Clean and standardise instagram entries

- Audit all person identifiers and their validation
- Check the domain and the username for valid entries

ynr/apps/people/README.md

@@ -10,3 +10,25 @@ standing in.
 Thing like their political party should be stored on the _candidacy_.
 This is currently represented as a `Membership` in the `popolo` app.
 
+# Person Identifiers
+
+`PersonIdentifiers` are used to store contact information and online presence for a candidate. Users can add/edit these identifiers in the person update form. The following fields are available in the `PersonIdentifier` model with the corresponding form labels and accepted formats:
+```
+email = "Email": Accepts a valid email address; raises an exception if the email is submitted with an invalid format. The email address must be unique.
+facebook_page_url = "Facebook Page": Accepts a valid URL but does not currently validate the domain.
+facebook_personal_url = "Facebook Personal": Same as above. 
+homepage_url = "Homepage": Accepts a valid URL
+blog_url = "Blog": Accepts a valid URL
+linkedin_url = "Linkedin": Accepts a valid URL; does not currently validate the domain or the format of the username.
+party_ppc_page_url = "Party Candidate Page": Accepts a valid URL
+twitter_username = "Twitter": Accepts a Twitter username and returns the full Twitter URL.No longer validates the username actually exists.
+mastodon_username = "Mastodon": Accepts and returns a valid Mastodon URL and validates the domain and username.
+wikipedia_url = "Wikipedia": Accepts a valid URL; does not currently validate the domain.
+wikidata_id = "Wikidata": Accepts a valid Wikidata ID
+youtube_profile = "YouTube Profile": Accepts a valid URL; does not currently validate the domain.
+instagram_url = "Instagram Profile": Accepts and returns a valid Instagram URL and validates the domain and format of the username.
+blue_sky_url = "Bluesky URL": Accepts and returns a valid URL
+threads_url = "Threads URL": Accepts and returns a valid URL
+tiktok_url = "TikTok URL": Accepts and returns a valid URL
+other_url = "Other URL": Accepts and returns a valid URL
+```

ynr/apps/people/forms/forms.py

@@ -19,6 +19,7 @@
     StrippedCharField,
 )
 from people.helpers import (
+    clean_instagram_url,
     clean_mastodon_username,
     clean_twitter_username,
     clean_wikidata_id,
@@ -116,6 +117,17 @@ def clean(self):
                     self.add_error(None, e)
         return self.cleaned_data
 
+    def clean_instagram_url(self, username):
+        if self.instance.value != username:
+            self.instance.internal_identifier = None
+        if self.instance.internal_identifier:
+            return username
+        try:
+            return clean_instagram_url(username)
+        except ValueError as e:
+            raise ValidationError(e)
+        return username
+
     def clean_twitter_username(self, username):
         if self.instance.value != username:
             self.instance.internal_identifier = None

ynr/apps/people/helpers.py

@@ -120,6 +120,30 @@ def clean_twitter_username(username):
     return username
 
 
+def clean_instagram_url(url):
+    parsed_username = urlparse(url)
+    if parsed_username.netloc not in [
+        "instagram.com",
+        "www.instagram.com",
+        "instagr.am",
+        "www.instagr.am",
+    ]:
+        raise ValueError(
+            "The Instagram URL must be from a valid Instagram domain."
+        )
+    username = parsed_username.path.strip("/")
+    if not re.search(
+        r"^(?!.*[_.]{2})(?!.*[_.]$)[a-zA-Z0-9._]{1,30}$",
+        username,
+    ):
+        raise ValueError(
+            "This is not a valid Instagram username. Please try again."
+        )
+    return "https://{domain}/{username}".format(
+        domain=parsed_username.netloc, username=username
+    )
+
+
 def clean_wikidata_id(identifier):
     identifier = identifier.strip().lower()
     m = re.search(r"^.*wikidata.org/(wiki|entity)/(\w+)", identifier)

ynr/apps/people/models.py

@@ -169,6 +169,8 @@ def get_value_html(self):
 
         if self.value_type == "twitter_username":
             url = format_html("https://twitter.com/{}", self.value)
+        if self.value_type == "instagram_url":
+            url = self.value
 
         if self.value.startswith("http"):
             url = format_html("{}", self.value)

ynr/apps/people/tests/test_person_form_identifier_crud.py

@@ -150,7 +150,6 @@ def _submit_values(self, value, value_type="twitter_username"):
         form["source"] = "They changed their username"
         form["tmp_person_identifiers-0-value_type"] = value_type
         form["tmp_person_identifiers-0-value"] = value
-        form.submit()
         return form.submit()
 
     def _submit_mastodon_values(self, value, value_type="mastodon_username"):
@@ -163,7 +162,30 @@ def _submit_mastodon_values(self, value, value_type="mastodon_username"):
         form["source"] = "They changed their username"
         form["tmp_person_identifiers-0-value_type"] = value_type
         form["tmp_person_identifiers-0-value"] = value
-        form.submit()
+        return form.submit()
+
+    def _submit_mastodon_values(self, value, value_type="mastodon_username"):
+        resp = self.app.get(
+            reverse("person-update", kwargs={"person_id": self.person.pk}),
+            user=self.user,
+        )
+
+        form = resp.forms[1]
+        form["source"] = "They changed their username"
+        form["tmp_person_identifiers-0-value_type"] = value_type
+        form["tmp_person_identifiers-0-value"] = value
+        return form.submit()
+
+    def _submit_instagram_values(self, value, value_type="instagram_url"):
+        resp = self.app.get(
+            reverse("person-update", kwargs={"person_id": self.person.pk}),
+            user=self.user,
+        )
+
+        form = resp.forms[1]
+        form["source"] = "They changed their username"
+        form["tmp_person_identifiers-0-value_type"] = value_type
+        form["tmp_person_identifiers-0-value"] = value
         return form.submit()
 
     def test_twitter_bad_url(self):
@@ -185,6 +207,41 @@ def test_twitter_full_url(self):
             PersonIdentifier.objects.get().value, "madeuptwitteraccount"
         )
 
+    def test_clean_instagram_url(self):
+        resp = self._submit_instagram_values(
+            "https://www.instagr.am/disco_dude"
+        )
+        self.assertEqual(resp.status_code, 302)
+        instagram_url_qs = PersonIdentifier.objects.filter(
+            value_type="instagram_url"
+        )
+        self.assertEqual(instagram_url_qs.count(), 1)
+        self.assertEqual(
+            instagram_url_qs[0].value,
+            "https://www.instagr.am/disco_dude",
+        )
+
+    def test_bad_instagram_domain(self):
+        resp = self._submit_instagram_values("www.instagl.am/blah")
+        form = resp.context["identifiers_formset"]
+        self.assertFalse(form.is_valid())
+        self.assertEqual(
+            form[0].non_field_errors(),
+            ["The Instagram URL must be from a valid Instagram domain."],
+        )
+
+    def test_bad_instagram_username(self):
+        resp = self._submit_instagram_values(
+            "https://www.instagr.am/________blah"
+        )
+        self.assertEqual(resp.status_code, 200)
+        form = resp.context["identifiers_formset"]
+        self.assertFalse(form.is_valid())
+        self.assertEqual(
+            form[0].non_field_errors(),
+            ["This is not a valid Instagram username. Please try again."],
+        )
+
     def test_mastodon_bad_url(self):
         # submit a username missing the `@` symbol
         resp = self._submit_mastodon_values("https://mastodon.social/joe")

ynr/apps/people/tests/test_person_identifiers.py

@@ -33,6 +33,23 @@ def test_get_value_html_twitter(self):
         # Test the value type HTML
         self.assertEqual(pi.get_value_type_html, "Twitter")
 
+    def test_get_value_html_instagram(self):
+        pi = PersonIdentifier.objects.create(
+            person=self.person,
+            value="https://www.instagram.com/democlub",
+            value_type="instagram_url",
+            internal_identifier="2324",
+        )
+
+        # Test the value HTML
+        self.assertEqual(
+            pi.get_value_html,
+            """<a href="https://www.instagram.com/democlub" rel="nofollow">https://www.instagram.com/democlub</a>""",
+        )
+
+        # Test the value type HTML
+        self.assertEqual(pi.get_value_type_html, "Instagram")
+
     def test_get_value_html_mastodon(self):
         pi = PersonIdentifier.objects.create(
             person=self.person,

ynr/apps/ynr_refactoring/settings.py

@@ -19,7 +19,7 @@ class PersonIdentifierFields(Enum):
     wikipedia_url = "Wikipedia"
     wikidata_id = "Wikidata"
     youtube_profile = "YouTube Profile"
-    instagram_url = "Instagram Profile"
+    instagram_url = "Instagram URL"
     blue_sky_url = "Bluesky URL"
     threads_url = "Threads URL"
     tiktok_url = "TikTok URL"