Fix linking in CJK text

URLs without a protocol, with a leading hyphen, with a short TLD inside CJK text
should no longer be linked

extract.yml

@@ -444,6 +444,10 @@ tests:
       text: "test http://twitter.com/."
       expected: ["http://twitter.com/"]
 
+    - description: "DO NOT extract URL with leading hyphen inside CJK text"
+      text: "ル-twpf.jpル"
+      expected: []
+
     - description: "Extract a URL with '?' in fragment"
       text: "http://tn.com.ar/show/00056158/la-estrella-del-certamen-el-turno-de-pamela-anderson?fb_xd_fragment#?=&cb=fe17523f223b7&relation=parent.parent&transport=fragment&type=resize&height=20&ackdata"
       expected: ["http://tn.com.ar/show/00056158/la-estrella-del-certamen-el-turno-de-pamela-anderson?fb_xd_fragment#?=&cb=fe17523f223b7&relation=parent.parent&transport=fragment&type=resize&height=20&ackdata"]