int websUrlParse(char_t *url, char_t **pbuf, char_t **phost, char_t **ppath,
char_t **pport, char_t **pquery, char_t **pproto, char_t **ptag,
char_t **pext)
{
char_t *tok, *cp, *host, *path, *port, *proto, *tag, *query, *ext;
char_t *last_delim, *hostbuf, *portbuf, *buf;
int c, len, ulen;
a_assert(url);
a_assert(pbuf);
ulen = gstrlen(url);
/*
* We allocate enough to store separate hostname and port number fields.
* As there are 3 strings in the one buffer, we need room for 3 null chars.
* We allocate MAX_PORT_LEN char_t's for the port number.
*/
len = ulen * 2 + MAX_PORT_LEN + 3;
if ((buf = balloc(B_L, len * sizeof(char_t))) == NULL) {
return -1;
}
portbuf = &buf[len - MAX_PORT_LEN - 1];
hostbuf = &buf[ulen+1];
gstrcpy(buf, url);
url = buf;
/*
* Convert the current listen port to a string. We use this if the URL has
* no explicit port setting
*/
stritoa(websGetPort(), portbuf, MAX_PORT_LEN);
port = portbuf;
path = T("/");
proto = T("http");
host = T("localhost");
query = T("");
ext = htmExt;
tag = T("");
if (gstrncmp(url, T("http://"), 7) == 0) {
tok = &url[7];
tok[-3] = '\0';
proto = url;
host = tok;
for (cp = tok; *cp; cp++) {
if (*cp == '/') {
break;
}
if (*cp == ':') {
*cp++ = '\0';
port = cp;
tok = cp;
}
}
if ((cp = gstrchr(tok, '/')) != NULL) {
/*
* If a full URL is supplied, we need to copy the host and port
* portions into static buffers.
*/
c = *cp;
*cp = '\0';
gstrncpy(hostbuf, host, ulen);
gstrncpy(portbuf, port, MAX_PORT_LEN);
*cp = c;
host = hostbuf;
port = portbuf;
path = cp;
tok = cp;
}
} else {
path = url;
tok = url;
}
/*
* Parse the query string
*/
if ((cp = gstrchr(tok, '?')) != NULL) {
*cp++ = '\0';
query = cp;
path = tok;
tok = query;
}
/*
* Parse the fragment identifier
*/
if ((cp = gstrchr(tok, '#')) != NULL) {
*cp++ = '\0';
if (*query == 0) {
path = tok;
}
}
/*
* Only do the following if asked for the extension
*/
if (pext) {
if ((cp = gstrrchr(path, '.')) != NULL) {
if ((last_delim = gstrrchr(path, '/')) != NULL) {
if (last_delim > cp) {
ext = htmExt;
} else {
ext = cp;
}
} else {
ext = cp;
}
} else {
if (path[gstrlen(path) - 1] == '/') {
ext = htmExt;
}
}
}
/*
* Pass back the fields requested (if not NULL)
*/
if (phost)
*phost = host;
if (ppath)
*ppath = path;
if (pport)
*pport = port;
if (pproto)
*pproto = proto;
if (pquery)
*pquery = query;
if (ptag)
*ptag = tag;
if (pext)
*pext = ext;
*pbuf = buf;
return 0;
}
int websUrlParse(char_t *url, char_t **pbuf, char_t **phost, char_t **ppath,
char_t **pport, char_t **pquery, char_t **pproto, char_t **ptag,
char_t **pext)
{
char_t *tok, *cp, *host, *path, *port, *proto, *tag, *query, *ext;
char_t *hostbuf, *portbuf, *buf;
int c, len, ulen;
a_assert(url);
a_assert(pbuf);
ulen = gstrlen(url);
/*
* We allocate enough to store separate hostname and port number fields.
* As there are 3 strings in the one buffer, we need room for 3 null chars.
* We allocate MAX_PORT_LEN char_t's for the port number.
*/
len = ulen * 2 + MAX_PORT_LEN + 3;
if ((buf = balloc(B_L, len * sizeof(char_t))) == NULL) {
return -1;
}
portbuf = &buf[len - MAX_PORT_LEN - 1];
hostbuf = &buf[ulen+1];
/*
Handle any URL encoding.
Otherwise a URL ending in ".as%70", for example, causes trouble.
*/
websDecodeUrl(buf, url, ulen);
url = buf;
/*
* Convert the current listen port to a string. We use this if the URL has
* no explicit port setting
*/
stritoa(websGetPort(), portbuf, MAX_PORT_LEN);
port = portbuf;
path = T("/");
proto = T("http");
host = T("localhost");
query = T("");
ext = htmExt;
tag = T("");
if (gstrncmp(url, T("http://"), 7) == 0) {
tok = &url[7];
tok[-3] = '\0';
proto = url;
host = tok;
for (cp = tok; *cp; cp++) {
if (*cp == '/') {
break;
}
if (*cp == ':') {
*cp++ = '\0';
port = cp;
tok = cp;
}
}
if ((cp = gstrchr(tok, '/')) != NULL) {
/*
* If a full URL is supplied, we need to copy the host and port
* portions into static buffers.
*/
c = *cp;
*cp = '\0';
gstrncpy(hostbuf, host, ulen);
gstrncpy(portbuf, port, MAX_PORT_LEN);
*cp = c;
host = hostbuf;
port = portbuf;
path = cp;
tok = cp;
}
} else {
path = url;
tok = url;
}
/*
* Parse the query string
*/
if ((cp = gstrchr(tok, '?')) != NULL) {
*cp++ = '\0';
query = cp;
path = tok;
tok = query;
}
/*
* Parse the fragment identifier
*/
if ((cp = gstrchr(tok, '#')) != NULL) {
*cp++ = '\0';
if (*query == 0) {
path = tok;
}
}
/*
* Only do the following if asked for the extension
*/
if (pext) {
/*
Later the path will be cleaned up for trailing slashes and so on.
To be ready, we need to clean up here, much as in websValidateUrl.
Otherwise a URL ending in "asp/" or "asP" sends Ejscript source
to the browser.
*/
if ((cp = gstrrchr(path, '.')) != NULL) {
const char_t* garbage = T("/\\");
int length = gstrcspn(cp, garbage);
int garbageLength = gstrspn(cp + length, garbage);
int ok = (length + garbageLength == (int) gstrlen(cp));
if (ok) {
cp[length] = '\0';
#ifdef WIN
strlower(cp);
#endif
ext = cp;
}
}
}
/*
* Pass back the fields requested (if not NULL)
*/
if (phost)
*phost = host;
if (ppath)
*ppath = path;
if (pport)
*pport = port;
if (pproto)
*pproto = proto;
if (pquery)
*pquery = query;
if (ptag)
*ptag = tag;
if (pext)
*pext = ext;
*pbuf = buf;
return 0;
}
