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

   1 #include "config.h"
   2 #include "ccanlint.h"
   3 #include <ccan/talloc/talloc.h>
   4 #include <ccan/str/str.h>
   5 #include <ccan/str_talloc/str_talloc.h>
   6 #include <ccan/talloc_link/talloc_link.h>
   7 #include <ccan/hash/hash.h>
   8 #include <ccan/htable/htable_type.h>
   9 #include <ccan/grab_file/grab_file.h>
  10 #include <ccan/noerr/noerr.h>
  11 #include <ccan/foreach/foreach.h>
  12 #include <ccan/asort/asort.h>
  13 #include "../tools.h"
  14 #include <unistd.h>
  15 #include <sys/types.h>
  16 #include <sys/stat.h>
  17 #include <fcntl.h>
  18 #include <err.h>
  19 #include <errno.h>
  20 #include <dirent.h>
  21 #include <ctype.h>
  22 #include <stdarg.h>
  23 #include <assert.h>
  24
  25 const char *ccan_dir;
  26
  27 static size_t dir_hash(const char *name)
  28 {
  29         return hash(name, strlen(name), 0);
  30 }
  31
  32 static const char *manifest_name(const struct manifest *m)
  33 {
  34         return m->dir;
  35 }
  36
  37 static bool dir_cmp(const struct manifest *m, const char *dir)
  38 {
  39         return strcmp(m->dir, dir) == 0;
  40 }
  41
  42 HTABLE_DEFINE_TYPE(struct manifest, manifest_name, dir_hash, dir_cmp, manifest);
  43 static struct htable_manifest *manifests;
  44
  45 const char *get_ccan_file_contents(struct ccan_file *f)
  46 {
  47         if (!f->contents) {
  48                 f->contents = grab_file(f, f->fullname, &f->contents_size);
  49                 if (!f->contents)
  50                         err(1, "Reading file %s", f->fullname);
  51         }
  52         return f->contents;
  53 }
  54
  55 char **get_ccan_file_lines(struct ccan_file *f)
  56 {
  57         if (!f->lines)
  58                 f->lines = strsplit(f, get_ccan_file_contents(f), "\n");
  59
  60         /* FIXME: is f->num_lines necessary? */
  61         f->num_lines = talloc_array_length(f->lines) - 1;
  62         return f->lines;
  63 }
  64
  65 struct list_head *get_ccan_file_docs(struct ccan_file *f)
  66 {
  67         if (!f->doc_sections) {
  68                 get_ccan_file_lines(f);
  69                 f->doc_sections = extract_doc_sections(f->lines);
  70         }
  71         return f->doc_sections;
  72 }
  73
  74 struct ccan_file *new_ccan_file(const void *ctx, const char *dir, char *name)
  75 {
  76         struct ccan_file *f;
  77
  78         assert(dir[0] == '/');
  79
  80         f = talloc(ctx, struct ccan_file);
  81         f->lines = NULL;
  82         f->line_info = NULL;
  83         f->doc_sections = NULL;
  84         f->compiled = NULL;
  85         f->name = talloc_steal(f, name);
  86         f->fullname = talloc_asprintf(f, "%s/%s", dir, f->name);
  87         f->contents = NULL;
  88         f->cov_compiled = NULL;
  89         f->simplified = NULL;
  90         return f;
  91 }
  92
  93 static void add_files(struct manifest *m, const char *dir)
  94 {
  95         DIR *d;
  96         struct dirent *ent;
  97         char **subs = NULL;
  98
  99         if (dir[0])
 100                 d = opendir(dir);
 101         else
 102                 d = opendir(".");
 103         if (!d)
 104                 err(1, "Opening directory %s", dir[0] ? dir : ".");
 105
 106         while ((ent = readdir(d)) != NULL) {
 107                 struct stat st;
 108                 struct ccan_file *f;
 109                 struct list_head *dest;
 110                 bool is_c_src;
 111
 112                 if (ent->d_name[0] == '.')
 113                         continue;
 114
 115                 f = new_ccan_file(m, m->dir,
 116                                   talloc_asprintf(m, "%s%s",
 117                                                   dir, ent->d_name));
 118                 if (lstat(f->name, &st) != 0)
 119                         err(1, "lstat %s", f->name);
 120
 121                 if (S_ISDIR(st.st_mode)) {
 122                         size_t len = talloc_array_length(subs);
 123                         subs = talloc_realloc(m, subs, char *, len+1);
 124                         subs[len] = talloc_append_string(f->name, "/");
 125                         continue;
 126                 }
 127                 if (!S_ISREG(st.st_mode)) {
 128                         talloc_free(f);
 129                         continue;
 130                 }
 131
 132                 if (streq(f->name, "_info")) {
 133                         m->info_file = f;
 134                         continue;
 135                 }
 136
 137                 is_c_src = strends(f->name, ".c");
 138                 if (!is_c_src && !strends(f->name, ".h")) {
 139                         dest = &m->other_files;
 140                 } else if (!strchr(f->name, '/')) {
 141                         if (is_c_src)
 142                                 dest = &m->c_files;
 143                         else
 144                                 dest = &m->h_files;
 145                 } else if (strstarts(f->name, "test/")) {
 146                         if (is_c_src) {
 147                                 if (strstarts(f->name, "test/api"))
 148                                         dest = &m->api_tests;
 149                                 else if (strstarts(f->name, "test/run"))
 150                                         dest = &m->run_tests;
 151                                 else if (strstarts(f->name, "test/compile_ok"))
 152                                         dest = &m->compile_ok_tests;
 153                                 else if (strstarts(f->name, "test/compile_fail"))
 154                                         dest = &m->compile_fail_tests;
 155                                 else
 156                                         dest = &m->other_test_c_files;
 157                         } else
 158                                 dest = &m->other_test_files;
 159                 } else
 160                         dest = &m->other_files;
 161
 162                 list_add(dest, &f->list);
 163         }
 164         closedir(d);
 165
 166         /* Before we recurse, sanity check this is a ccan module. */
 167         if (!dir[0]) {
 168                 size_t i;
 169
 170                 if (!m->info_file
 171                     && list_empty(&m->c_files)
 172                     && list_empty(&m->h_files))
 173                         errx(1, "No _info, C or H files found here!");
 174
 175                 for (i = 0; i < talloc_array_length(subs); i++)
 176                         add_files(m, subs[i]);
 177         }
 178         talloc_free(subs);
 179 }
 180
 181 static int cmp_names(struct ccan_file *const *a, struct ccan_file *const *b,
 182                      void *unused)
 183 {
 184         return strcmp((*a)->name, (*b)->name);
 185 }
 186
 187 static void sort_files(struct list_head *list)
 188 {
 189         struct ccan_file **files = NULL, *f;
 190         unsigned int i, num;
 191
 192         num = 0;
 193         while ((f = list_top(list, struct ccan_file, list)) != NULL) {
 194                 files = talloc_realloc(NULL, files, struct ccan_file *, num+1);
 195                 files[num++] = f;
 196                 list_del(&f->list);
 197         }
 198         asort(files, num, cmp_names, NULL);
 199
 200         for (i = 0; i < num; i++)
 201                 list_add_tail(list, &files[i]->list);
 202         talloc_free(files);
 203 }
 204
 205 struct manifest *get_manifest(const void *ctx, const char *dir)
 206 {
 207         struct manifest *m;
 208         char *olddir, *canon_dir;
 209         unsigned int len;
 210         struct list_head *list;
 211
 212         if (!manifests)
 213                 manifests = htable_manifest_new();
 214
 215         olddir = talloc_getcwd(NULL);
 216         if (!olddir)
 217                 err(1, "Getting current directory");
 218
 219         if (chdir(dir) != 0)
 220                 err(1, "Failed to chdir to %s", dir);
 221
 222         canon_dir = talloc_getcwd(olddir);
 223         if (!canon_dir)
 224                 err(1, "Getting current directory");
 225
 226         m = htable_manifest_get(manifests, canon_dir);
 227         if (m)
 228                 goto done;
 229
 230         m = talloc_linked(ctx, talloc(NULL, struct manifest));
 231         m->info_file = NULL;
 232         m->compiled = NULL;
 233         m->dir = talloc_steal(m, canon_dir);
 234         list_head_init(&m->c_files);
 235         list_head_init(&m->h_files);
 236         list_head_init(&m->api_tests);
 237         list_head_init(&m->run_tests);
 238         list_head_init(&m->compile_ok_tests);
 239         list_head_init(&m->compile_fail_tests);
 240         list_head_init(&m->other_test_c_files);
 241         list_head_init(&m->other_test_files);
 242         list_head_init(&m->other_files);
 243         list_head_init(&m->examples);
 244         list_head_init(&m->mangled_examples);
 245         list_head_init(&m->deps);
 246
 247         len = strlen(m->dir);
 248         while (len && m->dir[len-1] == '/')
 249                 m->dir[--len] = '\0';
 250
 251         m->basename = strrchr(m->dir, '/');
 252         if (!m->basename)
 253                 errx(1, "I don't expect to be run from the root directory");
 254         m->basename++;
 255
 256         /* We expect the ccan dir to be two levels above module dir. */
 257         if (!ccan_dir) {
 258                 char *p, *dir;
 259                 dir = talloc_strdup(NULL, m->dir);
 260                 p = strrchr(dir, '/');
 261                 if (!p)
 262                         errx(1, "I expect the ccan root directory in ../..");
 263                 *p = '\0';
 264                 p = strrchr(dir, '/');
 265                 if (!p)
 266                         errx(1, "I expect the ccan root directory in ../..");
 267                 *p = '\0';
 268                 ccan_dir = dir;
 269         }
 270
 271         add_files(m, "");
 272
 273         /* Nicer to run tests in a predictable order. */
 274         foreach_ptr(list, &m->api_tests, &m->run_tests, &m->compile_ok_tests,
 275                     &m->compile_fail_tests)
 276                 sort_files(list);
 277
 278         htable_manifest_add(manifests, m);
 279
 280 done:
 281         if (chdir(olddir) != 0)
 282                 err(1, "Returning to original directory '%s'", olddir);
 283         talloc_free(olddir);
 284
 285         return m;
 286 }
 287
 288
 289 /**
 290  * remove_comments - strip comments from a line, return copy.
 291  * @line: line to copy
 292  * @in_comment: are we already within a comment (from prev line).
 293  * @unterminated: are we still in a comment for next line.
 294  */
 295 static char *remove_comments(const char *line, bool in_comment,
 296                              bool *unterminated)
 297 {
 298         char *p, *ret = talloc_array(line, char, strlen(line) + 1);
 299
 300         p = ret;
 301         for (;;) {
 302                 if (!in_comment) {
 303                         /* Find first comment. */
 304                         const char *old_comment = strstr(line, "/*");
 305                         const char *new_comment = strstr(line, "//");
 306                         const char *comment;
 307
 308                         if (new_comment && old_comment)
 309                                 comment = new_comment < old_comment
 310                                         ? new_comment : old_comment;
 311                         else if (old_comment)
 312                                 comment = old_comment;
 313                         else if (new_comment)
 314                                 comment = new_comment;
 315                         else {
 316                                 /* Nothing more. */
 317                                 strcpy(p, line);
 318                                 *unterminated = false;
 319                                 break;
 320                         }
 321
 322                         /* Copy up to comment. */
 323                         memcpy(p, line, comment - line);
 324                         p += comment - line;
 325                         line += comment - line + 2;
 326
 327                         if (comment == new_comment) {
 328                                 /* We're done: goes to EOL. */
 329                                 p[0] = '\0';
 330                                 *unterminated = false;
 331                                 break;
 332                         }
 333                         in_comment = true;
 334                 }
 335
 336                 if (in_comment) {
 337                         const char *end = strstr(line, "*/");
 338                         if (!end) {
 339                                 *unterminated = true;
 340                                 p[0] = '\0';
 341                                 break;
 342                         }
 343                         line = end+2;
 344                         in_comment = false;
 345                 }
 346         }
 347         return ret;
 348 }
 349
 350 static bool is_empty(const char *line)
 351 {
 352         return strspn(line, " \r\t") == strlen(line);
 353 }
 354
 355 static bool continues(const char *line)
 356 {
 357         /* Technically, any odd number of these.  But who cares? */
 358         return strends(line, "\\");
 359 }
 360
 361 /* Get token if it's equal to token. */
 362 bool get_token(const char **line, const char *token)
 363 {
 364         unsigned int toklen;
 365
 366         *line += strspn(*line, " \t");
 367         if (cisalnum(token[0]) || token[0] == '_')
 368                 toklen = strspn(*line, IDENT_CHARS);
 369         else {
 370                 /* FIXME: real tokenizer handles ++ and other multi-chars.  */
 371                 toklen = strlen(token);
 372         }
 373
 374         if (toklen == strlen(token) && !strncmp(*line, token, toklen)) {
 375                 *line += toklen;
 376                 return true;
 377         }
 378         return false;
 379 }
 380
 381 char *get_symbol_token(void *ctx, const char **line)
 382 {
 383         unsigned int toklen;
 384         char *ret;
 385
 386         *line += strspn(*line, " \t");
 387         toklen = strspn(*line, IDENT_CHARS);
 388         if (!toklen)
 389                 return NULL;
 390         ret = talloc_strndup(ctx, *line, toklen);
 391         *line += toklen;
 392         return ret;
 393 }
 394
 395 static bool parse_hash_if(struct pp_conditions *cond, const char **line)
 396 {
 397         bool brackets, defined;
 398
 399         cond->inverse = get_token(line, "!");
 400         defined = get_token(line, "defined");
 401         brackets = get_token(line, "(");
 402         cond->symbol = get_symbol_token(cond, line);
 403         if (!cond->symbol)
 404                 return false;
 405         if (brackets && !get_token(line, ")"))
 406                 return false;
 407         if (!defined)
 408                 cond->type = PP_COND_IF;
 409
 410         /* FIXME: We just chain them, ignoring operators. */
 411         if (get_token(line, "||") || get_token(line, "&&")) {
 412                 struct pp_conditions *sub = talloc(cond, struct pp_conditions);
 413
 414                 sub->parent = cond->parent;
 415                 sub->type = PP_COND_IFDEF;
 416                 if (parse_hash_if(sub, line))
 417                         cond->parent = sub;
 418         }
 419
 420         return true;
 421 }
 422
 423 /* FIXME: Get serious! */
 424 static struct pp_conditions *analyze_directive(struct ccan_file *f,
 425                                                const char *line,
 426                                                struct pp_conditions *parent)
 427 {
 428         struct pp_conditions *cond = talloc(f, struct pp_conditions);
 429         bool unused;
 430
 431         line = remove_comments(line, false, &unused);
 432
 433         cond->parent = parent;
 434         cond->type = PP_COND_IFDEF;
 435
 436         if (!get_token(&line, "#"))
 437                 abort();
 438
 439         if (get_token(&line, "if")) {
 440                 if (!parse_hash_if(cond, &line))
 441                         goto unknown;
 442         } else if (get_token(&line, "elif")) {
 443                 /* Malformed? */
 444                 if (!parent)
 445                         return NULL;
 446                 cond->parent = parent->parent;
 447                 /* FIXME: Not quite true.  This implies !parent, but we don't
 448                  * do multiple conditionals yet. */
 449                 if (!parse_hash_if(cond, &line))
 450                         goto unknown;
 451         } else if (get_token(&line, "ifdef")) {
 452                 bool brackets;
 453                 cond->inverse = false;
 454                 brackets = get_token(&line, "(");
 455                 cond->symbol = get_symbol_token(cond, &line);
 456                 if (!cond->symbol)
 457                         goto unknown;
 458                 if (brackets && !get_token(&line, ")"))
 459                         goto unknown;
 460         } else if (get_token(&line, "ifndef")) {
 461                 bool brackets;
 462                 cond->inverse = true;
 463                 brackets = get_token(&line, "(");
 464                 cond->symbol = get_symbol_token(cond, &line);
 465                 if (!cond->symbol)
 466                         goto unknown;
 467                 if (brackets && !get_token(&line, ")"))
 468                         goto unknown;
 469         } else if (get_token(&line, "else")) {
 470                 /* Malformed? */
 471                 if (!parent)
 472                         return NULL;
 473
 474                 *cond = *parent;
 475                 cond->inverse = !cond->inverse;
 476                 return cond;
 477         } else if (get_token(&line, "endif")) {
 478                 talloc_free(cond);
 479                 /* Malformed? */
 480                 if (!parent)
 481                         return NULL;
 482                 /* Back up one! */
 483                 return parent->parent;
 484         } else {
 485                 /* Not a conditional. */
 486                 talloc_free(cond);
 487                 return parent;
 488         }
 489
 490         if (!is_empty(line))
 491                 goto unknown;
 492         return cond;
 493
 494 unknown:
 495         cond->type = PP_COND_UNKNOWN;
 496         return cond;
 497 }
 498
 499 /* This parser is rough, but OK if code is reasonably neat. */
 500 struct line_info *get_ccan_line_info(struct ccan_file *f)
 501 {
 502         bool continued = false, in_comment = false;
 503         struct pp_conditions *cond = NULL;
 504         unsigned int i;
 505
 506         if (f->line_info)
 507                 return f->line_info;
 508
 509         get_ccan_file_lines(f);
 510         f->line_info = talloc_array(f->lines, struct line_info, f->num_lines);
 511
 512         for (i = 0; i < f->num_lines; continued = continues(f->lines[i++])) {
 513                 char *p;
 514                 bool still_doc_line;
 515
 516                 /* Current conditions apply to this line. */
 517                 f->line_info[i].cond = cond;
 518                 f->line_info[i].continued = continued;
 519
 520                 if (continued) {
 521                         /* Same as last line. */
 522                         f->line_info[i].type = f->line_info[i-1].type;
 523                         /* Update in_comment. */
 524                         remove_comments(f->lines[i], in_comment, &in_comment);
 525                         continue;
 526                 }
 527
 528                 /* Preprocessor directive? */
 529                 if (!in_comment
 530                     && f->lines[i][strspn(f->lines[i], " \t")] == '#') {
 531                         f->line_info[i].type = PREPROC_LINE;
 532                         cond = analyze_directive(f, f->lines[i], cond);
 533                         continue;
 534                 }
 535
 536                 still_doc_line = (in_comment
 537                                   && f->line_info[i-1].type == DOC_LINE);
 538
 539                 p = remove_comments(f->lines[i], in_comment, &in_comment);
 540                 if (is_empty(p)) {
 541                         if (strstarts(f->lines[i], "/**") || still_doc_line)
 542                                 f->line_info[i].type = DOC_LINE;
 543                         else
 544                                 f->line_info[i].type = COMMENT_LINE;
 545                 } else
 546                         f->line_info[i].type = CODE_LINE;
 547                 talloc_free(p);
 548         }
 549         return f->line_info;
 550 }
 551
 552 struct symbol {
 553         struct list_node list;
 554         const char *name;
 555         const unsigned int *value;
 556 };
 557
 558 static struct symbol *find_symbol(struct list_head *syms, const char *sym)
 559 {
 560         struct symbol *i;
 561
 562         list_for_each(syms, i, list)
 563                 if (streq(sym, i->name))
 564                         return i;
 565         return NULL;
 566 }
 567
 568 static enum line_compiled get_pp(struct pp_conditions *cond,
 569                                  struct list_head *syms)
 570 {
 571         struct symbol *sym;
 572         unsigned int val;
 573         enum line_compiled parent, ret;
 574
 575         /* No conditions?  Easy. */
 576         if (!cond)
 577                 return COMPILED;
 578
 579         /* Check we get here at all. */
 580         parent = get_pp(cond->parent, syms);
 581         if (parent == NOT_COMPILED)
 582                 return NOT_COMPILED;
 583
 584         if (cond->type == PP_COND_UNKNOWN)
 585                 return MAYBE_COMPILED;
 586
 587         sym = find_symbol(syms, cond->symbol);
 588         if (!sym)
 589                 return MAYBE_COMPILED;
 590
 591         switch (cond->type) {
 592         case PP_COND_IF:
 593                 /* Undefined is 0. */
 594                 val = sym->value ? *sym->value : 0;
 595                 if (!val == cond->inverse)
 596                         ret = COMPILED;
 597                 else
 598                         ret = NOT_COMPILED;
 599                 break;
 600
 601         case PP_COND_IFDEF:
 602                 if (cond->inverse == !sym->value)
 603                         ret = COMPILED;
 604                 else
 605                         ret = NOT_COMPILED;
 606                 break;
 607
 608         default:
 609                 abort();
 610         }
 611
 612         /* If parent didn't know, NO == NO, but YES == MAYBE. */
 613         if (parent == MAYBE_COMPILED && ret == COMPILED)
 614                 ret = MAYBE_COMPILED;
 615         return ret;
 616 }
 617
 618 static void add_symbol(struct list_head *head,
 619                        const char *symbol, const unsigned int *value)
 620 {
 621         struct symbol *sym = talloc(head, struct symbol);
 622         sym->name = symbol;
 623         sym->value = value;
 624         list_add(head, &sym->list);
 625 }
 626
 627 enum line_compiled get_ccan_line_pp(struct pp_conditions *cond,
 628                                     const char *symbol,
 629                                     const unsigned int *value,
 630                                     ...)
 631 {
 632         enum line_compiled ret;
 633         struct list_head *head;
 634         va_list ap;
 635
 636         head = talloc(NULL, struct list_head);
 637         list_head_init(head);
 638
 639         va_start(ap, value);
 640         add_symbol(head, symbol, value);
 641
 642         while ((symbol = va_arg(ap, const char *)) != NULL) {
 643                 value = va_arg(ap, const unsigned int *);
 644                 add_symbol(head, symbol, value);
 645         }
 646         ret = get_pp(cond, head);
 647         talloc_free(head);
 648         return ret;
 649 }
 650
 651 void score_file_error(struct score *score, struct ccan_file *f, unsigned line,
 652                       const char *errorfmt, ...)
 653 {
 654         va_list ap;
 655
 656         struct file_error *fe = talloc(score, struct file_error);
 657         fe->file = f;
 658         fe->line = line;
 659         list_add_tail(&score->per_file_errors, &fe->list);
 660
 661         if (!score->error)
 662                 score->error = talloc_strdup(score, "");
 663
 664         if (verbose < 2 && strcount(score->error, "\n") > 5)
 665                 return;
 666
 667         if (line)
 668                 score->error = talloc_asprintf_append(score->error,
 669                                                       "%s:%u:",
 670                                                       f->fullname, line);
 671         else
 672                 score->error = talloc_asprintf_append(score->error,
 673                                                       "%s:", f->fullname);
 674
 675         va_start(ap, errorfmt);
 676         score->error = talloc_vasprintf_append(score->error, errorfmt, ap);
 677         va_end(ap);
 678         score->error = talloc_append_string(score->error, "\n");
 679
 680         if (verbose < 2 && strcount(score->error, "\n") > 5)
 681                 score->error = talloc_append_string(score->error,
 682                                     "... more (use -vv to see them all)\n");
 683 }