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