Log in
Seblog.nl

Seblog.nl

Day 22: read posts

In December I already posted some plain text notes tracking my reading progress in Stoner by John Williams. (The Dutch translation, by the way, because I got that one somewhere in a very portable version and I was on a city trip to Vilnius.) I already planned to upgrade those plain text notes to a more standarized format, so that I would have /read posts.

Before I could add them, though, I needed a way to give them URLs. So now I have this public list of books I own / owned / want to track, on my new library page: /bieb. All books in there are marked up as h-cites and have a url in the format https://seblog.nl/isbn/9780123456789. Their u-uid is set to isbn:9780123456789, because that seems to be a valid url scheme (but browsers don't know what to do with it).

The next step was simple: I just expanded my shortpost type with a 'read' type, that looks for a read-of field. Thanks to XRay and my existing code, the seblog.nl/isbn/-url gets expanded to a title and author. I also added a page field, because I posted when I was at certain points at the book, and I actually haven't finished it yet.

Now my new /gelezen page looks like this:

Inzichten op de late avond: het probleem is niet dat ik niet op Facebook zit. Het probleem is dat jullie er allemaal wel op zitten, en dat normaal vinden.

En, min of meer door Tegenlicht van vanavond: zwart-wit denken los je niet op met grijstinten. Het tegenovergestelde van zwart-wit is kleur, een totaal andere manier van kijken.

En dat gaat dan dus niet over ‘Trump denkt zwart-wit’ of ‘mensen die Trump erg vinden hebben een psychose’, want dat is in zichzelf nog steeds zwart-wit. Dat relativeren (wat ik net deed) is denk ik een grijstint. Kleur is de nuance proberen te zien.

Day 21: basic Micropub endpoint

Today I added a very basic Micropub endpoint to my Indieweb Toolkit. A Micropub endpoint is a hard thing to keep generic, because every site has a different way of storing data. The idea about the endpoint in the toolkit is that it does all the things that are the same about Micropub.

So here's an example of what it can do. This one simply writes a YAML file with the received fields to a folder. Please note that it overwrites any post with an existing slug. It's only an example.

endpoint::micropub([
  'create' => function($type, $fields, $slug) {
    $map = [
      'entry' => 'blog',
      'event' => 'events',
      'review' => 'reviews',
    ];

    if(!isset($map[$type])) return false;
    if(!$slug) $slug = uniqid();

    yaml::write(__DIR__ . DS . $map[$type] . DS . $slug . '.yml', $fields);

    return url::makeAbsolute('/'.$map[$type] . '/' . $slug . '.yml');
  }
]);

The idea is that you can put this code where-ever you want. It can be at /micropub.php or in some route you define. The new endpoint::micropub() does all the authentication for you and returns the appropriate header. All you need to do is pass in a 'create' callback function, which takes the fields and returns a URL. endpoint::micropub() will do the redirect with a 201 header.

At this moment it only supports 'create' with x-www-form-urlencoded, but JSON and 'delete', 'undelete' and 'update' will follow. I also want callbacks for different mp-synticate-tos, and of course there is the ?q=config query. But I like this idea, and will use this idea to clean out the code of my own endpoint.

Last caveat: at this moment I have not implemented the IndieAuth class in my toolkit. You can steal the class from my kirby-micropub plugin, but I want to clean it first before putting it in the toolkit.

Day 20: automatic liking on Twitter

Today I drafted a SiloAPI class in indieweb-toolkit. The idea is that it gives an easy way to interact with different Silo API's. In some ways this is a bad idea, because all API's work different and have different capabilities, but I like the idea, even if it's bad.

One API call that is available for both Twitter, Instagram and Facebook is to create a like. So that's where I focussed on first. Today, I made SiloAPI a wrapper for the TwitterOAuth PHP library, and it now supports ::like() just like micropub::like(). You still need to pass in the name of the silo though, but maybe I'll detecting the silo from the URL later.

siloapi::setToken('twitter', [
  'consumer_key'       => 'xxx',
  'consumer_secret'    => 'xxx',
  'oauth_token'        => 'xxx',
  'oauth_token_secret' => 'xxx',
]);

siloapi::like('https://twitter.com/_/status/123456', 'twitter');

Then I realised that Brid.gy probably can do this for me too, since it already syndicates my Tweets for me when I ask for it. So I just made my Micropub endpoint call Brid.gy publish for every like-of that is a twitter-URL.

I will add to SiloAPI though, because I want to autolike to Instagram too!

Day 19: reposts again

Today is a weird day. Fixed a lot of things, but also did kinda useless things. (I learned more about HTTP requests by trying to create a Micropub request using netcat, and at some point, I was chatting with myself in two Terminal windows with nc -l localhost 8080 in one and nc localhost 8080 in the other, completely by accident.)

Then it was 1:30 at night.

Earlier on in this series, I fixed my repost context. These contexts where based on a .json file I exported with my whole Twitter archive. A few days ago I made XRay look up likes for me. Today, I retweeted someone, and since .json files don't magically appear on your server (luckily), it didn't show any repost context.

So I combined the two and now XRay fetches the repost context for me. I don't use the .json file anymore, but keep the data the same way I keep likes: in a refs field.

And then it was 2:00 at night. I'm going to bed.

Day 18: Micropub helper class

Today I extended my new Indieweb Toolkit with a Micropub helper class. This allows you to easily send Micropub requests from your PHP code. Here’s an example:

// Set URL and access token of the endpoint to use
micropub::setEndpoint('http://yoursite.com/micropub', 'xxx');

micropub::reply('http://example.com/a-nice-post', "Oh what a post!");
micropub::like('http://example.com/another-post');
micropub::rsvp('http://example.com/an-event', 'maybe');

$newURL = micropub::post([
  'name' => 'Custom posts are possible!',
  'content' => 'This is a story about (...)',
  'category' => ['story', 'custom'],
  'mp-slug' => 'custom-posts-are-possible',
  'mp-syndicate-to' => 'https://twitter.com/example',
]);

go($newURL);

Note that this only creates posts, and relies havily on the server you are making requests to. The mp-syndicate-to in the above example indicates that the remote server needs to post the entry to Twitter also, this helper doesn’t do a thing for that.

A drop-in Micropub endpoint is hard, because every site stores it’s data different, but I intend to extend this toolkit so that the different parts involved will be easier. (De facto making my own code more re-usable.)


Posted this post with use of todays code :D

Day 17: okay, what now?

It’s only Day 17, and while there is allways stuff to do, I feel like I got to a point where everything more or less works like I think it should. To put it in another way: most of my itches have been rubbed.

At the same time, I feel like I have implemented a lot of stuff others have already build in some other form. Apart from my questionable new post type, I have not really build new things. And that’s no problem: nothing is entirely new, all things build on top of other things. My problem with it is that I keep building things other people build before me because they build it. I need to step back and think again: do I need this?

Another problem with the challenge is that I did a lot of ‘small’ things already, but keep seeing only big things on my list. This is both a problem (I keep postponing the big things) and the solution (this challenge forces me to break things down!).

There are a few general areas where I want to make progress that are just a bit to big:

  • Webmentions – I have them working, but I made some changes to the original Kirby plugin that might not be for everyone. I kind of want to start a new basic plugin, providing just sending and receiving Webmentions, possibly including a Panel Widget too (although I don’t use the panel myself). And then offer an extension to that plugin with Microformats parsing and display of comments.

  • Micropub – I wrote a Micropub plugin for Kirby, and it’s only the best out there because it’s the only one out there. I branched off the main branch to do some guild free drafting of the [update](https://www.w3.org/TR/micropub/#h-update) function. My own site uses the latest commit of that branch, the code is horrible and I have never merged it. The problem here is that there is quite a bit of technical debt in that plugin. It needs fixing anyway, although it does work for my site.

  • Importing old data – I’m just postponing on this one because it’s not very spectacular. But I think it’s okay to import some stuff every now and then and call it a day. I still have Strava, Hyves, my old Facebook, old blogposts from before april 2009 and three Vines. The problem here is that my current data is not really clean. I have tweets and blogposts that could be deduped, and the utf-8 conversion for some posts seems to be weird. Cleaning data doesn’t sound like a good thing within this challenge, so I’m postponing importing all together.

  • Private posts – This one might not be as big as the others in terms of work, but is does come in steps. I want to use IndieAuth, but also give Silo-friends a way of seeing posts, via Twitter OAuth and the like. Oh and I might want to write my own Authorization / Token Endpoint for Micropub, but that’s whole other point.

  • Multilingual / multi topic – This was an itch for me, but this is one of the things I’ve pretty much fixed for the moment. I have multiple feeds of posts, tags, and thanks to indexing it goes fast as well. I might open different blogs on different domains, at some point, and use them like I use Twitter now: post on Seblog and push the post to those blogs. But that’s not a real itch for the moment. It can wait.

All in all there is enough to do, but none of the above things fit in a ‘today I fixed my X’. I need to start doing ‘today I fixed Y of my X’.

As a start of these things, I made a first version for an Indieweb Toolkit. It’s inspired by and makes use of the Kirby Toolkit. I want to put some basic Indieweb stuff in this thing, so I can re-use it for different projects. At this moment it only consists of a Webmention Endpoint discovery function and a wrapper for the php-mf2 Microformats Parser, but more to come!

Day 16: Backing up my Gmail

With the recent events in the USA, I’ve decided that I want to move away from having data on American servers as much as possible. The first thing I want to tackle in that area is my e-mail. Although I do have an address at seblog.nl, I still just redirect it to Gmail.

My computer is currently in the process of backing up my main Gmail. I documented how I do it on the Indieweb Wiki:

Gmvault seems to be very simple and straight forward. It's on the command line, so it's scary for some users, but it does a good job of describing what it does. I did the following on my Mac, and since I can't remember installing pip, I think this works out of the box:

  • sudo pip install --upgrade pip
  • sudo pip install gmvault
  • gmvault sync example@gmail.com
  • Gmvault prompts for OAuth, with a description. Press enter to open the browser, and you have to make sure you are logged in at that browser to the Gmail account.
  • Do the OAuth in de browser and copy the key. Paste it in the Terminal
  • Gmvault does things! I got 6351 mails out of an old account in 16m 14s. It creates a folder called 'gmvault-db' in your home folder, with (in /db/) folders for every month. In those folders are, per e-mail, an '[id].meta' and an '[id].eml.gz'. The .meta is a JSON with info from Gmail (labels/tags, subject) and the .eml.gz is a gzipped .eml, which is just the plain-text e-mail with all the headers.

Having the data is just one step. I will need to think about how I want to manage my e-mail in the future. For now I’m on Gmail still, but I am making plans.


To make today a bit more IndieWeb-relevant (e-mail is not web), I backed Micro.blog, because today is the last day of their Kickstarter campaign.

Although I really like that the project gives Indieweb a lot of attention, it felt wrong to only give to Micro.blog. So I also backed Aaron Parecki with the same amount, for his ongoing 100daysofindieweb. I use a lot of things he made or did first.

Meer laden