Files with CRLF newlines fail to parse

[codejedi365]

In the latest release 0.15.0, tomlkit now throws an invalid control character exception for /r (carriage return) as a character in a file during parsing. This is problematic when the toml file is written or created on Windows where newlines are CRLFs (/r/n) instead of just a linux newline (/n).

This behavior was not existent in v0.13.0.

I use tomlkit within python-semantic-release and I found this bug during testing on my windows specific test environment with python 3.14 which pulls in the 0.15.0 version as opposed to windows running python 3.8 (which pulls in 0.13.0).

Pipeline failure logs:

• https://github.com/python-semantic-release/python-semantic-release/actions/runs/29205384608/job/86684342134 (logs expire in 90 days from 12 July 2026) - code commit hash abd942d71c077e012eb49c43e094c9bf4a352a44

InvalidControlChar('Control characters (codes less than 0x1f and 0x7f) are not allowed in comments, use \\u000d instead at line 43 col 0'

[dimbleby]

You probably don't have a CRLF at all, you probably have an invalid lone CR - which tomlkit correctly rejects

You should provide a file that you believe to be valid TOML and which reproduces the problem, else everyone is just guessing.

[codejedi365]

Unlikely as I am generating the TOML programmatically with no specific use of characters.