GitCommitBear.py: Require URL in commit body

This ensures that the git commit contains a URL
that relates to an issue.

Fixes coala#1112

.coafile

@@ -29,6 +29,7 @@ bears = SpaceConsistencyBear
 
 [commit]
 bears = GitCommitBear
+host_site = github
 shortlog_trailing_period = False
 shortlog_regex = ([^:]*|[^:]+: [A-Z0-9*].*)
 

bears/vcs/git/GitCommitBear.py

@@ -3,6 +3,8 @@
 import shutil
 import os
 
+from urllib.parse import urlparse
+
 from coalib.bears.GlobalBear import GlobalBear
 from coalib.bears.requirements.PipRequirement import PipRequirement
 from coala_utils.ContextManagers import change_directory
@@ -21,6 +23,9 @@ class GitCommitBear(GlobalBear):
     LICENSE = 'AGPL-3.0'
     ASCIINEMA_URL = 'https://asciinema.org/a/e146c9739ojhr8396wedsvf0d'
     CAN_DETECT = {'Formatting'}
+    GITHUB_KEYS = {'close', 'closes', 'closed', 'resolve',
+                   'resolves', 'resolved', 'fix', 'fixes', 'fixed'}
+    GITLAB_KEYS = GITHUB_KEYS.union({'closing', 'resolving', 'fixing'})
 
     @classmethod
     def check_prerequisites(cls):
@@ -175,6 +180,7 @@ def check_imperative(self, paragraph):
     def check_body(self, body,
                    body_line_length: int=72,
                    force_body: bool=False,
+                   host_site: str=None,
                    ignore_length_regex: typed_list(str)=()):
         """
         Checks the given commit body.
@@ -183,6 +189,10 @@ def check_body(self, body,
         :param body_line_length:    The maximum line-length of the body. The
                                     newline character at each line end does not
                                     count to the length.
+        :param host_site:           The website used for hosting the git
+                                    repository. Choosing this value checks the
+                                    presence of full URLs in issue related
+                                    commits.
         :param force_body:          Whether a body shall exist or not.
         :param ignore_length_regex: Lines matching each of the regular
                                     expressions in this list will be ignored.
@@ -197,6 +207,36 @@ def check_body(self, body,
                                'HEAD commit. Please add one.')
             return
 
+        if host_site:
+            body_string = '\n'.join(body).lower()
+            keys = set()
+
+            if 'github' in host_site.lower():
+                keys = self.GITHUB_KEYS
+            elif 'gitlab' in host_site.lower():
+                keys = self.GITLAB_KEYS
+
+            if any(key in body_string for key in keys):
+                urls = re.findall(r'(https?://\S+)', body_string)
+                good_url = False
+                for url in urls:
+                    (scheme, netloc, path) = urlparse(url)[:3]
+                    try:
+                        if all([scheme, netloc, path]):
+                            if host_site in netloc and '/issues/' in path:
+                                int(path.split('/')[-1])
+                                good_url = True
+                    except ValueError:
+                        yield Result(self, 'Invalid issue number present '
+                                           'in the body at HEAD commit.')
+                        return
+
+                if not good_url:
+                    yield Result(self, 'No issue related URL found in '
+                                       'the body at HEAD commit. '
+                                       'Please add one.')
+                    return
+
         ignore_regexes = [re.compile(regex) for regex in ignore_length_regex]
         if any((len(line) > body_line_length and
                 not any(regex.search(line) for regex in ignore_regexes))

tests/vcs/git/GitCommitBearTest.py

@@ -267,6 +267,76 @@ def test_body_checks(self):
                          [])
         self.assertTrue(self.msg_queue.empty())
 
+        # Commit has neither keywords nor issues
+        self.git_commit('Shortlog\n\n'
+                        'This line is ok.\n'
+                        'This line is by far too long (in this case).\n'
+                        'This one too, blablablablablablablablabla.')
+        self.assertEqual(self.run_uut(host_site='github',), [])
+
+        # GitHub chosen and has an issue
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Fix https://github.com/user/repo/issues/1234')
+        self.assertEqual(self.run_uut(host_site='github',), [])
+
+        # GitLab chosen and has an issue
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Closing https://gitlab.com/user/repo/issues/100')
+        self.assertEqual(self.run_uut(host_site='gitlab',), [])
+
+        # Invalid issue number in URL
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Closing https://gitlab.com/user/repo/issues/1af121')
+        self.assertEqual(self.run_uut(host_site='gitlab',),
+                         ['Invalid issue number present in the body '
+                          'at HEAD commit.'])
+        self.assert_no_msgs()
+
+        # GitHub chosen and has keywords but no url
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Fix #123')
+        self.assertEqual(self.run_uut(host_site='github',),
+                         ['No issue related URL found in '
+                          'the body at HEAD commit. '
+                          'Please add one.'])
+        self.assert_no_msgs()
+
+        # Git Host with no support
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Fix #123')
+        self.assertEqual(self.run_uut(host_site='bitbucket',),
+                         [])
+
+        # Has an invalid URL
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Fix http://google.com')
+        self.assertEqual(self.run_uut(host_site='github',),
+                         ['No issue related URL found in '
+                          'the body at HEAD commit. '
+                          'Please add one.'])
+
+        # Has a URL but not issue related
+        self.git_commit('Shortlog\n\n'
+                        'First line, blablablablablabla.\n'
+                        'Another line, blablablablablabla.\n'
+                        'Fix https://gitlab.com/user/repo')
+        self.assertEqual(self.run_uut(host_site='github',),
+                         ['No issue related URL found in '
+                          'the body at HEAD commit. '
+                          'Please add one.'])
+
     def test_different_path(self):
         no_git_dir = mkdtemp()
         self.git_commit('Add a very long shortlog for a bad project history.')