Refactor URI detection

Author: MGatner

system/HTTP/IncomingRequest.php

@@ -56,12 +56,26 @@ class IncomingRequest extends Request
 	protected $enableCSRF = false;
 
 	/**
-	 * A \CodeIgniter\HTTP\URI instance.
+	 * The URI representing the current URL for this request.
+	 *
+	 * Note: When this is returned as a shared instance via
+	 * Services::request() this property points to the shared
+	 * URI service and represents the primary "current URL"
+	 * for the main app.
+	 *
+	 * @see CodeIgniter\CodeIgniter::getRequestObject()
 	 *
 	 * @var URI
 	 */
 	public $uri;
 
+	/**
+	 * The detected path (relative to SCRIPT_NAME).
+	 *
+	 * @var string|null
+	 */
+	protected $path;
+
 	/**
 	 * File collection
 	 *
@@ -142,31 +156,16 @@ public function __construct($config, URI $uri = null, $body = 'php://input', Use
 			$body = file_get_contents('php://input');
 		}
 
-		$this->body      = ! empty($body) ? $body : null;
-		$this->config    = $config;
-		$this->userAgent = $userAgent;
+		$this->config       = $config;
+		$this->uri          = $uri;
+		$this->body         = ! empty($body) ? $body : null;
+		$this->userAgent    = $userAgent;
+		$this->validLocales = $config->supportedLocales;
 
 		parent::__construct($config);
 
 		$this->populateHeaders();
-
-		// Determine the current URI
-		// NOTE: This WILL NOT match the actual URL in the browser since for
-		// everything this cares about (and the router, etc) is the portion
-		// AFTER the script name. So, if hosted in a sub-folder this will
-		// appear different than actual URL. If you need that, use current_url().
-		$this->uri = $uri;
-
-		$this->detectURI($config->uriProtocol, $config->baseURL);
-
-		// Check if the baseURL scheme needs to be coerced into its secure version
-		if ($config->forceGlobalSecureRequests && $this->uri->getScheme() === 'http')
-		{
-			$this->uri->setScheme('https');
-		}
-
-		$this->validLocales = $config->supportedLocales;
-
+		$this->detectURL();
 		$this->detectLocale($config);
 	}
 
@@ -603,13 +602,67 @@ public function getFile(string $fileID)
 
 	//--------------------------------------------------------------------
 
+	/**
+	 * Sets the components of the underlying URI instance to
+	 * match the current URL as configured and detected.
+	 */
+	protected function detectURL()
+	{
+		// Require a baseURL for non-CLI requests
+		if ($this->config->baseURL === '')
+		{
+			if (is_cli())
+			{
+				$this->uri->setPath($this->getPath());
+				return;
+			}
+
+			die('You have an empty or invalid base URL. The baseURL value must be set in Config\App.php, or through the .env file.'); // @codeCoverageIgnore
+		}
+
+		// Build the full URL based on the config and relative path
+		$path = $this->getPath();
+		$url  = rtrim($this->config->baseURL, '/ ') . '/';
+
+		// Check for an index page
+		if ($this->config->indexPage !== '')
+		{
+			$url .= $this->config->indexPage;
+
+			// If the path has anything other than ? (for query) we need a separator
+			if ($path !== '' && strncmp($path, '?', 1) !== 0)
+			{
+				$url .= '/';
+			}
+		}
+
+		$url .= $path;
+
+		$this->uri->setURI($url);
+
+		// Check if the baseURL scheme needs to be coerced into its secure version
+		if ($this->config->forceGlobalSecureRequests && $this->uri->getScheme() === 'http')
+		{
+			$this->uri->setScheme('https');
+		}
+
+		// Set any remaining query vars
+		if (isset($_SERVER['QUERY_STRING']))
+		{
+			$this->uri->setQuery($_SERVER['QUERY_STRING']);
+		}
+	}
+
 	/**
 	 * Sets up our URI object based on the information we have. This is
 	 * either provided by the user in the baseURL Config setting, or
 	 * determined from the environment as needed.
 	 *
 	 * @param string $protocol
 	 * @param string $baseURL
+	 *
+	 * @deprecated         Use detectCurrentURI() instead
+	 * @codeCoverageIgnore
 	 */
 	protected function detectURI(string $protocol, string $baseURL)
 	{
@@ -632,20 +685,34 @@ protected function detectURI(string $protocol, string $baseURL)
 		}
 		else
 		{
-			// @codeCoverageIgnoreStart
 			if (! is_cli())
 			{
 				die('You have an empty or invalid base URL. The baseURL value must be set in Config\App.php, or through the .env file.');
 			}
-			// @codeCoverageIgnoreEnd
 		}
 	}
 
 	//--------------------------------------------------------------------
 
 	/**
-	 * Based on the URIProtocol Config setting, will attempt to
-	 * detect the path portion of the current URI.
+	 * Returns the path relative to SCRIPT_NAME,
+	 * running detection as necessary.
+	 *
+	 * @return string
+	 */
+	public function getPath(): string
+	{
+		if (is_null($this->path))
+		{
+			$this->detectPath($this->config->uriProtocol);
+		}
+
+		return $this->path;
+	}
+
+	/**
+	 * Detects the relative path based on
+	 * the URIProtocol Config setting.
 	 *
 	 * @param string $protocol
 	 *
@@ -661,18 +728,18 @@ public function detectPath(string $protocol = ''): string
 		switch ($protocol)
 		{
 			case 'REQUEST_URI':
-				$path = $this->parseRequestURI();
+				$this->path = $this->parseRequestURI();
 				break;
 			case 'QUERY_STRING':
-				$path = $this->parseQueryString();
+				$this->path = $this->parseQueryString();
 				break;
 			case 'PATH_INFO':
 			default:
-				$path = $this->fetchGlobal('server', $protocol) ?? $this->parseRequestURI();
+				$this->path = $this->fetchGlobal('server', $protocol) ?? $this->parseRequestURI();
 				break;
 		}
 
-		return $path;
+		return $this->path;
 	}
 
 	//--------------------------------------------------------------------
@@ -731,23 +798,23 @@ protected function parseRequestURI(): string
 		$query = $parts['query'] ?? '';
 		$uri   = $parts['path'] ?? '';
 
-		if (isset($_SERVER['SCRIPT_NAME'][0]) && pathinfo($_SERVER['SCRIPT_NAME'], PATHINFO_EXTENSION) === 'php')
+		// Strip the SCRIPT_NAME path from the URI
+		if ($uri !== '' && isset($_SERVER['SCRIPT_NAME'][0]) && pathinfo($_SERVER['SCRIPT_NAME'], PATHINFO_EXTENSION) === 'php')
 		{
-			// strip the script name from the beginning of the URI
-			if (strpos($uri, $_SERVER['SCRIPT_NAME']) === 0 && strpos($uri, '/index.php') === 0)
+			// Compare each segment, dropping them until there is no match
+			$segments = $keep = explode('/', $uri);
+			foreach (explode('/', $_SERVER['SCRIPT_NAME']) as $i => $segment)
 			{
-				$uri = (string) substr($uri, strlen($_SERVER['SCRIPT_NAME']));
-			}
-			// if the script is nested, strip the parent folder & script from the URI
-			elseif (strpos($uri, $_SERVER['SCRIPT_NAME']) > 0)
-			{
-				$uri = (string) substr($uri, strpos($uri, $_SERVER['SCRIPT_NAME']) + strlen($_SERVER['SCRIPT_NAME']));
-			}
-			// or if index.php is implied
-			elseif (strpos($uri, dirname($_SERVER['SCRIPT_NAME'])) === 0)
-			{
-				$uri = (string) substr($uri, strlen(dirname($_SERVER['SCRIPT_NAME'])));
+				// If these segments are not the same then we're done
+				if ($segment !== $segments[$i])
+				{
+					break;
+				}
+
+				array_shift($keep);
 			}
+
+			$uri = implode('/', $keep);
 		}
 
 		// This section ensures that even on servers that require the URI to contain the query string (Nginx) a correct