factor out request method from get method, make http method overwritable, fix Host header

2 files changed, 93 insertions, 75 deletions

diff --git a/it_url.class b/it_url.class
index a48013e..3caa6cb 100644
--- a/it_url.class
+++ b/it_url.class
@@ -238,8 +238,7 @@ function get($p=null, $timeout=5)
 {
 	if (!is_array($p))
 		$p = array('url' => $p, 'timeout' => $timeout);
-
-	$p += array('totaltimeout' => "999999", 'timeout' => 5, 'retries' => 1);
+	$p += array('retries' => 1);
 
 	if ($this instanceof it_url)
 	{
@@ -250,103 +249,120 @@ function get($p=null, $timeout=5)
 	else	# called statically
 		$url = new it_url($p['url']);
 
+	if ($url->protocol == 'http')
+	{
+		$result = $url->request($p);
+		if ($url->headers['Location'] && preg_match('#^(https?://[^/]*)?(/)?(.*)$#i', $url->headers['Location'], $parts) && ($parts[1] != $url->url)) # Handle redirects (supports relative and global)
+		{
+			unset($p['url'], $p['headers']['Host']);
+			$url->it_url($parts[1] ? $parts[1].$parts[2].$parts[3] : $url->protocol.'://'.$url->realhostname.($parts[2] ? $parts[2].$parts[3] : '/'.dirname($url->path).'/'.$parts[3]));
+			if (++$url->redir <= 4)  /* Avoid infinite redirects */
+				return $url->get($p);
+		}
+	}
+	else
+	{
+		$results = self::get_multi(array('urls' => array('one' => $p['url'])) + $p);
+		$result = $results['one'];
+	}
+
+	if (!$result && $p['retries'] > 0 && $url->result < 400)
+		$result = $url->get(array('retries' => $p['retries'] - 1) + $p);
+
+	return $result;
+}
+
+function request($p=array())
+{
+	$p += array('totaltimeout' => "999999", 'timeout' => 5);
+
+	$url = $this;
+	if ($p['url'])
+		$this->it_url($p['url']);
+
 	$url->result = $result = false;
 	unset($url->data);
 	$url->headers = array();
 	$p['timeout'] = min($p['timeout'], $p['totaltimeout']);	# No operation may be longer than totaltimeout
 	$endtime = time() + $p['totaltimeout'];
 
-	if ($url->protocol == 'http')
+	if ($fp = @fsockopen($url->realhostname, $url->port, $dummy_errno, $dummy_errstr, $p['timeout']))
 	{
-		if ($fp = @fsockopen($url->realhostname, $url->port, $dummy_errno, $dummy_errstr, $p['timeout']))
-		{
-			# urlencode data pairs if is array
-			if (is_array($p['data']))
-				$data = it_url::params($p['data']);
-
-			$p['headers'] = (array)$p['headers'] + array(
-				'Host' => $url->realhostname,
-				'User-Agent' => "Mozilla/4.0 (compatible; MSIE 7.0; ITools)",
-				'Accept-Language' => T_lang(),
-			);
+		# urlencode data pairs if is array
+		if (is_array($p['data']))
+			$data = it_url::params($p['data']);
+		else
+			$data = $p['data'];
 
-			if (is_int($p['filemtime']))
-				$p['headers']['If-Modified-Since'] = date("r", $p['filemtime']);
+		$p['headers'] = (array)$p['headers'] + array(
+			'Host' => $url->realhostname . ($url->port != 80 ? ":" . $url->port : ''),
+			'User-Agent' => "Mozilla/4.0 (compatible; MSIE 7.0; ITools)",
+			'Accept-Language' => T_lang(),
+		);
 
-			if ($datalen = strlen($data))
-			{
-				$method = "POST";
-				$p['headers'] += array(
-					'Content-Type' => "application/x-www-form-urlencoded",
-					'Content-Length' => $datalen,
-				);
-			}
-			else
-				$method = "GET";
+		if (is_int($p['filemtime']))
+			$p['headers']['If-Modified-Since'] = date("r", $p['filemtime']);
 
-			if ($url->user || $url->pass)
-				$p['headers'] += array('Authorization' => 'Basic ' . base64_encode($url->user . ':' . $url->pass));
+		if ($datalen = strlen($data))
+		{
+			$method = $p['method'] ?: "POST";
+			$p['headers'] += array(
+				'Content-Type' => "application/x-www-form-urlencoded",
+				'Content-Length' => $datalen,
+			);
+		}
+		else
+			$method = $p['method'] ?: "GET";
 
-			foreach ($p['headers'] as $header => $value)
-					$headers .= "$header: $value\r\n";
+		if ($url->user || $url->pass)
+			$p['headers'] += array('Authorization' => 'Basic ' . base64_encode($url->user . ':' . $url->pass));
 
-			stream_set_timeout($fp, intval($p['timeout']), intval(($p['timeout']*1000000)%1000000));
-			@fputs($fp, "$method /$url->path HTTP/1.0\r\n$headers\r\n$data");
+		foreach ($p['headers'] as $header => $value)
+				$headers .= "$header: $value\r\n";
 
-			while (!feof($fp) && ($line = @fgets($fp, 10240)) && ($line = trim($line)) && (time() < $endtime))
-			{
-				if (preg_match('#^(HTTP\S+)\s(\d+)#', $line, $parts)) # Parse result code
-					$url->headers[$parts[1]] = $url->result = $parts[2];
-				elseif (preg_match('#^Location: (https?://[^/]*)?(/)?(.*)$#i', $line, $parts) && ($parts[1] != $url->url)) # Handle redirects (supports relative and global)
-				{
-					unset($p['url'], $p['headers']['Host']);
-					$url->it_url($parts[1] ? $parts[1].$parts[2].$parts[3] : $url->protocol.'://'.$url->realhostname.($parts[2] ? $parts[2].$parts[3] : '/'.dirname($url->path).'/'.$parts[3]));
-					if (++$url->redir <= 4)  /* Avoid infinite redirects */
-						return $url->get($p);
-				}
-				elseif (preg_match('#^([^:]+): (.*)$#', $line, $parts))
-					$url->headers[$parts[1]] = $parts[2];
-			}
+		stream_set_timeout($fp, intval($p['timeout']), intval(($p['timeout']*1000000)%1000000));
+		@fputs($fp, "$method /$url->path HTTP/1.0\r\n$headers\r\n$data");
 
-			if ($url->result)
+		while (!feof($fp) && ($line = @fgets($fp, 10240)) && ($line = trim($line)) && (time() < $endtime))
+		{
+			if (preg_match('#^(HTTP\S+)\s(\d+)#', $line, $parts)) # Parse result code
+				$url->headers[$parts[1]] = $url->result = $parts[2];
+			elseif (preg_match('#^([^:]+): (.*)$#', $line, $parts))
+				$url->headers[$parts[1]] = $parts[2];
+		}
+
+		if ($url->result)
+		{
+			if ($url->headers['Transfer-Encoding'] == "chunked")	# Bogus HTTP/1.1 chunked answer from server (e.g. Wordpress/Apache2/PHP5)
 			{
-				if ($url->headers['Transfer-Encoding'] == "chunked")	# Bogus HTTP/1.1 chunked answer from server (e.g. Wordpress/Apache2/PHP5)
+				while ($len = hexdec(fgets($fp)))
 				{
-					while ($len = hexdec(fgets($fp)))
-					{
-						$chunk = "";
+					$chunk = "";
 
-						while (!feof($fp) && (strlen($chunk) < $len) && (time() < $endtime))
-							$chunk .= @fread($fp, $len - strlen($chunk));
+					while (!feof($fp) && (strlen($chunk) < $len) && (time() < $endtime))
+						$chunk .= @fread($fp, $len - strlen($chunk));
 
-						$url->data .= $chunk;
-					}
+					$url->data .= $chunk;
 				}
-				else
-				{
-					while (!feof($fp) && (time() < $endtime))
-						$url->data .= @fread($fp, 20480);
-				}
-
-				if ($p['filemtime'] && ($url->result == 304))
-					$result = true;	# Not modified, success but no data
-				else if ($url->result < 400)
-					$result =& $url->data;
+			}
+			else
+			{
+				while (!feof($fp) && (time() < $endtime))
+					$url->data .= @fread($fp, 20480);
 			}
 
-			@fclose($fp);
+			if ($p['filemtime'] && ($url->result == 304))
+				$result = true;	# Not modified, success but no data
+			else if ($url->result < 400)
+				$result =& $url->data;
 		}
-	} else {
-		$results = self::get_multi(array('urls' => array('one' => $p['url'])) + $p);
-		$result = $results['one'];
+
+		@fclose($fp);
 	}
 
 	if (time() >= $endtime)
 		$result = false;
 
-	if (!$result && $p['retries'] > 0 && $url->result < 400)
-		$result = $url->get(array('retries' => $p['retries'] - 1) + $p);
-
 	return $result;
 }
 
@@ -713,7 +729,7 @@ static function absolute($url=null)
 		{
 			$dir = preg_replace('#/[^/]*$#', '/', $_SERVER['PHP_SELF']);
 			$url = preg_match('#^/#', $url) ? $url : "$dir$url";
-			$url = "//" . $_SERVER['HTTP_HOST'] . ($_SERVER['SERVER_PORT'] == (isset($_SERVER['HTTPS']) ? 443 : 80) ? "" : ":{$_SERVER['SERVER_PORT']}") . $url;
+			$url = "//" . $_SERVER['HTTP_HOST'] . $url;
 		}
 		$url = "http" . (isset($_SERVER['HTTPS']) ? 's':'') . ":$url";
 	}
diff --git a/tests/it_url.t b/tests/it_url.t
index 45e882c..a1caaaa 100755
--- a/tests/it_url.t
+++ b/tests/it_url.t
@@ -66,11 +66,12 @@ is(
 $_SERVER['HTTP_HOST'] = "gna.ch";
 is(
 	it_url::absolute("/"),
-	'http://gna.ch:/',
+	'http://gna.ch/',
 	'it_url::absolute basic'
 );
 
 $_SERVER['SERVER_PORT'] = 42;
+$_SERVER['HTTP_HOST'] = "gna.ch:42";
 is(
 	it_url::absolute("/port"),
 	'http://gna.ch:42/port',
@@ -78,6 +79,7 @@ is(
 );
 
 $_SERVER['HTTPS'] = true;
+$_SERVER['HTTP_HOST'] = "gna.ch";
 $_SERVER['SERVER_PORT'] = 443;
 is(
 	it_url::absolute("/https"),
@@ -194,7 +196,7 @@ handle_server(
 	is(
 		it_url::get(U('http://localhost:8000/get_server_value', 'key' => 'HTTP_HOST')),
 		'localhost:8000',
-		'TODO it_url::get() sets correct Host header',
+		'it_url::get() sets correct Host header',
 	)
 );