Error parsing TOML on 1.1.386 if using comments in Arrays

[pavel-naumenko]

Description

After upgrading from 1.1.385 to 1.1.386 pyright raises errors "Pyproject file parse attempt 1 error:"

Issue is raised if pyproject.toml has comments in arrays

E.g. error in 1.1.386 (while OK in 1.1.385

[project]
name = "test"
version = "1.0.0"
dependencies = [
    "lib_a==1.1.1", # comment
    "lib_b==1.1.2"
]

If removing comments, parsing is OK in new version also

[project]
name = "test"
version = "1.0.0"
dependencies = [
    "lib_a==1.1.1", 
    "lib_b==1.1.2"
]

[InSyncWithFoo]

See #9297 for an explanation. Both @iarna/toml and iarna-toml-esm seem to be abandoned; it's probably better to use a maintained library, like smol-toml.

[erictraut]

Looks like I'll need to back out the change for #9286.

@InSyncWithFoo, I looked at smol-toml, but it unfortunately distributed as an ES Module rather than a CommonJS package. That means there's no straightforward way to use it in pyright.

I'll investigate other options.

[linde12]

Can confirm. Even comments above a list is failing for me

[erictraut]

I switched to a different toml parser library — one that claims to support the toml 1.0 standard. Let's see if this addresses all of the issues.

[Dr-Irv]

@erictraut I'm having the same issue with pandas-stubs, so if you use that as a test, you can see if the new TOML library handles it. Our pyproject.toml file has lots of comments in it.

[erictraut]

@Dr-Irv, good idea. I confirmed that the new toml parser works fine with pandas-stubs.

[plu-bgl]

I saw this one too late, but I created #9339 and I think the same issue was introduced in v1.1.285: #4367. It's all about using CRLF vs LF or CR.

[erictraut]

This is addressed in pyright 1.1.387.