Make ISEMAIL and ISURL more flexible for longer TLD (#834)

Allow TLD of length up to 24 in ISEMAIL

app/common/gutil.ts

@@ -23,7 +23,7 @@ export const UP_TRIANGLE = '\u25B2';
 export const DOWN_TRIANGLE = '\u25BC';
 
 const EMAIL_RE = new RegExp("^\\w[\\w%+/='-]*(\\.[\\w%+/='-]+)*@([A-Za-z0-9]([A-Za-z0-9-]*[A-Za-z" +
-  "0-9])?\\.)+[A-Za-z]{2,6}$", "u");
+  "0-9])?\\.)+[A-Za-z]{2,24}$", "u");
 
 // Returns whether str starts with prefix. (Note that this implementation avoids creating a new
 // string, and only checks a single location.)

sandbox/grist/functions/info.py

@@ -257,7 +257,8 @@ _email_regexp = re.compile(
   ([A-Za-z0-9]                    # Each part of hostname must start with alphanumeric
     ([A-Za-z0-9-]*[A-Za-z0-9])?\. # May have dashes inside, but end in alphanumeric
   )+
-  [A-Za-z]{2,6}$                  # Restrict top-level domain to length {2,6}. Google seems
+  [A-Za-z]{2,24}$                 # Restrict top-level domain to length {2,24} (theoretically,
+                                  # the max length is 63 bytes as per RFC 1034). Google seems
                                   # to use a whitelist for TLDs longer than 2 characters.
   """, re.UNICODE | re.VERBOSE)
 
@@ -289,7 +290,8 @@ def ISEMAIL(value):
   >>> ISEMAIL("john@aol...com")
   False
 
-  More tests:
+  More tests:                                             Google Sheets   Grist
+                                                          -------------   -----
   >>> ISEMAIL("Abc@example.com")                              # True,     True
   True
   >>> ISEMAIL("Abc.123@example.com")                          # True,     True
@@ -314,6 +316,10 @@ def ISEMAIL(value):
   True
   >>> ISEMAIL("Bob_O'Reilly+tag@example.com")                 # False,    True
   True
+  >>> ISEMAIL("marie@isola.corsica")                          # False,    True
+  True
+  >>> ISEMAIL("fabio@disapproved.solutions")                  # False,    True
+  True
   >>> ISEMAIL(u"фыва@mail.ru")                                # False,    True
   True
   >>> ISEMAIL("my@baddash.-.com")                             # True,     False
@@ -324,8 +330,6 @@ def ISEMAIL(value):
   False
   >>> ISEMAIL("john@-.com")                                   # True,     False
   False
-  >>> ISEMAIL("fabio@disapproved.solutions")                  # False,    False
-  False
   >>> ISEMAIL("!def!xyz%abc@example.com")                     # False,    False
   False
   >>> ISEMAIL("!#$%&'*+-/=?^_`.{|}~@example.com")             # False,    False
@@ -391,7 +395,8 @@ _url_regexp = re.compile(
   ([A-Za-z0-9]                    # Each part of hostname must start with alphanumeric
     ([A-Za-z0-9-]*[A-Za-z0-9])?\. # May have dashes inside, but end in alphanumeric
   )+
-  [A-Za-z]{2,6}                   # Restrict top-level domain to length {2,6}. Google seems
+  [A-Za-z]{2,24}                  # Restrict top-level domain to length {2,24} (theoretically,
+                                  # the max length is 63 bytes as per RFC 1034). Google seems
                                   # to use a whitelist for TLDs longer than 2 characters.
   ([/?][-\w!#$%&'()*+,./:;=?@~]*)?$ # Notably, this excludes <, >, and ".
   """, re.VERBOSE)
@@ -437,6 +442,8 @@ def ISURL(value):
   True
   >>> ISURL("http://foo.com/!#$%25&'()*+,-./=?@_~")
   True
+  >>> ISURL("http://collectivite.isla.corsica")
+  True
   >>> ISURL("http://../")
   False
   >>> ISURL("http://??/")

test/common/gutil.js

@@ -247,6 +247,7 @@ describe('gutil', function() {
       assert.isTrue(gutil.isEmail('email@subdomain.do-main.com'));
       assert.isTrue(gutil.isEmail('firstname+lastname@domain.com'));
       assert.isTrue(gutil.isEmail('email@domain.co.jp'));
+      assert.isTrue(gutil.isEmail('marie@isola.corsica'));
 
       assert.isFalse(gutil.isEmail('plainaddress'));
       assert.isFalse(gutil.isEmail('@domain.com'));