Do actual ISO date parsing

Fixes #187

.pre-commit-config.yaml

@@ -46,4 +46,11 @@ repos:
     rev: v0.991
     hooks:
       - id: mypy
-        additional_dependencies: [types-pyopenssl, types-bleach, types-mock, types-cachetools]
+        additional_dependencies:
+          [
+            types-pyopenssl,
+            types-bleach,
+            types-mock,
+            types-cachetools,
+            types-python-dateutil,
+          ]

activities/models/post.py

@@ -726,7 +726,7 @@ class Post(StatorModel):
                 # Resolve the author
                 author = Identity.by_actor_uri(data["attributedTo"], create=create)
                 # If the post is from a blocked domain, stop and drop
-                if author.domain.blocked:
+                if author.domain and author.domain.blocked:
                     raise cls.DoesNotExist("Post is from a blocked domain")
                 post = cls.objects.create(
                     object_uri=data["id"],

core/ld.py

@@ -2,6 +2,7 @@ import datetime
 import os
 import urllib.parse as urllib_parse
 
+from dateutil import parser
 from pyld import jsonld
 from pyld.jsonld import JsonLdError
 
@@ -366,6 +367,7 @@ schemas = {
 }
 
 DATETIME_FORMAT = "%Y-%m-%dT%H:%M:%SZ"
+DATETIME_TZ_FORMAT = "%Y-%m-%dT%H:%M:%S+00:00"
 DATETIME_MS_FORMAT = "%Y-%m-%dT%H:%M:%S.%fZ"
 
 
@@ -448,15 +450,7 @@ def format_ld_date(value: datetime.datetime) -> str:
 def parse_ld_date(value: str | None) -> datetime.datetime | None:
     if value is None:
         return None
-    try:
-        return datetime.datetime.strptime(value, DATETIME_FORMAT).replace(
-            tzinfo=datetime.timezone.utc
-        )
-    except ValueError:
-        return datetime.datetime.strptime(value, DATETIME_MS_FORMAT).replace(
-            tzinfo=datetime.timezone.utc,
-            microsecond=0,
-        )
+    return parser.isoparse(value).replace(microsecond=0)
 
 
 def get_first_image_url(data) -> str | None:

requirements.txt

@@ -20,6 +20,7 @@ pydantic~=1.10.2
 pyld~=2.0.3
 pylibmc~=1.6.3
 pymemcache~=4.0.0
+python-dateutil~=2.8.2
 python-dotenv~=0.21.0
 redis~=4.4.0
 sentry-sdk~=1.11.0

tests/core/test_ld.py

@@ -1,5 +1,7 @@
 import datetime
 
+from dateutil.tz import tzutc
+
 from core.ld import parse_ld_date
 
 
@@ -7,22 +9,35 @@ def test_parse_ld_date():
     """
     Tests that the various kinds of LD dates that we see will work
     """
-    assert parse_ld_date("2022-11-16T15:57:58Z") == datetime.datetime(
+    difference = parse_ld_date("2022-11-16T15:57:58Z") - datetime.datetime(
         2022,
         11,
         16,
         15,
         57,
         58,
-        tzinfo=datetime.timezone.utc,
+        tzinfo=tzutc(),
     )
+    assert difference.total_seconds() == 0
 
-    assert parse_ld_date("2022-11-16T15:57:58.123Z") == datetime.datetime(
+    difference = parse_ld_date("2022-11-16T15:57:58.123Z") - datetime.datetime(
         2022,
         11,
         16,
         15,
         57,
         58,
-        tzinfo=datetime.timezone.utc,
+        tzinfo=tzutc(),
     )
+    assert difference.total_seconds() == 0
+
+    difference = parse_ld_date("2022-12-16T13:32:08+00:00") - datetime.datetime(
+        2022,
+        12,
+        16,
+        13,
+        32,
+        8,
+        tzinfo=tzutc(),
+    )
+    assert difference.total_seconds() == 0