Incorrect type hint for `chunking_strategy` params on `File` class methods

[evangriffiths]

Confirm this is an issue with the Python library and not an underlying OpenAI API

• This is an issue with the Python library

Describe the bug

If I pass in a StaticFileChunkingStrategyParam to OpenAI.beta.vector_stores.create I get:

openai.BadRequestError: Error code: 400 - {'error': {'message': "Missing required parameter: 'chunking_strategy.type'.", 'type': 'invalid_request_error', 'param': 'chunking_strategy.type', 'code': 'missing_required_parameter'}}

To Reproduce

Run this repro:

import tempfile

from openai import OpenAI
from openai.types.beta.static_file_chunking_strategy_object import (
    StaticFileChunkingStrategyObject,
)
from openai.types.beta.static_file_chunking_strategy_param import (
    StaticFileChunkingStrategyParam,
)

client = OpenAI()
with tempfile.NamedTemporaryFile(suffix=".txt", mode="w+b") as temp_file:
    temp_file.write(b"foo bar")
    temp_file.flush()
    temp_file.seek(0)
    file = client.files.create(file=temp_file.file, purpose="assistants")

vector_store = client.beta.vector_stores.create(name="foo")
client.beta.vector_stores.files.create(
    vector_store_id=vector_store.id,
    file_id=file.id,
    chunking_strategy=StaticFileChunkingStrategyParam(
        max_chunk_size_tokens=250,
        chunk_overlap_tokens=10,
    ),
)

Based on type hints I'd expect this to work, but I get the above api error.

It works instead with:

from openai.types.beta.static_file_chunking_strategy_object import (
    StaticFileChunkingStrategyObject,
)
...
client.beta.vector_stores.files.create(
    vector_store_id=vector_store.id,
    file_id=file.id,
    chunking_strategy=StaticFileChunkingStrategyObject(
        type="static",
        static=StaticFileChunkingStrategyParam(
            max_chunk_size_tokens=250,
            chunk_overlap_tokens=10,
        ),
    ),
)

i.e. looks the type hint should be changed to FileChunkingStrategy. Note there are many such methods with this param in Files (e.g. create_and_poll, upload, ...). I haven't tested these, but this change may apply to them too.

Code snippets

No response

OS

macos 14.3.1

Python version

Python 3.10.14

Library version

openai==1.59.6

[Programmer-RD-AI]

Hi,

Thanks for bringing up this issue and providing such detailed information! 🙌 After investigating the problem further, I believe I found the root cause and made the following changes to resolve it.

Issue:

The current FileChunkingStrategyParam type hint in openai/types/beta/file_chunking_strategy_param.py includes:

FileChunkingStrategyParam: TypeAlias = Union[AutoFileChunkingStrategyParam, StaticFileChunkingStrategyParam]

However, as the API error suggests, StaticFileChunkingStrategyParam should be encapsulated in a StaticFileChunkingStrategyObject. This mismatch is why passing StaticFileChunkingStrategyParam directly results in a 400 - Missing required parameter: 'chunking_strategy.type' error.

---

Fix:

I updated the type hint to use StaticFileChunkingStrategyObject instead of StaticFileChunkingStrategyParam:

FileChunkingStrategyParam: TypeAlias = Union[AutoFileChunkingStrategyParam, StaticFileChunkingStrategyObject]

This change reflects the correct structure expected by the API and aligns the type hints with actual usage. Additionally, this resolves the error when using StaticFileChunkingStrategyParam.

---

Concern:

While this change resolves the immediate issue, I noticed that AutoFileChunkingStrategyParam doesn't include a static field, which StaticFileChunkingStrategyObject does. This inconsistency might cause issues if similar expectations arise for the AutoFileChunkingStrategyParam in the future. I plan to explore this further to ensure compatibility across the board.

---

Next Steps:

• I'll submit a PR with this fix for review. Fix type hint for FileChunkingStrategyParam to use StaticFileChunkingStrategyObject #2014
• If anyone has insights or additional feedback about the potential limitations with AutoFileChunkingStrategyParam or its expected structure, I'd love to collaborate further!

Thanks again for bringing this up, and I hope this update helps! Let me know if there’s anything else I should explore or test.

[RobertCraigie]

Thanks for the report! This will be fixed in the next release
#2013