HTMLParser fails to handle some characters in the starttag, e.g. <a$b> should see 'a$b' as name but currently stops at $.  The attached patch fixes the issue.

Attached a patch for 2.7.  The patch removes a "public" name/regex (tagfind_tolerant).  The name is not documented and it's supposed to be private like all the other top-level names on HTMLParser, and even creating an alias with tagfind_tolerant = tagfind won't work because the groups in the regex changed.

In the maintenance releases you should leave tagfind_tolerant defined with its old value, with a comment that it is internal, no longer used, and has been removed in 3.4.

For 2.7 that sounds like a reasonable option, for 3.3/3.4 however I'm keeping the name but I change the regex groups, so it might break if someone is using it with groups.  In theory I could add a third name and leave that unchanged, but I'm not sure it's worth it.

Then for 3.3 you are bug fixing the regex, so anyone using it ought to want the change, right? :)

If the same is true for 2.7, then creating an alias would probably be fine.  I haven't looked at the details, so I'll leave it to your judgment.  I just don't want the name going away in 2.7, it might be a gratuitous breaking of someone's code.

New changeset 695f988824bb by Ezio Melotti in branch '2.7':
#19480: HTMLParser now accepts all valid start-tag names as defined by the HTML5 standard.
http://hg.python.org/cpython/rev/695f988824bb

New changeset 9b9d188ed549 by Ezio Melotti in branch '3.3':
#19480: HTMLParser now accepts all valid start-tag names as defined by the HTML5 standard.
http://hg.python.org/cpython/rev/9b9d188ed549

New changeset 7d8a37020db9 by Ezio Melotti in branch 'default':
#19480: merge with 3.3.
http://hg.python.org/cpython/rev/7d8a37020db9

Fixed, thanks for the feedback!