PR welcome to jet to mention this tool :)

Thanks, I'm working on that exactly now :p

https://github.com/borkdude/jet/pull/125

Amazing stuff 👏

Can I somehow make keybindings for common pairs of options?

Eg, make an interactive command that sets input to JSON, output to EDN, and copies to the clipboard

This is possibly feedback to jet more than this glue, but it would be cool if jet was able to guess the input automatically - seems like it should be possible?

it's possible indeed, I'll add a example in the readme

👍

guess the input is a little bit harder I guess

I guess all json is valid transit, so how would you guess the difference?

try parsing it as transit first?

Sorry, I misunderstood

If transit can read both transit-specific JSON and just regular JSON, then there is no big problem? Default behavior with no -i could be to try to parse as edn, then as transit, then as yml.

I guess you could do this in emacs first and then see how it works out

nice I use jet extensively for formatting in emacs and converting between json and edn, I will go install, @U04V15CAJ curious if it would be possible to add in other map types like say python dict's as a quick way to convert, between languages I guess custom parsers would be needed ?

@UU67HFS2X What does that look like?

pretty similar to javascript maps: {'foo': 'bar'} IIRC

in that case converting from JSON should already work?

except that JSON doesn't support single quotes I guess

(defun jet-to-clipboard (thing &optional args)
  "Run jet for THING at cursor and ARGS copying to clipboard."
  ,,,)

Got me wondering now, sure I would have tried the json version, need to go find a python map now and try that

I think where the python map fails is on the values, so you can have functions for example, the json parser does not like those, there maybe other data values but I imagine most are compatible

Usually (jet--thing-at-point) @U9MKYDN4Q, the region selected or current form at point

Yeah, just found it 😅 Thanks 🙏

@U9MKYDN4Q I improved the readme mentioning how to customize the functions so you can keybind

Cool, thanks

I ended up with these:

(require 'jet)

(defun copy-edn-as-json ()
  (interactive)
  (jet-to-clipboard
   (jet--thing-at-point)
   '("--from=edn" "--to=json"))
  (keyboard-quit))

(defun copy-json-as-edn ()
  (interactive)
  (jet-to-clipboard
   (jet--thing-at-point)
   '("--from=json" "--to=edn" "--keywordize"))
  (keyboard-quit))

(global-set-key (kbd "C-c j e j") 'copy-edn-as-json)
(global-set-key (kbd "C-c j j e") 'copy-json-as-edn)

New Readme will save the next person some time 😄

Anyway, amazing stuff, thanks for your effort 👏

Added your example to the readme how to keybind, thank you!

this is honestly so damn useful @UKFSJSM38 thank you (and thanks again to borkers!)

nice! what lead you to write it? 🙂

Thanks! I'm releasing a library in the next couple of days for building Re-frame/React apps that includes a visual debugger that overlays graphical elements onto the live application. I needed a simple clustering algorithm for collecting and grouping certain visual elements on the screen.

Would you consider to "bind" your library closer to the upcoming Clojure data science stack by using a "tech.ml.dataset" (a column oriented table) as input data, instead of vector-of-maps TMD is living here: https://github.com/techascent/tech.ml.dataset If this is the case, you could as well considre to integrate the DBSCAN algorithm in one of the existing" collections of algorithms working on the same data format.

Sure, I actually use tech.ml.dataset quite regularly, it's fantastic. But never thought to connect the two! I'd have to think about what that would entail, are there examples in the code-base of other grouping or clustering bindings I could look at as a pattern to follow?

It should be very easy . The "core" of your algorithm could even stay as is. You would just convert from dataset to vector-of-maps, run your stuff and convert back to a tech.ml.dataset This might come with small performance penalty (versus rebuilding the algorithm using the TMD primitives), but probably negligible. So you just need a functions like yours : https://github.com/zalky/dbscan-clj/blob/bb7aa0fae89cd9c8477278b0aae70e443517284c/src/dbscan_clj/core.cljc#L88 but it takes a "tech.ml.dataset" as input. (and reurn such dataset as well), so things can be chained easely. Probably you need some options to define on "which columns" your clustering is performed.

This new code would be a kind of "adapter" to TMD. So where "physically" place your original code and teh adapter is a question to answer. Doing the adapter will require a dependency to tech.ml.dataset , so maybe you code in https://github.com/zalky/dbscan-clj wants to avoid that.

I would maybe even suggest to have the "adapter" standalone, depending on

io.zalky/dbscan-clj {:mvn/version "0.1.0"}

techascent/tech.ml.dataset {:mvn/version "7.000-beta-27"}

(one reason being ClojureScript compatibility). TMD is a Clojure library, not compatible with ClojureScript

Yes, I agree with these points. I'll take a bit of a closer look to see what alternatives there are for integrating with tech.ml.dataset's abstractions. There might be some protocols or interfaces that are easy to implement. Thanks for planting the seed, I don't think I would have made the connection otherwise.

what's a Turtle in this context?

https://www.w3.org/TR/turtle/

ahhh ok

I really don't know much about RDF

thanks 😊

maybe I should read about it

It’s been my career for over 20 years, so… 😄

so ask you instead?? 😜

I’m more than happy to help. But if you want something to read, you can start https://www.w3.org/TR/rdf11-primer/

cool name 🙂

Well, the partner library (for parsing Turtle files) is called https://github.com/quoll/raphael

makes sense 🙂