mdFilePattern instead or additionally to mdExtensions

[wenigwil]

What I would like to suggest is a new configuration option that will make the filetype detection of Vivify a little bit more configurable. I guess one could call it mdFilePattern such that it will enable the user to use a regular expression pattern in the determination what should be rendered as Markdown. Much like the filetype detection in Neovim with its pattern-argument of vim.filetype.add() just that it is fixed to the filetype markdown

Where could this be useful?
A limitation arises for me when I want to use Vivify together with calcurse which is a handy text-based calendar. You can have notes for every item you can put into calcurse. I edit those notes with nvim and would like to use vivify.vim to get my note rendered in Markdown on side :) BUT the notes, for the time of the edit, exist temporarily as /tmp/calcurse-note.Vf21ma (the file suffix are 6 alphanumeric characters). With mdFilePattern I could solve the filedetection just as I did in nvim by providing the a pattern for the whole path of the file.

[jannis-baum]

Hello! Is the suffix just any 6 alphanumerical characters? Not sure if you've already seen it but currently we have the config option "mdExtensions" to configure a list of fixed extensions that are considered as Markdown.

In case that doesn't do it for your use-case it'll be very straightforward to swap out the code that checks these literally to do pattern-matching instead. I could probably do that relatively soon but if you're interested in contributing and/or want it faster you're very welcome to give it a shot yourself! :) If you search for mdExtensions in the code you'll find what you'd have to modify

[wenigwil]

Thank you for the quick reply!
The suffix are any 6 alphanumeric characters. But for every built note in calcurse they are different. The lua-pattern in vim.filetype.add() I used to match the path was therefore

.*/calcurse%-note%.[%w]+

,where the modified character class [%w]+ stands for any alphnumeric character repeated more than once.

Unfortunately the mdExtensions option does not help unless it is possible to supply it with a pattern. I don't really know anything about TypeScript but I looked at src/parser/parser.ts and in the renderBody() function the fileEnding is split from the path and then checked using config.mdExtensions.includes(fileEnding). From what I looked up that does not support regex-patterns :(

It would be amazing if you can implement it, as soon as you see fit! I think it would make a great addition to Vivify :)

[jannis-baum]

Hey I implemented this in #285. For your use case you can set your config to the following which will match the whole path with exactly 6 alphanumerics at the end:

{
    "mdFilePatterns": ["^.*/calcurse-note\\.[a-zA-Z0-9]{6}$"]
}

Would you mind trying it out and seeing if it works for you like this? :) Let me know in case you need any help with building

[wenigwil]

Heyo! I just tested the new feature and it works!

I first had some trouble building Vivify with the system package nodejs 25.4.0-1 supplied by pacman on Arch. But a quick look into the PKGBUILD file for vivify 0.12.0-1 from the AUR confirmed that Node SEA is "still alpha and experimental" so I worked around it as in the PKGBUILD with a local node install with nvm. And that worked fine!

Thank you very much for implementing it so fast. I think it would be great for people outside of TypeScript to explain in the docs for mdFilePatterns which regex implementation/standard has to be used in the pattern. A link to a stable reference that explains how characters need to be escaped for example should be sufficient :).

[tuurep]

Great I tried this too and it works,

agreed it's a bit hard to use unless you have an exact use case in mind, although on the other hand that's the entire point of it. But could benefit from a practical example in the docs.

(nvim/vim): I notice docs matched this way won't have instant refresh because we'd need a similar feat in vivify.vim.

Edit: nevermind I think I'd need to set that up outside of the plugin:

the filetype detection in Neovim with its pattern-argument of vim.filetype.add()

I first had some trouble building Vivify with the system package nodejs 25.4.0-1 supplied by pacman on Arch. But a quick look into the PKGBUILD file for vivify 0.12.0-1 from the AUR confirmed that Node SEA is "still alpha and experimental" so I worked around it as in the PKGBUILD with a local node install with nvm. And that worked fine!

It would be great if this became more straightforward in the future :( right now, this PKGBUILD does seem like the best reference to build on Linux in general.

This is the script I use to build and test on a branch:

#!/bin/bash

source /usr/share/nvm/init-nvm.sh
nvm use node

yarn
make clean

./configure ~/.local/bin

make install

pkill -f vivify

[tuurep]

Makes me wonder, was there a reason we couldn't rely on (n)vim's filetype detection to determine if a file is markdown? 🤔

(note: offtopic on this repo)

Edit: we are using that, aren't we?

@wenigwil if you're using the (n)vim plugin, does the live refresh work for you with no changes to that plugin? (refresh on the browser on every inserted character)

Edit: ok read your opening post again, if the ft detection works it seems like all's good

[jannis-baum]

Thanks @tuurep @wenigwil ! I added some more details to the docs, what do you guys think? I couldn't really think of a sufficiently generic real example to include in the docs, if you have some better idea please let me know :)

EDIT I missed the obvious choice of using the calcurse example for the docs, we could of course do that, just not sure if that might be confusing for people who don't know calcurse; especially since it's a rather strange-looking pattern for a Markdown file and may not be super easy to read for everyone

EDIT 2 Went for the calcurse example

---

About the building: Yes, unfortunately the only Node versions that are reliably able to build Node SEAs are the NVM-distributed ones. SEA seemed like the great native option back when we initially switched to it but it seems like it has since been somewhat abandoned. Ultimately we should probably move away from Node entirely but that's more work and not worth the investment at the moment :(

Makes me wonder, was there a reason we couldn't rely on (n)vim's filetype detection to determine if a file is markdown? 🤔

(note: offtopic on this repo)

Edit: we are using that, aren't we?

Yes, we check the &filetype so if (n)vim knows it's Markdown so should vivify.vim