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