From 41a04af8dc5fb5f42642a60d1a550ded08c5d15a Mon Sep 17 00:00:00 2001
From: Jake Howard <git@theorangeone.net>
Date: Mon, 1 Jul 2024 22:34:58 +0100
Subject: [PATCH] Fix pickle errors for metadata

---
 website/blog/models.py  |  4 ++--
 website/blog/tests.py   | 14 +++++++++++++-
 website/common/utils.py | 11 ++++++++---
 3 files changed, 23 insertions(+), 6 deletions(-)

diff --git a/website/blog/models.py b/website/blog/models.py
index b33c1f3..97df387 100644
--- a/website/blog/models.py
+++ b/website/blog/models.py
@@ -7,7 +7,7 @@ from django.db.models.functions import Cast, Coalesce
 from django.http import HttpRequest, HttpResponse, HttpResponsePermanentRedirect
 from django.utils import timezone
 from django.utils.functional import cached_property
-from metadata_parser import MetadataParser
+from metadata_parser import ParsedResult
 from modelcluster.fields import ParentalManyToManyField
 from wagtail.admin.panels import FieldPanel
 from wagtail.models import Page, PageQuerySet, Site
@@ -239,7 +239,7 @@ class ExternalBlogPostPage(BaseContentPage):
         return tags
 
     @cached_property
-    def metadata(self) -> MetadataParser:
+    def metadata(self) -> ParsedResult:
         return get_page_metadata(self.external_url)
 
     @cached_property
diff --git a/website/blog/tests.py b/website/blog/tests.py
index fa66c55..848888e 100644
--- a/website/blog/tests.py
+++ b/website/blog/tests.py
@@ -1,3 +1,5 @@
+import pickle
+
 from django.test import TestCase
 from django.urls import reverse
 
@@ -96,7 +98,9 @@ class ExternalBlogPostPageTestCase(TestCase):
     def setUpTestData(cls) -> None:
         cls.home_page = HomePage.objects.get()
         cls.blog_post_list_page = BlogPostListPageFactory(parent=cls.home_page)
-        cls.page = ExternalBlogPostPageFactory(parent=cls.blog_post_list_page)
+        cls.page = ExternalBlogPostPageFactory(
+            parent=cls.blog_post_list_page, external_url="https://example.com"
+        )
 
     def test_redirects(self) -> None:
         with self.assertNumQueries(10):
@@ -107,3 +111,11 @@ class ExternalBlogPostPageTestCase(TestCase):
             status_code=301,
             fetch_redirect_response=False,
         )
+
+    def test_metadata(self) -> None:
+        metadata = self.page.metadata
+
+        self.assertIsNone(metadata.soup)
+
+        # Confirm it can pickle
+        pickle.dumps(metadata)
diff --git a/website/common/utils.py b/website/common/utils.py
index 2e83fd9..6aef1f2 100644
--- a/website/common/utils.py
+++ b/website/common/utils.py
@@ -11,7 +11,7 @@ from django.http import QueryDict
 from django.http.request import HttpRequest
 from django.utils.text import slugify
 from django_cache_decorator import django_cache_decorator
-from metadata_parser import MetadataParser
+from metadata_parser import MetadataParser, ParsedResult
 from wagtail.models import Page, Site
 from wagtail.models import get_page_models as get_wagtail_page_models
 
@@ -128,8 +128,13 @@ def get_ai_robots_txt() -> str:
 
 
 @django_cache_decorator(time=21600)
-def get_page_metadata(url: str) -> MetadataParser:
-    return MetadataParser(url=url, search_head_only=True)
+def get_page_metadata(url: str) -> ParsedResult:
+    metadata = MetadataParser(url=url, search_head_only=True).parsed_result
+
+    # HACK: BeautifulSoup doesn't pickle nicely, and so can't be cached
+    metadata.soup = None
+
+    return metadata
 
 
 def extend_query_params(url: str, params: dict[str, Any]) -> str: