GitCommitBear: Block invalid email addresses

bear-requirements.txt

@@ -8,6 +8,7 @@ cmakelint~=1.3
 cppclean~=0.12.0
 cpplint~=1.3
 dennis~=0.8
+dnspython~=1.15.0
 docutils-ast-writer~=0.1.2
 eradicate~=0.1.6
 guess-language-spirit~=0.5.2
@@ -24,6 +25,7 @@ proselint~=0.7.0
 pycodestyle~=2.2
 pydocstyle~=2.0
 pyflakes~=1.5.0
+pyisemail~=1.3.1
 pylint~=1.6
 pyroma~=2.2.0
 pyyaml~=3.12

bears/vcs/git/GitCommitBear.py

@@ -3,6 +3,7 @@
 import shutil
 import os
 from urllib.parse import urlparse
+from pyisemail import is_email
 
 from coalib.bears.GlobalBear import GlobalBear
 from dependency_management.requirements.PipRequirement import PipRequirement
@@ -15,7 +16,8 @@
 
 class GitCommitBear(GlobalBear):
     LANGUAGES = {'Git'}
-    REQUIREMENTS = {PipRequirement('nltk', '3.2')}
+    REQUIREMENTS = {PipRequirement('nltk', '3.2'),
+                    PipRequirement('pyisemail', '1.3.1')}
     AUTHORS = {'The coala developers'}
     AUTHORS_EMAILS = {'[email protected]'}
     LICENSE = 'AGPL-3.0'
@@ -223,7 +225,9 @@ def check_body(self, body,
                    body_line_length: int=72,
                    force_body: bool=False,
                    ignore_length_regex: typed_list(str)=(),
-                   body_regex: str=None):
+                   body_regex: str=None,
+                   valid_email_check: bool=True,
+                   email_dns_check: bool=False):
         """
         Checks the given commit body.
 
@@ -236,6 +240,10 @@ def check_body(self, body,
                                     expressions in this list will be ignored.
         :param body_regex:          If provided, checks the presence of regex
                                     in the commit body.
+        :param valid_email_check:   If True, it checks if emails are in valid
+                                    format or not
+        :param email_dns_check:     If True, it checks if domains in the emails
+                                    are valid or not
         """
         if len(body) == 0:
             if force_body:
@@ -260,6 +268,25 @@ def check_body(self, body,
                                'Commit body lines should not exceed {} '
                                'characters.'.format(body_line_length))
 
+        if valid_email_check:
+            result_message = 'Body contains these invalid emails:\n'
+            invalid_emails = False
+            for line in body:
+                for email in re.findall(r'\S*@\S*', line):
+                    email_check = re.match(r"(^[a-zA-Z0-9_.+-]"
+                                           +r"+@[a-zA-Z0-9-]+\."
+                                           +r"[a-zA-Z0-9-.]+$)", email)
+                    # if eamil do not match regex there is no need to
+                    # look up dns
+                    if email_dns_check and email_check:
+                        email_check = not is_email(email, diagnose=True,
+                                               check_dns=True).ERROR_CODES
+                    if not email_check:
+                        invalid_emails = True
+                        result_message += ' ' + email + '\n'
+            if invalid_emails:
+                yield Result(self, result_message)
+
     def check_issue_reference(self, body,
                               body_close_issue: bool=False,
                               body_close_issue_full_url: bool=False,

tests/vcs/git/GitCommitBearTest.py

@@ -288,6 +288,24 @@ def test_body_checks(self):
                              body_regex=r'TICKER\s*CLOSE\s+[1-9][0-9]*'), [])
         self.assert_no_msgs()
 
+        # Testing if check recognises invalid emails
+        self.git_commit('Shortlog\n\n'
+                        'invalid#!format!*[email protected]\n'
+                        'another.*invalid@format\n'
+                        '[email protected]\n'
+                        '[email protected]\n'
+                        'Fix 2017')
+        message = 'Body contains these invalid emails:\n'
+        message += ' invalid#!format!*[email protected]\n'
+        message += ' another.*invalid@format\n'
+        # Test with email_dns_check set to False
+        self.assertEqual(self.run_uut(), [message])
+        # Test with email_dns_check set to True
+        message += " [email protected]\n"
+        self.assertEqual(self.run_uut(email_dns_check=True), [message])
+        # Test with valid_email_check set to False
+        self.assertEqual(self.run_uut(valid_email_check=False), [])
+
     def test_check_issue_reference(self):
         # Commit with no remotes configured
         self.git_commit('Shortlog\n\n'