Merge pull request #2756 from alphagov/clean-up-domains

Clean up domains file

app/utils.py

@@ -563,7 +563,10 @@ class AgreementInfo:
 
     def _get_info(self):
 
-        details = self.domains.get(self._match) or {}
+        details = self.domains.get(self._match, {})
+
+        if details is None:
+            raise TypeError('Domain must have details ({})'.format(self._domain))
 
         if isinstance(details, str):
             self.is_canonical = False

domains.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+import os
+import yaml
+from itertools import chain
+from operator import itemgetter
+from sys import argv
+from app.utils import AgreementInfo
+
+_dir_path = os.path.dirname(os.path.realpath(__file__))
+
+
+if len(argv) < 2:
+    raise TypeError('Must specify `orgs` or `domains` as the first argument to this script')
+
+
+with open('{}/app/domains.yml'.format(_dir_path)) as source:
+
+    data = yaml.load(source)
+
+    for domain, details in data.items():
+        if isinstance(details, dict):
+            # We’re looking at the canonical domain
+            data[domain]['domains'] = [domain]
+
+    for domain, details in data.items():
+        if isinstance(details, str):
+            # This is an alias, let’s add it to the canonical domain
+            data[AgreementInfo(domain).canonical_domain]['domains'].append(domain)
+
+    out_data = [
+        details for domain, details in data.items()
+        if isinstance(details, dict)
+    ]
+
+    if argv[1] == 'orgs':
+        print(yaml.dump(out_data))  # noqa
+    elif argv[1] == 'domains':
+        print(  # noqa
+            sorted(
+                set(
+                    chain.from_iterable(
+                        map(
+                            itemgetter('domains'), out_data
+                        )
+                    )
+                )
+            )
+        )
+    else:
+        raise TypeError('Must specify `orgs` or `domains` as the first argument to this script')

tests/app/test_utils.py

@@ -1,4 +1,4 @@
-from collections import OrderedDict
+from collections import Counter, OrderedDict
 from csv import DictReader
 from io import StringIO
 from pathlib import Path
@@ -391,6 +391,11 @@ def test_get_domain_info_for_branding_request():
     )
 
 
+def test_domains_are_lowercased():
+    for domain in AgreementInfo.domains.keys():
+        assert domain == domain.lower()
+
+
 def test_validate_government_domain_data():
 
     for domain in AgreementInfo.domains.keys():
@@ -403,17 +408,25 @@ def test_validate_government_domain_data():
             True, False, None
         }
 
-        assert (
-            agreement_info.owner is None
-        ) or (
-            isinstance(agreement_info.owner, str)
-        )
+        assert isinstance(agreement_info.owner, str) and agreement_info.owner.strip()
 
         assert agreement_info.agreement_signed in {
             True, False, None
         }
 
 
+def test_domain_data_is_canonicalized():
+    for owner, count in Counter(
+        AgreementInfo(domain).owner
+        for domain in AgreementInfo.domains.keys()
+        if AgreementInfo(domain).is_canonical
+    ).most_common():
+        if count > 1:
+            raise ValueError(
+                '{} entries in domains.yml for {}'.format(count, owner)
+            )
+
+
 def test_validate_email_domain_data():
 
     for domain in GovernmentEmailDomain.domains.keys():