From 3deb4216cd947c4eb14b0a9e3b9f637ca25e77d3 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Roland=20H=C3=A4der?= <roland@mxchange.org>
Date: Sun, 22 Nov 2009 19:29:12 +0000
Subject: [PATCH] Fix for output to bots, 'slurp' is now detected

---
 inc/classes/rdf.class.php |  1 -
 inc/functions.php         | 41 +++++++++++++++++++--------------------
 inc/wrapper-functions.php | 12 ++++++++++--
 index.php                 | 12 ++++++------
 4 files changed, 36 insertions(+), 30 deletions(-)

diff --git a/inc/classes/rdf.class.php b/inc/classes/rdf.class.php
index 4ae354b702..95190ff85b 100644
--- a/inc/classes/rdf.class.php
+++ b/inc/classes/rdf.class.php
@@ -419,7 +419,6 @@ class fase4_rdf {
 		} else {
 			return $this->out;
 		}
-		flush();
 		$this->_garbage_collection();
 	}
 
diff --git a/inc/functions.php b/inc/functions.php
index 8b57a86193..d818478002 100644
--- a/inc/functions.php
+++ b/inc/functions.php
@@ -85,16 +85,7 @@ function outputHtml ($htmlCode, $newLine = true) {
 				app_die(__FUNCTION__, __LINE__, '<strong>{--FATAL_ERROR--}:</strong> {--LANG_NO_RENDER_DIRECT--}');
 				break;
 		} // END - switch
-	} elseif ((getPhpCaching() == 'on') && (isset($GLOBALS['footer_sent'])) && ($GLOBALS['footer_sent'] == 1)) {
-		// Headers already sent?
-		if (headers_sent()) {
-			// Log this error
-			logDebugMessage(__FUNCTION__, __LINE__, 'Headers already sent! We need debug backtrace here.');
-
-			// Trigger an user error
-			debug_report_bug('Headers are already sent!');
-		} // END - if
-
+	} elseif (getPhpCaching() == 'on') {
 		// Output cached HTML code
 		$GLOBALS['output'] = ob_get_contents();
 
@@ -142,7 +133,7 @@ function sendHttpHeaders () {
 	sendHeader('HTTP/1.1 200');
 
 	// General headers for no caching
-	sendHeader('Expired: ' . $now); // RFC2616 - Section 14.21
+	sendHeader('Expires: ' . $now); // RFC2616 - Section 14.21
 	sendHeader('Last-Modified: ' . $now);
 	sendHeader('Cache-Control: no-store, no-cache, must-revalidate, pre-check=0, post-check=0, max-age=0'); // HTTP/1.1
 	sendHeader('Pragma: no-cache'); // HTTP/1.0
@@ -177,8 +168,20 @@ function compileFinalOutput () {
 		$cnt++;
 	} // END - while
 
+	// Compress it?
+	if (!empty($_SERVER['HTTP_ACCEPT_ENCODING']) && (strpos('gzip', $_SERVER['HTTP_ACCEPT_ENCODING']) !== null)) {
+		// Compress it
+		$GLOBALS['output'] = gzencode($GLOBALS['output'], 9, true);
+
+		// Add header
+		sendHeader('Content-Encoding: gzip');
+	} // END - if
+
 	// Add final length
 	sendHeader('Content-Length: ' . strlen($GLOBALS['output']));
+
+	// Flush all headers
+	flushHeaders();
 }
 
 // Output the raw HTML code
@@ -918,14 +921,6 @@ function redirectToUrl ($URL) {
 		$rel = '';
 	} // END - if
 
-	// Get output buffer
-	$GLOBALS['output'] = ob_get_contents();
-
-	// Clear it only if there is content
-	if (!empty($GLOBALS['output'])) {
-		clearOutputBuffer();
-	} // END - if
-
 	// Three different ways to debug...
 	//* DEBUG: */ debug_report_bug(sprintf("%s[%s:] URL=%s", __FUNCTION__, __LINE__, $URL));
 	//* DEBUG: */ logDebugMessage(__FUNCTION__, __LINE__, 'URL=' . $URL);
@@ -3828,6 +3823,7 @@ function encodeUrl ($url, $outputMode = '0') {
 
 		// Add it to URL
 		if (session_id() != '') {
+			die($url);
 			$url .= $seperator . session_name() . '=' . session_id();
 		} // END - if
 	} // END - if
@@ -3844,11 +3840,14 @@ function encodeUrl ($url, $outputMode = '0') {
 
 // Simple check for spider
 function isSpider () {
+	// Get the UA
+	$userAgent = strtolower(detectUserAgent(true));
+
 	// It should not be empty, if so it is better a spider/bot
-	if (detectUserAgent(true) == '') return true;
+	if (empty($userAgent)) return true;
 
 	// Is it a spider?
-	return ((strpos('spider', strtolower(detectUserAgent(true))) !== false) || (strpos('bot', strtolower(detectUserAgent(true))) !== false));
+	return ((strpos($userAgent, 'spider') !== false) || (strpos($userAgent, 'slurp') !== false) || (strpos($userAgent, 'bot') !== false));
 }
 
 //////////////////////////////////////////////////
diff --git a/inc/wrapper-functions.php b/inc/wrapper-functions.php
index de2d8c0fdb..d3369e758f 100644
--- a/inc/wrapper-functions.php
+++ b/inc/wrapper-functions.php
@@ -379,14 +379,22 @@ function copyFileVerified ($source, $dest, $chmod = '') {
 // Wrapper function for header()
 // Send a header but checks before if we can do so
 function sendHeader ($header) {
+	// Send the header
+	$GLOBALS['header'][] = trim($header);
+}
+
+// Flushes all headers
+function flushHeaders () {
 	// Is the header already sent?
 	if (headers_sent()) {
 		// Then abort here
 		debug_report_bug('Headers already sent!');
 	} // END - if
 
-	// Send the header
-	header(trim($header));
+	// Flush all headers
+	foreach ($GLOBALS['header'] as $header) {
+		header($header);
+	} // END - foreach
 }
 
 // Wrapper function for chmod()
diff --git a/index.php b/index.php
index 4c859b6e3b..df23da6814 100644
--- a/index.php
+++ b/index.php
@@ -55,15 +55,15 @@ require('inc/config-global.php');
 // Set content type
 setContentType('text/html');
 
-// Header
-loadIncludeOnce('inc/header.php');
-
 // Fix missing array elements here
 if (!isConfigEntrySet('index_delay'))  setConfigEntry('index_delay' , 0);
 if (!isConfigEntrySet('index_cookie')) setConfigEntry('index_cookie', 0);
 
 // Check for cookies
 if ((isSessionVariableSet('visited')) || (getConfig('index_delay') > -1)) {
+	// Header
+	loadIncludeOnce('inc/header.php');
+
 	// Is the index page configured for redirect pr not?
 	if (getConfig('index_cookie') > 0) {
 		// Set cookie and remeber it for specified time
@@ -85,13 +85,13 @@ if ((isSessionVariableSet('visited')) || (getConfig('index_delay') > -1)) {
 		// Load template
 		loadTemplate('index_forward', false, $content);
 	} // END - if
+
+	// Footer
+	loadIncludeOnce('inc/footer.php');
 } else {
 	// Redirect to main page
 	redirectToUrl('modules.php?module=index');
 }
 
-// Footer
-loadIncludeOnce('inc/footer.php');
-
 // [EOF]
 ?>
-- 
2.39.5