Update to 5.1.3. From the changelog:

* Consolidated and simplified the character encoding detection code
* Issue 346 (the gb2312 encoding isn't always upgraded to gb18030)
* Issue 350 (HTTP Last-Modified example is incorrect in documentation)
* Issue 352 (importing lxml.etree changes what exceptions libxml2 throws)
* Issue 356 (add support for the HTML5 attributes `poster` and `preload`)
* Issue 364 (enclosure-sniffing microformat code can throw ValueError)
* Issue 373 (support RFC822-ish dates with swapped days and months)
* Issue 376 (uppercase 'X' in hex character references cause ValueError)
* Issue 382 (don't strip inline user:password credentials from FTP URL's)

(schmonz)