/**
* soup_uri_normalize:
* @part: a URI part
* @unescape_extra: (allow-none): reserved characters to unescape (or %NULL)
*
* %<!-- -->-decodes any "unreserved" characters (or characters in
* @unescape_extra) in @part, and %<!-- -->-encodes any non-ASCII
* characters, spaces, and non-printing characters in @part.
*
* "Unreserved" characters are those that are not allowed to be used
* for punctuation according to the URI spec. For example, letters are
* unreserved, so soup_uri_normalize() will turn
* <literal>http://example.com/foo/b%<!-- -->61r</literal> into
* <literal>http://example.com/foo/bar</literal>, which is guaranteed
* to mean the same thing. However, "/" is "reserved", so
* <literal>http://example.com/foo%<!-- -->2Fbar</literal> would not
* be changed, because it might mean something different to the
* server.
*
* In the past, this would return %NULL if @part contained invalid
* percent-encoding, but now it just ignores the problem (as
* soup_uri_new() already did).
*
* Return value: the normalized URI part
*/
char *
soup_uri_normalize (const char *part, const char *unescape_extra)
{
g_return_val_if_fail (part != NULL, NULL);
return uri_normalized_copy (part, strlen (part), unescape_extra);
}
/**
* soup_uri_normalize:
* @part: a URI part
* @unescape_extra: reserved characters to unescape (or %NULL)
*
* %<!-- -->-decodes any "unreserved" characters (or characters in
* @unescape_extra) in @part.
*
* "Unreserved" characters are those that are not allowed to be used
* for punctuation according to the URI spec. For example, letters are
* unreserved, so soup_uri_normalize() will turn
* <literal>http://example.com/foo/b%<!-- -->61r</literal> into
* <literal>http://example.com/foo/bar</literal>, which is guaranteed
* to mean the same thing. However, "/" is "reserved", so
* <literal>http://example.com/foo%<!-- -->2Fbar</literal> would not
* be changed, because it might mean something different to the
* server.
*
* In the past, this would return %NULL if @part contained invalid
* percent-encoding, but now it just ignores the problem (as
* soup_uri_new() already did).
*
* Return value: the normalized URI part
*/
char *
soup_uri_normalize (const char *part, const char *unescape_extra)
{
return uri_normalized_copy (part, strlen (part), unescape_extra);
}
/**
* soup_uri_new_with_base:
* @base: a base URI
* @uri_string: the URI
*
* Parses @uri_string relative to @base.
*
* Return value: a parsed #SoupURI.
**/
SoupURI *
soup_uri_new_with_base (SoupURI *base, const char *uri_string)
{
SoupURI *uri;
const char *end, *hash, *colon, *at, *path, *question;
const char *p, *hostend;
gboolean remove_dot_segments = TRUE;
int len;
/* First some cleanup steps (which are supposed to all be no-ops,
* but...). Skip initial whitespace, strip out internal tabs and
* line breaks, and ignore trailing whitespace.
*/
while (g_ascii_isspace (*uri_string))
uri_string++;
len = strcspn (uri_string, "\t\n\r");
if (uri_string[len]) {
char *clean = g_malloc (strlen (uri_string) + 1), *d;
const char *s;
for (s = uri_string, d = clean; *s; s++) {
if (*s != '\t' && *s != '\n' && *s != '\r')
*d++ = *s;
}
*d = '\0';
uri = soup_uri_new_with_base (base, clean);
g_free (clean);
return uri;
}
end = uri_string + len;
while (end > uri_string && g_ascii_isspace (end[-1]))
end--;
uri = g_slice_new0 (SoupURI);
/* Find fragment. */
hash = strchr (uri_string, '#');
if (hash) {
uri->fragment = uri_normalized_copy (hash + 1, end - hash + 1,
NULL);
end = hash;
}
/* Find scheme: initial [a-z+.-]* substring until ":" */
p = uri_string;
while (p < end && (g_ascii_isalnum (*p) ||
*p == '.' || *p == '+' || *p == '-'))
p++;
if (p > uri_string && *p == ':') {
uri->scheme = soup_uri_parse_scheme (uri_string, p - uri_string);
uri_string = p + 1;
}
if (uri_string == end && !base && !uri->fragment)
return uri;
/* Check for authority */
if (strncmp (uri_string, "//", 2) == 0) {
uri_string += 2;
path = uri_string + strcspn (uri_string, "/?#");
if (path > end)
path = end;
at = strchr (uri_string, '@');
if (at && at < path) {
colon = strchr (uri_string, ':');
if (colon && colon < at) {
uri->password = uri_decoded_copy (colon + 1,
at - colon - 1);
} else {
uri->password = NULL;
colon = at;
}
uri->user = uri_decoded_copy (uri_string,
colon - uri_string);
uri_string = at + 1;
} else
uri->user = uri->password = NULL;
/* Find host and port. */
if (*uri_string == '[') {
uri_string++;
hostend = strchr (uri_string, ']');
if (!hostend || hostend > path) {
soup_uri_free (uri);
return NULL;
}
if (*(hostend + 1) == ':')
colon = hostend + 1;
else
colon = NULL;
} else {
colon = memchr (uri_string, ':', path - uri_string);
hostend = colon ? colon : path;
}
uri->host = uri_decoded_copy (uri_string, hostend - uri_string);
if (colon && colon != path - 1) {
char *portend;
uri->port = strtoul (colon + 1, &portend, 10);
if (portend != (char *)path) {
soup_uri_free (uri);
return NULL;
}
}
uri_string = path;
}
/* Find query */
question = memchr (uri_string, '?', end - uri_string);
if (question) {
uri->query = uri_normalized_copy (question + 1,
end - (question + 1),
NULL);
end = question;
}
if (end != uri_string) {
uri->path = uri_normalized_copy (uri_string, end - uri_string,
NULL);
}
/* Apply base URI. This is spelled out in RFC 3986. */
if (base && !uri->scheme && uri->host)
uri->scheme = base->scheme;
else if (base && !uri->scheme) {
uri->scheme = base->scheme;
uri->user = g_strdup (base->user);
uri->password = g_strdup (base->password);
uri->host = g_strdup (base->host);
uri->port = base->port;
if (!uri->path) {
uri->path = g_strdup (base->path);
if (!uri->query)
uri->query = g_strdup (base->query);
remove_dot_segments = FALSE;
} else if (*uri->path != '/') {
char *newpath, *last;
last = strrchr (base->path, '/');
if (last) {
newpath = g_strdup_printf ("%.*s%s",
(int)(last + 1 - base->path),
base->path,
uri->path);
} else
newpath = g_strdup_printf ("/%s", uri->path);
g_free (uri->path);
uri->path = newpath;
}
}
if (remove_dot_segments && uri->path && *uri->path) {
char *p, *q;
/* Remove "./" where "." is a complete segment. */
for (p = uri->path + 1; *p; ) {
if (*(p - 1) == '/' &&
*p == '.' && *(p + 1) == '/')
memmove (p, p + 2, strlen (p + 2) + 1);
else
p++;
}
/* Remove "." at end. */
if (p > uri->path + 2 &&
*(p - 1) == '.' && *(p - 2) == '/')
*(p - 1) = '\0';
/* Remove "<segment>/../" where <segment> != ".." */
for (p = uri->path + 1; *p; ) {
if (!strncmp (p, "../", 3)) {
p += 3;
continue;
}
q = strchr (p + 1, '/');
if (!q)
break;
if (strncmp (q, "/../", 4) != 0) {
p = q + 1;
continue;
}
memmove (p, q + 4, strlen (q + 4) + 1);
p = uri->path + 1;
}
/* Remove "<segment>/.." at end where <segment> != ".." */
q = strrchr (uri->path, '/');
if (q && !strcmp (q, "/..")) {
p = q - 1;
while (p > uri->path && *p != '/')
p--;
if (strncmp (p, "/../", 4) != 0)
*(p + 1) = 0;
}
/* Remove extraneous initial "/.."s */
while (!strncmp (uri->path, "/../", 4))
memmove (uri->path, uri->path + 3, strlen (uri->path) - 2);
if (!strcmp (uri->path, "/.."))
uri->path[1] = '\0';
}
/* HTTP-specific stuff */
if (uri->scheme == SOUP_URI_SCHEME_HTTP ||
uri->scheme == SOUP_URI_SCHEME_HTTPS) {
if (!uri->path)
uri->path = g_strdup ("/");
if (!SOUP_URI_VALID_FOR_HTTP (uri)) {
soup_uri_free (uri);
return NULL;
}
}
if (uri->scheme == SOUP_URI_SCHEME_FTP) {
if (!uri->host) {
soup_uri_free (uri);
return NULL;
}
}
if (!uri->port)
uri->port = soup_scheme_default_port (uri->scheme);
if (!uri->path)
uri->path = g_strdup ("");
return uri;
}
/**
* soup_uri_new_with_base:
* @base: a base URI
* @uri_string: the URI
*
* Parses @uri_string relative to @base.
*
* Return value: a parsed #SoupURI.
**/
SoupURI *
soup_uri_new_with_base (SoupURI *base, const char *uri_string)
{
SoupURI *uri;
const char *end, *hash, *colon, *at, *path, *question;
const char *p, *hostend;
gboolean remove_dot_segments = TRUE;
uri = g_slice_new0 (SoupURI);
/* See RFC 3986 for details. IF YOU CHANGE ANYTHING IN THIS
* FUNCTION, RUN tests/uri-parsing AFTERWARDS.
*/
/* Find fragment. */
end = hash = strchr (uri_string, '#');
if (hash && hash[1]) {
uri->fragment = uri_normalized_copy (hash + 1, strlen (hash + 1),
NULL, FALSE);
if (!uri->fragment) {
soup_uri_free (uri);
return NULL;
}
} else
end = uri_string + strlen (uri_string);
/* Find scheme: initial [a-z+.-]* substring until ":" */
p = uri_string;
while (p < end && (g_ascii_isalnum (*p) ||
*p == '.' || *p == '+' || *p == '-'))
p++;
if (p > uri_string && *p == ':') {
uri->scheme = soup_uri_get_scheme (uri_string, p - uri_string);
if (!uri->scheme) {
soup_uri_free (uri);
return NULL;
}
uri_string = p + 1;
}
if (!*uri_string && !base)
return uri;
/* Check for authority */
if (strncmp (uri_string, "//", 2) == 0) {
uri_string += 2;
path = uri_string + strcspn (uri_string, "/?#");
at = strchr (uri_string, '@');
if (at && at < path) {
colon = strchr (uri_string, ':');
if (colon && colon < at) {
uri->password = uri_decoded_copy (colon + 1,
at - colon - 1);
if (!uri->password) {
soup_uri_free (uri);
return NULL;
}
} else {
uri->password = NULL;
colon = at;
}
uri->user = uri_decoded_copy (uri_string,
colon - uri_string);
if (!uri->user) {
soup_uri_free (uri);
return NULL;
}
uri_string = at + 1;
} else
uri->user = uri->password = NULL;
/* Find host and port. */
if (*uri_string == '[') {
uri_string++;
hostend = strchr (uri_string, ']');
if (!hostend || hostend > path) {
soup_uri_free (uri);
return NULL;
}
if (*(hostend + 1) == ':')
colon = hostend + 1;
else
colon = NULL;
} else {
colon = memchr (uri_string, ':', path - uri_string);
hostend = colon ? colon : path;
}
uri->host = uri_decoded_copy (uri_string, hostend - uri_string);
if (!uri->host) {
soup_uri_free (uri);
return NULL;
}
if (colon && colon != path - 1) {
char *portend;
uri->port = strtoul (colon + 1, &portend, 10);
if (portend != (char *)path) {
soup_uri_free (uri);
return NULL;
}
}
uri_string = path;
}
/* Find query */
question = memchr (uri_string, '?', end - uri_string);
if (question) {
if (question[1]) {
uri->query = uri_normalized_copy (question + 1,
end - (question + 1),
NULL, TRUE);
if (!uri->query) {
soup_uri_free (uri);
return NULL;
}
}
end = question;
}
if (end != uri_string) {
uri->path = uri_normalized_copy (uri_string, end - uri_string,
NULL, TRUE);
if (!uri->path) {
soup_uri_free (uri);
return NULL;
}
}
/* Apply base URI. Again, this is spelled out in RFC 3986. */
if (base && !uri->scheme && uri->host)
uri->scheme = base->scheme;
else if (base && !uri->scheme) {
uri->scheme = base->scheme;
uri->user = g_strdup (base->user);
uri->password = g_strdup (base->password);
uri->host = g_strdup (base->host);
uri->port = base->port;
if (!uri->path) {
uri->path = g_strdup (base->path);
if (!uri->query)
uri->query = g_strdup (base->query);
remove_dot_segments = FALSE;
} else if (*uri->path != '/') {
char *newpath, *last;
last = strrchr (base->path, '/');
if (last) {
newpath = g_strdup_printf ("%.*s/%s",
(int)(last - base->path),
base->path,
uri->path);
} else
newpath = g_strdup_printf ("/%s", uri->path);
g_free (uri->path);
uri->path = newpath;
}
}
if (remove_dot_segments && uri->path && *uri->path) {
char *p = uri->path, *q;
/* Remove "./" where "." is a complete segment. */
for (p = uri->path + 1; *p; ) {
if (*(p - 1) == '/' &&
*p == '.' && *(p + 1) == '/')
memmove (p, p + 2, strlen (p + 2) + 1);
else
p++;
}
/* Remove "." at end. */
if (p > uri->path + 2 &&
*(p - 1) == '.' && *(p - 2) == '/')
*(p - 1) = '\0';
/* Remove "<segment>/../" where <segment> != ".." */
for (p = uri->path + 1; *p; ) {
if (!strncmp (p, "../", 3)) {
p += 3;
continue;
}
q = strchr (p + 1, '/');
if (!q)
break;
if (strncmp (q, "/../", 4) != 0) {
p = q + 1;
continue;
}
memmove (p, q + 4, strlen (q + 4) + 1);
p = uri->path + 1;
}
/* Remove "<segment>/.." at end where <segment> != ".." */
q = strrchr (uri->path, '/');
if (q && !strcmp (q, "/..")) {
p = q - 1;
while (p > uri->path && *p != '/')
p--;
if (strncmp (p, "/../", 4) != 0)
*(p + 1) = 0;
}
/* Remove extraneous initial "/.."s */
while (!strncmp (uri->path, "/../", 4))
memmove (uri->path, uri->path + 3, strlen (uri->path) - 2);
if (!strcmp (uri->path, "/.."))
uri->path[1] = '\0';
}
/* HTTP-specific stuff */
if (uri->scheme == SOUP_URI_SCHEME_HTTP ||
uri->scheme == SOUP_URI_SCHEME_HTTPS) {
if (!SOUP_URI_VALID_FOR_HTTP (uri)) {
soup_uri_free (uri);
return NULL;
}
if (!uri->path)
uri->path = g_strdup ("/");
}
if (!uri->port)
uri->port = soup_scheme_default_port (uri->scheme);
if (!uri->path)
uri->path = g_strdup ("");
return uri;
}