src/procfile.c

   1 #include "common.h"
   2
   3 #define PF_PREFIX "PROCFILE"
   4
   5 #define PFWORDS_INCREASE_STEP 200
   6 #define PFLINES_INCREASE_STEP 10
   7 #define PROCFILE_INCREMENT_BUFFER 512
   8
   9 int procfile_adaptive_initial_allocation = 0;
  10
  11 // if adaptive allocation is set, these store the
  12 // max values we have seen so far
  13 uint32_t procfile_max_lines = PFLINES_INCREASE_STEP;
  14 uint32_t procfile_max_words = PFWORDS_INCREASE_STEP;
  15 size_t procfile_max_allocation = PROCFILE_INCREMENT_BUFFER;
  16
  17 // ----------------------------------------------------------------------------
  18 // An array of words
  19
  20
  21 pfwords *pfwords_add(pfwords *fw, char *str) {
  22         // debug(D_PROCFILE, PF_PREFIX ":       adding word No %d: '%s'", fw->len, str);
  23
  24         if(unlikely(fw->len == fw->size)) {
  25                 // debug(D_PROCFILE, PF_PREFIX ":       expanding words");
  26
  27                 fw = reallocz(fw, sizeof(pfwords) + (fw->size + PFWORDS_INCREASE_STEP) * sizeof(char *));
  28                 fw->size += PFWORDS_INCREASE_STEP;
  29         }
  30
  31         fw->words[fw->len++] = str;
  32
  33         return fw;
  34 }
  35
  36 pfwords *pfwords_new(void) {
  37         // debug(D_PROCFILE, PF_PREFIX ":       initializing words");
  38
  39         uint32_t size = (procfile_adaptive_initial_allocation) ? procfile_max_words : PFWORDS_INCREASE_STEP;
  40
  41         pfwords *new = mallocz(sizeof(pfwords) + size * sizeof(char *));
  42         new->len = 0;
  43         new->size = size;
  44         return new;
  45 }
  46
  47 void pfwords_reset(pfwords *fw) {
  48         // debug(D_PROCFILE, PF_PREFIX ":       reseting words");
  49         fw->len = 0;
  50 }
  51
  52 void pfwords_free(pfwords *fw) {
  53         // debug(D_PROCFILE, PF_PREFIX ":       freeing words");
  54
  55         freez(fw);
  56 }
  57
  58
  59 // ----------------------------------------------------------------------------
  60 // An array of lines
  61
  62 pflines *pflines_add(pflines *fl, uint32_t first_word) {
  63         // debug(D_PROCFILE, PF_PREFIX ":       adding line %d at word %d", fl->len, first_word);
  64
  65         if(unlikely(fl->len == fl->size)) {
  66                 // debug(D_PROCFILE, PF_PREFIX ":       expanding lines");
  67
  68                 fl = reallocz(fl, sizeof(pflines) + (fl->size + PFLINES_INCREASE_STEP) * sizeof(ffline));
  69                 fl->size += PFLINES_INCREASE_STEP;
  70         }
  71
  72         fl->lines[fl->len].words = 0;
  73         fl->lines[fl->len++].first = first_word;
  74
  75         return fl;
  76 }
  77
  78 pflines *pflines_new(void) {
  79         // debug(D_PROCFILE, PF_PREFIX ":       initializing lines");
  80
  81         uint32_t size = (unlikely(procfile_adaptive_initial_allocation)) ? procfile_max_words : PFLINES_INCREASE_STEP;
  82
  83         pflines *new = mallocz(sizeof(pflines) + size * sizeof(ffline));
  84         new->len = 0;
  85         new->size = size;
  86         return new;
  87 }
  88
  89 void pflines_reset(pflines *fl) {
  90         // debug(D_PROCFILE, PF_PREFIX ":       reseting lines");
  91
  92         fl->len = 0;
  93 }
  94
  95 void pflines_free(pflines *fl) {
  96         // debug(D_PROCFILE, PF_PREFIX ":       freeing lines");
  97
  98         freez(fl);
  99 }
 100
 101
 102 // ----------------------------------------------------------------------------
 103 // The procfile
 104
 105 #define PF_CHAR_IS_SEPARATOR    ' '
 106 #define PF_CHAR_IS_NEWLINE              'N'
 107 #define PF_CHAR_IS_WORD                 'W'
 108 #define PF_CHAR_IS_QUOTE        'Q'
 109 #define PF_CHAR_IS_OPEN         'O'
 110 #define PF_CHAR_IS_CLOSE        'C'
 111
 112 void procfile_close(procfile *ff) {
 113         debug(D_PROCFILE, PF_PREFIX ": Closing file '%s'", ff->filename);
 114
 115         if(likely(ff->lines)) pflines_free(ff->lines);
 116         if(likely(ff->words)) pfwords_free(ff->words);
 117
 118         if(likely(ff->fd != -1)) close(ff->fd);
 119         freez(ff);
 120 }
 121
 122 procfile *procfile_parser(procfile *ff) {
 123         debug(D_PROCFILE, PF_PREFIX ": Parsing file '%s'", ff->filename);
 124
 125         char *s = ff->data, *e = &ff->data[ff->len], *t = ff->data, quote = 0;
 126         uint32_t l = 0, w = 0;
 127         int opened = 0;
 128
 129         ff->lines = pflines_add(ff->lines, w);
 130         if(unlikely(!ff->lines)) goto cleanup;
 131
 132         while(likely(s < e)) {
 133                 // we are not at the end
 134
 135                 switch(ff->separators[(uint8_t)(*s)]) {
 136                         case PF_CHAR_IS_OPEN:
 137                                 if(s == t) {
 138                                         opened++;
 139                                         t = ++s;
 140                                 }
 141                                 else if(opened) {
 142                                         opened++;
 143                                         s++;
 144                                 }
 145                                 else
 146                                         s++;
 147                                 continue;
 148
 149                         case PF_CHAR_IS_CLOSE:
 150                                 if(opened) {
 151                                         opened--;
 152
 153                                         if(!opened) {
 154                                                 *s = '\0';
 155                                                 ff->words = pfwords_add(ff->words, t);
 156                                                 if(unlikely(!ff->words)) goto cleanup;
 157
 158                                                 ff->lines->lines[l].words++;
 159                                                 w++;
 160
 161                                                 t = ++s;
 162                                         }
 163                                         else
 164                                                 s++;
 165                                 }
 166                                 else
 167                                         s++;
 168                                 continue;
 169
 170                         case PF_CHAR_IS_QUOTE:
 171                                 if(unlikely(!quote && s == t)) {
 172                                         // quote opened at the beginning
 173                                         quote = *s;
 174                                         t = ++s;
 175                                 }
 176                                 else if(unlikely(quote && quote == *s)) {
 177                                         // quote closed
 178                                         quote = 0;
 179
 180                                         *s = '\0';
 181                                         ff->words = pfwords_add(ff->words, t);
 182                                         if(unlikely(!ff->words)) goto cleanup;
 183
 184                                         ff->lines->lines[l].words++;
 185                                         w++;
 186
 187                                         t = ++s;
 188                                 }
 189                                 else
 190                                         s++;
 191                                 continue;
 192
 193                         case PF_CHAR_IS_SEPARATOR:
 194                                 if(unlikely(quote || opened)) {
 195                                         // we are inside a quote
 196                                         s++;
 197                                         continue;
 198                                 }
 199
 200                                 if(unlikely(s == t)) {
 201                                         // skip all leading white spaces
 202                                         t = ++s;
 203                                         continue;
 204                                 }
 205
 206                                 // end of word
 207                                 *s = '\0';
 208
 209                                 ff->words = pfwords_add(ff->words, t);
 210                                 if(unlikely(!ff->words)) goto cleanup;
 211
 212                                 ff->lines->lines[l].words++;
 213                                 w++;
 214
 215                                 t = ++s;
 216                                 continue;
 217
 218                         case PF_CHAR_IS_NEWLINE:
 219                                 // end of line
 220                                 *s = '\0';
 221
 222                                 ff->words = pfwords_add(ff->words, t);
 223                                 if(unlikely(!ff->words)) goto cleanup;
 224
 225                                 ff->lines->lines[l].words++;
 226                                 w++;
 227
 228                                 // debug(D_PROCFILE, PF_PREFIX ":       ended line %d with %d words", l, ff->lines->lines[l].words);
 229
 230                                 ff->lines = pflines_add(ff->lines, w);
 231                                 if(unlikely(!ff->lines)) goto cleanup;
 232                                 l++;
 233
 234                                 t = ++s;
 235                                 continue;
 236
 237                         default:
 238                                 s++;
 239                                 continue;
 240                 }
 241         }
 242
 243         if(likely(s > t && t < e)) {
 244                 // the last word
 245                 if(likely(ff->len < ff->size))
 246                         *s = '\0';
 247                 else {
 248                         // we are going to loose the last byte
 249                         ff->data[ff->size - 1] = '\0';
 250                 }
 251
 252                 ff->words = pfwords_add(ff->words, t);
 253                 if(unlikely(!ff->words)) goto cleanup;
 254
 255                 ff->lines->lines[l].words++;
 256                 w++;
 257         }
 258
 259         return ff;
 260
 261 cleanup:
 262         error(PF_PREFIX ": Failed to parse file '%s'", ff->filename);
 263         procfile_close(ff);
 264         return NULL;
 265 }
 266
 267 procfile *procfile_readall(procfile *ff) {
 268         debug(D_PROCFILE, PF_PREFIX ": Reading file '%s'.", ff->filename);
 269
 270         ssize_t s, r = 1, x;
 271         ff->len = 0;
 272
 273         while(likely(r > 0)) {
 274                 s = ff->len;
 275                 x = ff->size - s;
 276
 277                 if(!x) {
 278                         debug(D_PROCFILE, PF_PREFIX ": Expanding data buffer for file '%s'.", ff->filename);
 279
 280                         ff = reallocz(ff, sizeof(procfile) + ff->size + PROCFILE_INCREMENT_BUFFER);
 281                         ff->size += PROCFILE_INCREMENT_BUFFER;
 282                 }
 283
 284                 debug(D_PROCFILE, "Reading file '%s', from position %ld with length %lu", ff->filename, s, ff->size - s);
 285                 r = read(ff->fd, &ff->data[s], ff->size - s);
 286                 if(unlikely(r == -1)) {
 287                         if(unlikely(!(ff->flags & PROCFILE_FLAG_NO_ERROR_ON_FILE_IO))) error(PF_PREFIX ": Cannot read from file '%s'", ff->filename);
 288                         procfile_close(ff);
 289                         return NULL;
 290                 }
 291
 292                 ff->len += r;
 293         }
 294
 295         debug(D_PROCFILE, "Rewinding file '%s'", ff->filename);
 296         if(unlikely(lseek(ff->fd, 0, SEEK_SET) == -1)) {
 297                 if(unlikely(!(ff->flags & PROCFILE_FLAG_NO_ERROR_ON_FILE_IO))) error(PF_PREFIX ": Cannot rewind on file '%s'.", ff->filename);
 298                 procfile_close(ff);
 299                 return NULL;
 300         }
 301
 302         pflines_reset(ff->lines);
 303         pfwords_reset(ff->words);
 304
 305         ff = procfile_parser(ff);
 306
 307         if(unlikely(procfile_adaptive_initial_allocation)) {
 308                 if(unlikely(ff->len > procfile_max_allocation)) procfile_max_allocation = ff->len;
 309                 if(unlikely(ff->lines->len > procfile_max_lines)) procfile_max_lines = ff->lines->len;
 310                 if(unlikely(ff->words->len > procfile_max_words)) procfile_max_words = ff->words->len;
 311         }
 312
 313         debug(D_PROCFILE, "File '%s' updated.", ff->filename);
 314         return ff;
 315 }
 316
 317 static void procfile_set_separators(procfile *ff, const char *separators) {
 318         static char def[256] = { [0 ... 255] = 0 };
 319         int i;
 320
 321         if(unlikely(!def[255])) {
 322                 // this is thread safe
 323                 // we check that the last byte is non-zero
 324                 // if it is zero, multiple threads may be executing this at the same time
 325                 // setting in def[] the exact same values
 326                 for(i = 0; likely(i < 256) ;i++) {
 327                         if(unlikely(i == '\n' || i == '\r')) def[i] = PF_CHAR_IS_NEWLINE;
 328                         else if(unlikely(isspace(i) || !isprint(i))) def[i] = PF_CHAR_IS_SEPARATOR;
 329                         else def[i] = PF_CHAR_IS_WORD;
 330                 }
 331         }
 332
 333         // copy the default
 334         char *ffs = ff->separators, *ffd = def, *ffe = &def[256];
 335         while(likely(ffd != ffe)) *ffs++ = *ffd++;
 336
 337         // set the separators
 338         if(unlikely(!separators))
 339                 separators = " \t=|";
 340
 341         ffs = ff->separators;
 342         const char *s = separators;
 343         while(likely(*s))
 344                 ffs[(int)*s++] = PF_CHAR_IS_SEPARATOR;
 345 }
 346
 347 void procfile_set_quotes(procfile *ff, const char *quotes) {
 348         // remove all quotes
 349         int i;
 350         for(i = 0; i < 256 ; i++)
 351                 if(ff->separators[i] == PF_CHAR_IS_QUOTE)
 352                         ff->separators[i] = PF_CHAR_IS_WORD;
 353
 354         // if nothing given, return
 355         if(unlikely(!quotes || !*quotes))
 356                 return;
 357
 358         // set the quotes
 359         char *ffs = ff->separators;
 360         const char *s = quotes;
 361         while(likely(*s))
 362                 ffs[(int)*s++] = PF_CHAR_IS_QUOTE;
 363 }
 364
 365 void procfile_set_open_close(procfile *ff, const char *open, const char *close) {
 366         // remove all open/close
 367         int i;
 368         for(i = 0; i < 256 ; i++)
 369                 if(ff->separators[i] == PF_CHAR_IS_OPEN || ff->separators[i] == PF_CHAR_IS_CLOSE)
 370                         ff->separators[i] = PF_CHAR_IS_WORD;
 371
 372         // if nothing given, return
 373         if(unlikely(!open || !*open || !close || !*close))
 374                 return;
 375
 376         // set the openings
 377         char *ffs = ff->separators;
 378         const char *s = open;
 379         while(likely(*s))
 380                 ffs[(int)*s++] = PF_CHAR_IS_OPEN;
 381
 382         s = close;
 383         while(likely(*s))
 384                 ffs[(int)*s++] = PF_CHAR_IS_CLOSE;
 385 }
 386
 387 procfile *procfile_open(const char *filename, const char *separators, uint32_t flags) {
 388         debug(D_PROCFILE, PF_PREFIX ": Opening file '%s'", filename);
 389
 390         int fd = open(filename, O_RDONLY, 0666);
 391         if(unlikely(fd == -1)) {
 392                 if(unlikely(!(flags & PROCFILE_FLAG_NO_ERROR_ON_FILE_IO))) error(PF_PREFIX ": Cannot open file '%s'", filename);
 393                 return NULL;
 394         }
 395
 396         size_t size = (unlikely(procfile_adaptive_initial_allocation)) ? procfile_max_allocation : PROCFILE_INCREMENT_BUFFER;
 397         procfile *ff = mallocz(sizeof(procfile) + size);
 398         strncpyz(ff->filename, filename, FILENAME_MAX);
 399
 400         ff->fd = fd;
 401         ff->size = size;
 402         ff->len = 0;
 403         ff->flags = flags;
 404
 405         ff->lines = pflines_new();
 406         ff->words = pfwords_new();
 407
 408         if(unlikely(!ff->lines || !ff->words)) {
 409                 error(PF_PREFIX ": Cannot initialize parser for file '%s'", filename);
 410                 procfile_close(ff);
 411                 return NULL;
 412         }
 413
 414         procfile_set_separators(ff, separators);
 415
 416         debug(D_PROCFILE, "File '%s' opened.", filename);
 417         return ff;
 418 }
 419
 420 procfile *procfile_reopen(procfile *ff, const char *filename, const char *separators, uint32_t flags) {
 421         if(unlikely(!ff)) return procfile_open(filename, separators, flags);
 422
 423         if(likely(ff->fd != -1)) close(ff->fd);
 424
 425         ff->fd = open(filename, O_RDONLY, 0666);
 426         if(unlikely(ff->fd == -1)) {
 427                 procfile_close(ff);
 428                 return NULL;
 429         }
 430
 431         strncpyz(ff->filename, filename, FILENAME_MAX);
 432
 433         ff->flags = flags;
 434
 435         // do not do the separators again if NULL is given
 436         if(likely(separators)) procfile_set_separators(ff, separators);
 437
 438         return ff;
 439 }
 440
 441 // ----------------------------------------------------------------------------
 442 // example parsing of procfile data
 443
 444 void procfile_print(procfile *ff) {
 445         uint32_t lines = procfile_lines(ff), l;
 446         uint32_t words, w;
 447         char *s;
 448
 449         debug(D_PROCFILE, "File '%s' with %u lines and %u words", ff->filename, ff->lines->len, ff->words->len);
 450
 451         for(l = 0; likely(l < lines) ;l++) {
 452                 words = procfile_linewords(ff, l);
 453
 454                 debug(D_PROCFILE, "     line %u starts at word %u and has %u words", l, ff->lines->lines[l].first, ff->lines->lines[l].words);
 455
 456                 for(w = 0; likely(w < words) ;w++) {
 457                         s = procfile_lineword(ff, l, w);
 458                         debug(D_PROCFILE, "             [%u.%u] '%s'", l, w, s);
 459                 }
 460         }
 461 }