convert_to_openai_function will drop title field in output_schema if it's a Pydantic model.

[elonzh]

Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

from pprint import pprint
from langchain_core.utils.function_calling import convert_to_openai_function
from pydantic import v1

class BiblioExtraction(v1.BaseModel):
    title: str
    authors: list[str]

f = convert_to_openai_function(BiblioExtraction)
pprint(f)

Error Message and Stack Trace (if applicable)

{'description': '',
 'name': 'BiblioExtraction',
 'parameters': {'properties': {'authors': {'items': {'type': 'string'},
                                           'type': 'array'}},
                'required': ['title', 'authors'],
                'type': 'object'}}

Description

I am trying to use langchain.chains.openai_functions.create_structured_output_runnable to extract citation information from the text, and then I found that I cannot extract the title field.

After investigation, I discovered that langchain introduced an undocumented behavior in version 0.1.4, which deletes the title field in schema.

System Info

❯ python -m langchain_core.sys_info

System Information
------------------
> OS:  Linux
> OS Version:  #1 SMP Thu Oct 5 21:02:42 UTC 2023
> Python Version:  3.10.13 (main, Nov 16 2023, 15:58:41) [GCC 11.4.0]

Package Information
-------------------
> langchain_core: 0.1.27
> langchain: 0.1.9
> langchain_community: 0.0.24
> langsmith: 0.1.5
> langchain_openai: 0.0.5

Packages not installed (Not Necessarily a Problem)
--------------------------------------------------
The following packages were not found:

> langgraph
> langserve

[keenborder786]

The unexpected behaviour is caused by rm_titles being set to True by default. Also, if you urgently want to use the tool then why don't you change the spelling of your parameter e.g title to title_extracted since that will also fix your isse.

[keenborder786]

@elonzh Actually I am closing the PR because that bring in unexpected breaking changes in other parts of code. I would recommend you to use alternative name for tile e.g tile_extracted. That will serve your purpose:

from pprint import pprint
from langchain_core.utils.function_calling import convert_to_openai_function
from pydantic import BaseModel

class BiblioExtraction(BaseModel):
    title_extracted: str
    authors: list[str]

f = convert_to_openai_function(BiblioExtraction)
pprint(f)

[Gregory-Gerard]

Same here. The pain is that it's not documented, and can be time-consuming to find the actual problem. For the moment I've renamed my title, but it would be nice to be able to modify this behavior via an argument, for example.