# HG changeset patch
# User Matti Hamalainen <ccr@tnsp.org>
# Date 1518022896 -7200
# Node ID 506977ea9d0c5b838c21ee54ab9e997cc25d6514
# Parent  fe478b7bd80ee030f4d1d507e2380d59a7fc31ee
urllog: Improve URL validation.

diff -r fe478b7bd80e -r 506977ea9d0c urllog.tcl
--- a/urllog.tcl	Mon Dec 18 14:54:18 2017 +0200
+++ b/urllog.tcl	Wed Feb 07 19:01:36 2018 +0200
@@ -290,7 +290,7 @@
 
   ### Try to guess the URL protocol component (if it is missing)
   set u_checktld 1
-  if {![string match "http://*" $urlStr] && ![string match "https://*" $urlStr] && ![string match "ftp://*" $urlStr]} {
+  if {![string match "http://*" $urlStr] && ![string match "https://*" $urlStr] && ![string match "ftp://*" $urlStr] && ![string match "*://*" $urlStr]} {
     if {[string match "*www.*" $urlStr]} {
       set urlStr "http://$urlStr"
     } elseif {[string match "*ftp.*" $urlStr]} {
@@ -323,7 +323,16 @@
 
   ### Get URL protocol component
   set urlProto ""
-  regexp "(\[a-z\]+)://" $urlStr urlMatch urlProto
+  if  {[regexp "(\[a-z\]+)://" $urlStr urlMatch urlProto]} {
+    ### Is it a http or ftp url?
+    if {$urlProto != "http" && $urlProto != "https" && $urlProto != "ftp"} {
+      urllog_log "Broken URL from $urlNick: ($urlStr) UNSUPPORTED protocol class ($urlProto)."
+      return 0
+    }
+  } else {
+    urllog_log "Broken URL from $urlNick: ($urlStr), no protocol specifier."
+    return 0
+  }
 
   ### Check the PORT (if the ":" is there)
   set urlRecord [split $urlStr "/"]
@@ -555,7 +564,7 @@
   if {[utl_match_delim_list $urllog_log_channels $uchan]} {
     ### Do the URL checking
     foreach str [split $utext " "] {
-      if {[regexp "((ftp|http|https)://\[^\[:space:\]\]+|^(www|ftp)\.\[^\[:space:\]\]+)" $str ulink]} {
+      if {[regexp "(\[a-z]+://\[^\[:space:\]\]+|^(www|ftp)\.\[^\[:space:\]\]+)" $str ulink]} {
         urllog_check_url $str $unick $uhost $uchan
       }
     }