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