src/procfile.c

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