reformatting fields

version gcc 2.95.3
SunOS 5.6

Hi,

I'm a bit stuck, I have to open a text file, read records and write them out to a new text file. The records are delimited by a semi-colon, there are between 80 and 90 fields, and about 40/50,000 records per file per month. A little less than half of them have to be formatted slightly differently, for example

the fifth field can look like this

;1724470.00;
;774375.00;
;17022.00;
;3824571.00;

and needs to look like an integer(20)

;1724470;
;774375;
;17022;
;3824571;

Should I write out each field in a seperate fprintf statement ?


tia,

>>Should I write out each field in a seperate fprintf statement ?
Either way is fine, but you may find that performance is affected when you run large amounts of data. Design your program the way you feel most comfortable with, then if it's too slow, start the fine tuning process. Most times you'll find that a program runs quick enough, and spending hours of coding time to save 2 seconds of runtime is a waste (not all the time though ).


>>and needs to look like an integer(20)
Remember, you don't need to convert the input from a char array to an in to do that. You are simply ending the string at the decimal point. This *may* save you some time, depending on how you're processing the line (ie how you read it in).

thanks Hammer

don't know if you can help me with this,

Code:

void populate_output_record(RecordIn * input, RecordOut * output)
{
#ifdef DEBUG
    printf("In populate_output_record().\n");
    fflush(NULL);
#endif

   memset(output, SPACE_INT, sizeof(RecordOut));

   memcpy(output->country,                  input->country,                  I_COUNTRY_LEN);
   memcpy(output->region_code,              input->region_code,              I_REGION_CODE_LEN);
   memcpy(output->region_name,              input->region_name,              I_REGION_NAME_LEN);
   memcpy(output->country_iso3_code,        input->country_iso3_code,        I_COUNTRY_ISO3_CODE_LEN];
   memcpy(output->country_iso2_code,        input->country_iso2_code,        I_COUNTRY_ISO2_CODE_LEN];
   memcpy(output->entity_code,              input->entity_code,              I_ENTITY_CODE_LEN];
   memcpy(output->entity_name,              input->entity_name,              I_ENTITY_NAME_LEN];

   memcpy(output->cirprecno,                fprint(input, "%d", cirprecno),                I_CIRPRECNO_LEN];

   memcpy(output->customer_defined_code,    input->customer_defined_code     I_CUSTOMER_DEFINED_CODE_LEN];
   memcpy(output->service_order_ref,        input->service_order_ref,        I_SERVICE_ORDER_REF_LEN];

etc.,

cirprecno is the field thats looks like 1724470.00
and needs to look like 1724470

but I cannot workout how to format it while processing


tia,

You can't really use memcpy for this, as the length of the fields varies. For example, country names are different length, so copying a fixed amount of data won't work. I am presuming your input isn't padded with blanks to fixed field lengths, that would be kinda pointless in a comma seperated file.

You might find this of use:
http://faq.cprogramming.com/cgi-bin/...&id=1044780608

Also, this might give you some clues:

Code:

#include <stdio.h>
#include <string.h>

int main(void)
{
  char a[] = "blah;one;two";
  char *p1, *p2;
  
  p1 = strchr(a, ';');
  if (p1)
  {
    p1++;
    puts (p1);
    p2 = strchr(p1, ';');
    if (p2)
    {
      printf ("%.*s\n", p2 - p1, p1);
    }
  }
    
  return(0);
}

/* Output:
  
  ;one;two
  one
  */

aah -

the penny drops



thanks Hammer