356                 return true;
 357         }
 358         return false;
 359 }
 360
 361 char *get_symbol_token(void *ctx, const char **line)
 362 {
 363         unsigned int toklen;
 364         char *ret;
 365
 366         *line += strspn(*line, " \t");
 367         toklen = strspn(*line, IDENT_CHARS);
 368         if (!toklen)
 369                 return NULL;
 370         ret = talloc_strndup(ctx, *line, toklen);
 371         *line += toklen;
 372         return ret;
 373 }
 374
 375 static bool parse_hash_if(struct pp_conditions *cond, const char **line)
 376 {
 377         bool brackets, defined;
 378
 379         cond->inverse = get_token(line, "!");
 380         defined = get_token(line, "defined");
 381         brackets = get_token(line, "(");
 382         cond->symbol = get_symbol_token(cond, line);
 383         if (!cond->symbol)
 384                 return false;
 385         if (brackets && !get_token(line, ")"))
 386                 return false;
 387         if (!defined)
 388                 cond->type = PP_COND_IF;
 389         return true;
 390 }
 391
 392 /* FIXME: Get serious! */
 393 static struct pp_conditions *analyze_directive(struct ccan_file *f,
 394                                                const char *line,
 395                                                struct pp_conditions *parent)
 396 {
 397         struct pp_conditions *cond = talloc(f, struct pp_conditions);
 398         bool unused;
 399
 400         line = remove_comments(line, false, &unused);
 401
 402         cond->parent = parent;
 403         cond->type = PP_COND_IFDEF;
 404
 405         if (!get_token(&line, "#"))
 406                 abort();
 407
 408         if (get_token(&line, "if")) {
 409                 if (!parse_hash_if(cond, &line))
 410                         goto unknown;
 411         } else if (get_token(&line, "elif")) {
 412                 /* Malformed? */
 413                 if (!parent)
 414                         return NULL;
 415                 cond->parent = parent->parent;
 416                 /* FIXME: Not quite true.  This implies !parent, but we don't
 417                  * do multiple conditionals yet. */
 418                 if (!parse_hash_if(cond, &line))
 419                         goto unknown;
 420         } else if (get_token(&line, "ifdef")) {
 421                 bool brackets;
 422                 cond->inverse = false;
 423                 brackets = get_token(&line, "(");
 424                 cond->symbol = get_symbol_token(cond, &line);
 425                 if (!cond->symbol)
 426                         goto unknown;
 427                 if (brackets && !get_token(&line, ")"))
 428                         goto unknown;
 429         } else if (get_token(&line, "ifndef")) {
 430                 bool brackets;
 431                 cond->inverse = true;
 432                 brackets = get_token(&line, "(");
 433                 cond->symbol = get_symbol_token(cond, &line);
 434                 if (!cond->symbol)
 435                         goto unknown;
 436                 if (brackets && !get_token(&line, ")"))
 437                         goto unknown;
 438         } else if (get_token(&line, "else")) {
 439                 /* Malformed? */
 440                 if (!parent)
 441                         return NULL;
 442
 443                 *cond = *parent;
 444                 cond->inverse = !cond->inverse;
 445                 return cond;
 446         } else if (get_token(&line, "endif")) {
 447                 talloc_free(cond);
 448                 /* Malformed? */
 449                 if (!parent)
 450                         return NULL;
 451                 /* Back up one! */
 452                 return parent->parent;
 453         } else {
 454                 /* Not a conditional. */
 455                 talloc_free(cond);
 456                 return parent;
 457         }
 458
 459         if (!is_empty(line))
 460                 goto unknown;
 461         return cond;
 462
 463 unknown:
 464         cond->type = PP_COND_UNKNOWN;
 465         return cond;
 466 }
 467
 468 /* This parser is rough, but OK if code is reasonably neat. */
 469 struct line_info *get_ccan_line_info(struct ccan_file *f)
 470 {
 471         bool continued = false, in_comment = false;
 472         struct pp_conditions *cond = NULL;
 473         unsigned int i;
 474
 475         if (f->line_info)
 476                 return f->line_info;
 477
 478         get_ccan_file_lines(f);
 479         f->line_info = talloc_array(f->lines, struct line_info, f->num_lines);
 480
 481         for (i = 0; i < f->num_lines; continued = continues(f->lines[i++])) {
 482                 char *p;
 483                 bool still_doc_line;
 484
 485                 /* Current conditions apply to this line. */
 486                 f->line_info[i].cond = cond;
 487                 f->line_info[i].continued = continued;
 488
 489                 if (continued) {
 490                         /* Same as last line. */
 491                         f->line_info[i].type = f->line_info[i-1].type;
 492                         /* Update in_comment. */
 493                         remove_comments(f->lines[i], in_comment, &in_comment);
 494                         continue;
 495                 }
 496
 497                 /* Preprocessor directive? */
 498                 if (!in_comment
 499                     && f->lines[i][strspn(f->lines[i], " \t")] == '#') {
 500                         f->line_info[i].type = PREPROC_LINE;
 501                         cond = analyze_directive(f, f->lines[i], cond);
 502                         continue;
 503                 }
 504
 505                 still_doc_line = (in_comment
 506                                   && f->line_info[i-1].type == DOC_LINE);
 507
 508                 p = remove_comments(f->lines[i], in_comment, &in_comment);
 509                 if (is_empty(p)) {
 510                         if (strstarts(f->lines[i], "/**") || still_doc_line)
 511                                 f->line_info[i].type = DOC_LINE;
 512                         else
 513                                 f->line_info[i].type = COMMENT_LINE;
 514                 } else
 515                         f->line_info[i].type = CODE_LINE;
 516                 talloc_free(p);
 517         }
 518         return f->line_info;
 519 }
 520
 521 struct symbol {
 522         struct list_node list;
 523         const char *name;
 524         const unsigned int *value;
 525 };
 526
 527 static struct symbol *find_symbol(struct list_head *syms, const char *sym)
 528 {
 529         struct symbol *i;
 530
 531         list_for_each(syms, i, list)
 532                 if (streq(sym, i->name))
 533                         return i;
 534         return NULL;
 535 }
 536
 537 static enum line_compiled get_pp(struct pp_conditions *cond,
 538                                  struct list_head *syms)
 539 {
 540         struct symbol *sym;
 541         unsigned int val;
 542         enum line_compiled parent, ret;
 543
 544         /* No conditions?  Easy. */
 545         if (!cond)
 546                 return COMPILED;
 547
 548         /* Check we get here at all. */
 549         parent = get_pp(cond->parent, syms);
 550         if (parent == NOT_COMPILED)
 551                 return NOT_COMPILED;
 552
 553         if (cond->type == PP_COND_UNKNOWN)
 554                 return MAYBE_COMPILED;
 555
 556         sym = find_symbol(syms, cond->symbol);
 557         if (!sym)
 558                 return MAYBE_COMPILED;
 559
 560         switch (cond->type) {
 561         case PP_COND_IF:
 562                 /* Undefined is 0. */
 563                 val = sym->value ? *sym->value : 0;
 564                 if (!val == cond->inverse)
 565                         ret = COMPILED;
 566                 else
 567                         ret = NOT_COMPILED;
 568                 break;
 569
 570         case PP_COND_IFDEF:
 571                 if (cond->inverse == !sym->value)
 572                         ret = COMPILED;
 573                 else
 574                         ret = NOT_COMPILED;
 575                 break;
 576
 577         default:
 578                 abort();
 579         }
 580
 581         /* If parent didn't know, NO == NO, but YES == MAYBE. */
 582         if (parent == MAYBE_COMPILED && ret == COMPILED)
 583                 ret = MAYBE_COMPILED;
 584         return ret;
 585 }
 586
 587 static void add_symbol(struct list_head *head,
 588                        const char *symbol, const unsigned int *value)
 589 {
 590         struct symbol *sym = talloc(head, struct symbol);
 591         sym->name = symbol;
 592         sym->value = value;
 593         list_add(head, &sym->list);
 594 }
 595
 596 enum line_compiled get_ccan_line_pp(struct pp_conditions *cond,
 597                                     const char *symbol,
 598                                     const unsigned int *value,
 599                                     ...)
 600 {
 601         enum line_compiled ret;
 602         struct list_head *head;
 603         va_list ap;
 604
 605         head = talloc(NULL, struct list_head);
 606         list_head_init(head);
 607
 608         va_start(ap, value);
 609         add_symbol(head, symbol, value);
 610
 611         while ((symbol = va_arg(ap, const char *)) != NULL) {
 612                 value = va_arg(ap, const unsigned int *);
 613                 add_symbol(head, symbol, value);
 614         }
 615         ret = get_pp(cond, head);
 616         talloc_free(head);
 617         return ret;
 618 }
 619
 620 void score_file_error(struct score *score, struct ccan_file *f, unsigned line,
 621                       const char *errorfmt, ...)
 622 {
 623         va_list ap;
 624
 625         struct file_error *fe = talloc(score, struct file_error);
 626         fe->file = f;
 627         fe->line = line;
 628         list_add_tail(&score->per_file_errors, &fe->list);
 629
 630         if (!score->error)
 631                 score->error = talloc_strdup(score, "");
 632
 633         if (verbose < 2 && strcount(score->error, "\n") > 5)
 634                 return;
 635
 636         if (line)
 637                 score->error = talloc_asprintf_append(score->error,
 638                                                       "%s:%u:",
 639                                                       f->fullname, line);
 640         else
 641                 score->error = talloc_asprintf_append(score->error,
 642                                                       "%s:", f->fullname);
 643
 644         va_start(ap, errorfmt);
 645         score->error = talloc_vasprintf_append(score->error, errorfmt, ap);
 646         va_end(ap);
 647         score->error = talloc_append_string(score->error, "\n");
 648
 649         if (verbose < 2 && strcount(score->error, "\n") > 5)
 650                 score->error = talloc_append_string(score->error,
 651                                     "... more (use -vv to see them all)\n");
 652 }