Line data    Source code

       1             : /*********************************************************************
       2             :  *
       3             :  * File        :  $Source: /cvsroot/ijbswa/current/ssplit.c,v $
       4             :  *
       5             :  * Purpose     :  A function to split a string at specified delimiters.
       6             :  *
       7             :  * Copyright   :  Written by and Copyright (C) 2001-2012 the
       8             :  *                Privoxy team. https://www.privoxy.org/
       9             :  *
      10             :  *                Based on the Internet Junkbuster originally written
      11             :  *                by and Copyright (C) 1997 Anonymous Coders and
      12             :  *                Junkbusters Corporation.  http://www.junkbusters.com
      13             :  *
      14             :  *                This program is free software; you can redistribute it
      15             :  *                and/or modify it under the terms of the GNU General
      16             :  *                Public License as published by the Free Software
      17             :  *                Foundation; either version 2 of the License, or (at
      18             :  *                your option) any later version.
      19             :  *
      20             :  *                This program is distributed in the hope that it will
      21             :  *                be useful, but WITHOUT ANY WARRANTY; without even the
      22             :  *                implied warranty of MERCHANTABILITY or FITNESS FOR A
      23             :  *                PARTICULAR PURPOSE.  See the GNU General Public
      24             :  *                License for more details.
      25             :  *
      26             :  *                The GNU General Public License should be included with
      27             :  *                this file.  If not, you can view it at
      28             :  *                http://www.gnu.org/copyleft/gpl.html
      29             :  *                or write to the Free Software Foundation, Inc., 59
      30             :  *                Temple Place - Suite 330, Boston, MA  02111-1307, USA.
      31             :  *
      32             :  *********************************************************************/
      33             : 
      34             : 
      35             : #include "config.h"
      36             : 
      37             : #include <string.h>
      38             : #include <stdlib.h>
      39             : 
      40             : #include "ssplit.h"
      41             : #include "miscutil.h"
      42             : 
      43             : 
      44             : /*********************************************************************
      45             :  *
      46             :  * Function    :  ssplit
      47             :  *
      48             :  * Description :  Split a string using delimiters in `delim'.  Results
      49             :  *                go into `vec'.
      50             :  *
      51             :  * Parameters  :
      52             :  *          1  :  str = string to split.  Will be split in place
      53             :  *                (i.e. do not free until you've finished with vec,
      54             :  *                previous contents will be trashed by the call).
      55             :  *          2  :  delim = array of delimiters (if NULL, uses " \t").
      56             :  *          3  :  vec[] = results vector (aka. array) [out]
      57             :  *          4  :  vec_len = number of usable slots in the vector (aka. array size)
      58             :  *
      59             :  * Returns     :  -1 => Error: vec_len is too small to hold all the
      60             :  *                      data, or str == NULL.
      61             :  *                >=0 => the number of fields put in `vec'.
      62             :  *                On error, vec and str may still have been overwritten.
      63             :  *
      64             :  *********************************************************************/
      65      137742 : int ssplit(char *str, const char *delim, char *vec[], size_t vec_len)
      66             : {
      67             :    unsigned char is_delim[256];
      68             :    unsigned char char_type;
      69      137742 :    int vec_count = 0;
      70             :    enum char_type {
      71             :       WANTED     = 0,
      72             :       SEPARATOR  = 1,
      73             :       TERMINATOR = 2,
      74             :    };
      75             : 
      76             : 
      77      137742 :    if (!str)
      78             :    {
      79           0 :       return(-1);
      80             :    }
      81             : 
      82             : 
      83             :    /* Build is_delim array */
      84             : 
      85      137742 :    memset(is_delim, '\0', sizeof(is_delim));
      86             : 
      87      137742 :    if (!delim)
      88             :    {
      89           0 :       delim = " \t";  /* default field separators */
      90             :    }
      91             : 
      92      350628 :    while (*delim)
      93             :    {
      94      212886 :       is_delim[(unsigned)(unsigned char)*delim++] = SEPARATOR;
      95             :    }
      96             : 
      97      137742 :    is_delim[(unsigned)(unsigned char)'\0'] = TERMINATOR;
      98      137742 :    is_delim[(unsigned)(unsigned char)'\n'] = TERMINATOR;
      99             : 
     100             : 
     101             :    /* Parse string */
     102             : 
     103             :    /* Skip leading separators. XXX: Why do they matter? */
     104      158103 :    while (is_delim[(unsigned)(unsigned char)*str] == SEPARATOR)
     105             :    {
     106       20361 :       str++;
     107             :    }
     108             : 
     109             :    /* The first pointer is the beginning of string */
     110      137742 :    if (is_delim[(unsigned)(unsigned char)*str] == WANTED)
     111             :    {
     112             :       /*
     113             :        * The first character in this field is not a
     114             :        * delimiter or the end of string, so save it.
     115             :        */
     116      128636 :       if (vec_count >= vec_len)
     117             :       {
     118           0 :          return(-1); /* overflow */
     119             :       }
     120      128636 :       vec[vec_count++] = str;
     121             :    }
     122             : 
     123     6436497 :    while ((char_type = is_delim[(unsigned)(unsigned char)*str]) != TERMINATOR)
     124             :    {
     125     6298841 :       if (char_type == SEPARATOR)
     126             :       {
     127             :          /* the char is a separator */
     128             : 
     129             :          /* null terminate the substring */
     130      342196 :          *str++ = '\0';
     131             : 
     132             :          /* Check if we want to save this field */
     133      342196 :          if (is_delim[(unsigned)(unsigned char)*str] == WANTED)
     134             :          {
     135             :             /*
     136             :              * The first character in this field is not a
     137             :              * delimiter or the end of string. So save it.
     138             :              */
     139      264081 :             if (vec_count >= vec_len)
     140             :             {
     141          86 :                return(-1); /* overflow */
     142             :             }
     143      263995 :             vec[vec_count++] = str;
     144             :          }
     145             :       }
     146             :       else
     147             :       {
     148     5956645 :          str++;
     149             :       }
     150             :    }
     151             :    /* null terminate the substring */
     152      137656 :    *str = '\0';
     153             : 
     154      137656 :    return(vec_count);
     155             : }
     156             : 
     157             : 
     158             : /*
     159             :   Local Variables:
     160             :   tab-width: 3
     161             :   end:
     162             : */