01229: IDN links are incorrectly encoded

Summary: IDN links are incorrectly encoded
Created: 2010-10-13 17:40
Status: Open
Category: Bug
From: Petko
Priority: 22
Version: 2.2.18

Description: Domain names containing international characters use a different standard encoding (Punycode) than the URL path (URL encoding).


For exemple,

  • the domain www.акцент.bg should be encoded to www.xn--80akonzv.bg or left as-is in the (UTF-8) page text, but it is currently incorrectly percent-encoded.
  • a path like /Петко/ is correctly percent-encoded to /%D0%9F%D0%B5%D1%82%D0%BA%D0%BE/

Currently PmWiki URL-encodes the full address -- it should either apply the Punycode function to the domain part, or leave the domain part as text and let the browser figure it out. Most browsers as of 2010 are able to follow such links.

References :