lib/csv.c

   1 #include <stdio.h>
   2 #include <stdlib.h>
   3 #include <string.h>
   4 #include <stdarg.h>
   5 #include <assert.h>
   6 #include <sys/queue.h>
   7 #include <fcntl.h>
   8 #include <unistd.h>
   9 #include "csv.h"
  10
  11 #define DEBUG_E 1
  12 #define DEBUG_V 1
  13
  14 #define log_error(fmt, ...) \
  15   do { if (DEBUG_E) fprintf(stderr, "%s:%d:%s(): " fmt, __FILE__, \
  16                             __LINE__, __func__, ##__VA_ARGS__); } while (0)
  17
  18 #define log_verbose(fmt, ...) \
  19   do { if (DEBUG_V) fprintf(stderr, "%s:%d:%s(): " fmt, __FILE__, \
  20                             __LINE__, __func__, __VA_ARGS__); } while (0)
  21
  22 struct _csv_field_t_ {
  23   TAILQ_ENTRY(_csv_field_t_) next_field;
  24   char *field;
  25   int field_len;
  26 };
  27
  28 struct _csv_record_t_ {
  29   TAILQ_HEAD(, _csv_field_t_) fields;
  30   TAILQ_ENTRY(_csv_record_t_) next_record;
  31   char *record;
  32   int rec_len;
  33 };
  34
  35 struct _csv_t_ {
  36   TAILQ_HEAD(, _csv_record_t_) records;
  37   char *buf;
  38   int buflen;
  39   int csv_len;
  40   int pointer;
  41   int num_recs;
  42 };
  43
  44
  45 int
  46 csvlen (csv_t *csv)
  47 {
  48   return (csv->csv_len);
  49 }
  50
  51 csv_t *
  52 csv_init (csv_t *csv,
  53           char *buf,
  54           int buflen)
  55 {
  56   if (csv == NULL) {
  57     csv = malloc(sizeof(csv_t));
  58     if (csv == NULL) {
  59       log_error("CSV Malloc failed\n");
  60       return (NULL);
  61     }
  62   }
  63   memset(csv, 0, sizeof(csv_t));
  64
  65   csv->buf = buf;
  66   csv->buflen = buflen;
  67   TAILQ_INIT(&(csv->records));
  68   return (csv);
  69 }
  70
  71 void
  72 csv_clean (csv_t *csv)
  73 {
  74   csv_record_t *rec;
  75   csv_record_t *rec_n;
  76
  77   rec = TAILQ_FIRST(&(csv->records));
  78   while (rec != NULL) {
  79     rec_n = TAILQ_NEXT(rec, next_record);
  80     csv_remove_record(csv, rec);
  81     rec = rec_n;
  82   }
  83 }
  84
  85 void
  86 csv_free (csv_t *csv)
  87 {
  88   if (csv != NULL) {
  89     free(csv);
  90   }
  91 }
  92
  93 static void
  94 csv_init_record (csv_record_t *record)
  95 {
  96   TAILQ_INIT(&(record->fields));
  97   record->rec_len = 0;
  98 }
  99
 100 csv_record_t *
 101 csv_record_iter (csv_t *csv)
 102 {
 103   return(TAILQ_FIRST(&(csv->records)));
 104 }
 105
 106 csv_record_t *
 107 csv_record_iter_next (csv_record_t *rec)
 108 {
 109   if(!rec) return NULL;
 110   return(TAILQ_NEXT(rec, next_record));
 111 }
 112
 113 char *
 114 csv_field_iter (csv_record_t *rec,
 115                 csv_field_t **fld)
 116 {
 117   if(!rec) return NULL;
 118   *fld = TAILQ_FIRST(&(rec->fields));
 119   return ((*fld)->field);
 120 }
 121
 122 char *
 123 csv_field_iter_next (csv_field_t **fld)
 124 {
 125   *fld = TAILQ_NEXT(*fld, next_field);
 126   if ((*fld) == NULL) {
 127     return (NULL);
 128   }
 129   return ((*fld)->field);
 130 }
 131
 132 int
 133 csv_field_len(csv_field_t *fld)
 134 {
 135   if (fld) {
 136     return fld->field_len;
 137   }
 138   return 0;
 139 }
 140
 141 static void
 142 csv_decode_record(csv_record_t *rec)
 143 {
 144     char *curr = rec->record;
 145     char *field;
 146     csv_field_t *fld;
 147
 148     field = strpbrk(curr, ",");
 149     while (field != NULL) {
 150         fld = malloc(sizeof(csv_field_t));
 151         if (fld) {
 152             TAILQ_INSERT_TAIL(&(rec->fields), fld, next_field);
 153             fld->field = curr;
 154             fld->field_len = field-curr;
 155         }
 156         curr = field + 1;
 157         field = strpbrk(curr, ",");
 158     }
 159     field = strstr(curr, "\n");
 160     fld = malloc(sizeof(csv_field_t));
 161     if (field && fld) {
 162         fld->field = curr;
 163         fld->field_len = field-curr;
 164         TAILQ_INSERT_TAIL(&(rec->fields), fld, next_field);
 165     }
 166 }
 167
 168 static csv_field_t *
 169 csv_add_field_to_record(csv_t *csv,
 170                         csv_record_t *rec,
 171                         char *col)
 172 {
 173     csv_field_t *fld;
 174     char *str = rec->record;
 175     int rlen = rec->rec_len;
 176     int blen = csv->buflen;
 177
 178     fld = malloc(sizeof(csv_field_t));
 179     if (!fld) {
 180         log_error("field malloc failed\n");
 181         /* more cleanup needed */
 182         return (NULL);
 183     }
 184     TAILQ_INSERT_TAIL(&(rec->fields), fld, next_field);
 185     fld->field = str+rlen;
 186     fld->field_len = snprintf((str+rlen), (blen - rlen), "%s", col);
 187     rlen += fld->field_len;
 188     rec->rec_len = rlen;
 189     return fld;
 190 }
 191
 192 csv_record_t *
 193 csv_encode (csv_t *csv,
 194             int count,
 195             ...)
 196 {
 197   int tempc;
 198   va_list list;
 199   char *buf = csv->buf;
 200   int len = csv->buflen;
 201   int pointer = csv->pointer;
 202   char *str = NULL;
 203   char *col;
 204   csv_record_t *rec;
 205   csv_field_t *fld;
 206
 207   if (buf) {
 208      str = buf + pointer;
 209   } else {
 210      /* allocate sufficient buffer */
 211      str = (char *)malloc(csv->buflen);
 212      if (!str) {
 213         log_error("field str malloc failed\n");
 214         return (NULL);
 215      }
 216   }
 217
 218   va_start(list, count);
 219   rec = malloc(sizeof(csv_record_t));
 220   if (!rec) {
 221     log_error("record malloc failed\n");
 222     return (NULL);
 223   }
 224   csv_init_record(rec);
 225   rec->record = str;
 226   TAILQ_INSERT_TAIL(&(csv->records), rec, next_record);
 227   csv->num_recs++;
 228
 229   /**
 230    * Iterate through the fields passed as a variable list and add them
 231    */
 232   for (tempc = 0; tempc < count; tempc++) {
 233     col = va_arg(list, char *);
 234     fld = csv_add_field_to_record(csv, rec, col);
 235     if (!fld) {
 236       log_error("fld malloc failed\n");
 237       csv_remove_record(csv, rec);
 238       return (NULL);
 239     }
 240     if (tempc < (count - 1)) {
 241       rec->rec_len += snprintf((str+rec->rec_len), (len - rec->rec_len), ",");
 242     }
 243   }
 244   rec->rec_len += snprintf((str+rec->rec_len), (len - rec->rec_len), "\n");
 245   va_end(list);
 246   csv->csv_len += rec->rec_len;
 247   csv->pointer += rec->rec_len;
 248   return (rec);
 249 }
 250
 251 int
 252 csv_num_records (csv_t *csv)
 253 {
 254     if (csv) {
 255         return csv->num_recs;
 256     }
 257     return 0;
 258 }
 259
 260 csv_record_t *
 261 csv_encode_record (csv_t *csv,
 262                    csv_record_t *rec,
 263                    int count,
 264                    ...)
 265 {
 266   int tempc;
 267   va_list list;
 268   char *str;
 269   char *col;
 270   csv_field_t *fld = NULL;
 271   int i;
 272
 273   va_start(list, count);
 274   str = csv_field_iter(rec, &fld);
 275   for (tempc = 0; tempc < count; tempc++) {
 276     col = va_arg(list, char *);
 277     for (i = 0; i < fld->field_len; i++) {
 278       str[i] = col[i];
 279     }
 280     str = csv_field_iter_next(&fld);
 281   }
 282   va_end(list);
 283   return (rec);
 284 }
 285
 286 csv_record_t *
 287 csv_append_record (csv_t *csv,
 288                    csv_record_t *rec,
 289                    int count,
 290                    ...)
 291 {
 292     int tempc;
 293     va_list list;
 294     int len = csv->buflen, tlen;
 295     char *str;
 296     csv_field_t *fld;
 297     char *col;
 298
 299     if (csv->buf) {
 300         /* not only works with discrete bufs */
 301         return NULL;
 302     }
 303
 304     if (!rec) {
 305         /* create a new rec */
 306         rec = calloc(1, sizeof(csv_record_t));
 307         if (!rec) {
 308             log_error("record malloc failed\n");
 309             return NULL;
 310         }
 311         csv_init_record(rec);
 312         rec->record = calloc(1, csv->buflen);
 313         if (!rec->record) {
 314             log_error("field str malloc failed\n");
 315             free(rec);
 316             return NULL;
 317         }
 318         csv_insert_record(csv, rec);
 319     }
 320
 321     str = rec->record;
 322
 323     va_start(list, count);
 324
 325     if (rec->rec_len && (str[rec->rec_len-1] == '\n'))
 326             str[rec->rec_len-1] = ',';
 327
 328     /**
 329      * Iterate through the fields passed as a variable list and add them
 330      */
 331     tlen = rec->rec_len;
 332     for (tempc = 0; tempc < count; tempc++) {
 333         col = va_arg(list, char *);
 334         fld = csv_add_field_to_record(csv, rec, col);
 335         if (!fld) {
 336             log_error("fld malloc failed\n");
 337             break;
 338         }
 339         if (tempc < (count - 1)) {
 340             rec->rec_len += snprintf((str+rec->rec_len),
 341                                      (len - rec->rec_len), ",");
 342         }
 343     }
 344     rec->rec_len += snprintf((str+rec->rec_len),
 345                              (len - rec->rec_len), "\n");
 346     va_end(list);
 347     csv->csv_len += (rec->rec_len - tlen);
 348     csv->pointer += (rec->rec_len - tlen);
 349     return (rec);
 350 }
 351
 352 int
 353 csv_serialize(csv_t *csv, char *msgbuf, int msglen)
 354 {
 355   csv_record_t *rec;
 356   int offset = 0;
 357
 358   if (!csv || !msgbuf) return -1;
 359
 360   rec = csv_record_iter(csv);
 361   while (rec != NULL) {
 362     if ((offset + rec->rec_len) >= msglen)
 363         return -1;
 364     offset += sprintf(&msgbuf[offset], "%s", rec->record);
 365     rec = csv_record_iter_next(rec);
 366   }
 367
 368   return 0;
 369 }
 370
 371 void
 372 csv_clone_record (csv_t *csv, csv_record_t *in_rec, csv_record_t **out_rec)
 373 {
 374     char *curr;
 375     csv_record_t *rec;
 376
 377     /* first check if rec belongs to this csv */
 378     if(!csv_is_record_valid(csv, in_rec)){
 379         log_error("rec not in this csv\n");
 380         return;
 381     }
 382
 383     /* only works with csv with discrete bufs */
 384     if (csv->buf) {
 385         log_error("un-supported for this csv type - single buf detected\n");
 386         return;
 387     }
 388
 389     /* create a new rec */
 390     rec = calloc(1, sizeof(csv_record_t));
 391     if (!rec) {
 392         log_error("record malloc failed\n");
 393         return;
 394     }
 395     csv_init_record(rec);
 396     curr = calloc(1, csv->buflen);
 397     if (!curr) {
 398         log_error("field str malloc failed\n");
 399         return;
 400     }
 401     rec->record = curr;
 402     rec->rec_len = in_rec->rec_len;
 403     strcpy(rec->record, in_rec->record);
 404
 405     /* decode record into fields */
 406     csv_decode_record(rec);
 407
 408     *out_rec = rec;
 409 }
 410
 411 void
 412 csv_remove_record (csv_t *csv, csv_record_t *rec)
 413 {
 414   csv_field_t *fld, *p_fld;
 415
 416   /* first check if rec belongs to this csv */
 417   if(!csv_is_record_valid(csv, rec)){
 418     log_error("rec not in this csv\n");
 419     return;
 420   }
 421
 422   /* remove fields */
 423   csv_field_iter(rec, &fld);
 424   while(fld) {
 425     p_fld = fld;
 426     csv_field_iter_next(&fld);
 427     TAILQ_REMOVE(&(rec->fields), p_fld, next_field);
 428     free(p_fld);
 429   }
 430
 431   TAILQ_REMOVE(&(csv->records), rec, next_record);
 432
 433   csv->num_recs--;
 434   csv->csv_len -= rec->rec_len;
 435   csv->pointer -= rec->rec_len;
 436   if (!csv->buf)
 437     free(rec->record);
 438   free(rec);
 439 }
 440
 441 void
 442 csv_insert_record (csv_t *csv, csv_record_t *rec)
 443 {
 444   /* first check if rec already in csv */
 445   if(csv_is_record_valid(csv, rec)){
 446     log_error("rec already in this csv\n");
 447     return;
 448   }
 449
 450   /* we can only insert records if no buf was supplied during csv init */
 451   if (csv->buf) {
 452     log_error("un-supported for this csv type - single buf detected\n");
 453     return;
 454   }
 455
 456   /* do we go beyond the max buf set for this csv ?*/
 457   if ((csv->csv_len + rec->rec_len) > csv->buflen ) {
 458     log_error("cannot insert - exceeded buf size\n");
 459     return;
 460   }
 461
 462   TAILQ_INSERT_TAIL(&(csv->records), rec, next_record);
 463   csv->num_recs++;
 464   csv->csv_len += rec->rec_len;
 465   csv->pointer += rec->rec_len;
 466 }
 467
 468 csv_record_t *
 469 csv_concat_record (csv_t *csv,
 470                            csv_record_t *rec1,
 471                            csv_record_t *rec2)
 472 {
 473   char *curr;
 474   char *ret;
 475   csv_record_t *rec;
 476
 477   /* first check if rec1 and rec2 belong to this csv */
 478   if(!csv_is_record_valid(csv, rec1) ||
 479      !csv_is_record_valid(csv, rec2)) {
 480     log_error("rec1 and/or rec2 invalid\n");
 481     return (NULL);
 482   }
 483
 484   /* we can only concat records if no buf was supplied during csv init */
 485   if (csv->buf) {
 486     log_error("un-supported for this csv type - single buf detected\n");
 487     return (NULL);
 488   }
 489
 490   /* create a new rec */
 491   rec = calloc(1, sizeof(csv_record_t));
 492   if (!rec) {
 493     log_error("record malloc failed\n");
 494     return (NULL);
 495   }
 496   csv_init_record(rec);
 497
 498   curr = (char *)calloc(1, csv->buflen);
 499   if (!curr) {
 500     log_error("field str malloc failed\n");
 501     return (NULL);
 502   }
 503   rec->record = curr;
 504
 505   /* concat the record string */
 506   ret = strstr(rec1->record, "\n");
 507   if (!ret) {
 508     log_error("rec1 str not properly formatted\n");
 509     return (NULL);
 510   }
 511
 512   snprintf(curr, (int)(ret - rec1->record + 1), "%s", rec1->record);
 513   strcat(curr, ",");
 514
 515   ret = strstr(rec2->record, "\n");
 516   if (!ret) {
 517     log_error("rec2 str not properly formatted\n");
 518     return (NULL);
 519   }
 520
 521   snprintf((curr+strlen(curr)), (int)(ret - rec2->record + 1), "%s",
 522            rec2->record);
 523   strcat(curr, "\n");
 524   rec->rec_len = strlen(curr);
 525
 526   /* paranoia */
 527   assert(csv->buflen >
 528          (csv->csv_len - rec1->rec_len - rec2->rec_len + rec->rec_len));
 529
 530   /* decode record into fields */
 531   csv_decode_record(rec);
 532
 533   /* now remove rec1 and rec2 and insert rec into this csv */
 534   csv_remove_record(csv, rec1);
 535   csv_remove_record(csv, rec2);
 536   csv_insert_record(csv, rec);
 537
 538   return rec;
 539 }
 540
 541 void
 542 csv_decode (csv_t *csv, char *inbuf)
 543 {
 544   char *buf;
 545   char *pos;
 546   csv_record_t *rec;
 547
 548   buf = (inbuf)? inbuf:csv->buf;
 549   pos = strpbrk(buf, "\n");
 550   while (pos != NULL) {
 551     rec = calloc(1, sizeof(csv_record_t));
 552     csv_init_record(rec);
 553     TAILQ_INSERT_TAIL(&(csv->records), rec, next_record);
 554     csv->num_recs++;
 555     if (csv->buf)
 556       rec->record = buf;
 557     else {
 558       rec->record = calloc(1, csv->buflen);
 559       if (!rec->record) {
 560         log_error("field str malloc failed\n");
 561         return;
 562       }
 563       strncpy(rec->record, buf, pos-buf+1);
 564     }
 565     rec->rec_len = pos-buf+1;
 566     /* decode record into fields */
 567     csv_decode_record(rec);
 568     buf = pos+1;
 569     pos = strpbrk(buf, "\n");
 570   }
 571 }
 572
 573 int
 574 csv_is_record_valid(csv_t *csv, csv_record_t *in_rec)
 575 {
 576   csv_record_t *rec;
 577   int valid = 0;
 578
 579   rec = csv_record_iter(csv);
 580   while (rec) {
 581     if(rec == in_rec) {
 582       valid = 1;
 583       break;
 584     }
 585     rec = csv_record_iter_next(rec);
 586   }
 587
 588   return valid;
 589 }
 590
 591 void
 592 csv_dump (csv_t *csv)
 593 {
 594   csv_record_t *rec;
 595   csv_field_t *fld;
 596   char *str;
 597
 598   rec = csv_record_iter(csv);
 599   while (rec != NULL) {
 600     str = csv_field_iter(rec, &fld);
 601     while (str != NULL) {
 602       fprintf(stderr, "%s\n", str);
 603       str = csv_field_iter_next(&fld);
 604     }
 605     rec = csv_record_iter_next(rec);
 606   }
 607 }
 608
 609 #ifdef TEST_CSV
 610
 611 static int
 612 get_memory_usage (pid_t pid)
 613 {
 614   int fd, data, stack;
 615   char buf[4096], status_child[BUFSIZ];
 616   char *vm;
 617
 618   sprintf(status_child, "/proc/%d/status", pid);
 619   if ((fd = open(status_child, O_RDONLY)) < 0)
 620     return -1;
 621
 622   read(fd, buf, 4095);
 623   buf[4095] = '\0';
 624   close(fd);
 625
 626   data = stack = 0;
 627
 628   vm = strstr(buf, "VmData:");
 629   if (vm) {
 630     sscanf(vm, "%*s %d", &data);
 631   }
 632   vm = strstr(buf, "VmStk:");
 633   if (vm) {
 634     sscanf(vm, "%*s %d", &stack);
 635   }
 636
 637   return data + stack;
 638 }
 639
 640 int main ()
 641 {
 642   char buf[10000];
 643   csv_t csv;
 644   int p;
 645   int i, j;
 646   csv_record_t *rec;
 647   csv_field_t *fld;
 648   char *str;
 649   char hdr1[32], hdr2[32];
 650
 651   log_verbose("Mem: %ld\n", get_memory_usage(getpid()));
 652   csv_init(&csv, buf, 256);
 653   sprintf(hdr1, "%4u", 0);
 654   sprintf(hdr2, "%4u", 1);
 655   log_verbose("(%d/%d/%d/%d)\n", strlen(hdr1),
 656           strlen(hdr2), atoi(hdr1), atoi(hdr2));
 657   rec = csv_encode(&csv, 2, hdr1, hdr2);
 658   csv_encode(&csv, 4, "name", "age", "sex", "hei");
 659   csv_encode(&csv, 3, NULL, "0", NULL);
 660   csv_encode(&csv, 2, "p", "35");
 661   for (i=0; i < 50; i++) {
 662     csv_encode(&csv, 2, "p", "10");
 663   }
 664   csv_encode(&csv, 2, "pdfadfadfadsadsaddfdfdsfdsd",
 665                  "35444554545454545");
 666   log_verbose("%s\n", buf);
 667   sprintf(hdr1, "%4u", csv.csv_len);
 668   sprintf(hdr2, "%4u", 1);
 669   log_verbose("(%d/%d/%d/%d)\n", strlen(hdr1),
 670           strlen(hdr2), atoi(hdr1), atoi(hdr2));
 671   rec = csv_encode_record(&csv, rec, 2, hdr1, hdr2);
 672   log_verbose("(%d/%d)\n%s\n", rec->rec_len, csv.csv_len, buf);
 673
 674   log_verbose("Mem: %ld\n", get_memory_usage(getpid()));
 675   csv_clean(&csv);
 676   log_verbose("Mem: %ld\n", get_memory_usage(getpid()));
 677   csv_init(&csv, buf, 256);
 678   csv_decode(&csv, NULL);
 679   log_verbose("AFTER DECODE\n");
 680   csv_dump(&csv);
 681   csv_clean(&csv);
 682   log_verbose("Mem: %ld\n", get_memory_usage(getpid()));
 683 }
 684 #endif