Add data integrity check for bad emails and malformed repositories (#…

…1379)

1. Make sure that repository isn't accidentally annotated as
`bioregistry` or given not as a URL. Fixes 3 issues here
2. Make sure that emails curated as contacts aren't used that look like
group emails (e.g. contains `help@`, `discuss@`)

src/bioregistry/data/bioregistry.json

@@ -38378,7 +38378,6 @@
     "homepage": "https://www.fao.org/fishery/en/collection/asfis/en",
     "name": "Aquatic Sciences and Fisheries Information System",
     "pattern": "^\\d+$",
-    "repository": "bioregistry",
     "reviewer": {
       "email": "[email protected]",
       "github": "cthoyt",
@@ -97696,7 +97695,7 @@
     "name": "The data cube vocabulary",
     "pattern": "^\\w+$",
     "preferred_prefix": "qb",
-    "repository": "UKGovLD/publishing-statistical-data",
+    "repository": "https://github.com/UKGovLD/publishing-statistical-data",
     "uri_format": "http://purl.org/linked-data/cube#$1",
     "zazuko": {
       "prefix": "qb",
@@ -109711,7 +109710,6 @@
     "homepage": "https://www.w3.org/TR/vocab-ssn/",
     "license": "http://www.w3.org/Consortium/Legal/2015/copyright-software-and-document",
     "name": "System capabilities, operating ranges, and survival ranges ontology",
-    "repository": "bioregistry",
     "reviewer": {
       "email": "[email protected]",
       "github": "cthoyt",

tests/test_data.py

@@ -29,6 +29,14 @@
 
 logger = logging.getLogger(__name__)
 
+disallowed_email_parts = {
+    "contact@",
+    "help@",
+    "helpdesk@",
+    "discuss@",
+    "support@",
+}
+
 
 class TestRegistry(unittest.TestCase):
     """Tests for the registry."""
@@ -770,6 +778,14 @@ def assert_contact_metadata(self, author: Attributable):
             self.assertNotIn(" ", author.orcid)
         if author.email:
             self.assertRegex(author.email, EMAIL_RE)
+            self.assertFalse(
+                any(
+                    disallowed_email_part in author.email
+                    for disallowed_email_part in disallowed_email_parts
+                ),
+                msg=f"Bioregistry policy states that an email must correspond to a single person. "
+                f"The email provided appears to be for a group/mailing list: {author.email}",
+            )
 
     def test_contributors(self):
         """Check contributors have minimal metadata."""
@@ -1038,3 +1054,19 @@ def test_resolvable_annotation(self):
                     resource.comment,
                     msg="Any resource with a non-resolvable URI format needs a comment as to why",
                 )
+
+    def test_repository(self) -> None:
+        """Test the repository annotation."""
+        for prefix, resource in self.registry.items():
+            if resource.repository is None:
+                continue
+            with self.subTest(prefix=prefix):
+                self.assertNotEqual(
+                    "bioregistry",
+                    resource.repository,
+                    msg="repository accidentally kept flag from GitHub",
+                )
+                self.assertTrue(
+                    resource.repository.startswith("http"),
+                    msg=f"repository is not a valid URL: {resource.repository}",
+                )