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