Merge ~cjwatson/launchpad:loosen-valid-email into launchpad:master

[Colin Watson]

Loosen valid_email to accept digits and hyphens in TLDs

This normally comes up in practice when processing package uploads from people who've forgotten to set their email address properly so dch(1) generated one from their hostname; these are syntactically well-formed despite being incorrect, but dak accepts them and we don't have a principled reason to reject the upload on that basis.

Digits and hyphens also appear in the IDNA encoding of internationalised domain names. This commit doesn't constitute full support for IDNs, but it makes it possible to work around some issues.

[Thiago F. Pappacena (pappacena)]

LGTM

review: Approve

diff --git a/lib/lp/app/validators/email.py b/lib/lp/app/validators/email.py
index d54e084..d3789bc 100644
--- a/lib/lp/app/validators/email.py
+++ b/lib/lp/app/validators/email.py
@@ -52,8 +52,13 @@ def valid_email(emailaddr):
     False
     >>> valid_email('[e-mail]')
     False
+
+    The IDNA encoding of internationalised domain names is also accepted.
+
+    >>> valid_email('[e-mail]--deba0ad')
+    True
     """
-    email_re = r"^[_\.0-9a-zA-Z-+=]+@(([0-9a-zA-Z-]{1,}\.)*)[a-zA-Z]{2,}$"
+    email_re = r"^[_\.0-9a-zA-Z-+=]+@(([0-9a-zA-Z-]{1,}\.)*)[0-9a-zA-Z-]{2,}$"
     email_match = re.match(email_re, emailaddr)
     if not email_match:
         return False