Removing single dot path names in URL in C - c

Question

I'm making a function in an apache module which is supposed to fix URLs that are thrown at it. Currently I'm trying to remove single dot path names.
For example, if my URL is:
http://example.com/1/./2/./3/./4.php
Then I want the URL to be:
http://example.com/1/2/3/4.php
However I'm stuck with the logic. I'm using pointers in an effort to make this function run as fast as possible. I'm confused at the logic I should apply at the lines with //? added to the end of them.
Can someone give me advice on how to proceed? Even if its some hidden manual online? I searched bing and google for answers with no success.
static long fixurl(char *u){
char u1[10000];
char *u11=u1,*uu1=u;
long ct=0,fx=0;
while (*uu1){
*u11=*uu1;
if (*uu1=='/'){
ct++;
if (ct >=2){
uu1++;
break;
}
} else {
ct=0;
}
}
while (*uu1){
if (*uu1!='/') { //?
if (*uu1!='.') {
*u11=*uu1;
u11++;
} //?
} //?
uu1++;
}
*u11='\0';
strcpy(u,u1);
return fx;
}

Answer 1

You forget to look ahead one character here:
if (*uu1!='/') { //?
if (*uu1!='.') {
– you are checking the same character twice (against a 'not', so it could have some use, but your question marks indicate you are not sure what to do there and further on).
Note that you actually need to look ahead two characters. If you encounter a slash, test the next character for a . and the one after that for another /.
Rather than trying to fix your code (what is fx, the returned value, supposed to be?), I'd rewrite it from scratch to copy from source to dest and skip the offending sections. The continue makes sure that a sequence /1/././2 gets cleansed correctly to just /1/2 – it needs a chance to check the second slash again, so I just throw it back into the loop.
void fixurl (char *theUrl)
{
char *source, *dest;
source = dest = theUrl;
while (*source)
{
if (source[0] == '/' && source[1] == '.' && source[2] == '/')
{
source += 2; /* effectively, 'try again on the next slash' */
} else
{
*dest = *source;
source++;
dest++;
}
}
*dest = 0;
}
(Afterthought:)
Interestingly, adding proper support for removal of /../ is fairly trivial. If you test for that sequence, you should search backwards for the last / before it and reset dest to that position. You'll want to make sure the path is still valid, though.

Answer 2

This code is untested. In short, it is iterating the string (until the next character is the end sign, since if there is no next character, then you can no longer have a problem) and searches for '/'. When it finds one, analyzes the next character and handles it.
static long fixurl(char *u){
char u1[10000];
int currentIndex = 0;
if (*u == '\0') {
return 0;
}
for (; *(u + 1) != '\0'; u++){
if (*u == '/') {
if (*(u + 1) == '/') {
continue;
} else if ((*(u + 1) == '.') && (*(u + 2) == '.')) {
u++;
continue;
}
}
u1[currentIndex] = *u;
}
strcpy(u,u1);
return currentIndex;
}

Answer 3

here is a version of the code that works
Note it will remove all '.' that follow a '/'
However, it does not check for extraneous '/' characters being inserted into the output as the OPs posted code does not make that check.
Notice the proper formatting of the for() statement
Notice the use of meaningful names, removal of code clutter,
inclusion of a few key comments, etc
Notice the literal characters are placed on the left side of a comparison so writing a '=' when it should be '==' is caught by the compiler.
#include <string.h>
long fixurl( char * );
long fixurl(char *rawURL)
{
char cookedURL[10000] = {'\0'}; // assure new string is terminated
int currentIndex = 0;
cookedURL[currentIndex] = rawURL[0];
rawURL++;
for ( ; *rawURL; rawURL++)
{
// if prior saved char was / and current char is .
// then skip current char
if( ( '/' != cookedURL[currentIndex] )
||
( '.' != *rawURL ))
{
// copy input char to out buffer
currentIndex++;
cookedURL[currentIndex] = *rawURL;
}
} // end for
// copy modified URL back to caller's buffer
strcpy(rawURL, cookedURL);
return currentIndex+1; // number of characters in modified buffer
} // end function: fixurl