X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=application%2FHttpUtils.php;h=2e0792f97a5d2fdbfdf2b8978e2012300f519e54;hb=fd5ac47ea27353e641a521b5e1867bfd20ff0512;hp=499220c596c201c098bc2178417c09b7d89f8be1;hpb=482d67bd523bf12f36508a0131d09fe299ee02f4;p=github%2Fshaarli%2FShaarli.git

diff --git a/application/HttpUtils.php b/application/HttpUtils.php
index 499220c5..2e0792f9 100644
--- a/application/HttpUtils.php
+++ b/application/HttpUtils.php
@@ -13,7 +13,7 @@
  *  [1] = URL content (downloaded data)
  *
  * Example:
- *  list($headers, $data) = get_http_url('http://sebauvage.net/');
+ *  list($headers, $data) = get_http_response('http://sebauvage.net/');
  *  if (strpos($headers[0], '200 OK') !== false) {
  *      echo 'Data type: '.htmlspecialchars($headers['Content-Type']);
  *  } else {
@@ -24,31 +24,106 @@
  * @see http://php.net/manual/en/function.stream-context-create.php
  * @see http://php.net/manual/en/function.get-headers.php
  */
-function get_http_url($url, $timeout = 30, $maxBytes = 4194304)
+function get_http_response($url, $timeout = 30, $maxBytes = 4194304)
 {
+    $urlObj = new Url($url);
+    $cleanUrl = $urlObj->idnToAscii();
+
+    if (! filter_var($cleanUrl, FILTER_VALIDATE_URL) || ! $urlObj->isHttp()) {
+        return array(array(0 => 'Invalid HTTP Url'), false);
+    }
+
     $options = array(
         'http' => array(
             'method' => 'GET',
             'timeout' => $timeout,
-            'user_agent' => 'Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:23.0)'
-                         .' Gecko/20100101 Firefox/23.0'
+            'user_agent' => 'Mozilla/5.0 (X11; Fedora; Linux x86_64; rv:45.0)'
+                         .' Gecko/20100101 Firefox/45.0',
+            'accept_language' => substr(setlocale(LC_COLLATE, 0), 0, 2) . ',en-US;q=0.7,en;q=0.3',
         )
     );
 
-    $context = stream_context_create($options);
+    stream_context_set_default($options);
+    list($headers, $finalUrl) = get_redirected_headers($cleanUrl);
+    if (! $headers || strpos($headers[0], '200 OK') === false) {
+        $options['http']['request_fulluri'] = true;
+        stream_context_set_default($options);
+        list($headers, $finalUrl) = get_redirected_headers($cleanUrl);
+    }
+
+    if (! $headers || strpos($headers[0], '200 OK') === false) {
+        return array($headers, false);
+    }
 
     try {
         // TODO: catch Exception in calling code (thumbnailer)
-        $content = file_get_contents($url, false, $context, -1, $maxBytes);
+        $context = stream_context_create($options);
+        $content = file_get_contents($finalUrl, false, $context, -1, $maxBytes);
     } catch (Exception $exc) {
         return array(array(0 => 'HTTP Error'), $exc->getMessage());
     }
 
-    if (!$content) {
-        return array(array(0 => 'HTTP Error'), '');
+    return array($headers, $content);
+}
+
+/**
+ * Retrieve HTTP headers, following n redirections (temporary and permanent ones).
+ *
+ * @param string $url              initial URL to reach.
+ * @param int    $redirectionLimit max redirection follow.
+ *
+ * @return array HTTP headers, or false if it failed.
+ */
+function get_redirected_headers($url, $redirectionLimit = 3)
+{
+    $headers = get_headers($url, 1);
+    if (!empty($headers['location']) && empty($headers['Location'])) {
+        $headers['Location'] = $headers['location'];
+    }
+
+    // Headers found, redirection found, and limit not reached.
+    if ($redirectionLimit-- > 0
+        && !empty($headers)
+        && (strpos($headers[0], '301') !== false || strpos($headers[0], '302') !== false)
+        && !empty($headers['Location'])) {
+
+        $redirection = is_array($headers['Location']) ? end($headers['Location']) : $headers['Location'];
+        if ($redirection != $url) {
+            $redirection = getAbsoluteUrl($url, $redirection);
+            return get_redirected_headers($redirection, $redirectionLimit);
+        }
+    }
+
+    return array($headers, $url);
+}
+
+/**
+ * Get an absolute URL from a complete one, and another absolute/relative URL.
+ *
+ * @param string $originalUrl The original complete URL.
+ * @param string $newUrl      The new one, absolute or relative.
+ *
+ * @return string Final URL:
+ *   - $newUrl if it was already an absolute URL.
+ *   - if it was relative, absolute URL from $originalUrl path.
+ */
+function getAbsoluteUrl($originalUrl, $newUrl)
+{
+    $newScheme = parse_url($newUrl, PHP_URL_SCHEME);
+    // Already an absolute URL.
+    if (!empty($newScheme)) {
+        return $newUrl;
     }
 
-    return array(get_headers($url, 1), $content);
+    $parts = parse_url($originalUrl);
+    $final = $parts['scheme'] .'://'. $parts['host'];
+    $final .= (!empty($parts['port'])) ? $parts['port'] : '';
+    $final .= '/';
+    if ($newUrl[0] != '/') {
+        $final .= substr(ltrim($parts['path'], '/'), 0, strrpos($parts['path'], '/'));
+    }
+    $final .= ltrim($newUrl, '/');
+    return $final;
 }
 
 /**
@@ -71,11 +146,21 @@ function server_url($server)
     // Shaarli is served behind a proxy
     if (isset($server['HTTP_X_FORWARDED_PROTO'])) {
         // Keep forwarded scheme
-        $scheme = $server['HTTP_X_FORWARDED_PROTO'];
+        if (strpos($server['HTTP_X_FORWARDED_PROTO'], ',') !== false) {
+            $schemes = explode(',', $server['HTTP_X_FORWARDED_PROTO']);
+            $scheme = trim($schemes[0]);
+        } else {
+            $scheme = $server['HTTP_X_FORWARDED_PROTO'];
+        }
 
         if (isset($server['HTTP_X_FORWARDED_PORT'])) {
             // Keep forwarded port
-            $port = ':'.$server['HTTP_X_FORWARDED_PORT'];
+            if (strpos($server['HTTP_X_FORWARDED_PORT'], ',') !== false) {
+                $ports = explode(',', $server['HTTP_X_FORWARDED_PORT']);
+                $port = ':' . trim($ports[0]);
+            } else {
+                $port = ':' . $server['HTTP_X_FORWARDED_PORT'];
+            }
         }
 
         return $scheme.'://'.$server['SERVER_NAME'].$port;
@@ -108,7 +193,7 @@ function server_url($server)
 function index_url($server)
 {
     $scriptname = $server['SCRIPT_NAME'];
-    if (endswith($scriptname, 'index.php')) {
+    if (endsWith($scriptname, 'index.php')) {
         $scriptname = substr($scriptname, 0, -9);
     }
     return server_url($server) . $scriptname;