src/libs/ck-libs/metis/GKlib/tokenizer.c

   1 /*!
   2 \file  tokenizer.c
   3 \brief String tokenization routines
   4
   5 This file contains various routines for splitting an input string into
   6 tokens and returning them in form of a list. The goal is to mimic perl's
   7 split function.
   8
   9 \date   Started 11/23/04
  10 \author George
  11 \version\verbatim $Id: tokenizer.c 10711 2011-08-31 22:23:04Z karypis $ \endverbatim
  12 */
  13
  14
  15 #include <GKlib.h>
  16
  17
  18 /************************************************************************
  19 * This function tokenizes a string based on the user-supplied delimiters
  20 * list. The resulting tokens are returned into an array of strings.
  21 *************************************************************************/
  22 void gk_strtokenize(char *str, char *delim, gk_Tokens_t *tokens)
  23 {
  24   int i, ntoks, slen;
  25
  26   tokens->strbuf = gk_strdup(str);
  27
  28   slen  = strlen(str);
  29   str   = tokens->strbuf;
  30
  31   /* Scan once to determine the number of tokens */
  32   for (ntoks=0, i=0; i<slen;) {
  33     /* Consume all the consecutive characters from the delimiters list */
  34     while (i<slen && strchr(delim, str[i]))
  35       i++;
  36
  37     if (i == slen)
  38       break;
  39
  40     ntoks++;
  41
  42     /* Consume all the consecutive characters from the token */
  43     while (i<slen && !strchr(delim, str[i]))
  44       i++;
  45   }
  46
  47
  48   tokens->ntoks = ntoks;
  49   tokens->list  = (char **)gk_malloc(ntoks*sizeof(char *), "strtokenize: tokens->list");
  50
  51
  52   /* Scan a second time to mark and link the tokens */
  53   for (ntoks=0, i=0; i<slen;) {
  54     /* Consume all the consecutive characters from the delimiters list */
  55     while (i<slen && strchr(delim, str[i]))
  56       str[i++] = '\0';
  57
  58     if (i == slen)
  59       break;
  60
  61     tokens->list[ntoks++] = str+i;
  62
  63     /* Consume all the consecutive characters from the token */
  64     while (i<slen && !strchr(delim, str[i]))
  65       i++;
  66   }
  67 }
  68
  69
  70 /************************************************************************
  71 * This function frees the memory associated with a gk_Tokens_t
  72 *************************************************************************/
  73 void gk_freetokenslist(gk_Tokens_t *tokens)
  74 {
  75   gk_free((void *)&tokens->list, &tokens->strbuf, LTERM);
  76 }
  77