utf-8 chars in name of file in post requests

I have some code like this:

files = {"file": ("some utf-8 chars like '№'", open("file.jpg", "rb"), "image/jpeg")}
r = requests.post("localhost", files=files)

and I receive broken jpg file (actually, 2x size of original)

Expected Result

I expect to receive file.jpg

Actual Result

But requests send to me broken file.

System Information

{
  "chardet": {
    "version": "3.0.4"
  },
  "cryptography": {
    "version": "2.7"
  },
  "idna": {
    "version": "2.8"
  },
  "implementation": {
    "name": "CPython",
    "version": "3.6.9"
  },
  "platform": {
    "release": "10",
    "system": "Windows"
  },
  "pyOpenSSL": {
    "openssl_version": "1010103f",
    "version": "19.0.0"
  },
  "requests": {
    "version": "2.22.0"
  },
  "system_ssl": {
    "version": "1000211f"
  },
  "urllib3": {
    "version": "1.24.2"
  },
  "using_pyopenssl": true
}

Start by running the provided Python reproduction with requests.post and a UTF-8 filename such as “№”, then inspect the multipart upload and resulting file bytes. Compare the behavior with an ASCII filename and verify that the uploaded JPEG remains valid and matches the original file; add or update regression coverage if the relevant test location is identified.