What file extensions are supported ?

[motorto]

Summary

The harper_ls language server, as documented in the [nvim-lspconfig documentation](https://github.com/neovim/nvim-lspconfig/blob/master/doc/configs.md#harper_ls), is supposed to support the following file types:

{ "c", "cpp", "cs", "gitcommit", "go", "html", "java", "javascript", "lua", "markdown", "nix", "python", "ruby", "rust", "swift", "toml", "typescript", "typescriptreact" }

However, while it works correctly for markdown and lua files, no errors are being reported for other file types, even when no other LSP is active for the file.

Steps to Reproduce

1. Open a supported file type (e.g., python, c, etc.) in Neovim.
2. Ensure that no other LSP server is active for the file.
3. Introduce a syntax error or invalid code.
4. Observe the absence of error reporting by harper_ls.

Observed Behavior

• harper_ls works correctly for markdown files (errors are reported as expected).
• No errors or diagnostics are reported for other supported file types.

Expected Behavior

• harper_ls should provide diagnostics and error reporting for all supported file types listed in the documentation.

Environment

• Neovim Version: NVIM v0.10.2
• OS: fedora 41

Additional Notes

• The issue does not appear to be configuration-specific, as the server initializes without errors.
• This behavior persists across multiple supported file types (python, c, etc.).

[grantlemons]

Harper doesn't aim to correct syntax errors or invalid code, that is the role of a language specific lsp. Those languages are "supported" because it will spellcheck comments on files of those types.

[elijah-potter]

Hey @motorto,

It sounds like you're concerned because you are not getting language diagnostics when you provide invalid code to harper-ls. Does that sound right?

That's actually entirely expected. Harper doesn't aim to fix you're code, but rather your comments. Harper is actually a grammar checker, designed to make your non-code text easier to read and be understood by other people on your team.

If you want confirmation that it's working, try adding a comment containing "This is an test". You should be getting diagnostics to fix it.

Hope your day goes well :). Let me know if that answered your question so I can go ahead and close this ticket.

[motorto]

Hey, I was expecting that it would give grammar errors only on comments which it doesn't. I tried the sentence: # This are error. And get nothing. The LspLogs doesn't give me nothing, and I can confirm that Harper is running because of lspinfo. Happy to help more, at the end of the day it might be a config problem on my end. Can you confirm that it gives you grammar corrections in python comments ?

…

On Fri, Dec 20, 2024, 19:38 Elijah Potter ***@***.***> wrote: Hey @motorto <https://github.com/motorto>, It sounds like you're concerned because you are not getting language diagnostics when you provide invalid code to harper-ls. Does that sound right? That's actually entirely expected. Harper doesn't aim to fix you're code, but rather your comments. Harper <https://writewithharper.com> is actually a grammar checker, designed to make your non-code text easier to read and be understood by other people on your team. If you want confirmation that it's working, try adding a comment containing "This is an test". You should be getting diagnostics to fix it. Hope your day goes well :). Let me know if that answered your question so I can go ahead and close this ticket. — Reply to this email directly, view it on GitHub <#320 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMRTQPOVYM3EIKZXOMUUJ5T2GRW43AVCNFSM6AAAAABT5DCIVWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNJXGYYTMMJWHA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[andremarianiello]

I am running into a similar problem. I am expecting spell-checkable regions of my code (e.g. comments) to be analyzed by harper, but when I try in this example python file test.py

# Ths is is main

def main():
    print("hello world")

neovim's spell checker catches Ths, but harper doesn't catch the duplicated words even though it is running on this file. If I rename the file to test.md, then harper catches the misspelling and duplicated word on the first line. At this point I assumed that my expectations about how harper decides what to analyze are simply unfounded assumptions, so I came to the repo looking for an explanation in the issues, and that is how I found this one detailing the same problem. Since this is relevant here I might as well simply ask: how does harper decide what portions of the active buffer to analyze diagnostics for?

EDIT: If I port this to Go

// Ths is is main

func main() {
  fmt.Println("hello world")
}

I do get harper diagnostics for the comment on the first line, so this might be a python specific issue?

[grantlemons]

Hmm, this could be an issue with the comment parser (harper-comments/src/comment_parser.rs). Go is a weird case as it has a custom handler, and markdown isn't comment parsing, so any issues here wouldn't apply.

Weirdly, this seems to only apply to is for me.

Raw Parsing Output:

{"span":{"start":2,"end":6},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":null,"is_possessive":null,"is_pronoun":true},"verb":null,"adjective":null,"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":true}}}
{"span":{"start":6,"end":7},"kind":{"kind":"Space","value":1}}
{"span":{"start":7,"end":11},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":null,"is_possessive":null,"is_pronoun":true},"verb":null,"adjective":null,"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":true}}}
{"span":{"start":11,"end":12},"kind":{"kind":"Space","value":1}}
{"span":{"start":12,"end":14},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":true,"is_possessive":null,"is_pronoun":true},"verb":{"is_linking":true,"tense":null},"adjective":null,"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":false}}}
{"span":{"start":14,"end":15},"kind":{"kind":"Space","value":1}}
{"span":{"start":15,"end":17},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":true,"is_possessive":null,"is_pronoun":true},"verb":{"is_linking":true,"tense":null},"adjective":null,"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":false}}}
{"span":{"start":17,"end":18},"kind":{"kind":"Space","value":1}}
{"span":{"start":18,"end":19},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":null,"is_possessive":null,"is_pronoun":true},"verb":{"is_linking":null,"tense":null},"adjective":{},"adverb":null,"conjunction":{},"swear":null,"article":false,"preposition":false,"common":true}}}
{"span":{"start":19,"end":20},"kind":{"kind":"Space","value":1}}
{"span":{"start":20,"end":27},"kind":{"kind":"Word","value":{"noun":null,"verb":null,"adjective":{},"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":false}}}
{"span":{"start":27,"end":28},"kind":{"kind":"Space","value":1}}
{"span":{"start":28,"end":35},"kind":{"kind":"Word","value":{"noun":null,"verb":null,"adjective":{},"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":false}}}
{"span":{"start":35,"end":36},"kind":{"kind":"Space","value":1}}
{"span":{"start":36,"end":41},"kind":{"kind":"Word","value":{"noun":{"is_proper":null,"is_plural":null,"is_possessive":null,"is_pronoun":null},"verb":{"is_linking":null,"tense":null},"adjective":null,"adverb":null,"conjunction":null,"swear":null,"article":false,"preposition":false,"common":false}}}

[grantlemons]

Ah, nevermind, the issue I observed is unrelated, and is because is is considered a likely homograph, so the repetition rule doesn't fire. (New issue #333)

Going back to your python vs. go example, I don't observe the same thing.

Advice: 
   ╭─[test.py:1:1]
   │
 1 │ # Ths is is main
   │   ─┬─  
   │    ╰─── Did you mean to spell “Ths” this way?

This might be a difference of version. We haven't been able to do a proper release in a while, so if you are using the Mason version that might be causing this. There should be a new release very soon, so I'll ping y'all to check if you still observe this after the next release.

[andremarianiello]

I installed from the latest github release.

[grantlemons]

I installed from the latest github release.

Testing on commit 38d76e5 (last release) I observe your issue, but it seems to be fixed already on master. Hopefully we can get this release out ASAP to get this fixed!

[andremarianiello]

Awesome!