handle case-insensitive LinK, relative URLs

handle case-insensitive LinK header; added optional $url param to http_rels so it can turn relative Link header URLs into absolute URLs; disable CURLOPT_FOLLOWLOCATION because it causes PHP warnings
indieweb · 1558cf145091268f95c9a7e4f2813d5e2a9fa0d1 · tantek committed Apr 11, 2016 · Apr 11, 2016 · 1558cf1
1 parent eda9800
commit 1558cf145091268f95c9a7e4f2813d5e2a9fa0d1
Unified Split

Showing with 16 additions and 10 deletions.

+16 −10 src/IndieWeb/link_rel_parser.php
@@ -7,19 +7,20 @@
 */
 
 /**
- * @param string $h  HTTP headers
- * @return array $rels rel values as indices to arrays of URLs
+ * @param string $h  HTTP headers as a string
+ * @param string $url optional base URL to resolve relative URLs
+ * @return array $rels rel values as indices to arrays of URLs, empty array if no rels at all
  */
-function http_rels($h) {
+function http_rels($h, $url = '') {
   $h = preg_replace("/(\r\n|\r)/", "\n", $h);
   $h = explode("\n", preg_replace("/(\n)[ \t]+/", " ", $h));
   $rels = array();
   foreach ($h as $f) {
-    if (!strncmp($f, 'X-Pingback: ', 12)) {
+    if (!strncasecmp($f, 'X-Pingback: ', 12)) {
       // convert to a link header and have common code handle it
       $f = 'Link: <' . trim(substr($f, 12)) . '>; rel="pingback"';
     }
-    if (!strncmp($f, 'Link: ', 6)) {
+    if (!strncasecmp($f, 'Link: ', 6)) {
       $links = explode(', ', trim(substr($f, 6)));
       foreach ($links as $link) {
         $hrefandrel = explode('; ', $link);
@@ -38,6 +39,9 @@ function http_rels($h) {
               if (!array_key_exists($rel, $rels)) {
                 $rels[$rel] = array();
               }
+              if ($url) { 
+                $href = get_absolute_uri($href, $url); 
+              }
               if (!in_array($href, $rels[$rel])) {
                 $rels[$rel][] = $href;
               }
@@ -52,7 +56,7 @@ function http_rels($h) {
 
 /**
  * @param $url  URL to get HTTP HEAD Link (and effective/x-extended) rels
- * @return array  "status"=> HTTP status code, "type"=> HTTP Content-Type, "rels" array with http_rels return value
+ * @return array  "status"=> HTTP status code, "type"=> HTTP Content-Type, "rels" array with http_rels return value. empty array if no rels
  */
 function head_http_rels($url) {
   $c = curl_init();
@@ -61,20 +65,22 @@ function head_http_rels($url) {
   curl_setopt($c, CURLOPT_CONNECTTIMEOUT, 2);
   curl_setopt($c, CURLOPT_TIMEOUT, 4);
   curl_setopt($c, CURLOPT_USERAGENT, 'head_http_rels function');
-  curl_setopt($c, CURLOPT_FOLLOWLOCATION, 1);
+// curl_setopt($c, CURLOPT_FOLLOWLOCATION, 1);
+// commented out due to:
+// Warning: curl_setopt(): CURLOPT_FOLLOWLOCATION cannot be activated when an open_basedir is set
   curl_setopt($c, CURLOPT_SSL_VERIFYPEER , false );
   curl_setopt($c, CURLOPT_SSL_VERIFYHOST , false );
   curl_setopt($c, CURLOPT_HEADER, true);
   curl_setopt($c, CURLOPT_NOBODY, true);
   $h = curl_exec($c);
-	$i = curl_getinfo($c);
+  $i = curl_getinfo($c);
   curl_close($c);
   unset($c);
 
   $r = array();
-  $r['status'] = $i['http_code'];
+  $r['status'] = string($i['http_code']);
   $r['type'] = $i['content_type'];
-  $r['rels'] = http_rels($h);
+  $r['rels'] = http_rels($h, $url);
   return $r;
 }
 ?>