Replied to Yes to ActivityPub, but no to Friends by Shelley Powers (Burningbird)
I decided to disable the Friends plug-in when I realized it was inserting every new feed item as a new post in my database. This could easily become unmanageable. Considering you can use a feed reader to read weblogs AND Mastodon accounts, it just didn’t seem worth the database burden.

I’ve also been mess­ing with the Friends and ActivityPub plu­g­ins for WordPress on my blog, and I share Shelley’s con­cerns about the for­mer bloat­ing the data­base with feed items. You can con­trol this some­what by set­ting reten­tion val­ues in days or a num­ber of posts, but you have to go into each friend’s Feeds tab and do it manually–there’s no default setting.

After read­ing that post, I’m also con­sid­er­ing dis­abling Friends in favor of a feed read­er, espe­cial­ly because (as Shelley also not­ed) there are gaps when with favorites and com­ment con­ver­sa­tions bridg­ing between WordPress and Mastodon servers. Like her, I’m not keen on installing a single-​user Mastodon instance or oth­er fedi­verse serv­er that requires man­ag­ing an unfa­mil­iar pro­gram­ming language.

I’m also try­ing to do this in tan­dem with a suite of IndieWeb plu­g­ins, and I’m run­ning into an issue with my friends feed page not show­ing any posts when the Post Kinds plu­g­in is acti­vat­ed. I real­ly want to keep this plu­g­in because it lets me inter­act bet­ter with oth­er IndieWeb sites as well as the Bridgy POSSE/​back­feed ser­vice con­nect­ing me to oth­er social networks.

My ide­al is a per­son­al web­site where I write every­thing, includ­ing long-​form arti­cles, short sta­tus­es, and replies like these. Folks can then find me via a sin­gle iden­ti­fi­able address and then subscribe/​follow the entire fire­hose of con­tent or choose sub­sets accord­ing to post types, top­ics, or tags. They’d then be able to reply or react on my site or their favored plat­form, which my site would col­lect regard­less of ori­gin, with sub­se­quent replies and reac­tions get­ting pushed out to them. Oh, and it should work with both ActivityPub clients and servers, IndieWeb sites, and syndicate/​backfeed to oth­er social net­works either with or akin to the Bridgy ser­vice I men­tioned above.

So far I haven’t seen any­thing that ticks all these box­es, and I’m get­ting itchy to write my own. Perl is my favorite pro­gram­ming lan­guage, so I’m look­ing at the Yancy CMS as a base. But I know that it would still be a hell of a project, and one of the rea­sons I chose WordPress for blog­ging was that it was well-​established and ‑sup­port­ed but still eas­i­ly exten­si­ble so that I could con­cen­trate on writ­ing instead of end­less­ly tweak­ing the engine. Unfortunately, I’m start­ing to fall into that trap anyway.

Quoted 5 Reasons Why Using AI to Generate Blog Posts Can Destroy Your SEO by Dave Cross (Medium)
Using artificial intelligence (AI) to generate blog posts can be bad for search engine optimization (SEO) for several reasons. First and foremost, AI-generated content is often low quality and lacks…


(reads last line)

Wait a minute…

Liked snarfed/bridgy (GitHub)
Connects your web site to social media. Likes, retweets, mentions, cross-posting, and more... - GitHub - snarfed/bridgy: 📣 Connects your web site to social media. Likes, retweets, mentions, cross...

Been wiring up this blog for IndieWeb and Bridgy is the spe­cial sauce that con­nects it to more siloed social networks

IKEA BLÅHAJ shark toys

IKEA’s toy BLÅHAJ shark has become a beloved Internet icon over the past sev­er­al years. I thought it might be cute to write a lit­tle Perl to get info about it and even dis­play a cud­dly pic­ture right in the ter­mi­nal where I’m run­ning the code. Maybe this will give you some ideas for your own quick web clients. Of course, you could accom­plish all of these things using a pipeline of indi­vid­ual command-​line util­i­ties like curl, jq, and GNU core­uti­lsbase64. These exam­ples focus on Perl as the glue, though.

Warning: dodgy API ahead

I haven’t found a publicly-​documented and ‑sup­port­ed offi­cial API for query­ing IKEA prod­uct infor­ma­tion but oth­ers have decon­struct­ed the company’s web site AJAX requests so we can use that instead. The alter­na­tive would be to scrape the IKEA web site direct­ly which, although pos­si­ble, would be more tedious and prone to fail­ure should their design change. An unof­fi­cial API is also unre­li­able but the sim­pler client code is eas­i­er to change should any errors surface.

Enter the Mojolicious

My orig­i­nal goal was to do this in a sin­gle line issued to the perl com­mand, and luck­i­ly the Mojolicious framework’s ojo mod­ule is tailor-​made for such things. By adding a -Mojo switch to the perl com­mand, you get over a dozen quick single-​character func­tions for spin­ning up a quick web appli­ca­tion or, in our case, mak­ing and inter­pret­ing web requests with­out a lot of cer­e­mo­ny. Here’s the start of my one-​line request to the IKEA API for infor­ma­tion on their BLÅHAJ prod­uct, using ojo’s g func­tion to per­form an HTTP GET and dis­play­ing the JSON from the response body to the terminal.

$ perl -Mojo -E 'say g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->body'

This cur­rent­ly returns over 2,400 lines of data, so after read­ing it over I’ll con­vert the response body JSON to a Perl data struc­ture and dump only the main prod­uct infor­ma­tion using ojo’s r func­tion:

$ perl -Mojo -E 'say r g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}'
  "availability" => [],
  "breathTaking" => bless( do{\(my $o = 0)}, 'JSON::PP::Boolean' ),
  "colors" => [
      "hex" => "0058a3",
      "id" => 10007,
      "name" => "blue"
      "hex" => "ffffff",
      "id" => 10156,
      "name" => "white"
  "contextualImageUrl" => "",
  "currencyCode" => "USD",
  "discount" => "",
  "features" => [],
  "gprDescription" => {
    "numberOfVariants" => 0,
    "variants" => []
  "id" => 90373590,
  "itemMeasureReferenceText" => "39 \x{bc} \"",
  "itemNo" => 90373590,
  "itemNoGlobal" => 30373588,
  "itemType" => "ART",
  "lastChance" => $VAR1->{"breathTaking"},
  "mainImageAlt" => "BL\x{c5}HAJ Soft toy, shark, 39 \x{bc} \"",
  "mainImageUrl" => "",
  "name" => "BL\x{c5}HAJ",
  "onlineSellable" => bless( do{\(my $o = 1)}, 'JSON::PP::Boolean' ),
  "pipUrl" => "",
  "price" => {
    "decimals" => 99,
    "isRegularCurrency" => $VAR1->{"breathTaking"},
    "prefix" => "\$",
    "separator" => ".",
    "suffix" => "",
    "wholeNumber" => 19
  "priceNumeral" => "19.99",
  "quickFacts" => [],
  "tag" => "NONE",
  "typeName" => "Soft toy"

If I just want the price I can do:

$ perl -Mojo -E 'say g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}->@{qw(currencyCode priceNumeral)}'

That ->@{qw(currencyCode priceNumeral)} towards the end uses the post­fix ref­er­ence slic­ing syn­tax intro­duced exper­i­men­tal­ly in Perl v5.20 and made offi­cial in v5.24. If you’re using an old­er perl, you’d say:

$ perl -Mojo -E 'say @{g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}}{qw(currencyCode priceNumeral)}'

I pre­fer the for­mer, though, because it’s eas­i­er to read left-to-right.

But I’m not in the United States! Where’s my native currency?

You can either replace the us/en” in the URL above or use the core I18N::LangTags::Detect mod­ule added in Perl v5.8.5 if you’re real­ly deter­mined to be portable across dif­fer­ent users’ locales. This is real­ly stretch­ing the def­i­n­i­tion of one-​liner,” though.

$ LANG=de_DE.UTF-8 perl -Mojo -MI18N::LangTags::Detect -E 'my @lang = (split /-/, I18N::LangTags::Detect::detect)[1,0]; say g("" . join("/", @lang == 2 ? @lang : ("us", "en")) . "/search-result-page", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}->@{qw(currencyCode priceNumeral)}'

Window dressing

It’s hard to envi­sion cud­dling a num­ber, but luck­i­ly the prod­uct infor­ma­tion returned above links to a JPEG file in the mainImageUrl key. My favorite ter­mi­nal app iTerm2 can dis­play images inline from either a file or Base64 encod­ed data, so adding an extra HTTP request and encod­ing from the core MIME::Base64 mod­ule yields:

$ perl -Mojo -MMIME::Base64 -E 'say "\c[]1337;File=inline=1;width=100%:", encode_base64(g(g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}{mainImageUrl})->body), "\cG"'

(You could just send the image URL to iTerm2’s bun­dled imgcat util­i­ty, but where’s the fun in that?)

$ imgcat --url `perl -Mojo -E 'print g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}{mainImageUrl}'`

But I don’t have iTerm2 or a Mac!

I got you. At the expense of a num­ber of oth­er depen­den­cies, here’s a ver­sion that will work on any ter­mi­nal that sup­ports 256-​color mode with ANSI codes using Image::Term256Color from CPAN and a Unicode font with block char­ac­ters. I’ll also use Term::ReadKey to size the image for the width of your win­dow. (Again, this stretch­es the def­i­n­i­tion of one-​liner.”)

$ perl -Mojo -MImage::Term256Color -MTerm::ReadKey -E 'say for Image::Term256Color::convert(g(g("", form => {types => "PRODUCT", q => "BLÅHAJ"})->json->{searchResultPage}{products}{main}{items}[0]{product}{mainImageUrl})->body, {scale_x => (GetTerminalSize)[0], utf8 => 1})'

I hate Mojolicious! Can’t you just use core modules?

Fine. Here’s retriev­ing the prod­uct price using HTTP::Tiny and the pure-​Perl JSON pars­er JSON::PP, which were added to core in ver­sion 5.14.

$ perl -MHTTP::Tiny -MJSON::PP -E 'say @{decode_json(HTTP::Tiny->new->get("ÅHAJ")->{content})->{searchResultPage}{products}{main}{items}[0]{product}}{qw(currencyCode priceNumeral)}'

Fetching and dis­play­ing a pic­ture of the hug­gable shark using MIME::Base64 or Image::Term256Color as above is left as an exer­cise to the reader.