how to parse URLs in c using sscanf() ?

Question

Hi,

this is my c code that reads a list of URLs from a file , and tries to separate the various parts of the URL.. This is just rough parsing , i'm not bothered about special cases.. I guess there is some fault with the sscanf() statement , when i run this , i get " segmentation FAULT" .. and moreover , the full url is being assigned to the "proto" string.. can anyone please help me figure out the problem.. please , it's urgent..

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

# define BAG_SIZE 14

char bag[117][30];

void initbag()
{
strcpy(bag[0],"account");
strcpy(bag[1],"audit");
strcpy(bag[2],"ad");
strcpy(bag[3],"advertising");
strcpy(bag[4],"marketing");
strcpy(bag[5],"application");
strcpy(bag[6],"banking");
strcpy(bag[7],"barter");
strcpy(bag[8],"business");
strcpy(bag[9],"econo");
strcpy(bag[10],"commerce");
strcpy(bag[11],"communication");
strcpy(bag[12],"computer");
strcpy(bag[13],"processing");
}
/*
 other bag[] values will be later copied
*/

void substr(char dest[10],char src[200],int start,int len)
{
int i,j;

for(i=start,j=0;i<start+len;i++,j++)
dest[j]=src[i];
dest[j]='\0';

}

int found(char* word)
{
   int i;
   for(i=0;i<BAG_SIZE;i++)
   if((!strcmp(word,bag[i]))||(strstr(bag[i],word)!=NULL)) return 1;
   return 0;
}

void main()
{
int i,j,k;

char buff[10],fullurl[100];
char proto[5],www[4],host[100],tokens[200],tld[4];
float feature[11];for(i=0;i<11;i++) feature[i]=0;
FILE *furl,*fop;
furl=fopen("bizurls.txt","r");
fop=fopen("urlsvm.txt","w");
initbag();
printf("\nbag initialised");fflush(stdout);

while(!feof(furl))
{
   fscanf(furl,"%s",fullurl);
   printf("%s",fullurl);
   sscanf(fullurl,"%s://%s.%s.%s/%s\n",proto,www,host,tld,tokens);// this line isnt working properly
   printf("2hi");fflush(stdout);
   printf("proto : %s\nwww:%s\nhost :%s\ntld:%s\ntokens:%s\n",proto,www,host,tld,tokens);fflush(stdout);


   for( i=4;i<=8;i++)
   {
       for(j=0;j<strlen(host)-i+1;j++)
           {
                substr(buff,host,j,i);
                if(found(buff)) feature[i-3]++;

           }
   }
  if((!strcmp(tld,"biz"))||(!strcmp(tld,"org"))||(!strcmp(tld,"com"))||(!strcmp(tld,"jobs")))   
        feature[0]=1;
  else if((!strcmp(tld,"info"))||(!strcmp(tld,"coop"))||(!strcmp(tld,"net")))
        feature[0]=0.5;
  else
    feature[0]=0;


   for( i=4;i<=8;i++)
   {
       for(j=0;j<strlen(tokens)-i+1;j++)
           {
                substr(buff,tokens,j,i);
                if(found(buff)) feature[i+2]++;

           }
   }

/*.biz · .com · .info · .name · .net · .org · .pro
.aero, .coop, .jobs, .travel */

for(i=0;i<11;i++) fprintf(fop," %d:%f",i,feature[i]);
fprintf(fop,"\n");


}
fflush(fop);
fclose(furl);
fclose(fop);
}

Answer 1

It won't work because proto would match the whole fullurl and the rest will be unmatched. You should use a proper URL parsing function or regex for this.

Answer 2

%s in sscanf will only stop when it hits the first white-space character, the end of the string or the specified maximum length. Seeing as an URL has no whitespace, that's the reason why proto becomes fullurl.

For the segmentation fault: as proto can only hold 5 bytes (including the trailing null, so only 4 bytes of data which would not cover e.g. https), putting the full URL into it will cause a buffer overflow / segmentation fault. sscanf is rather problematic in this regards. Documentation requests that each char buffer receiving a %s should be big enough to hold the full string (plus \0).

Answer 3

Many answers here:
http://stackoverflow.com/questions/726122/best-ways-of-parsing-a-url-using-c