Disallow periods at the start and end of new usernames

Work towards resolving an ambiguity when parsing mentions out of text like "I
like this @bob." by disallowing periods at the start and end of usernames.

There are a small number (~2000 or <0.1%) of existing accounts which do contain
periods at the start or end of usernames.

Fixes #9335

docs/_extra/api-reference/hypothesis-v1.yaml

@@ -194,7 +194,7 @@ components:
         The userID should be of the format `acct:<username>@<authority>`
       schema:
         type: string
-        pattern: 'acct:^[A-Za-z0-9._]{3,30}@.*$'
+        pattern: 'acct:^[A-Za-z0.9_][A-Za-z0-9._]{1,28}[A-Za-z0-9_]@.*$'
 
   # -------------------------
   # Reusable responses

docs/_extra/api-reference/hypothesis-v2.yaml

@@ -192,7 +192,7 @@ components:
         The userID should be of the format `acct:<username>@<authority>`
       schema:
         type: string
-        pattern: 'acct:^[A-Za-z0-9._]{3,30}@.*$'
+        pattern: 'acct:^[A-Za-z0.9_][A-Za-z0-9._]{1,28}[A-Za-z0-9_]@.*$'
 
   # -------------------------
   # Reusable responses

h/accounts/schemas.py

@@ -160,7 +160,9 @@ class RegisterSchema(CSRFSchema):
             validators.Length(min=USERNAME_MIN_LENGTH, max=USERNAME_MAX_LENGTH),
             colander.Regex(
                 USERNAME_PATTERN,
-                msg=_("Must have only letters, numbers, periods, and underscores."),
+                msg=_(
+                    "Must have only letters, numbers, periods and underscores. May not start or end with period."
+                ),
             ),
             unique_username,
             unblacklisted_username,

h/models/user.py

@@ -19,7 +19,11 @@
 
 USERNAME_MIN_LENGTH = 3
 USERNAME_MAX_LENGTH = 30
-USERNAME_PATTERN = "(?i)^[A-Z0-9._]+$"
+
+# nb. This pattern is used in Python code, JSON schemas and HTML forms, so it
+# needs to use portable syntax.
+USERNAME_PATTERN = "^[A-Za-z0-9_][A-Za-z0-9._]+[A-Za-z0-9_]$"
+
 EMAIL_MAX_LENGTH = 100
 DISPLAY_NAME_MAX_LENGTH = 30
 USER_PUBID_LENGTH = 12

h/schemas/api/user.py

@@ -3,6 +3,7 @@
     EMAIL_MAX_LENGTH,
     USERNAME_MAX_LENGTH,
     USERNAME_MIN_LENGTH,
+    USERNAME_PATTERN,
 )
 from h.schemas.base import JSONSchema
 
@@ -18,7 +19,7 @@ class CreateUserAPISchema(JSONSchema):
                 "type": "string",
                 "minLength": USERNAME_MIN_LENGTH,
                 "maxLength": USERNAME_MAX_LENGTH,
-                "pattern": "^[A-Za-z0-9._]+$",
+                "pattern": USERNAME_PATTERN,
             },
             "email": {
                 "type": "string",

h/templates/admin/users.html.jinja2

@@ -73,7 +73,7 @@
       <form method="POST" action="{{request.route_path('admin.users_rename')}}" class="form-inline">
         <input type="hidden" name="csrf_token" value="{{ get_csrf_token() }}">
         <input type="hidden" name="userid" value="{{user.userid}}">
-        <input type="text" name="new_username" placeholder="New Username" pattern="^[A-Za-z0-9._]+$">
+        <input type="text" name="new_username" placeholder="New Username" pattern="{{ username_pattern }}">
         <button class="btn" type="submit">Change username</button>
       </form>
 

h/views/admin/users.py

@@ -5,6 +5,7 @@
 from h import models
 from h.accounts.events import ActivationEvent
 from h.i18n import TranslationString as _
+from h.models.user import USERNAME_PATTERN
 from h.security import Permission
 from h.services.user_rename import UserRenameError, UserRenameService
 
@@ -69,6 +70,7 @@ def users_index(request):
         "authority": authority,
         "user": user,
         "user_meta": user_meta,
+        "username_pattern": USERNAME_PATTERN,
         "format_date": format_date,
     }
 

tests/unit/h/accounts/schemas_test.py

@@ -86,8 +86,8 @@ def test_it_is_invalid_when_username_too_short(self, pyramid_request):
         schema = schemas.RegisterSchema().bind(request=pyramid_request)
 
         with pytest.raises(colander.Invalid) as exc:
-            schema.deserialize({"username": "a"})
-        assert exc.value.asdict()["username"] == ("Must be 3 characters or more.")
+            schema.deserialize({"username": "ab"})
+        assert "Must be 3 characters or more." in exc.value.asdict()["username"]
 
     def test_it_is_invalid_when_username_too_long(self, pyramid_request):
         schema = schemas.RegisterSchema().bind(request=pyramid_request)
@@ -102,7 +102,7 @@ def test_it_is_invalid_with_invalid_characters_in_username(self, pyramid_request
         with pytest.raises(colander.Invalid) as exc:
             schema.deserialize({"username": "Fred Flintstone"})
         assert exc.value.asdict()["username"] == (
-            "Must have only letters, numbers, periods, and underscores."
+            "Must have only letters, numbers, periods and underscores. May not start or end with period."
         )
 
     def test_it_is_invalid_with_false_privacy_accepted(self, pyramid_request):

tests/unit/h/schemas/api/user_test.py

@@ -47,8 +47,16 @@ def test_it_raises_when_username_too_long(self, schema, payload):
         with pytest.raises(ValidationError):
             schema.validate(payload)
 
-    def test_it_raises_when_username_format_invalid(self, schema, payload):
-        payload["username"] = "dagr!un"
+    @pytest.mark.parametrize(
+        "username",
+        [
+            "invalid!chars",
+            ".starts_with_period",
+            "ends_with_period.",
+        ],
+    )
+    def test_it_raises_when_username_format_invalid(self, schema, payload, username):
+        payload["username"] = username
 
         with pytest.raises(ValidationError):
             schema.validate(payload)

tests/unit/h/views/admin/users_test.py

@@ -2,6 +2,7 @@
 from unittest import mock
 
 import pytest
+from h_matchers import Any
 from pyramid import httpexceptions
 
 from h.models import Annotation
@@ -35,6 +36,7 @@ def test_users_index(pyramid_request):
         "username": None,
         "authority": None,
         "user": None,
+        "username_pattern": Any.string(),
         "user_meta": {},
         "format_date": format_date,
     }
@@ -89,6 +91,7 @@ def test_users_index_no_user_found(models, pyramid_request):
         "authority": "foo.org",
         "user": None,
         "user_meta": {},
+        "username_pattern": Any.string(),
         "format_date": format_date,
     }
 
@@ -107,6 +110,7 @@ def test_users_index_user_marked_as_deleted(models, pyramid_request, factories):
         "authority": "foo.org",
         "user": None,
         "user_meta": {},
+        "username_pattern": Any.string(),
         "format_date": format_date,
     }
 
@@ -128,6 +132,7 @@ def test_users_index_user_found(
         "authority": "foo.org",
         "user": user,
         "user_meta": {"annotations_count": 8},
+        "username_pattern": Any.string(),
         "format_date": format_date,
     }