ship!

im getting tangled with exams , so i vibecoded the last few touchups w claude, like ipc integration, and the q quitting when the search was active, also a slight Ui change and made the results look good! i do plan on maintaining this project for longer and actually making it something big!

artwork

i wanted to get the artwork rendered in a ascii style, and after a bit of googling i found that braille rendering would be a good option as its sharp, then, i copiloted. a function for braille, as i wasnt really getting the logic behind it, then i wired the artwork and messed around with the left panel layout, but braille ddnt work well, then i generated a ascii one, tldr; that was broken asw. then i chose unicode with color, squared it out, displayed it across two seperate cards for alignemtn and clarity and it works pretty well! only need to work on the ipc connections, and i might have to vibecode alot of it now due to exams and ft ending .

progress bar and deezer

got all mpv instances to quit when q is clicked, and when a new song is played. also i vibecoded for the api to use the deezer endpoint as its actually a streaming servce unlike discogs which was for like vinyls, albums rather than singles, also, updated the ui, i have a UI design in mind, so ill sketch it, and also updated the ui layout and got progress bar working! i do plan on adding ipc next! i had to get duration just like i was sending search results with a sender and a reciever, also i love how rust does f strings!

tons of shit breaking

So, at first, i got the tui working using ratatui, i havent really used it properly, apart from a simple weather app, although it was really fun to build with! then, i checked out monochrome instances and added more playback and search bases just so if one fails there would be another to keep the app running, but, randomly, all of them 404d, api.monochrome got suspended and the qqdl endpoints all starting throwing 504s, so then i refactored the code to use yt-dlp to get music results, just like downloadarr, still using mpv to stream the music, getting results from yt-dlp was really slow, so then i found discogs in fmhy and decided to use it, so then i changed a few things in the api and added models for discogs which replicated the json it provides when you curl it, then i used claude to change the main.rs’s functionality to get the results accordingly, there was this issue with the results not splitting properly which was messing up the yt results, which is now fixed, and playback works! i tried using the download link so its easier, but seems like mpv can just play from the yt video link of the music! currently, when the app exits, the mpv instance is still running, but im way to tired to fix it today

Changelog

• using result yt url for playback
• fixed text splitting
• doscorgs and yt
• tui base

harp.rs - api intergration

I had a old python project that used hifi-api to get music to play, but that was really barebones and the api didnt work so well, so now im monochrome and qqdl to get music from tidal! built models for tracks, artists, search data, albums and whatnot, kind of like in swift, seperated the core and the cli packages like seperate files and function ins python! it is a bit harder to build in rust, as ive not really built a large scale project in rust yet!

Changelog

• added mpv integration
• cli and core api

ready to ship!

i forgot to mention the caching system which makes loading much easier and faster! also got the homebrew packaging to work and also uploaded the project to pypi for windows and linux! the caching system saves the file using a .json with versions, the json is encrypted asw using a key!

Changelog

• readme
• updates for versioning

Packaging

I really wanted to try to publish it to homebrew and get it to like a real cli tool, tried publishing it but kept getting issued with dependencies for the app itself. i plan on packaing for linux using pipx as its pretty simple to do so!

Changelog

• 0.1.1 release
• fix struct

Textual

made a textual app with proper rows and preview added! instead of the rich cli, also TIL that you can skip all the class init jargon using a dataclass also, removed the main, and had to make a shit ton of functions to get the picker to work preoperl.

Changelog

• removed main
• textual app
• arguments working

ready to ship

copied demo website from downloadarr, made a simple banner on photopea, linked helpme-site, and the time spent debugging on the exec (helpme-macos) to the project. also recorded a final demo!

helpme - executable

made a executable using pyinstaller, and the project would only work on macos, as it uses specific macos only utils like screencapture, and applescript! i hope the readme is well descripted, and makes it easy to run, fixed bugs wihth the release, it now works! a venv is needed with f5_tts_mlx installed, the readme better mentions it all.

Changelog

• packaing + readme
• remvoed ds store, added readme

tui interface

made a simple tui interface using rich, i did have to change the two files a tiny bit, to make imports from the cli work properly, i dont plan on really making this project more advanced, would probably ship after packaging this, also rich is fire.

Changelog

• remvoed ds store, added readme
• cli mvp
• clean up, working on tui

helpme

so, basically this is a tool which would make life very easy, I just take a screenshot of a question, then the question is passed on to hcapi, and it gets the answer, clones my voice, says the answer in my voice.

so far, i have the basic loop ready, where i take a screenshot (native macos screenshot ), and the image gets passed on to the hc api gemini model, and it gets the answer. and then the model says the answer. the model is a bit finnicky, but it works.

Changelog

• speak with f5 mlx
• setup main loop

whereisthis - an human based cli file finder

i wanted to build a tool that would make file finding easy, something like fzf, but enhanced, so it would work with human language, and give me the results and actions for what to do with the file. I did want to build this in rust, but then i ditched the rust idea, as it was getting to complicated and i wasnt understanding alot of the syntax and had to chatgpt alot of it, so i switched to python! built the cli func, and got the scanner to scan files and match to find results from the files

Changelog

• ditch rust, python setup
• scanner added walkdir imp
• cmd parsing and arguments
• init

docker app!

made it a docker app so its easy to run on home servers! completed the web UI, the downloads is a bit finnicky, as some downloads complete really fast and just update in the library, i do plan on adding a settings page, but currently, its pretty awesome, users can change the music dir and add spotify credentials in the .env or in the docker compose. i did have a problem with audio files not being found by yt-dlp and had to spam copilot, and found out it was a version issue with the requirements! ready to ship!

web app done! messed around with the UI alot, ended up to scratch the orignal idea and use amp to give it a hacker look, made the downloads tracking work, and got the artwork to show instead of a emoji on the main cards, also, this is one of the few times ive used js + python together, pretty cool though, it was a pain in the ass to get the library to work properly, because i previously used snake cases instead of camel cases which i was using in js! i do plan on making this a docker app!

got the js functions ready, the core functions were pretty similar to the swift ones, so they were pretty simple to js copy over, still need to complete the UI, want a minimal style, the functions work as expected! TIL how to get a static file on the / of a fastapi!

pausing work on the ios app cuz its becoming a pain in the ass, started work on a web interface, built the simple html nav, will work on the main func tomm, made a simple ui, wanted to keep it simple!

made a simple donwload manager, which would create a job, then send it to the api, and then track progress, then it would ping the /library endpoint and check if the song is downloaded, and show completed! i couldnt get the download view to update, even when the song was downloading, so used copilot to fix that.

downloadarr-ios

started working on the ios app a few days ago, but didnt make a lot of progress to devlog it, tday i got the models down. pretty simple as they were just like the python fyunctions, then worked on the API’s, currently the api is hardcoded, but ill start working on the download manager and settings soon! currently the app is connecting with the api, and getting deezer queries and listing them! i also used sent a color pallete to claude, and asked it to generate a theme with button styles and it worked so well! saved a ton oftime!

set up a server, i plan on making a ios app for it, where the user enters the ip, and i also plan on making this an docker app, so its easy to setup on servers, rn the server is pretty simple, just called functions from the main file and exposed endpoints.

made a demo website, with a video demo and github release page, also used github actions for the first time to make excetuables from pyinstalled for macos, windows and linux, i couldnt get mac intel working, used copliot for the yml.

ready to ship! updated readme for installation and shipping and removed the test code from the main file

worked on interface, and got a mvp, i do plan on exposing the functions to a api, then making an ios app! but for now, this works pretty well, and is ready to ship! now the app has a terminal interface, you can set up a spotify app and then export your client secret and id and the app would recognize it, but it does use deezer as a fallback, which works well, you can also just skip the metadata entirely! overwrite functions also work, and orgnanize works too! js need to pull this on the pi now!

debugged and rewrote the download audio functions after tons of ai slop copy and paste, couldnt get proper downloads to work, seemed to get node and cookie errors, also was not getting audio files for downloading, just got artwork in sdb’s tagging also works, just need to make a organize function, and a frontend!

basic setup, got the spotify search working, it is a bit finnicky, asking for a premium account, and i might implement deezer, cus it seems to have good coverage also got tagging ready acc to jellyfin’s metadata

basic project setup, first got the epub’s raw data and chapter extraction. then got the data scraped and then presented using textual. need to fix the scrolling though.

worked on the setup screen ui, i also added the color assets and the font, just took assets from the site, and then messed around with the layout, also made the error UI, but currently the loading data is pretty slow, and i think its because of the way that i am paginating and searching for the name.

app init, connected the api, also, i stole a bit of code from my hackatime-ios project, keychain and api to be specific, then, wired the api to flavortown, and then to the home view, i didnt expect pagination, but it was fun to implement in the app! getting a summary, but the UI is really demo right now, using a navigation stack so i dont have to worry about alignments, will change. but im getting an summary of the user!

added a few touch ups like auto hiding for passwords and a open meme button, which messed with the css, but i did end up adjusting it. also packaged and uploaded to pypi! and now vaultic is a package! also updated the readme, ready to ship!

implemented basic crud functions, i added update functions and vault functions to the vault. and then simply wired them to the get screen, i did avoid making another screen, since i wanted to keep the app minimal, but, without another screen, the app was just becoming more congested and complicated. so i made a seperate update screen, for the user to change the password of a specific service. the delete function works the same way, with it getting the master password, and verifying and deleting the password, i did have to use a bit of chatgpt, since i was getting issues with how to get the current pwd field showing and then toggling and all that, but i just made it reveal at the end.

added a generate password function so a secure password can be made, using secrets + string instead of random… also i added copy functions to both the store screen and get screen so the user doesnt have to reveal just to get the password, i also need to add managing passwords, as in deletion and edition.

made a meme preview to make the app more appealing, used rich-pixels since textual doesnt allow image rendering, the image looks a bit wonky, but thats how it looks pixelated, i tried to fix it by making the image bigger, and it did work a bit. but it slowly became in the pain the bum bum to auto size, which i did end up adding, so the app is more responsive, another fix that i added was that the app now automatically refreshes the list of saved passwords when the getscreen is loaded

added a multiscreen view, and learnt that textual screens work alot like swiftui! also, i setup a get password system, which lists the current passwords listed and then lets you verify with your master password to get your pass, not instead of the vault being stored in vault.bin, it is now saved in a meme! the meme is randomly pulled from meme-api.com/gimme, and is forced in png for stego to work! i also implemented this unlock system so its actually like a vault!

made a basic textual app and its proper input fields with minimal css, then i added the master password and vault logic, so now to add a password, you use your master password, which has a vault and is stored with an salt in the home directory. it uses cryptography for the encryptions and decryption! need to add the meme integration and make the other passwords more encrypted. currently i have loading meme commented, since it just slows the current flow, but will add meme integration soon.