Kirby3 feeds

Requirement: Kirby 3.0.0-RC3 or better

Coffee, Beer, etc.

This started as a simple plugin and morphed into something quite complex to cope with a combination of trying to get something sensible out of Kirby while also trying to generate something acceptable within the constraints of the different syndication formats.

This plugin is free but if you use it in a commercial project to show your support you are welcome to:

• make a donation 🍻 or
• buy me ☕☕ or
• buy a Kirby license using this affiliate link

Documentation

Purpose

For a kirby3 site, this plugin (omz13/feeds) generates syndication feeds in RSS, ATOM, and JSON formats based on a Kirby3 "collection". If a page is given in a collection it will be included in a syndication feed; therefore, ensure that the collection filters "published" or "draft" per your requirements because this plugin does no filtering whatsoever (because you may want to do something like have a feed for drafts). You are in control of what does or does not get put into a feed (and this can be different from what is publically shown on a site).

• Kirby3 "collection" to atom/json/rss-feed methodology.
• Generates a RSS 2.0 syndication feed (at /feeds/rss) based on a default collection, c.f. firehose in Configuration.
• Generates an ATOM syndication feed (at /feeds/atom) based on a default collection.
• Generates a JSON 1 syndication feed (at /feeds/json) based on a default collection.
• Generates sub-setted syndication feeds at feeds/<category>/atom|json|rss) where <category> is mapped to a kirby collection, c.f. categories in Configuration.
• A feed will have a maximum of 60 items.
• To mitigate server load, and to return a speedy response, the generated feeds are cached in the server for a determined amount of time, c.f. cacheTTL in Configuration.
• Supports HTTP conditional get (with strong validation): i.e. respects If-Modified-Since and If-None-Match headers and will return a 304 Not Modified response if appropriate (saves your bandwidth).
• The following data is derrived as follows from fields in the pages (and their use depends on the feed format used):
  • the author(s) to be attributed is taken from the author field: this field can be either a structured field or a text field, but it should be the email address of the author (and will be used to map to a kirby user for their details). If an author can not be mapped to a user, a default is applied, c.f. author in Configuration.
  • the title from title, a string.
  • the modified date from, in order of preference: updatedat, date; note: if neither of these can be resolved, the the page's underlying file's modified-at attribute is used.
  • the published date from in order of preference: date, updatedat; note: if neither of these can be resolved, the the page's underlying file's modified-at attribute is used.
• For ATOM and JSON feeds, per author details are supplemented by their user (in order of preference) website or twitter or instagram data mapping to uri.
• If an item has multiple authors:
  • For RSS and JSON feeds, the author will be given a concatenation of the authors' names (e.g. if the authors are Ford and Zaphod, they would be attributed as "Ford and Zaphod").
  • For RSS, each author is additionally listed individually in <dc:creator>.
  • For JSON, each author's details are given in authors per brentsimmons/JSONFeed#120. This is all highly experimental so may break things - your feedback is appreciated.
• For a RSS feed, also provides <dc:date> for each item, and <dc:rights> for the channel.
• For a JSON feed, in addition to providing html content in content_html a markdown version is also provied in content_text; note that the markdown version is generated from the html representation and is not simply the raw (markdown + kirbytext) page content.
• For debugging purposes, the generated sitemap can include additional information as comments; c.f. debugqueryvalue in Configuration. The output may still work with a reader, but it is intended for debugging not consuming.

Caveat:

• Withdrawn pages (i.e. with a method issunset that returns true) can be excluded by appropriate configuration within a collection.
• Pages under an embargo (i.e. with a method isunderembargo that returns true) can similarly be managed.
• May contains bugs because there is a lot of nasty code to deal with edge cases in the various syndication specifications; where possible it tries to generate things as best it can (i.e. sensible assumptions).

Client Support

RSS feeds are generally well supported by clients; the ATOM feed offers some extra technical niceties, but client support varies; the JSON format is very new, and client support varies. Here are details of a few clients that I have used:

In terms of client for OS X:

• Readkit, does RSS and ATOM, but not JSON; it is über-fussy with the way it resolves names to hosts (and will just moan about "invalid feed" instead of really saying what it thinks is wrong).
• ReederApp supoorts all three formats.
• Leaf supports RSS and ATOM - note: not updated since Nov 2017.
• NetNewsWire - supoorts all three formats - note: v5 has gone back to Brent, so very alpha at the moment.
• NewsExplorer - supoorts all three formats.
• Vienna does RSS and ATOM but not JSON.

In terms of iOS:

• Readkit - see OS X
• ReederApp - see OS X
• NewsExplorer (does Apple TV too!)

In terms of generic services/clients:

• FeedBin
• Feedly supports RSS
• FeedWrangler
• NewsBlur supports all three formats.

In terms of Windows clients: sorry, I have no idea as inhabit an OS X / iOS world.

Related plugins

For a plugin that provides the methods issunset and isunderembargo, kindly see omz13/kirby3-suncyclepages.

Roadmap

The non-binding list of planned features and implementation notes are:

• MVP
• Linkblogs
• Splitting
• Source by collection
• Languages
• Author into item stream
• Category into item stream
• Generate feed discovery headers
• dc for RSS

Installation

Pick one of the following per your epistemological model:

• composer require omz13/kirby3-feeds; the plugin will automagically appear in site/plugins.
• Download a zip of the latest release - master.zip - and copy the contents to your site/plugins/kirby3-feeds.
• git submodule add https://github.com/omz13/kirby3-feeds.git site/plugins/kirby3-feeds.

For the record: installation by composer is cool; supporting installation by zip and submodule was an absolute pain, especially as I am an installation by composer person, so do feel guilted into getting me Coffee, Beer, etc., because this is for your benefit and not mine (and yes, I would have have preferred to spend my time somewhere warm and sunny instead of being hunched over a keyboard while the snow falls outside and the thermometer shows no inclination to get above 0C).

Configuration

There are four aspects that need configuration:

• site/config/config.php
• content/site.txt via blueprint (optional)
• Feed discovery
• Collections

site/config/config.php

In your site's site/config/config.php the following entries prefixed with omz13.feeds. can be used:

• disable - optional - boolean - default false - a boolean which, if true, to disable serving syndication feeds. Requests to such pages will receive a 503 Service Unavailable.
• cacheTTL - optional - integer - default 10 - the number of minutes that a feed should be cached before being regenerated; if explicitly set to zero, the cache is disabled. If not specified a default of 10 minutes is assumed.
• firehose - optional - string - default 'articles' - the name of the kirby collection to be used for the 'firehose' feed /feeds/atom|json|rss.
• categories - optional - array - default [ 'articles' ] - the names of the kirby collections that can be accessed using the uri /feeds/<category>/atom|json|rss. If an empty array is specifed ([]) then this feature is disabled.
• debugqueryvalue - optional - string - the value for the query parameter debug to return a feed with debugging information (as comments within response). The global kirby debug configuration must also be true for this to work. Be aware that the debugging information will show, if applicable, details of any pages that have been excluded (so if you are using this in production and you don't want things to leak, set debugqueryvalue to something random). Furthermore, the site debug flag needs to be set too (i.e. the debug flag in site/config.php).
• author - optional - string - default 'Staff Writer' - the name to be used for each item in a feed when either the author is unknown or the syndication format does not allow an author name (looking at you RSS2).

content/site.txt (via blueprint site/blueprints/site.yml)

The plugin can be explicitly disabled in content/site.txt by having an entry for feeds and setting this to false. This could be achieved through the panel by adding the following into site/blueprints/site.yml:

type:          fields
fields:
  feeds:
    label:     Syndication
    type:      toggle
    default:   off
    text:
      - RSS disabled
      - RSS enabled

If any entry for feeds is not present in content/site.txt, it is assumed that the plugin is not be be disabled (i.e. failsafe to false).

Feed discovery

At a minimum you need to add appropriate feed-discovery links to your site's homepage.

This plugin provides two snippets to dynamically generate the necessary meta data:

• feeds-header - to provide the links for all feeds (firehose and category-based)
• firehose-feeds-header - to provide the links for the firehose feed only.

Or for those who prefer to use pageMethods, the equivalent to the above are:

• headFeeds
• headFirehoseFeeds

Example:

In site/snippets/header.php - or whatever generates your <head> data - simply add <?php snippet('feeds-header') ?> or <?= $page->headFeeds() ?> in the best applicable place.

If the plugin is disabled, these snippets will not generate the appropriate feed-discovery links.

Collections

For the default (firehose) syndication feed, you need to ensure that you have a collection and have configured it. Typically this would be an 'articles' collection, and an example would be:

<?php

return function ($site) {
    return $site->find('blog')->children()->listed()->flip();
};

If using omz13/suncyclepages for embargo and withdraw, this would then be:

<?php

return function ($site) {
	    return $site->find('blog')->children()->listed()->isunderembargo(false)->issunset(false)->flip();
};

Use

For example, for the Kirby Starter Kit, the following would be applicable:

<?php

return [
  'omz13.feeds.cacheTTL' => 60,
  'omz13.feeds.firehose' => 'articles',
  ],
];

For experimental purposes this plugin implements a single-level pseudo-namespace. You can mix discrete vs array options, but try not to, and be aware that priority is given to the array variant. The above discrete configuration would therefore become:

<?php

return [
  'omz13.feeds' => [
    'cacheTTL' => 60,
    'firehose' => 'articles',
  ],
];

See Kirby3's issue #761 for more about namespaced options.

Disclaimer

This plugin is provided "as is" with no guarantee. Use it at your own risk and always test it yourself before using it in a production environment. If you find any issues, please create a new issue.