mirror of
https://github.com/hwchase17/langchain.git
synced 2026-02-21 22:56:05 +00:00
8e31a5d7bd
- **Description:** The root cause of this issue is that when a user
defines `model_config` in a `BaseModel`, the `{"type": <tool_name>}`
value is derived from the title specified in `model_config` when the
results are parsed
[here](https://vscode.dev/github/keenborder786/langchain/blob/fix/tool_name_dict/libs/core/langchain_core/output_parsers/openai_tools.py#L199).
However,
[tool.__name__](https://vscode.dev/github/keenborder786/langchain/blob/fix/tool_name_dict/libs/core/langchain_core/output_parsers/openai_tools.py#L331)
uses the class name (in uppercase) of the `BaseModel`, resulting in a
`KeyError` when a custom title is provided in `model_config`.
The Best Solution will be to use the title provided in `model_config`
attribute if provided one since that is what `type` will be parsed to,
if not then use `tool.__name__`. But need to make sure that this works
only for Pydantic V2.
- **Issue:** #27260
---------
Co-authored-by: Mason Daugherty <mason@langchain.dev>