strtok with comma delimited string

[jtrapat1]

I'm trying to parse a comma delimited string with strtok -
It's working fine but I have a couple of special cases:
I'm reading from a file and one line of my 320-character string looks like this:
JONES, SMITH, ABBEY, JOHNSON, LONG ISLAND, STATEN ISLAND, ...
My program can handle most of these values but it has a problem with items such as LONG ISLAND.
It treats these as two separate tokens: LONG and ISLAND.
Is there any way to modify my code so I can handle these values?
Here's my code:
-------------------------------------------------
/* declared in main() */
char seps[] = ", ";
char *token;
char req_tmp[320+1];
/* where I read the string from the file */
strcpy(req_tmp, adds.req_by);
token = strtok(req_tmp, seps);

while(token != NULL)
{
strcpy(bp70.last_name,token);
get_mbr_id();
token = strtok(NULL, seps);
}
----------------------------------------------------
I tried different delimiters but because of my file layout, a blank space and then a comma separating each item, it gets confused near LONG ISLAND and treats this as two separate items.
Is there a way to modify my code to correctly parse this field?

Thanks in Advance
John

 

[glathaa]

John, you have given the separators as "," and " " in the strtok(). That is why you did not get the expected result.
strtok() function accepts a single character delimiters.

Write your own program using strstr() to do the name separation.

 

[nkm]

Hi

Small problem. The delimiter you have specified is ", ".
It is a comma and a space both are taken as delimiters so when you have a word like LONG ISLAND it is taking them as two tokens.
Use
char seps[] = ",";

It will work fine...

 

[marsd]

If these suggestions don't work I have a string splitting
function that does what you ask and returns an array of
the splits.

If you want I'll post it.

 

[jtrapat1]

marsd,
Yes, I'm still having a problem parsing thru my string.
Since my data was entered as: comma then space, it's a problem.
For example, I have JONES, SMITH, ALLEN, PARSON, ...

If you could post that string splitting function, I would appreciate it.

Thanks

John

 

[PieterWinter]

Hi jtrapat1

You could try to "strip" the result of strtok of its spaces (at the front and the back).

 

[jtrapat1]

Thanks for the posts.
I finally fixed the problem.
Had to trim the white space from both ends of the tokens.
Here's the function I ended up using:
------------------------------
char* strtrim(char* str) {
int i = 0;
int j;
while (str!= '\0' && str == ' ')
i = i + 1;
j = i;
while (str[j]!= '\0' && str[j] != ' ')
j = j + 1;
str[j] = '\0';
return &str;
}
------------------------------

Thanks Again.
John

 

[marsd]

Here they are.
Your method looks like it's better in any case.
I hacked them up so don't expect a lot.


char **retSplit(char *str, char sep) {
int cur = 0, acnt = 0, p, len;
char **new = NULL;
char *ppt = str,*tpt;


/*init*/
len = strlen(str);
if ( (p = numSeps(str,sep)) > 0) {
ppt = str;
new = malloc(p * sizeof(char *));
if (!new) {
return NULL;
}
}

/* main work here*/
p = 0;
while (*str) {
cur++;
if (*str == sep) {
printf("Found separator at: %d\n String is: %s\n",cur,ppt + cur);
new[acnt] = malloc((cur - p) + 1 * sizeof(char));
/*new string boundary*/
tpt = ppt + p;
printf("New string marker: %s\n", tpt);
strncpy(new[acnt],tpt,(cur - p) - 1);
termStr(new[acnt]);
p = cur;
acnt++;
}
*str++;
}

/*final cut*/
tpt = ppt + p ;
new[acnt] = malloc((len - p) + 1 * sizeof(char));
strncpy(new[acnt],tpt,(len - p));
/*return the resulting array*/
return new;
}

NumSeps() just goes through the char array looking
for sep and returns the number of matches + 1 for the
array allocation.

int numSeps(char *str, char sep) {
int m = 0;

while (*str) {
if (*str == sep) {
m++;
}
*str++;
}
return m + 1;
}

termStr() takes a char array, reallocs the array size + 1
and terminates len + 1. Like strtok it modifies the
original char array.
void termStr(char *str) {
int m = 0;
char *sp = str;
while (*sp++) {
m++;
}
realloc(str,(m + 1) * sizeof(char));
str[m + 1] = '\0';
}

Sample run without debugging printf's reading from
stdin.
>tstr "this, is a tes,tof the,emergency"
this at 0
is a tes at 1
tof the at 2
emergency at 3