Merge branch 'email-validation-glab' of 'https://github.com/sduenas/s…

…ortinghat'

Merges #111
Closes #111
Fixes #107

misc/grimoirelab2sh

@@ -42,8 +42,9 @@ def main():
     args = parse_args()
 
     try:
+        email_validation = not args.no_email_validation
         parser = parse_grimoirelab_file(args.identities, args.organizations,
-                                        args.source)
+                                        args.source, email_validation)
     except (IOError, UnicodeDecodeError, InvalidFormatError) as e:
         raise RuntimeError(str(e))
 
@@ -71,6 +72,9 @@ def parse_args():
     parser.add_argument('-o', '--outfile', nargs='?', type=argparse.FileType('w'),
                         default=sys.stdout,
                         help='Sorting Hat JSON output filename')
+    parser.add_argument('--no-email-validation', dest='no_email_validation',
+                        action='store_true',
+                        help="do not email addresses validation")
 
     args = parser.parse_args()
 
@@ -80,14 +84,16 @@ def parse_args():
     return args
 
 
-def parse_grimoirelab_file(identities, organizations, source):
+def parse_grimoirelab_file(identities, organizations, source, email_validation):
     """Parse GrimoireLab JSON file"""
 
     content_id = read_file(identities) if identities else None
     content_org = read_file(organizations) if organizations else None
 
     try:
-        parser = GrimoireLabParser(content_id, content_org, source=source)
+        parser = GrimoireLabParser(content_id, content_org,
+                                   source=source,
+                                   email_validation=email_validation)
     except ValueError:
         s = "Error: Empty input file(s)\n"
         sys.stdout.write(s)

sortinghat/parsing/grimoirelab.py

@@ -47,26 +47,30 @@ class GrimoireLabParser(object):
     The unique identities are stored in an object named 'uidentities'.
     The keys of this object are the UUID of the unique identities.
     Each unique identity object stores a list of identities and
-    enrollments.
+    enrollments. Email addresses will not be validated when `email_validation`
+    is set to `False`.
 
     Organizations are stored in 'organizations' object. Its keys
     are the name of the organizations and each organization object is
     related to a list of domains.
 
-    :param stream: stream to parse
+    :param identities: stream of identities to parse
+    :param organizations: stream of organizations to parse
+    :param source: source of the data
+    :param email_validation: validate email addresses; set to True by default
 
     :raises InvalidFormatError: raised when the format of the stream is
         not valid.
     """
-
     EMAIL_ADDRESS_REGEX = r"^(?P<email>[^\s@]+@[^\s@.]+\.[^\s@]+)$"
     GRIMOIRELAB_INVALID_FORMAT = "invalid GrimoireLab yaml format. %(error)s"
 
     def __init__(self, identities=None, organizations=None,
-                 source='grimoirelab'):
+                 source='grimoirelab', email_validation=True):
         self._identities = {}
         self._organizations = {}
         self.source = source
+        self.email_validation = email_validation
 
         if not (identities or organizations):
             raise ValueError('Null identities and organization streams')
@@ -155,7 +159,7 @@ def __create_sh_identities(name, emails, yaml_entry):
 
                 emails = yid.get('email', None)
 
-                if emails:
+                if emails and self.email_validation:
                     self.__validate_email(emails[0])
 
                 enrollments = yid.get('enrollments', None)

tests/test_parser_grimoirelab.py

@@ -283,6 +283,37 @@ def test_enrollments_parser(self):
         # Unknown organization is ignored during the parsing process
         self.assertEqual(len(uid.enrollments), 0)
 
+    def test_email_validation(self):
+        """Check wheter it raises an error on invalid email addresses"""
+
+        stream_ids = self.read_file('data/grimoirelab_invalid_email.yml')
+
+        with self.assertRaisesRegexp(InvalidFormatError, '^.+Invalid email address: lcanas__at__bitergia.com$'):
+            GrimoireLabParser(stream_ids, email_validation=True)
+
+    def test_supress_email_validation(self):
+        """Check wheter it ignores invalid email addresses"""
+
+        stream_ids = self.read_file('data/grimoirelab_invalid_email.yml')
+        parser = GrimoireLabParser(stream_ids, email_validation=False)
+
+        uids = parser.identities
+        self.assertEqual(len(uids), 3)
+
+        uid = uids[1]
+        self.assertIsInstance(uid, UniqueIdentity)
+        self.assertFalse(uid.profile.is_bot)
+        self.assertEqual(uid.profile.name, 'Luis Cañas-Díaz')
+
+        # This identity has an invalid email address
+        id1 = uid.identities[1]
+        self.assertIsInstance(id1, Identity)
+        self.assertEqual(id1.name, None)
+        self.assertEqual(id1.email, 'lcanas__at__bitergia.com')
+        self.assertEqual(id1.username, None)
+        self.assertEqual(id1.source, 'grimoirelab')
+        self.assertEqual(id1.uuid, None)
+
     def test_not_valid_organizations_stream(self):
         """Check whether it parses invalid organizations files"""