Fix: tailwindcss utilities no longer break microformat parsing.

• Fixed #54: Broken on wagtail>=6.2.
  • Issue: In wagtail==6.2, the implementation of RoutablePageMixin.route was changed to add the routable_resolver_match attribute to the given request object. get_model_for_url_by_wagtail previously called that method with request=None (because there is no relevant request object available at the time webmentions are processed), so trying to set that attribute on the request fails.
  • Fix: get_model_for_url_by_wagtail now creates and passes a dummy instance of HttpRequest when calling that method.

• Added Webmention.has_been_read: bool field.

  • New context processor mentions.context_processors.unread_webmentions adds unread_webmentions field to template context, containing the queryset of unread Webmention objects.
  • Added admin actions for marking as read/unread.

• New objects manager for Webmention with some common filters and actions.

• New settings for allowing or disabling webmentions to/from a set of domains:
  • WEBMENTIONS_DOMAINS_INCOMING_ALLOW: Iterable[str] = None
  • WEBMENTIONS_DOMAINS_INCOMING_DENY: Iterable[str] = None
  • WEBMENTIONS_DOMAINS_OUTGOING_ALLOW: Iterable[str] = None
  • WEBMENTIONS_DOMAINS_OUTGOING_DENY: Iterable[str] = None
• New settings for tags that can be added to your HTML links to allow/disable sending webmentions for just that specific link. (Overrides above allow/deny lists)
  • WEBMENTIONS_DOMAINS_OUTGOING_TAG_ALLOW: str = None
  • WEBMENTIONS_DOMAINS_OUTGOING_TAG_DENY: str = None

Resolves #53: Compatibility with dependency mf2py>=2.0.

• Resolves #50: broken search field on QuotableAdmin.
• Added tests for admin pages to avoid that sort of thing happening again.
• Minor touch-ups for the admin pages.
  • Source and target URL fields are now read-only.
  • Added appropriate search fields and list filters for each model.
  • quote field now uses a textarea widget for comfier editing.

• Added management command mentions_reverify [filters ...] [--all]

  • Allows you to reprocess received Webmentions to see if they are still ‘live’.
  • Accepts a space-separated list of field=value queryset filters, or --all to reprocess all of them.

• Added management command mentions_pending to replace pending_mentions for naming consistency with other commands.

  • pending_mentions still works though.

• Now compatible with Wagtail>=3.0.3 (previously >=4.0).

  • mentions_wagtail_path still requires >=4.0: use mentions_wagtail_route or mentions_wagtail_re_path.

• MentionableMixin:

  • allow_outgoing_webmentions default now configurable via settings.WEBMENTIONS_ALLOW_OUTGOING_DEFAULT.

  • Removed slug field. If you use this field you can restore the previous behaviour by adding the following to your model.

    class MyModel(MentionableMixin, models.Model):
        slug = models.SlugField(unique=True)
    
        @classmethod
        def resolve_from_url_kwargs(cls, slug, **url_kwargs):
            return cls.objects.get(slug=slug)
    

  • Deprecated methodall_text, replaced by get_content_html. Overriding all_text still works for now but will log a warning asking you to rename the method.

• Moved template files to mentions sub-directory. If you have custom overrides of these templates in your root templates/ directory please move them to templates/mentions/.

• Added user agent header to all network requests.

  • Customisable via settings.WEBMENTIONS_USER_AGENT: str.

• Admin-facing strings are now translatable.

• Added urlpatterns helper functions mentions_path, mentions_re_path for (hopefully) simpler setup.

  • More straightforward view-to-model mapping.
  • Removes the need to implement resolve_from_url_kwargs on your MentionableMixin implementation.
  • See the wiki for example setup instructions.

• Support for Wagtail.

  • Page models should implement MentionableMixin as usual.
  • RoutablePageMixin should use the new @mentions_wagtail_path, @mentions_wagtail_re_path decorators in place of the Wagtail equivalents @path, @re_path.
    • These work essentially the same mentions_path and mentions_re_path.

  • If using RichTextField call richtext on any RichTextField values in get_content_html:

    from wagtail.templatetags.wagtailcore_tags import richtext
    
    class MyModel(MentionableMixin, Page):
        ...
        def get_content_html(self) -> str:
            return f"{richtext(self.overview)} {richtext(self.body)}"
    

• Wiki pages are now live! These will be kept up-to-date going forwards but may not be useful for pre-4.0 versions.

• Fix: Relative URLs in h-card homepage or avatar are now resolved to absolute URLs.

Fixes #43: outgoing webmention being resubmitted continuously.

• Resolves #38: Revalidate target URLs when handling pending mentions

  • Should be unnecessary generally (they are also validated at discovery time when parsed from HTML) but important if validation checks are updated.

• Resolves #41: Find correct endpoint when multiple links in HTTP header.

• Added settings.WEBMENTIONS_INCOMING_TARGET_MODEL_REQUIRED [bool | default=False]. If True, incoming mentions are only accepted if their target resolves to a MentionableMixin instance.

• Added settings.WEBMENTIONS_ALLOW_SELF_MENTIONS [bool | default=True].

  • If False, outgoing links that target your own domain (as specified by settings.DOMAIN_NAME) will be ignored - you will only submit mentions to other domains.
  • If True, outgoing links that use a relative path (e.g. href="/article/1/") are now supported.

• Fix: WebmentionHeadMiddleware no longer overwrites existing links in HTTP header when adding webmention endpoint.

• Fix: Webmention ’notes’ no longer persists across instances.

Upgrade warning

If upgrading from an older version please be aware of these changes:

• Unused MentionableMixin.allow_incoming_webmentions field has been removed.
• Any existing instances of PendingIncomingWebmention and PendingOutgoingContent will be deleted.
  • These models have new constraints so it is necessary to recreate them.
  • If this is problematic for you please don’t upgrade yet. Contact me or create an issue and I will make a tool to persist these between versions.

Changes

Thanks to @philgyford for reporting most of the issues referenced here.

• Resolves #25.

  • Added QuotableMixin.post_type field.

  • Incoming webmentions are now checked for the following microformat properties that describe the type of mention they are:

    • u-bookmark-of
    • u-like-of
    • u-listen-of
    • u-in-reply-to
    • u-repost-of
    • u-translation-of
    • u-watch-of

  • The /webmention/get endpoint serializes these values in the type field respectively as:

    • bookmark
    • like
    • listen
    • reply
    • repost
    • translation
    • watch
    • If no specific type is specified this defaults to webmention.

• Resolves #30

  • Added MentionableMixin.should_process_webmentions() -> bool method to enable custom logic.

• Resolves #31

  • Success message now rendered via template, enabling override by user.

• Resolves #32

  • Same-page #anchor links no longer treated as webmention targets.

• Resolves #33, #34

  • Much improved parsing of h-card.
    • Now tries to find anh-card that is directly related to the mention link (embedded in p-author of a parent h-entry or h-feed container).
    • If that doesn’t yield a result, try to find a top-level h-card on the page.
  • Removed HCard.from_soup() classmethod. Parsing logic moved to tasks.incoming.parsing package.

• Resolves #36

  • Added constraints to PendingIncomingWebmention and PendingOutgoingContent to avoid duplication.
    • Warning: If updating from an older version of django-wm this will delete any existing Pending... model instances.
  • Request timeouts are now handled gracefully.
  • PendingIncomingWebmention and OutgoingWebmentionStatus now implement the new RetryableMixin.
  • Reworked webmention processing tasks to allow failed webmentions to be retried periodically.
    • See settings WEBMENTIONS_MAX_RETRIES, WEBMENTIONS_RETRY_INTERVAL, WEBMENTIONS_TIMEOUT below for customisation details.

• New dashboard/ view: a simple overview of recent mentions.

  • Shows the latest instances of Webmention, OutgoingWebmentionStatus, PendingIncomingWebmention, PendingOutgoingContent and info on their current status.
  • By default, restricted to users with mentions.view_webmention_dashboard permission.
  • Can be made public via settings.WEBMENTIONS_DASHBOARD_PUBLIC = True.

• New optional settings:

  • settings.WEBMENTIONS_TIMEOUT [float | default=10] specifies the time (in seconds) to wait for network calls to resolve.
  • settings.WEBMENTIONS_RETRY_INTERVAL [int | default=600] specifies the minimum time (in seconds) to wait before retrying to process a webmention.
  • settings.WEBMENTIONS_MAX_RETRIES [int | default=5] specifies how many times we can attempt to process a mention before giving up.
  • settings.WEBMENTIONS_DASHBOARD_PUBLIC [bool | default=False] specifies whether the the dashboard/ view can be viewed by anyone. If False (default), the dashboard/ view is only available to users with mentions.view_webmention_dashboard permission.

• WebmentionHeadMiddleware

  • Removed port from generated endpoint URL in HTTP headers as request.META.SERVER_PORT may not be reliable depending on reverse proxy configuration.

• MentionableMixin:

  • mentions() is now a method, not a property.
  • Removed field allow_incoming_webmentions as it has never been used for anything.

• Streamlined template tags

  • {% load webmentions_endpoint %} replaced by {% load webmentions %}.
  • {% webmention_endpoint %} replaced by {% webmentions_endpoint %} (used in HTML <head>).
  • New tag {% webmentions_dashboard %} creates a <a>link</a> to your webmentions/dashboard/ view.

Resolves #28

New MentionableMixin classmethod: resolve_from_url_kwargs(**url_kwargs)

• This method receives captured parameters from an urlpatterns path.
• By default, it uses <slug:slug> to look up your object by a unique slug field.
• You can customize this by overriding the classmethod in your MentionableMixin implementation

e.g.

# urls.py
urlpatterns = [
    path(
        fr"<int:year>/<int:month>/<int:day>/<slug:post_slug>/",
        MyBlogPostView.as_view(),
        name="my-blog",
        kwargs={
            "model_name": "myapp.MyBlog",
        },
    ),
]

# models.py
class MyBlog(MentionableMixin, models.Model):
    date = models.DateTimeField(default=timezone.now)
    slug = models.SlugField()
    content = models.TextField()

    def all_text(self):
        return self.content

    def get_absolute_url(self):
        return reverse(
            "my-blog",
            kwargs={
                "year": self.date.strftime("%Y"),
                "month": self.date.strftime("%m"),
                "day": self.date.strftime("%d"),
                "post_slug": self.slug,
            }
        )

    @classmethod
    def resolve_from_url_kwargs(cls, year, month, day, post_slug, **url_kwargs):
        return cls.objects.get(
            date__year=year,
            date__month=month,
            date__day=day,
            slug=post_slug,
        )

• mentions.resolution.get_model_for_url_path now delegates to MentionableMixin.resolve_from_url_kwargs to resolve captured URL parameters to a model instance.

Merges #24: QuotableMixin.published can now be overriden - thanks @garrettc.

Fixes #26: requests 2.20 or greater (until version 3) are now allowed. Likewise for beautifulsoup4 4.6 and mf2py 1.1.

Added get_mentions_for_view(HttpRequest) -> Iterable[QuotableMixin] convenience method. This may be used to retrieve mentions for rendering directly in a Django template, as an alternative to using the webmention/get endpoint from a frontend script.

• Added setting WEBMENTIONS_USE_CELERY (boolean, default True)

  • If False:
    • celery does not need to be installed
    • New models PendingIncomingWebmention and PendingOutgoingContent are created to store the required data for later batch-processing.
    • New management command: manage.py pending_mentions can be used to process these data.

• /get endpoint:

  • Now returns results for SimpleMention objects as well as Webmentions.
  • Added field type with value webmention or simple so they can be differentiated when displaying.

• Updated instructions for installation with or without celery.

Breaking Changes

• Migrations are now included. If you are upgrading from any 1.x.x version please follow these instructions to avoid data loss. Thanks to @GriceTurrble for providing these instructions.

• requirements.txt celery version updated to 5.2.2 due to CVE-2021-23727. If you are upgrading from 4.x please follow the upgrade instructions provided by Celery.

Web API changes:

• /get endpoint:
  • Removed status from JSON object - now uses HTTP response codes 200 if the target url was resolved correctly or 404 otherwise.
  • Missing HCards are now serialized as null instead of an empty dict

// https://example.org/webmention/get?url=my-article
// Old 1.x.x response
{
  "status": 1,
  "target_url": "https://example.org/my-article",
  "mentions": [
    {
      "hcard": {},
      "quote": null,
      "source_url": "https://another-example.org/their-article",
      "published": "2020-01-17T21:45:24.542Z"
    }
  ]
}

// https://example.org/webmention/get?url=my-article
// New 2.0.0 response with HTTP status 200 (or 404 if target_url does not exist)
{
  "target_url": "https://example.org/my-article",
  "mentions": [
    {
      "hcard": null,
      "quote": null,
      "source_url": "https://another-example.org/their-article",
      "published": "2020-01-17T21:45:24.542Z"
    }
  ]
}

New

• Use{% webmention_endpoint %} template tag to include your Webmentions endpoint in your Django template to help other sites find it easily.

  {% load webmention_endpoint %}
  <!-- my-template.html -->
  ...
  <head>
    <!-- Rendered as <link rel="webmention" href="/webmention/" /> -->
    {% webmention_endpoint %}
  </head>
  ...
  

Incoming webmentions may now target any path, not just those that map directly to a mentionable model.

The app is now detected correctly by makemigrations