add support for username:password@, user it_url::params to encode params in get(), minor cleanup, add test suite

2 files changed, 89 insertions, 27 deletions

diff --git a/tests/url.t b/tests/url.t
new file mode 100755
index 0000000..bc63bcb
--- /dev/null
+++ b/tests/url.t
@@ -0,0 +1,70 @@
+#!/www/server/bin/php -qC
+<?php
+
+# Tests for url.class, currently only constructor's parser
+
+require 'searchlib/search_test.class';
+
+# Create object and parse url
+$url = new it_url('HTTP://falcon:joshua@www.Relog.CH:80/default.asp');
+
+plan(9);
+
+is(
+	$url->url,
+	'http://www.relog.ch/',
+	'$url->url'
+);
+
+is(
+	$url->protocol,
+	'http',
+	'$url->protocol'
+);
+
+is(
+	$url->hostname,
+	'relog.ch',
+	'$url->hostname'
+);
+
+is(
+	$url->realhostname,
+	'www.relog.ch',
+	'$url->realhostname'
+);
+
+is(
+	$url->port,
+	80,
+	'$url->port'
+);
+
+is(
+	$url->path,
+	'',
+	'$url->path'
+);
+
+is(
+	$url->user,
+	'falcon',
+	'$url->user'
+);
+
+is(
+	$url->pass,
+	'joshua',
+	'$url->pass'
+);
+
+# and now check for path
+$url = new it_url('HTTP://falcon:joshua@www.Relog.CH:80/foo/bar.html');
+
+is(
+	$url->path,
+	'foo/bar.html',
+	'$url->path'
+);
+
+?>
diff --git a/url.class b/url.class
index e7aeb57..575f8ca 100644
--- a/url.class
+++ b/url.class
@@ -4,25 +4,27 @@
 **
 **	ITools - the Internet Tools Library
 **
-**	Copyright (C) 1995-2003 by the ITools Authors.
+**	Copyright (C) 1995-2007 by the ITools Authors.
 **	This program is free software; you can redistribute it and/or
 **	modify it under the terms of either the GNU General Public License
 **	or the GNU Lesser General Public License, as published by the Free
 **	Software Foundation. See http://www.gnu.org/licenses/ for details.
 **
-**	url.class - Create an URL object and canonize it
+**	url.class - URL parsing, retrieval and caching functions
 */
 
 class it_url
 {
-	/* E.g. HTTP://www.Relog.CH.:80/default.asp */
+	/* E.g. HTTP://falcon:joshua@www.Relog.CH.:80/default.asp */
 	var $url;		/* E.g. http://www.relog.ch/ */
 	var $protocol;		/* E.g. http */
 	var $hostname;		/* E.g. relog.ch */
 	var $realhostname;	/* E.g. www.relog.ch */
 	var $port;		/* E.g. 80 */
 	var $path;		/* E.g. / */
-	var $rawurl;		/* E.g. HTTP://www.Relog.CH.:80/default.asp */
+	var $rawurl;		/* E.g. HTTP://falcon:joshua@www.Relog.CH.:80/default.asp */
+	var $user;		/* E.g. falcon */
+	var $pass;		/* E.g. joshua */
 
 	var $page;		/* Page or empty */
 	var $page_read;		/* true if page read */
@@ -42,10 +44,12 @@ function it_url($url, $options = array())
 {
 	$this->rawurl = $url;
 
-	if (eregi('^([a-z]+):/+(.*)$', $url, $regs))
+	if (preg_match('#^([a-z]+):/+(?:([^:]*):([^@]*)@)?(.*)$#i', $url, $regs))
 	{
 		$this->protocol = strtolower($regs[1]);
-		$url = $regs[2];
+		$this->user = strtolower($regs[2]);
+		$this->pass = strtolower($regs[3]);
+		$url = $regs[4];
 	}
 	else if (ereg('^[a-z]:', $url) || ereg('^/', $url))
 	{
@@ -60,7 +64,7 @@ function it_url($url, $options = array())
 	else if ($this->protocol == 'https')
 		$protoport = 443;
 
-	$this->port = $protoport;
+	$this->port = intval($protoport);
 
 	if (class_exists('Net_IDNA'))
 		$idn = Net_IDNA::getInstance();
@@ -77,23 +81,15 @@ function it_url($url, $options = array())
 		$this->realhostname = strtolower($hostname);
 
 		if ($port)
-			$this->port = $port;
+			$this->port = intval($port);
 
 		$url = $regs[2];
 	}
 
-	if (ereg('^www\.(.*)$', $this->realhostname, $regs))
-		$this->hostname = $regs[1];
-	else
-		$this->hostname = $this->realhostname;
-
-	$index_files = array('index.html', 'index.htm', 'index.phtml', 'index.shtml', 'index.php3', 'index.php', 'default.asp');
+	$this->hostname = preg_replace('/^www\./', '', $this->realhostname);
 
-	for ($i = 0; $i < count($index_files); $i++)
-	{
-		$url = eregi_replace("^$index_files[$i]\$", '', $url);
-		$url = eregi_replace("/$index_files[$i]\$", '', $url);
-	}
+	# Get rid of common index file names
+	$url = preg_replace('#(^|/)(index\.[ps]?html?|index\.php[34]?|default\.aspx?)$#', '', $url);
 
 	$this->path = ereg_replace('^/$', '', $url);
 
@@ -245,14 +241,7 @@ function get($p=null, $timeout=5000)
 		{
 			# urlencode data pairs if is array
 			if (is_array($p['data']))
-			{
-				$data_pairs = array();
-
-				foreach ($p['data'] as $key => $value)
-					$data_pairs[] = "$key=".urlencode($value);
-
-				$data = implode('&', $data_pairs);
-			}
+				$data = it_url::params($p['data']);
 
 			$p['headers'] = (array)$p['headers'] + array(
 				'Host' => $url->realhostname,
@@ -271,6 +260,9 @@ function get($p=null, $timeout=5000)
 			else
 				$method = "GET";
 
+			if ($url->user || $url->pass)
+				$p['headers'] += array('Authorization' => 'Basic ' . base64_encode($url->user . ':' . $url->pass));
+
 			foreach ($p['headers'] as $header => $value)
 					$headers .= "$header: $value\r\n";