/* COMMAND "getUser": Extract the user identity from the URL
* SIGNATURE: getUser(str) : str; */
str
URLgetUser(str *retval, url *val)
{
const char *s;
const char *p;
const char *u;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getUser", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(p = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(s = skip_path(p, NULL, NULL)) == NULL)
throw(ILLARG, "url.getUser", "bad url");
if (p == s || *p != '/' || p[1] != '~') {
*retval = GDKstrdup(str_nil);
} else {
size_t l;
u = p + 2;
for (p = u; p < s && *p != '/'; p++)
;
l = p - u;
if ((*retval = GDKmalloc(l + 1)) != NULL) {
strncpy(*retval, u, l);
(*retval)[l] = 0;
}
}
if (*retval == NULL)
throw(MAL, "url.getUser", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getQuery": Extract the query part from the URL
* SIGNATURE: getQuery(str) : str; */
str
URLgetQuery(str *retval, url *val)
{
const char *s;
const char *q;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getQuery", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(q = skip_path(s, NULL, NULL)) == NULL ||
(s = skip_search(q)) == NULL)
throw(ILLARG, "url.getQuery", "bad url");
if (*q == '?') {
size_t l;
q++;
l = s - q;
if ((*retval = GDKmalloc(l + 1)) != NULL) {
strncpy(*retval, q, l);
(*retval)[l] = 0;
}
} else {
*retval = GDKstrdup(str_nil);
}
if (*retval == NULL)
throw(MAL, "url.getQuery", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getDomain": Extract the Internet domain from the URL
* SIGNATURE: getDomain(str) : str; */
str
URLgetDomain(str *retval, url *val)
{
const char *s;
const char *h = NULL;
const char *p = NULL;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getDomain", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, &h, &p)) == NULL)
throw(ILLARG, "url.getDomain", "bad url");
if (h == NULL) {
*retval = GDKstrdup(str_nil);
} else {
size_t l;
if (p != NULL)
p--;
else
p = s;
l = 0;
while (p > h && p[-1] != '.') {
p--;
l++;
}
if ((*retval = GDKmalloc(l + 1)) != NULL) {
strncpy(*retval, p, l);
(*retval)[l] = 0;
}
}
if (*retval == NULL)
throw(MAL, "url.getDomain", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getPort": Extract the port id from the URL
* SIGNATURE: getPort(str) : str; */
str
URLgetPort(str *retval, url *val)
{
const char *s;
const char *p = NULL;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getPort", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, &p)) == NULL)
throw(ILLARG, "url.getPort", "bad url");
if (p == NULL) {
*retval = GDKstrdup(str_nil);
} else {
size_t l = s - p;
if ((*retval = GDKmalloc(l + 1)) != NULL) {
strncpy(*retval, p, l);
(*retval)[l] = 0;
}
}
if (*retval == NULL)
throw(MAL, "url.getPort", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getBasename": Extract the base of the last file name of the URL,
* thus, excluding the file extension.
* SIGNATURE: getBasename(str) : str; */
str
URLgetBasename(str *retval, url *val)
{
const char *s;
const char *b = NULL;
const char *e = NULL;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getBasename", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(s = skip_path(s, &b, &e)) == NULL)
throw(ILLARG, "url.getBasename", "bad url");
if (b == NULL) {
*retval = GDKstrdup(str_nil);
} else {
size_t l;
if (e != NULL) {
l = e - b;
} else {
l = s - b;
}
if ((*retval = GDKmalloc(l + 1)) != NULL) {
strncpy(*retval, b, l);
(*retval)[l] = 0;
}
}
if (*retval == NULL)
throw(MAL, "url.getBasename", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getExtension": Extract the file extension of the URL
* SIGNATURE: getExtension(str) : str; */
str
URLgetExtension(str *retval, url *val)
{
const char *s;
const char *e = NULL;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getExtension", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(s = skip_path(s, NULL, &e)) == NULL)
throw(ILLARG, "url.getExtension", "bad url");
if (e == NULL) {
*retval = GDKstrdup(str_nil);
} else {
size_t l = s - e;
assert(*e == '.');
if ((*retval = GDKmalloc(l)) != NULL) {
strncpy(*retval, e + 1, l - 1);
(*retval)[l - 1] = 0;
}
}
if (*retval == NULL)
throw(MAL, "url.getExtension", "Allocation failed");
return MAL_SUCCEED;
}
int
xps_url_is_remote(char *path)
{
char *p = skip_authority(skip_scheme(path));
return p != path;
}
static char *
xps_clean_path(char *name)
{
char *p, *q, *dotdot, *start;
int rooted;
start = skip_scheme(name);
start = skip_authority(start);
rooted = start[0] == '/';
/*
* invariants:
* p points at beginning of path element we're considering.
* q points just past the last path element we wrote (no slash).
* dotdot points just past the point where .. cannot backtrack
* any further (no slash).
*/
p = q = dotdot = start + rooted;
while (*p)
{
if(p[0] == '/') /* null element */
p++;
else if (p[0] == '.' && SEP(p[1]))
p += 1; /* don't count the separator in case it is nul */
else if (p[0] == '.' && p[1] == '.' && SEP(p[2]))
{
p += 2;
if (q > dotdot) /* can backtrack */
{
while(--q > dotdot && *q != '/')
;
}
else if (!rooted) /* /.. is / but ./../ is .. */
{
if (q != start)
*q++ = '/';
*q++ = '.';
*q++ = '.';
dotdot = q;
}
}
else /* real path element */
{
if (q != start+rooted)
*q++ = '/';
while ((*q = *p) != '/' && *q != 0)
p++, q++;
}
}
if (q == start) /* empty string is really "." */
*q++ = '.';
*q = '\0';
return name;
}
/* COMMAND "getRobotURL": Extract the location of the robot control file
* SIGNATURE: getRobotURL(str) : str; */
str
URLgetRobotURL(str *retval, url *val)
{
const char *s;
size_t l;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getQuery", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL)
throw(ILLARG, "url.getQuery", "bad url");
l = s - *val;
if ((*retval = GDKmalloc(l + sizeof("/robots.txt"))) == NULL)
throw(MAL, "url.getQuery", "Allocation failed");
sprintf(*retval, "%.*s/robots.txt", (int) l, *val);
return MAL_SUCCEED;
}
void
xps_resolve_url(char *output, char *base_uri, char *path, int output_size)
{
char *p = skip_authority(skip_scheme(path));
if (p != path || path[0] == '/')
{
fz_strlcpy(output, path, output_size);
}
else
{
int len = fz_strlcpy(output, base_uri, output_size);
if (len == 0 || output[len-1] != '/')
fz_strlcat(output, "/", output_size);
fz_strlcat(output, path, output_size);
}
xps_clean_path(output);
}
/* COMMAND "getAnchor": Extract an anchor (reference) from the URL
* SIGNATURE: getAnchor(url) : str; */
str
URLgetAnchor(str *retval, url *val)
{
const char *s;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getAnchor", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(s = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(s = skip_path(s, NULL, NULL)) == NULL ||
(s = skip_search(s)) == NULL)
throw(ILLARG, "url.getAnchor", "bad url");
if (*s == '#')
s++;
else
s = str_nil;
if ((*retval = GDKstrdup(s)) == NULL)
throw(MAL, "url.getAnchor", "Allocation failed");
return MAL_SUCCEED;
}
/* COMMAND "getContext": Extract the path context from the URL
* SIGNATURE: getContext(str) : str; */
str
URLgetContext(str *retval, url *val)
{
const char *s;
const char *p;
if (val == NULL || *val == NULL)
throw(ILLARG, "url.getContext", "url missing");
if ((s = skip_scheme(*val)) == NULL ||
(p = skip_authority(s, NULL, NULL, NULL, NULL)) == NULL ||
(s = skip_path(p, NULL, NULL)) == NULL)
throw(ILLARG, "url.getContext", "bad url");
if (p == s) {
*retval = GDKstrdup(str_nil);
} else if ((*retval = GDKmalloc(s - p + 1)) != NULL) {
strncpy(*retval, p, s - p);
(*retval)[s - p] = 0;
}
if (*retval == NULL)
throw(MAL, "url.getContext", "Allocation failed");
return MAL_SUCCEED;
}