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