When I post a JSON while using Mercado Libre's API all accented letters are removed and I don't think this is the default behavior

Question

PROBLEM

I am using the Brazilian version of the Mercado Libre API to publish products. Whenever I post a JSON, the accents are being removed by their system and this, I expect, should not at all be the default behavior because the Portuguese language has all of these accents. Their API documentation does not mention anything about this.

WHAT I HAVE TRIED

I tried to encode and decode it multiple times in various ways and using UTF-8 and latin1 with and without json.dumps().

My last attempt was adding:

headers['Content-Type'] = "application/json;charset=UTF-8"

All this did was, instead of vanishing completely, the accented letters were replaced by escape sequences such as:

Curto a Médio -> Curto a M\xe9dio

I also tried without the "charset=UTF-8" part and the accents were still escaped.

CODE

headers = {
                'Authorization': f"Bearer {ACCESS_TOKEN}",
                }

    if request_type == 'Product upload':
        headers['Content-Type'] = "application/json;"

response = requests.post(url, headers=headers, data=data)

DATA

[
   {
      "id":"GENDER",
      "value_name":"Mulher"
   },
   {
      "id":"BRAND",
      "value_name":"Torricella"
   },
   {
      "id":"MODEL",
      "value_name":"65019A"
   },
   {
      "id":"FOOTWEAR_TYPE",
      "value_name":"Botas"
   },
   {
      "id":"FOOTWEAR_STYLE",
      "value_name":"Bota Bico Fino"
   },
   {
      "id":"SHAFT_TYPE",
      "value_name":"Curto a Médio"
   },
   {
      "id":"RELEASE_SEASON",
      "value_name":"Primavera/Verão"
   },
   {
      "id":"RELEASE_YEAR",
      "value_name":2021
   }
]

Unfortunately they do not provide support of any kind to people who are not official partners.

What else could I try? I think it's unlikely that this is a problem with their API since they are worth 4 freaking billion dollars and their API has been around for ages.

Answer 1

Using json.dumps(data) without ensure_ascii=False worked!

Indeed, if ensure_ascii=True (the default), the output is guaranteed to have all incoming non-ASCII characters correctly escaped.