Support linking out to external posts

requirements.txt

@@ -29,6 +29,7 @@ django-enforce-host==1.1.0
 django-proxy==1.2.2
 wagtail-lite-youtube-embed==0.1.0
 django-minify-html==1.7.1
+metadata-parser==0.12.1
 
 # DRF OpenAPI dependencies
 uritemplate

static/src/scss/_listing.scss

@@ -18,6 +18,17 @@
 
   .title {
     margin-bottom: 0;
+
+    a {
+      display: flex;
+      flex-direction: row;
+      align-items: center;
+      justify-content: space-between;
+
+      i {
+        font-size: 50%;
+      }
+    }
   }
 
   .content-details {

website/blog/factories.py

@@ -1,3 +1,5 @@
+import factory
+
 from website.common.factories import BaseContentFactory, BaseListingFactory
 
 from . import models
@@ -11,3 +13,11 @@ class BlogPostListPageFactory(BaseListingFactory):
 class BlogPostPageFactory(BaseContentFactory):
     class Meta:
         model = models.BlogPostPage
+
+
+class ExternalBlogPostPageFactory(BaseContentFactory):
+    external_url = factory.Faker("url")
+
+    class Meta:
+        model = models.ExternalBlogPostPage
+        exclude = ["subtitle"]

website/blog/migrations/0006_externalblogpostpage.py

@@ -0,0 +1,45 @@
+# Generated by Django 5.0.4 on 2024-05-29 21:10
+
+import django.db.models.deletion
+import django.utils.timezone
+import modelcluster.fields
+import wagtailmetadata.models
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("blog", "0005_auto_20230602_1236"),
+        ("wagtailcore", "0089_log_entry_data_json_null_to_object"),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="ExternalBlogPostPage",
+            fields=[
+                (
+                    "page_ptr",
+                    models.OneToOneField(
+                        auto_created=True,
+                        on_delete=django.db.models.deletion.CASCADE,
+                        parent_link=True,
+                        primary_key=True,
+                        serialize=False,
+                        to="wagtailcore.page",
+                    ),
+                ),
+                ("external_url", models.URLField()),
+                ("date", models.DateField(default=django.utils.timezone.now)),
+                (
+                    "tags",
+                    modelcluster.fields.ParentalManyToManyField(
+                        blank=True, to="blog.blogposttagpage"
+                    ),
+                ),
+            ],
+            options={
+                "abstract": False,
+            },
+            bases=("wagtailcore.page", wagtailmetadata.models.MetadataMixin),
+        ),
+    ]

website/blog/models.py

@@ -1,18 +1,26 @@
 from typing import Any, Optional
+from urllib.parse import urlsplit
 
 from django.contrib.postgres.search import TrigramSimilarity
 from django.db import models
-from django.db.models.functions import Cast
+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 modelcluster.fields import ParentalManyToManyField
 from wagtail.admin.panels import FieldPanel
-from wagtail.models import PageQuerySet
+from wagtail.models import Page, PageQuerySet, Site
 from wagtail.search import index
 from wagtailautocomplete.edit_handlers import AutocompletePanel
 
-from website.common.models import BaseContentPage, BaseListingPage
-from website.common.utils import TocEntry
+from website.common.models import BaseContentPage, BaseListingPage, BasePage
+from website.common.utils import (
+    TocEntry,
+    extend_query_params,
+    get_page_metadata,
+    get_url_mime_type,
+)
 from website.contrib.singleton_page.utils import SingletonPageCache
 
 
@@ -23,6 +31,8 @@ class BlogPostListPage(BaseListingPage):
         "blog.BlogPostTagListPage",
         "blog.BlogPostCollectionListPage",
         "blog.BlogPostCollectionPage",
+        "blog.BlogPostCollectionPage",
+        "blog.ExternalBlogPostPage",
     ]
 
     @cached_property
@@ -31,9 +41,12 @@ class BlogPostListPage(BaseListingPage):
 
     def get_listing_pages(self) -> models.QuerySet:
         return (
-            BlogPostPage.objects.descendant_of(self)
-            .live()
+            Page.objects.live()
             .public()
+            .annotate(date=Coalesce("blogpostpage__date", "externalblogpostpage__date"))
+            .descendant_of(self)
+            .type(BlogPostPage, ExternalBlogPostPage)
+            .specific()
             .order_by("-date", "title")
         )
 
@@ -87,10 +100,6 @@ class BlogPostPage(BaseContentPage):
 
         similar_posts = listing_pages.exclude(id=self.id).alias(
             title_similarity=TrigramSimilarity("title", self.title),
-            # If this page has no subtitle, ignore it as part of similarity
-            subtitle_similarity=TrigramSimilarity("subtitle", self.subtitle)
-            if self.subtitle
-            else models.Value(1),
         )
 
         page_tags = list(self.tags.public().live().values_list("id", flat=True))
@@ -109,7 +118,6 @@ class BlogPostPage(BaseContentPage):
         similar_posts = similar_posts.annotate(
             similarity=(models.F("tag_similarity") * 2)
             + (models.F("title_similarity") * 10)
-            + (models.F("subtitle_similarity"))
         ).order_by("-similarity")[:3]
 
         return similar_posts
@@ -137,7 +145,20 @@ class BlogPostTagPage(BaseListingPage):
 
     def get_listing_pages(self) -> models.QuerySet:
         blog_list_page = BlogPostListPage.objects.get()
-        return blog_list_page.get_listing_pages().filter(tags=self)
+        listing_pages = blog_list_page.get_listing_pages()
+        blog_post_tags = list(
+            BlogPostPage.objects.filter(id__in=listing_pages, tags=self).values_list(
+                "id", flat=True
+            )
+        )
+        external_post_tags = list(
+            ExternalBlogPostPage.objects.filter(
+                id__in=listing_pages, tags=self
+            ).values_list("id", flat=True)
+        )
+        return listing_pages.filter(
+            id__in=blog_post_tags + external_post_tags
+        ).specific()
 
 
 class BlogPostCollectionListPage(BaseListingPage):
@@ -167,3 +188,101 @@ class BlogPostCollectionPage(BaseListingPage):
             .public()
             .order_by("-date", "title")
         )
+
+
+class ExternalBlogPostPage(BaseContentPage):
+    subpage_types: list[Any] = []
+    parent_page_types = [BlogPostListPage]
+    preview_modes: list[Any] = []
+
+    is_external = True
+
+    # Some `BaseContentPage` fields aren't relevant
+    body = None
+    subtitle = None
+    hero_image = None
+    hero_unsplash_photo = None
+
+    external_url = models.URLField()
+
+    tags = ParentalManyToManyField("blog.BlogPostTagPage", blank=True)
+    date = models.DateField(default=timezone.now)
+
+    content_panels = BasePage.content_panels + [FieldPanel("external_url")]
+
+    promote_panels = BaseContentPage.promote_panels + [
+        FieldPanel("date"),
+        AutocompletePanel("tags"),
+    ]
+
+    search_fields = BaseContentPage.search_fields + [
+        index.RelatedFields("tags", [index.SearchField("title", boost=1)]),
+        index.SearchField("external_url"),
+    ]
+
+    @cached_property
+    def tag_list_page_url(self) -> Optional[str]:
+        return SingletonPageCache.get_url(BlogPostTagListPage)
+
+    @cached_property
+    def tags_list(self) -> models.QuerySet:
+        """
+        Use this to get a page's tags.
+        """
+        tags = self.tags.order_by("slug")
+
+        # In drafts, `django-modelcluster` doesn't support these filters
+        if isinstance(tags, PageQuerySet):
+            return tags.public().live()
+
+        return tags
+
+    @cached_property
+    def metadata(self) -> MetadataParser:
+        return get_page_metadata(self.external_url)
+
+    @cached_property
+    def _body_html(self) -> str:
+        try:
+            return self.metadata.get_metadatas("description")[0]
+        except (KeyError, IndexError, TypeError):
+            return ""
+
+    @cached_property
+    def plain_text(self) -> str:
+        # The metadata is already just text
+        return self._body_html
+
+    def hero_url(
+        self, image_size: str, wagtail_image_spec_extra: Optional[str] = None
+    ) -> Optional[str]:
+        try:
+            return self.metadata.get_metadatas("image")[0]
+        except (KeyError, IndexError, TypeError):
+            return None
+
+    @cached_property
+    def hero_image_url(self) -> str:
+        return ""
+
+    @cached_property
+    def hero_image_alt(self) -> str:
+        return ""
+
+    def get_meta_image_mime(self) -> Optional[str]:
+        return get_url_mime_type(self.hero_url(""))
+
+    def get_url(
+        self, request: HttpRequest | None = None, current_site: Site | None = None
+    ) -> str:
+        return self.get_full_url(request)
+
+    def get_full_url(self, request: HttpRequest | None = None) -> str:
+        full_url = urlsplit(super().get_full_url(request))
+        return extend_query_params(self.external_url, {"utm_source": full_url.netloc})
+
+    def serve(self, request: HttpRequest, *args: tuple, **kwargs: dict) -> HttpResponse:
+        """
+        Send the user directly to the external page
+        """
+        return HttpResponsePermanentRedirect(self.get_full_url(request))

website/blog/templates/blog/external_blog_post_page.html

@@ -0,0 +1,16 @@
+{% comment %}
+This template is never used, but exists just in case.
+{% endcomment %}
+
+<!DOCTYPE html>
+<html lang="en-GB">
+  <head>
+    <title>Redirecting...</title>
+    <link rel="canonical" href="{{ page.external_url }}" />
+    <meta charset="utf-8" />
+    <meta http-equiv="refresh" content="0; url={{ page.external_url }}" />
+  </head>
+  <body>
+    <p>Redirecting...</p>
+  </body>
+</html>

website/blog/tests.py

@@ -3,7 +3,11 @@ from django.urls import reverse
 
 from website.home.models import HomePage
 
-from .factories import BlogPostListPageFactory, BlogPostPageFactory
+from .factories import (
+    BlogPostListPageFactory,
+    BlogPostPageFactory,
+    ExternalBlogPostPageFactory,
+)
 
 
 class BlogPostPageTestCase(TestCase):
@@ -69,14 +73,15 @@ class BlogPostListPageTestCase(TestCase):
 
         BlogPostPageFactory(parent=cls.page)
         BlogPostPageFactory(parent=cls.page)
+        ExternalBlogPostPageFactory(parent=cls.page, external_url="https://example.com")
 
     def test_accessible(self) -> None:
         response = self.client.get(self.page.url)
         self.assertEqual(response.status_code, 200)
-        self.assertEqual(len(response.context["listing_pages"]), 2)
+        self.assertEqual(len(response.context["listing_pages"]), 3)
 
     def test_queries(self) -> None:
-        with self.assertNumQueries(39):
+        with self.assertNumQueries(43):
             self.client.get(self.page.url)
 
     def test_feed_accessible(self) -> None:
@@ -84,3 +89,21 @@ class BlogPostListPageTestCase(TestCase):
         self.assertRedirects(
             response, reverse("feed"), status_code=301, fetch_redirect_response=True
         )
+
+
+class ExternalBlogPostPageTestCase(TestCase):
+    @classmethod
+    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)
+
+    def test_redirects(self) -> None:
+        with self.assertNumQueries(10):
+            response = self.client.get(self.page.url)
+        self.assertRedirects(
+            response,
+            self.page.external_url + "?utm_source=localhost",
+            status_code=301,
+            fetch_redirect_response=False,
+        )

website/common/models.py

@@ -1,7 +1,6 @@
 from datetime import timedelta
 from math import ceil
 from typing import Any, Optional
-from urllib.parse import urlencode
 
 from django.core.paginator import EmptyPage, Paginator
 from django.core.paginator import Page as PaginatorPage
@@ -31,6 +30,7 @@ from .serializers import PaginationSerializer
 from .streamfield import add_heading_anchors, get_blocks, get_content_html
 from .utils import (
     TocEntry,
+    extend_query_params,
     extract_text,
     get_site_title,
     get_table_of_contents,
@@ -181,6 +181,7 @@ class BaseContentPage(BasePage, MetadataMixin):
             for size, width in UNSPLASH_SIZES.items()
         }
 
+    @cached_property
     def hero_image_url(self) -> Optional[str]:
         return self.hero_url("regular")
 
@@ -286,10 +287,7 @@ class BaseListingPage(RoutablePageMixin, BaseContentPage):
 
         url = super().get_meta_url()
 
-        if not query_data:
-            return url
-
-        return url + "?" + urlencode(query_data)
+        return extend_query_params(url, query_data)
 
     @route(r"^feed/$")
     def feed(self, request: HttpRequest) -> HttpResponse:

website/common/templates/common/listing-item.html

@@ -16,7 +16,10 @@
           {% include "common/breadcrumbs.html" with parents=page.get_parent_pages %}
         {% endif %}
         <h2 class="title is-3">
-          <a href="{% pageurl page %}">{{ page.title }}</a>
+          <a href="{% pageurl page %}">
+            {{ page.title }}
+            {% if page.is_external %}<i class="fa-solid fa-arrow-up-right-from-square" title="This page is from a external source"></i>{% endif %}
+          </a>
         </h2>
         {% include "common/content-details.html" %}
         <p>{{ page.summary }}</p>

website/common/tests/test_utils.py

@@ -3,6 +3,7 @@ from django.test import SimpleTestCase
 from wagtail.rich_text import features as richtext_feature_registry
 
 from website.common.utils import (
+    extend_query_params,
     extract_text,
     get_table_of_contents,
     heading_id,
@@ -111,3 +112,25 @@ class HeadingIDTestCase(SimpleTestCase):
         self.assertEqual(heading_id("123"), "ref-123")
         self.assertEqual(heading_id("test"), "test")
         self.assertEqual(heading_id("Look, a title!"), "look-a-title")
+
+
+class ExtendQueryParamsTestCase(SimpleTestCase):
+    def test_params(self) -> None:
+        self.assertEqual(
+            extend_query_params("https://example.com", {"foo": "bar"}),
+            "https://example.com?foo=bar",
+        )
+        self.assertEqual(
+            extend_query_params("https://example.com?foo=bar", {"bar": "foo"}),
+            "https://example.com?foo=bar&bar=foo",
+        )
+        self.assertEqual(
+            extend_query_params("https://example.com?foo=baz", {"foo": "baz"}),
+            "https://example.com?foo=baz",
+        )
+
+    def test_removes_param(self) -> None:
+        self.assertEqual(
+            extend_query_params("https://example.com?foo=bar", {"foo": None}),
+            "https://example.com",
+        )

website/common/utils.py

@@ -1,14 +1,17 @@
 from dataclasses import dataclass
 from itertools import pairwise
-from typing import Optional, Type
+from typing import Any, Optional, Type
+from urllib.parse import urlsplit, urlunsplit
 
 import requests
 from bs4 import BeautifulSoup, SoupStrainer
 from django.conf import settings
 from django.db import models
+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 wagtail.models import Page, Site
 from wagtail.models import get_page_models as get_wagtail_page_models
 
@@ -122,3 +125,21 @@ def get_ai_robots_txt() -> str:
     return requests_session.get(
         "https://raw.githubusercontent.com/ai-robots-txt/ai.robots.txt/main/robots.txt"
     ).content.decode()
+
+
+@django_cache_decorator(time=21600)
+def get_page_metadata(url: str) -> MetadataParser:
+    return MetadataParser(url=url, search_head_only=True)
+
+
+def extend_query_params(url: str, params: dict[str, Any]) -> str:
+    scheme, netloc, path, query, fragment = urlsplit(url)
+    query_dict = QueryDict(query, mutable=True)
+
+    for k, v in params.items():
+        if v is None:
+            del query_dict[k]
+        else:
+            query_dict[k] = v
+
+    return urlunsplit((scheme, netloc, path, query_dict.urlencode(), fragment))

website/common/views.py

@@ -23,7 +23,7 @@ from website.search.models import SearchPage
 
 from .feed_generators import CustomFeed
 from .models import BaseListingPage, BasePage
-from .utils import get_ai_robots_txt
+from .utils import extend_query_params, get_ai_robots_txt
 
 
 class Error404View(TemplateView):
@@ -116,7 +116,9 @@ class AllPagesFeed(Feed):
         return item.title
 
     def item_link(self, item: BasePage) -> str:
-        return item.get_full_url(request=self.request) + "?utm_medium=rss"
+        return extend_query_params(
+            item.get_full_url(request=self.request), {"utm_medium": "rss"}
+        )
 
     def item_pubdate(self, item: BasePage) -> datetime:
         if item_date := getattr(item, "date", None):

website/search/views.py

@@ -7,7 +7,7 @@ from wagtail.search.utils import parse_query_string
 from wagtail_favicon.models import FaviconSettings
 from wagtail_favicon.utils import get_rendition_url
 
-from website.common.utils import get_or_none, get_site_title
+from website.common.utils import extend_query_params, get_or_none, get_site_title
 from website.contrib.singleton_page.utils import SingletonPageCache
 
 from .models import SearchPage
@@ -87,4 +87,4 @@ class GoView(RedirectView):
         if slug_match := get_or_none(pages.filter(slug__iexact=query)):
             return slug_match.get_url(request=self.request)
 
-        return f"{search_page_url}?{self.request.GET.urlencode()}"
+        return extend_query_params(search_page_url, self.request.GET)

website/settings.py

@@ -391,6 +391,11 @@ LOGGING = {
             "level": "WARNING",
             "propagate": False,
         },
+        "metadata_parser": {
+            "handlers": ["console"],
+            "level": "CRITICAL",
+            "propagate": False,
+        },
     },
 }