vimwiki

commit 50c95f1beffa3d80a0e636a28c23dd51a6bd7803
parent 4d4e88bfa26de6a638e4f3c0e3b7c315d29e9e50
Author: Michael F. Schönitzer <michael@schoenitzer.de>
Date:   Sun, 17 Jun 2018 16:10:44 +0200

Remove all protocols and tlds on URL-cleaning

Also make removal of protocol, tld and 'www' more robust against
fails-positives by considering the position in the url.

Diffstat:
M
autoload/vimwiki/base.vim
 | 
23
++++++++++-------------

1 file changed, 10 insertions(+), 13 deletions(-)
diff --git a/autoload/vimwiki/base.vim b/autoload/vimwiki/base.vim
@@ -1858,20 +1858,17 @@ endfunction
 
 
 function! s:clean_url(url)
-  let url = split(a:url, '/\|=\|-\|&\|?\|\.')
+  " remove protocol and tld
+  let url = substitute(a:url, '^\a\+://', '', '')
+  let url = substitute(url, '^\([^/]\+\).\a\{2,4}/', '\1/', '')
+  let url = split(url, '/\|=\|-\|&\|?\|\.')
   let url = filter(url, 'v:val !=# ""')
-  let url = filter(url, 'v:val !=# "www"')
-  let url = filter(url, 'v:val !=# "com"')
-  let url = filter(url, 'v:val !=# "org"')
-  let url = filter(url, 'v:val !=# "net"')
-  let url = filter(url, 'v:val !=# "edu"')
-  let url = filter(url, 'v:val !=# "http\:"')
-  let url = filter(url, 'v:val !=# "https\:"')
-  let url = filter(url, 'v:val !=# "file\:"')
-  let url = filter(url, 'v:val !=# "xml\:"')
-  let url = filter(url, 'v:val !=# "html"')
-  let url = filter(url, 'v:val !=# "htm"')
-  let url = filter(url, 'v:val !=# "php"')
+  if url[0] == "www"
+    let url = url[1:]
+  endif
+  if url[-1] =~ '^\(htm\|html\|php\)$'
+    let url = url[0:-2]
+  endif
   " remove words consisting of only hexadecimal digits or non-word characters
   let url = filter(url, 'v:val !~  "^\\A\\{4,}$"')
   let url = filter(url, 'v:val !~  "^\\x\\{4,}$" || v:val !~ "\\d"')