fix page/image mime content type detection

crontab/crawler.php

@@ -279,11 +279,19 @@ try {
     }
 
     // Skip image processing on MIME type not allowed in settings
-    if (false === strpos(CRAWL_IMAGE_MIME, $hostImageContentType)) {
+    $hostImageBanned = true;
+    foreach ((array) explode(',', CRAWL_IMAGE_MIME) as $mime) {
 
-      $hostImagesBanned += $db->updateHostImageTimeBanned($queueHostImage->hostImageId, time());
+      if (false !== strpos($hostImageContentType, trim($mime))) {
 
-      continue;
+        $hostImageBanned = false;
+        break;
+      }
+    }
+
+    if ($hostImageBanned) {
+
+      $hostImagesBanned += $db->updateHostImageTimeBanned($queueHostImage->hostImageId, time());
     }
 
     // Convert remote image data to base64 string
@@ -359,11 +367,19 @@ try {
     }
 
     // Skip page processing on MIME type not allowed in settings
-    if (false === strpos(CRAWL_PAGE_MIME, $contentType)) {
+    $hostPageBanned = true;
+    foreach ((array) explode(',', CRAWL_PAGE_MIME) as $mime) {
 
-      $hostPagesBanned += $db->updateHostPageTimeBanned($queueHostPage->hostPageId, time());
+      if (false !== strpos($contentType, trim($mime))) {
 
-      continue;
+        $hostPageBanned = false;
+        break;
+      }
+    }
+
+    if ($hostPageBanned) {
+
+      $hostPagesBanned += $db->updateHostPageTimeBanned($queueHostPage->hostPageId, time());
     }
 
     // Skip page processing without returned data