Bug 566530 - Handle (illegal) unencoded spaces in URIs

	* libsoup/soup-uri.c (uri_normalized_copy): optionally fix up
	parts with unencoded spaces in them.
	(soup_uri_new_with_base): tell uri_normalized_copy() to fix up
	spaces in the path and query components

	* tests/uri-parsing.c (abs_tests): test parsing and unparsing a
	URI with an unencoded space in it.

	* tests/redirect-test.c (tests, server_callback): add a test of
	redirecting to a URI with an unencoded space in it.

svn path=/trunk/; revision=1252

4 files changed, 65 insertions, 8 deletions

diff --git a/ChangeLog b/ChangeLog
index 5581c492..3477375e 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,18 @@
+2009-03-23  Dan Winship  <danw@gnome.org>
+
+	Bug 566530 - Handle (illegal) unencoded spaces in URIs
+
+	* libsoup/soup-uri.c (uri_normalized_copy): optionally fix up
+	parts with unencoded spaces in them.
+	(soup_uri_new_with_base): tell uri_normalized_copy() to fix up
+	spaces in the path and query components
+
+	* tests/uri-parsing.c (abs_tests): test parsing and unparsing a
+	URI with an unencoded space in it.
+
+	* tests/redirect-test.c (tests, server_callback): add a test of
+	redirecting to a URI with an unencoded space in it.
+
 2009-03-15  Dan Winship  <danw@gnome.org>
 
 	* configure.in: 2.26.0
diff --git a/libsoup/soup-uri.c b/libsoup/soup-uri.c
index ff923adb..2743576c 100644
--- a/libsoup/soup-uri.c
+++ b/libsoup/soup-uri.c
@@ -93,7 +93,7 @@
 
 static void append_uri_encoded (GString *str, const char *in, const char *extra_enc_chars);
 static char *uri_decoded_copy (const char *str, int length);
-static char *uri_normalized_copy (const char *str, int length, const char *unescape_extra);
+static char *uri_normalized_copy (const char *str, int length, const char *unescape_extra, gboolean fixup);
 
 const char *_SOUP_URI_SCHEME_HTTP, *_SOUP_URI_SCHEME_HTTPS;
 
@@ -152,7 +152,7 @@ soup_uri_new_with_base (SoupURI *base, const char *uri_string)
 	end = hash = strchr (uri_string, '#');
 	if (hash && hash[1]) {
 		uri->fragment = uri_normalized_copy (hash + 1, strlen (hash + 1),
-						     NULL);
+						     NULL, FALSE);
 		if (!uri->fragment) {
 			soup_uri_free (uri);
 			return NULL;
@@ -249,7 +249,7 @@ soup_uri_new_with_base (SoupURI *base, const char *uri_string)
 		if (question[1]) {
 			uri->query = uri_normalized_copy (question + 1,
 							  end - (question + 1),
-							  NULL);
+							  NULL, TRUE);
 			if (!uri->query) {
 				soup_uri_free (uri);
 				return NULL;
@@ -260,7 +260,7 @@ soup_uri_new_with_base (SoupURI *base, const char *uri_string)
 
 	if (end != uri_string) {
 		uri->path = uri_normalized_copy (uri_string, end - uri_string,
-						 NULL);
+						 NULL, TRUE);
 		if (!uri->path) {
 			soup_uri_free (uri);
 			return NULL;
@@ -654,10 +654,12 @@ soup_uri_decode (const char *part)
 }
 
 static char *
-uri_normalized_copy (const char *part, int length, const char *unescape_extra)
+uri_normalized_copy (const char *part, int length,
+		     const char *unescape_extra, gboolean fixup)
 {
 	unsigned char *s, *d, c;
 	char *normalized = g_strndup (part, length);
+	gboolean need_fixup = FALSE;
 
 	s = d = (unsigned char *)normalized;
 	do {
@@ -678,10 +680,27 @@ uri_normalized_copy (const char *part, int length, const char *unescape_extra)
 				*d++ = g_ascii_toupper (*s++);
 				*d++ = g_ascii_toupper (*s);
 			}
-		} else
+		} else {
+			if (*s == ' ')
+				need_fixup = TRUE;
 			*d++ = *s;
+		}
 	} while (*s++);
 
+	if (fixup && need_fixup) {
+		char *tmp, *sp;
+		/* This code is lame, but so are people who put
+		 * unencoded spaces in URLs!
+		 */
+		while ((sp = strchr (normalized, ' '))) {
+			tmp = g_strdup_printf ("%.*s%%20%s",
+					       (int)(sp - normalized),
+					       normalized, sp + 1);
+			g_free (normalized);
+			normalized = tmp;
+		};
+	}
+
 	return normalized;
 }
 
@@ -709,7 +728,7 @@ uri_normalized_copy (const char *part, int length, const char *unescape_extra)
 char *
 soup_uri_normalize (const char *part, const char *unescape_extra)
 {
-	return uri_normalized_copy (part, strlen (part), unescape_extra);
+	return uri_normalized_copy (part, strlen (part), unescape_extra, FALSE);
 }
 
 
diff --git a/tests/redirect-test.c b/tests/redirect-test.c
index 5bb37e95..e863068c 100644
--- a/tests/redirect-test.c
+++ b/tests/redirect-test.c
@@ -93,6 +93,12 @@ static struct {
 	    { "GET", "/", 200 },
 	    { NULL } } },
 	{ { { "POST", "/307", 307 },
+	    { NULL } } },
+
+	/* Test behavior with recoverably-bad Location header
+	 */
+	{ { { "GET", "/bad", 302 },
+	    { "GET", "/bad%20with%20spaces", 200 },
 	    { NULL } } }
 };
 static const int n_tests = G_N_ELEMENTS (tests);
@@ -206,6 +212,19 @@ server_callback (SoupServer *server, SoupMessage *msg,
 	char *remainder;
 	guint status_code;
 
+	if (g_str_has_prefix (path, "/bad")) {
+		if (!strcmp (path, "/bad")) {
+			soup_message_set_status (msg, SOUP_STATUS_FOUND);
+			soup_message_headers_replace (msg->response_headers,
+						      "Location",
+						      "/bad with spaces");
+		} else if (!strcmp (path, "/bad with spaces"))
+			soup_message_set_status (msg, SOUP_STATUS_OK);
+		else
+			soup_message_set_status (msg, SOUP_STATUS_NOT_FOUND);
+		return;
+	}
+
 	if (!strcmp (path, "/")) {
 		if (msg->method != SOUP_METHOD_GET &&
 		    msg->method != SOUP_METHOD_HEAD) {
diff --git a/tests/uri-parsing.c b/tests/uri-parsing.c
index 52d52717..407ddfd7 100644
--- a/tests/uri-parsing.c
+++ b/tests/uri-parsing.c
@@ -57,7 +57,11 @@ static struct {
 	{ "http://[::FFFF:129.144.52.38]:80/index.html",
 	  "http://[::FFFF:129.144.52.38]/index.html" },
 	{ "http://[2010:836B:4179::836B:4179]",
-	  "http://[2010:836B:4179::836B:4179]/" }
+	  "http://[2010:836B:4179::836B:4179]/" },
+
+	/* Try to recover certain kinds of invalid URIs */
+	{ "http://host/path with spaces",
+	  "http://host/path%20with%20spaces" }
 };
 static int num_abs_tests = G_N_ELEMENTS(abs_tests);