A developer's toolkit for OPML support. Node and browser-based JavaScript code that reads and writes OPML.

OPML is an XML-based format designed to store and exchange outlines with attributes.

It's been around since the early 2000s, and is widely used in the RSS world to exchange subscription lists.

It's also a standard for interop among outliners. If you support OPML, our products will interop, and our users will be able to use all our products on their data.

I wanted to make it really easy for developers to add basic OPML support to their apps.

So I put the basic code you need to read and write OPML files, code that's debugged, used in other apps, including my own. It's known to work, to respect the format, and be easy to deploy.

There are other ways to read OPML, and that's very good. The more support there is, the more interop and that's the goal. I'm going to accumulate links to resources for OPML developers. If you have something you think they could use, send me an email at the address in the package.json file above.

I recorded a podcast explaining all this. :-)

JavaScript code to parse and stringify OPML.

• opml.parse -- turns OPML text into a JavaScript structure representing the OPML.

• opml.stringify -- takes the JavaScript structure and turns it into OPML text.

• opml.htmlify -- a simple routine to display outlines in HTML.

• opml.visitAll -- a routine that visits all the nodes in an outline.

It's on NPM, it's called OPML.

Here's a snippet that reads an OPML file, converts it to a JavaScript object, displays it to the console via JSON.stringify.

const fs = require ("fs");
const opml = require ("opml");
fs.readFile ("states.opml", function (err, opmltext) {
	if (!err) {
		opml.parse (opmltext, function (err, theOutline) {
			if (!err) {
				console.log (JSON.stringify (theOutline, undefined, 4));
				}
			});
		}
	});

The full Node example is here.

The same routines are available for JavaScript code running in the browser.

See the example. You have to include opml.js in your app, as the example does.

You can run the example right now, without downloading the repo.

The example app includes the Bootstrap Toolkit, the Ubuntu font and jQuery. The OPML parsing and generating code does not depend on the first two, they're just used in the example app. opml.js uses jQuery, but otherwise should be completely self-contained, i.e. it does not depend on any other files.

I have a few OPML-related projects on GitHub and on the web.

• Drummer -- Browser and Electron-based outliner that uses OPML as its native format.

• OPML Developer checklist. Examples, source code and advice for people adapting outliners to read and write OPML.

• instantOutlines -- Example code for sharing live-updated outlines between users, using websockets as the notification system.

• OPML 2.0 spec.

• etc.

New function -- opml.readOutline. A simple bit of recurring code. Reads an outline over the web returns a standard outline object.

It's time to start a fresh sequence of versions with 0.5.0. No breakage, of course. ;-)

New function -- opml.expandIncludes.

Takes two params, an outline that could possibly have include nodes, and a callback, that receives a copy of the outline with includes expanded.

It doesn't stop for errors. This was much-debated internally, but there is linkrot and an outline with includes should work as well as it possibly can, as a blog with one broken link still works.

It is only available in the Node version, but it could be adapted to work in the client if opml.expandInclude (sic) is converted.

BTW, sorry for the closeness in the names, expandInclude and expandIncludes, but it is correct, one is singular and the other is plural.

opml.visitAll is now defined in both the Node and browser versions. Previously it was only defined in browser version.

New optional param on markdownToOutline, options, an object.

And options.flAddUnderscores, defaults true. If true, we add underscores before attribute names coming from the markdown, so we know to restore them when converting back to markdown.

But sometimes the outline is going to OPML, and on to a processor where you want it to recognize its name without the underscore.

When we're publishing a blog from a LogSeq markdown outline is an example.

There was some confusion about whether or not we should try to handle head-level atts in the markdown format, and in the end I decided not to try to do that at this time. Before doing this I have to understand much better what's going on on the other side of the interop. At this stage, whatever I do is going to be wrong, and will have to be grandfathered in for perpetuity. We have a pretty good ability to interop on the content of the outlines, but different products see the file-level metadata very differently. If there ever is an agreement on how this should work it's going to happen later.

However I did fix a problem, if a head-level att does appear as we import, we don't try to attach it to undefined.

Added expandInclude in Node package.

Last night's release only worked in the client version. The Node package was broken. It should now be fixed.

Also added a new example app that demonstrates the reading and writing of markdown/outline files in a Node app.

Two new routines, opml.markdownToOutline and opml.outlineToMarkdown, to read and write markdown files that are used to represent outlines. This is an extended Markdown that LogSeq generates. The format does not have a name at this time, or as far as I know, a spec, but at least now there is JavaScript code that reads and writes the format.

We are using this code in a new version of Drummer in the works.

It is being discussed in this thread.

New entry-point in the client, opml.read.

Reads an OPML file, returns a JavaScript object with the outline head and structure.

If options.flSubscribe is true, we ask to be notified when the file changes over a websocket.

We call back to the same routine we called when the file was read, assuming it will do the same thing with the updated OPML.

If you have any questions or comments please post an issue here.