summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--Auth/OpenID/URINorm.php252
-rw-r--r--Tests/Auth/OpenID/URINorm.php68
-rw-r--r--Tests/Auth/OpenID/data/urinorm.txt79
-rw-r--r--Tests/TestDriver.php3
4 files changed, 401 insertions, 1 deletions
diff --git a/Auth/OpenID/URINorm.php b/Auth/OpenID/URINorm.php
new file mode 100644
index 0000000..a26f5df
--- /dev/null
+++ b/Auth/OpenID/URINorm.php
@@ -0,0 +1,252 @@
+<?php
+
+/**
+ * URI normalization routines.
+ *
+ * @package OpenID
+ * @author JanRain, Inc. <openid@janrain.com>
+ * @copyright 2005 Janrain, Inc.
+ * @license http://www.gnu.org/copyleft/lesser.html LGPL
+ */
+
+// from appendix B of rfc 3986 (http://www.ietf.org/rfc/rfc3986.txt)
+$__uri_pattern = '&^(([^:/?#]+):)?(//([^/?#]*))?([^?#]*)(\?([^#]*))?(#(.*))?&';
+$__authority_pattern = '/^([^@]*@)?([^:]*)(:.*)?/';
+$__pct_encoded_pattern = '/%([0-9A-Fa-f]{2})/';
+
+$__UCSCHAR = array(
+ array(0xA0, 0xD7FF),
+ array(0xF900, 0xFDCF),
+ array(0xFDF0, 0xFFEF),
+ array(0x10000, 0x1FFFD),
+ array(0x20000, 0x2FFFD),
+ array(0x30000, 0x3FFFD),
+ array(0x40000, 0x4FFFD),
+ array(0x50000, 0x5FFFD),
+ array(0x60000, 0x6FFFD),
+ array(0x70000, 0x7FFFD),
+ array(0x80000, 0x8FFFD),
+ array(0x90000, 0x9FFFD),
+ array(0xA0000, 0xAFFFD),
+ array(0xB0000, 0xBFFFD),
+ array(0xC0000, 0xCFFFD),
+ array(0xD0000, 0xDFFFD),
+ array(0xE1000, 0xEFFFD)
+ );
+
+$__IPRIVATE = array(
+ array(0xE000, 0xF8FF),
+ array(0xF0000, 0xFFFFD),
+ array(0x100000, 0x10FFFD)
+ );
+
+$_unreserved = array();
+for ($i = 0; $i < 256; $i++) {
+ $_unreserved[$i] = false;
+}
+
+for ($i = ord('A'); $i <= ord('Z'); $i++) {
+ $_unreserved[$i] = true;
+}
+
+for ($i = ord('0'); $i <= ord('9'); $i++) {
+ $_unreserved[$i] = true;
+}
+
+for ($i = ord('a'); $i <= ord('z'); $i++) {
+ $_unreserved[$i] = true;
+}
+
+$_unreserved[ord('-')] = true;
+$_unreserved[ord('.')] = true;
+$_unreserved[ord('_')] = true;
+$_unreserved[ord('~')] = true;
+
+$parts = array();
+foreach (array_merge($__UCSCHAR, $__IPRIVATE) as $pair) {
+ list($m, $n) = $pair;
+ $parts[] = sprintf("%s-%s", $m, $n);
+}
+
+$_escapeme_re = sprintf('[%s]', implode('', $parts));
+
+function startswith($s, $stuff)
+{
+ return strpos($s, $stuff) === 0;
+}
+
+function _pct_escape_unicode($char_match)
+{
+ $c = $char_match[0];
+ $result = "";
+ for ($i = 0; $i < strlen($c); $i++) {
+ $result .= "%".sprintf("%X", ord($c[$i]));
+ }
+ return $result;
+}
+
+function _pct_encoded_replace_unreserved($mo)
+{
+ global $_unreserved;
+
+ $i = intval($mo[1], 16);
+ if ($_unreserved[$i]) {
+ return chr($i);
+ } else {
+ return strtoupper($mo[0]);
+ }
+
+ return $mo[0];
+}
+
+function _pct_encoded_replace($mo)
+{
+ return chr(intval($mo[1], 16));
+}
+
+function remove_dot_segments($path)
+{
+ $result_segments = array();
+
+ while ($path) {
+ if (startswith($path, '../')) {
+ $path = substr($path, 3);
+ } else if (startswith($path, './')) {
+ $path = substr($path, 2);
+ } else if (startswith($path, '/./')) {
+ $path = substr($path, 2);
+ } else if ($path == '/.') {
+ $path = '/';
+ } else if (startswith($path, '/../')) {
+ $path = substr($path, 3);
+ if ($result_segments) {
+ array_pop($result_segments);
+ }
+ } else if ($path == '/..') {
+ $path = '/';
+ if ($result_segments) {
+ array_pop($result_segments);
+ }
+ } else if (($path == '..') ||
+ ($path == '.')) {
+ $path = '';
+ } else {
+ $i = 0;
+ if ($path[0] == '/') {
+ $i = 1;
+ }
+ $i = strpos($path, '/', $i);
+ if ($i === false) {
+ $i = strlen($path);
+ }
+ $result_segments[] = substr($path, 0, $i);
+ $path = substr($path, $i);
+ }
+ }
+
+ return implode('', $result_segments);
+}
+
+function Auth_OpenID_urinorm($uri)
+{
+ global $__uri_pattern, $__authority_pattern, $__pct_encoded_pattern;
+
+ $uri_matches = array();
+ preg_match($__uri_pattern, $uri, $uri_matches);
+
+ if (count($uri_matches) < 9) {
+ for ($i = count($uri_matches); $i <= 9; $i++) {
+ $uri_matches[] = '';
+ }
+ }
+
+ print_r($uri_matches);
+
+ $scheme = $uri_matches[2];
+ if ($scheme) {
+ $scheme = strtolower($scheme);
+ }
+
+ $scheme = $uri_matches[2];
+ if ($scheme === '') {
+ // No scheme specified
+ return null;
+ }
+
+ $scheme = strtolower($scheme);
+ if (!in_array($scheme, array('http', 'https'))) {
+ // Not an absolute HTTP or HTTPS URI
+ return null;
+ }
+
+ $authority = $uri_matches[4];
+ if ($authority === '') {
+ // Not an absolute URI
+ return null;
+ }
+
+ $authority_matches = array();
+ preg_match($__authority_pattern, $authority, $authority_matches);
+ if (count($authority_matches) === 0) {
+ // URI does not have a valid authority
+ return null;
+ }
+
+ if (count($authority_matches) < 4) {
+ for ($i = count($authority_matches); $i <= 4; $i++) {
+ $authority_matches[] = '';
+ }
+ }
+
+ list($_whole, $userinfo, $host, $port) = $authority_matches;
+
+ if ($userinfo === null) {
+ $userinfo = '';
+ }
+
+ if (strpos($host, '%') !== -1) {
+ $host = strtolower($host);
+ $host = preg_replace_callback(
+ $__pct_encoded_pattern, '_pct_encoded_replace', $host);
+ // NO IDNA.
+ // $host = unicode($host, 'utf-8').encode('idna');
+ } else {
+ $host = strtolower($host);
+ }
+
+ if ($port) {
+ if (($port == ':') ||
+ ($scheme == 'http' && $port == ':80') ||
+ ($scheme == 'https' && $port == ':443')) {
+ $port = '';
+ }
+ } else {
+ $port = '';
+ }
+
+ $authority = $userinfo . $host . $port;
+
+ $path = $uri_matches[5];
+ $path = preg_replace_callback(
+ $__pct_encoded_pattern,
+ '_pct_encoded_replace_unreserved', $path);
+
+ $path = remove_dot_segments($path);
+ if (!$path) {
+ $path = '/';
+ }
+
+ $query = $uri_matches[6];
+ if ($query === null) {
+ $query = '';
+ }
+
+ $fragment = $uri_matches[8];
+ if ($fragment === null) {
+ $fragment = '';
+ }
+
+ return $scheme . '://' . $authority . $path . $query . $fragment;
+}
+
+?> \ No newline at end of file
diff --git a/Tests/Auth/OpenID/URINorm.php b/Tests/Auth/OpenID/URINorm.php
new file mode 100644
index 0000000..02263af
--- /dev/null
+++ b/Tests/Auth/OpenID/URINorm.php
@@ -0,0 +1,68 @@
+<?php
+
+/**
+ * Tests for the URI normalization routines used by the OpenID
+ * library.
+ *
+ * PHP versions 4 and 5
+ *
+ * LICENSE: See the COPYING file included in this distribution.
+ *
+ * @package OpenID
+ * @author JanRain, Inc. <openid@janrain.com>
+ * @copyright 2005 Janrain, Inc.
+ * @license http://www.gnu.org/copyleft/lesser.html LGPL
+ */
+
+require_once 'PHPUnit.php';
+require_once 'Auth/OpenID/URINorm.php';
+require_once 'Tests/Auth/OpenID/TestUtil.php';
+
+class Tests_Auth_OpenID_URINorm_TestCase extends PHPUnit_TestCase {
+ function Tests_Auth_OpenID_URINorm_TestCase(
+ $name, $uri, $expected)
+ {
+
+ $this->setName($name);
+ $this->uri = $uri;
+ $this->expected = $expected;
+ }
+
+ function runTest()
+ {
+ $actual = Auth_OpenID_urinorm($this->uri);
+ $this->assertEquals($this->expected, $actual);
+ }
+}
+
+class Tests_Auth_OpenID_URINorm extends PHPUnit_TestSuite {
+ function _readTestCases()
+ {
+ $lines = Tests_Auth_OpenID_readlines('urinorm.txt');
+ $cases = array();
+ $case = array();
+ for ($i = 0; $i < count($lines) && ($i + 3 <= count($lines)); $i += 4) {
+ $name = trim($lines[$i]);
+ $uri = trim($lines[$i + 1]);
+ $expected = trim($lines[$i + 2]);
+ if ($expected == 'fail') {
+ $expected = null;
+ }
+ $cases[] = array($name, $uri, $expected);
+ }
+
+ return $cases;
+ }
+
+ function Tests_Auth_OpenID_URINorm($name)
+ {
+ $this->setName($name);
+ $cases = $this->_readTestCases();
+ foreach ($cases as $case) {
+ list($name, $uri, $expected) = $case;
+ $this->addTest(new Tests_Auth_OpenID_URINorm_TestCase($name, $uri, $expected));
+ }
+ }
+}
+
+?> \ No newline at end of file
diff --git a/Tests/Auth/OpenID/data/urinorm.txt b/Tests/Auth/OpenID/data/urinorm.txt
new file mode 100644
index 0000000..95262fe
--- /dev/null
+++ b/Tests/Auth/OpenID/data/urinorm.txt
@@ -0,0 +1,79 @@
+Already normal form
+http://example.com/
+http://example.com/
+
+Add a trailing slash
+http://example.com
+http://example.com/
+
+Remove an empty port segment
+http://example.com:/
+http://example.com/
+
+Remove a default port segment
+http://example.com:80/
+http://example.com/
+
+Capitalization in host names
+http://wWw.exaMPLE.COm/
+http://www.example.com/
+
+Capitalization in scheme names
+htTP://example.com/
+http://example.com/
+
+Capitalization in percent-escaped reserved characters
+http://example.com/foo%2cbar
+http://example.com/foo%2Cbar
+
+Unescape percent-encoded unreserved characters
+http://example.com/foo%2Dbar%2dbaz
+http://example.com/foo-bar-baz
+
+remove_dot_segments example 1
+http://example.com/a/b/c/./../../g
+http://example.com/a/g
+
+remove_dot_segments example 2
+http://example.com/mid/content=5/../6
+http://example.com/mid/6
+
+remove_dot_segments: single-dot
+http://example.com/a/./b
+http://example.com/a/b
+
+remove_dot_segments: double-dot
+http://example.com/a/../b
+http://example.com/b
+
+remove_dot_segments: leading double-dot
+http://example.com/../b
+http://example.com/b
+
+remove_dot_segments: trailing single-dot
+http://example.com/a/.
+http://example.com/a/
+
+remove_dot_segments: trailing double-dot
+http://example.com/a/..
+http://example.com/
+
+remove_dot_segments: trailing single-dot-slash
+http://example.com/a/./
+http://example.com/a/
+
+remove_dot_segments: trailing double-dot-slash
+http://example.com/a/../
+http://example.com/
+
+Test of all kinds of syntax-based normalization
+hTTPS://a/./b/../b/%63/%7bfoo%7d
+https://a/b/c/%7Bfoo%7D
+
+Unsupported scheme
+ftp://example.com/
+fail
+
+Non-absolute URI
+http:/foo
+fail \ No newline at end of file
diff --git a/Tests/TestDriver.php b/Tests/TestDriver.php
index 4827c3e..0fa94d9 100644
--- a/Tests/TestDriver.php
+++ b/Tests/TestDriver.php
@@ -117,7 +117,8 @@ $_test_names = array(
'Server',
'TrustRoot',
'Discover',
- 'OpenID_Yadis'
+ 'OpenID_Yadis',
+ 'URINorm'
);
function selectTests($names)
generated by cgit v1.1 at 2025-06-08 18:27:55 +0000