Rework regex in xx-Remove-emails

This changes the regex to be stricter about what it matches in terms of
email addresses.  That should cut down on the amount of false positives
from censored swearing (e.g., "f!@#$%%").  Unscientifically, it's also
faster, at least on significant sized archives.

Author: hlieberman

xx-Remove-emails-from-Open-Doors-Tables.py

@@ -6,8 +6,11 @@
 from prompt_toolkit.formatted_text import FormattedText
 from prompt_toolkit.shortcuts import clear
 
+# This regex is pulled from the HTML5 spec. Though it is technically not
+# compliant with RFC 5322 ("a willful violation"), it's good enough for our
+# purposes.
 email_regex = re.compile(
-    r"([-!#-'*+/-9=?A-Z^-~]+(\.[-!#-'*+/-9=?A-Z^-~]+)*|\"([]!#-[^-~ \t]|(\\[\t -~]))+\")@([-!#-'*+/-9=?A-Z^-~]+(\.[-!#-'*+/-9=?A-Z^-~]+)*|\[[\t -Z^-~]*])"
+    r"([a-zA-Z0-9.!#$%&'*+\/=?^_`{|}~-]+)@([a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?(?:\.[a-zA-Z0-9](?:[a-zA-Z0-9-]{0,61}[a-zA-Z0-9])?)*)"
 )
 
 
@@ -44,7 +47,7 @@ def is_mailto(match) -> bool:
 def ask_user_for_action(match) -> str:
     start, end = match.span()
     raw_email = match.string[start:end]
-    domain = match.group(5)
+    domain = match.group(2)
     clear()
     print_context(match, 50)
     while True:
@@ -85,7 +88,7 @@ def return_from_list(match) -> str:
         return raw_email
     elif address_entry is not None:
         return address_entry
-    domain = match.group(5)
+    domain = match.group(2)
     domain_entry = domains.get(domain)
     if domain_entry is True:
         return raw_email