Youtube link regex fails for youtu.be links.

[edelooff]

Given the example link "http://youtu.be/watch?v=COiIC3A0ROM"
and the _VALID_URL regex present in the Jan 7 checkout, the groups() results are as follows:
('http://youtu.be/', 'watch')

The fix for the regex is a parenthesis that is placed too late. There should be an extra one after the ".com/" part, closing the optional matching group for youtu.be | \w+.youtube.com.

The regular expression that "works for me" is as follows:
_VALID_URL = r'^((?:https?://)?(?:youtu.be/|(?:\w+.)?youtube(?:-nocookie)?.com/)(?:(?:v/)|(?:(?:watch(?:popup)?(?:.php)?)?(?:?|#!?)(?:.+&)?v=)))?([0-9A-Za-z-]+)(?(1).+)?$'

[edelooff]

An annotated version of the above, which is not as verbose as it could be, but attempts some explanation of parts:

_VALID_URL = re.compile(r"""
^( # Start matching an optional URL.
(?:https?://)? # The scheme may be 'http', 'https' or missing.
(?:youtu.be/| # Domain can be youtu.be ...
(?:\w+.)?youtube # ... or any youtube subdomain ...
(?:-nocookie)?.com/) # with or without -nocookie in the domain name.
(?:(?:v/)| # The path might start with 'v', ...
(?:(?:watch(?:_popup)? # ... 'watch', with or without 'popup' ...
(?:.php)?)? # ... and might have a trailing '.php'.
(?:?|#!?) # Info may be in the fragment or in the path ...
(?:.+&)? # ... and may contain leading query arguments.
v=)))? # but it WILL contain a 'v=' for the video ID!
([0-9A-Za-z-]+) # Capture the video_id (group #2).
(?(1).+)?$ # If the video ID was in an URL, capture the rest.
""", re.VERBOSE)


[rg3]

Fixed. Thanks.