ignore embedded photo binary blobs in n-s-f-w processing

[friendica-addons.git] / nsfw / nsfw.php

diff --git a/nsfw/nsfw.php b/nsfw/nsfw.php
index 0f94158bc83b19dd2d4b92623d808d9bf2be72fc..96e1a9bf8fcc312b396968e3cc106bb857abc720 100755 (executable)
--- a/nsfw/nsfw.php
+++ b/nsfw/nsfw.php
@@ -10,7 +10,7 @@
  */
 
 function nsfw_install() {
-       register_hook('prepare_body', 'addon/nsfw/nsfw.php', 'nsfw_prepare_body');
+       register_hook('prepare_body', 'addon/nsfw/nsfw.php', 'nsfw_prepare_body', 10);
        register_hook('plugin_settings', 'addon/nsfw/nsfw.php', 'nsfw_addon_settings');
        register_hook('plugin_settings_post', 'addon/nsfw/nsfw.php', 'nsfw_addon_settings_post');
 
@@ -24,6 +24,37 @@ function nsfw_uninstall() {
 
 }
 
+// This function isn't perfect and isn't trying to preserve the html structure - it's just a 
+// quick and dirty filter to pull out embedded photo blobs because 'nsfw' seems to come up 
+// inside them quite often. We don't need anything fancy, just pull out the data blob so we can
+// check against the rest of the body. 
+ 
+function nsfw_extract_photos($body) {
+
+       $new_body = '';
+       
+       $img_start = strpos($body,'src="data:');
+       $img_end = (($img_start !== false) ? strpos(substr($body,$img_start),'>') : false);
+
+       $cnt = 0;
+
+       while($img_end !== false) {
+               $img_end += $img_start;
+               $new_body = $new_body . substr($body,0,$img_start);
+       
+               $cnt ++;
+               $body = substr($body,0,$img_end);
+
+               $img_start = strpos($body,'src="data:');
+               $img_end = (($img_start !== false) ? strpos(substr($body,$img_start),'>') : false);
+
+       }
+
+       if(! $cnt)
+               return $body;
+
+       return $new_body;
+}
 
 
 
@@ -77,6 +108,7 @@ function nsfw_addon_settings_post(&$a,&$b) {
 
 function nsfw_prepare_body(&$a,&$b) {
 
+
        $words = null;
        if(get_pconfig(local_user(),'nsfw','disable'))
                return;
@@ -93,18 +125,22 @@ function nsfw_prepare_body(&$a,&$b) {
 
        $found = false;
        if(count($arr)) {
+
+               $body = nsfw_extract_photos($b['html']);
+
                foreach($arr as $word) {
-                       if(! strlen(trim($word))) {
+                       $word = trim($word);
+                       if(! strlen($word)) {
                                continue;
                        }
                        if(strpos($word,'/') === 0) {
-                               if(preg_match($word,$b['html'])) {
+                               if(preg_match($word,$body)) {
                                        $found = true;
                                        break;
                                }
                        }
                        else {
-                               if(stristr($b['html'],$word)) {
+                               if(stristr($body,$word)) {
                                        $found = true;
                                        break;
                                }
@@ -114,6 +150,7 @@ function nsfw_prepare_body(&$a,&$b) {
                                }
                        } 
                }
+               
        }
        if($found) {
                $rnd = random_string(8);