Skip to content

_api_encode encodes bool/datetime list elements raw (e.g. flags[0]=True instead of true) #1956

Open

Description

@alphasafal

Describe the bug

stripe._encode._api_encode converts bool and datetime values to the forms the API expects at the top level and inside nested dicts (bool → "true"/"false", datetime → unix timestamp), but yields them raw when they are direct elements of a list.

The request body/query string is built with urlencode(list(_api_encode(params))) (in _api_requestor.py), so a list parameter containing a boolean is serialized as "True"/"False" (Python str(bool)) and a list containing a datetime is serialized as its str() repr instead of a unix timestamp. Multipart form encoding is affected too, since MultipartDataGenerator.add_params also goes through _api_encode.

To Reproduce

import datetime
from urllib.parse import urlencode
from stripe._encode import _api_encode

dt = datetime.datetime(2024, 1, 2, 3, 4, 5, tzinfo=datetime.timezone.utc)

print(urlencode(list(_api_encode({"active": True}))))          # correct
print(urlencode(list(_api_encode({"flags": [True, False]}))))  # wrong
print(urlencode(list(_api_encode({"created": dt}))))           # correct
print(urlencode(list(_api_encode({"created": [dt]}))))         # wrong

Output:

active=true
flags%5B0%5D=True&flags%5B1%5D=False
created=1704164645
created%5B0%5D=2024-01-02+03%3A04%3A05%2B00%3A00

The top-level bool/datetime are encoded correctly (active=true, created=1704164645); the same values inside a list are not.

Expected behavior

A bool or datetime inside a list should be encoded the same way as at the top level:

  • flags[0]=true&flags[1]=false
  • created[0]=1704164645

Code snippets

OS

Any (reproduced on Linux)

Language version

Python 3.12 (version-independent)

Library version

main (89e8528)

API version

N/A — client-side request encoding, independent of API version

Additional context

Root cause

In _api_encode (stripe/_encode.py), the list branch handles dict/_data elements and nested list/tuple elements, but every other element — including bool and datetime.datetime — falls through to else: yield (encoded_key, sv) raw. The top-level code has explicit datetime and bool branches; the list branch does not.

(Nested dict values are unaffected because they recurse back through _api_encode; nested lists are unaffected because they go through _encode_nested_dict + recursion. Only direct scalar elements of a list are affected.)

Proposed fix

Mirror the top-level handling in the list branch:

elif isinstance(sv, datetime.datetime):
    yield (encoded_key, _encode_datetime(sv))
elif isinstance(sv, bool):
    yield (encoded_key, str(sv).lower())
else:
    yield (encoded_key, sv)

I've prepared this fix with regression tests on a branch: https://github.com/alphasafal/stripe-python/tree/fix/encode-list-bool-datetime — happy to open a PR if a maintainer would like one (the repo currently limits PR creation to collaborators).

Activity

  1. pcbeingused333 commented on Nov 2, 2026

    @pcbeingused333

    The list branch of _api_encode yields scalar elements as-is, so the bool and datetime conversions that apply at the top level and inside nested dicts are skipped for direct list elements ({"flags": [True, False]} is sent as flags[0]=True&flags[1]=False).

    Opening pull requests here is limited to collaborators, so the fix is on a branch instead, in case it is useful: pcbeingused333/stripe-python@master...c0486-encode-list-scalars. It adds the two missing conversions in the list branch, two tests in tests/test_encode.py that fail without the change, and a .hark changefile.

    This change was prepared with the help of an AI coding assistant (Claude Code).

  2. ramya-stripe commented on Nov 10, 2026

    @ramya-stripe
    Contributor

    Thank you for reporting @alphasafal and @pcbeingused333
    We will try and fix this by the end of the year

  3. alphasafal commented on Nov 11, 2026

    @alphasafal
    Author

    Thanks @ramya-stripe. The branch linked in the issue has the fix with regression tests if it saves you time; feel free to take it as is.

  4. ramya-stripe commented on Nov 12, 2026

    @ramya-stripe
    Contributor

    Hey @alphasafal
    We just temporarily opened up community contributions. So if you want to send a PR, please do!

  5. alphasafal commented on Nov 12, 2026

    @alphasafal
    Author

    Thanks @ramya-stripe, and thank you for opening this up to community contributions! I've opened #1978 with the fix, regression tests, and a changefile.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions