URL parser in C

Question

Title should be self-explenatory. This is a simple URL parser I wrote in C. The function takes a URL from the user and produces a struct that contains the information that can be used to request the resource at the URL over http(s).

enum url_protocol {
 PROTOCOL_HTTP,
 PROTOCOL_HTTPS
};
struct url {
 /* Protocol to use */
 enum url_protocol protocol;
 /* Credentials */
 char *username;
 char *password;
 /* Host and service */
 char *host;
 char *service;
 /* Path to request */
 char *path;
};
void
free_url(struct url *url);
int
parse_url(char *start, struct url *url)
{
 char *end, *delim;
 memset(url, 0, sizeof(*url));
 /* Find the end of the scheme */
 end = strchr(start, ':');
 if (!end)
 goto err;
 /* Find protocol from scheme */
 if (!strncmp(start, "http", end - start))
 url->protocol = PROTOCOL_HTTP;
 else if (!strncmp(start, "https", end - start))
 url->protocol = PROTOCOL_HTTPS;
 else
 goto err;
 /* URLs must begin with // */
 if (*++end != '/' || *++end != '/')
 goto err;
 /* Parse credentials */
 start = ++end;
 end = strchr(start, '@');
 if (end) {
 delim = strchr(start, ':');
 if (delim && delim < end) { /* Username and password */
 if (!(delim - start))
 goto err;
 url->username = strndup(start, delim - start);
 ++delim;
 if (!(end - delim))
 goto err;
 url->password = strndup(delim, end - delim);
 } else { /* Only username */
 if (!(end - start))
 goto err;
 url->username = strndup(start, end - start);
 }
 start = ++end;
 }
 /* Host till / or end of string */
 end = strchrnul(start, '/');
 /* Skip IPv6 literals before : search */
 if (*start == '[' && (delim = strchr(start, ']')))
 delim = strchr(delim, ':'); /* Found IPv6 literal */
 else
 delim = strchr(start, ':'); /* Normal search */
 if (delim && delim < end) { /* Host and service */
 if (!(delim - start))
 goto err;
 url->host = strndup(start, delim - start);
 ++delim;
 if (!(end - delim))
 goto err;
 url->service = strndup(delim, end - delim);
 } else { /* Only host */
 if (!(end - start))
 goto err;
 url->host = strndup(start, end - start);;
 }
 /* Default path is a single / */
 url->path = *end ? strdup(end) : strdup("/");
 return 0;
err:
 free_url(url);
 return -1;
}
void
free_url(struct url *url)
{
 /* Assume free ignores NULL */
 free(url->username);
 free(url->password);
 free(url->host);
 free(url->service);
 free(url->path);
}
```

• Sugar: does it meant to be rfc 1738 compatible?

Answer 1

Just some minor things in your code. Is a bit strange that you have a function call free_url and you dont have one called init_url where you alloc and do the memset. My suggestion is that you have another function for that and you move your memset of the function

int
parse_url(char *start, struct url *url)
{
 char *end, *delim;
 memset(url, 0, sizeof(*url)); <- Move this to init_url

And your function

 struct url *init_url()
 {
 struct url *u = (struct url*)malloc(sizeof(struct url));
 if (u != NULL)
 memset(u, 0, sizeof(struct url));
 return u
 } 

My recommendation is that you should have a good suite of unit tests for parsing things for the user, so you will make your code more robust and with less errors.

Hope it helps

• mateass: There is no init_url on purpose, the struct how its used now is allocated on the callers stack instead of on the heap. The only job of the free_url is to free the strings.