Let’s see how tags work for Latin-8 characters first (for example, the French word français) and then for Chinese.

In Flickr

Let’s look at http://flickr.com/photos/tags/fran%C3%A7ais/. There is no collapsing of français to francais. See the photo at http://flickr.com/photos/raymondyee/368644336/ to see that I can have both a français and francais tag; invoking the API^[57]

It seems that Chinese works in a similar way. I don’t know much Chinese, but I do know my name in Chinese (余俊雄). I managed to add it as a tag for one of my pictures.^[58]

http://flickr.com/photos/raymondyee/tags/%E4%BD%99%E4%BF%8A%E9%9B%84/

Again, you can confirm that %E4%BD%99%E4%BF%8A%E9%9B%84 is a URL-­encoded UTF-8 representation of my Chinese name. With Python, here’s that code:

               import urllib
               print urllib.unquote('%E4%BD%99%E4%BF%8A%E9%9B%84').decode('utf-8')
               u'\u4f59\u4fca\u96c4'
            

And you can see that the Unicode character point 4f59 is indeed .^[59]

In del.icio.us

I added a URL for the France-­Berkeley program:^[60]

http://del.icio.us/tag/fran%C3%A7ais ?does come up with many links with the tag français, as well as the corresponding full-­text search for français.^[61]

To test Chinese functionality in del.icio.us, I added my picture,^[62][63][64]fran%C3%A7ais from français? With a bit of Python programming, you can convince yourself that it’s a URL-­encoding of the UTF-8 encoding of français:

               >>> print chr(231)
               ç
               >>> print urllib.urlencode({'q':chr(231).decode('ISO-8859-1').Â
               encode('utf-8')})
               q=%C3%A7
            

YouTube copes well with Chinese characters too: I can find a video tagged with my Chinese name: http://www.youtube.com/results?search_query=%E4%BD%99%E4%BF%8A%E9%9B%84.

In rel-­tag

The rel-­tag specification gives the following example of how to encode tags:^[65]

               <a href="http://technorati.com/tag/Sant%C3%A9+et+bien-%C3%AAtre" rel="tag">
               Santé et bien-être</a>
            

You can verify that the tag is the URL encoding of the UTF-8 encoding of the tag string. In Python, the following code:

               import urllib
               s = "Santé et bien-être"
               u = s.decode('iso-8859-1')
               print urllib.urlencode({'q':u.encode('utf8')})
            

returns the following:

               q=Sant%C3%A9+et+bien-%C3%AAtre
               http://technorati.com/tag/Sant%C3%A9+et+bien-%C3%AAtre
            

which is a search on Santé et bien-être.