BUG: construct_type() crashes with ValueError on bare dict annotation (no type args)

[devteamaegis]

What breaks

construct_type() in src/openai/_models.py (line 660) does _, items_type = get_args(type_) after checking origin == dict. When type_ is the bare, unparameterised dict class (no [K, V] type arguments), get_args(dict) returns an empty tuple (). Unpacking it with two targets raises ValueError: not enough values to unpack (expected 2, got 0). This crashes whenever a Pydantic BaseModel field or any response type is annotated as plain dict and the library tries to deserialise a mapping value into it.

from openai._models import construct_type
result = construct_type(value={"key": "value"}, type_=dict)
# ValueError: not enough values to unpack (expected 2, got 0)

Traceback

File "/tmp/repro.py", line 2, in <module>
    result = construct_type(value={"key": "value"}, type_=dict)
  File ".../src/openai/_models.py", line 660, in construct_type
    _, items_type = get_args(type_)  # Dict[_, items_type]
    ^^^^^^^^^^^^^
ValueError: not enough values to unpack (expected 2, got 0)

[nightcityblade]

Hi, I'd like to work on this. I'll submit a PR shortly.

[Xsidz]

Working on this in PR #3625.

[marcuswood-oai]

Thank you for the report and follow-ups. Bare dict annotations have no type arguments, which caused the unconditional unpacking in construct_type() to raise ValueError. #3760 now handles bare dictionaries and lists, with coverage for model construction and sync/async response deserialization.

The merge has resolved and automatically closed this issue. Please use a release containing #3760 when available; the fix is in main and is not included in v3.11.0.