fixV2(core): Resolve PydanticSerializationError for Google FunctionCall

[iamzimozic]

Description

The core fix is to enhance the Pydantic serialization within the base ChatMessage class to explicitly handle the Google SDK's FunctionCall object. The _recursive_serialization method in llama_index/core/base/llms/types.py is updated to check for this specific external object type and convert it into a standard Python dictionary using its internal .to_dict() method. This prevents the PydanticSerializationError from being thrown when the ChatMessage is dumped to JSON for storage.

Fixes # (issue)

New Package?

Did I fill in the tool.llamahub section in the pyproject.toml and provide a detailed README.md for my new integration or package?

• Yes
• No

Version Bump?

Did I bump the version in the pyproject.toml file of the package I am updating? (Except for the llama-index-core package)

• Yes
• No

Type of Change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Your pull-request will likely not be merged unless it is covered by some form of impactful unit testing.

• I added new unit tests to cover this change
• I believe this change is already covered by existing unit tests

Suggested Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added Google Colab support for the newly added notebooks.
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes
• I ran uv run make format; uv run make lint to appease the lint gods

[logan-markewich]

Did you mean to commit this file?

[iamzimozic]

no it was intented for local testing only. I have deleted the file now.

[logan-markewich]

There's no need to import types from google here

[logan-markewich]

We don't need to check for FunctionCall, we should be checking for whatever the root type of FunctionCall is

Looking at the source code, its a base model. Which is actually already being checked in self._recursive_serialization(value) -- but I think its not recursing nicely into the pydantic object?

[iamzimozic]

1. Core: Fixed Generic Pydantic Serialization in types.py
   Problem: The initial issue was a PydanticSerializationError when handling nested Pydantic models (like Google's FunctionCall) inside additional_kwargs.

Solution: I fixed a recursion bug in the _recursive_serialization helper function. The corrected logic now properly handles any nested BaseModel, making the serialization robust and generic. This is a more durable fix than adding a special case for a single type.

2. memory-mem0: Resolved Pydantic Compatibility Issues
   Problem: Tests for mem0 failed on all Python versions with an AttributeError. Debugging revealed that despite running on Pydantic v2, the model validator was being called in a backward-compatible v1 style, receiving a plain dict instead of a ValidationInfo object.

Solution: I updated the validator logic in mem0/base.py to work with a dict (changing values.data.get() to values.get()), aligning the code with its actual runtime behavior and resolving the crash.

3. readers-web: Added Dependencies & Fixed PackageNotFoundError
   Problem: The test suite for readers-web was failing due to numerous ModuleNotFoundError errors and a PackageNotFoundError in the OxylabsWebReader test.

Solution:

I added all missing development dependencies (markdownify, lxml[html_clean], playwright, selenium, oxylabs, etc.) to the package's pyproject.toml, which are now tracked in the updated poetry.lock file.

To fix the PackageNotFoundError, I made the OxylabsWebReader code more robust by wrapping the version() call in a try...except block. It now defaults to a "local" string instead of crashing when run in a local test environment.

4. graph_rag/cognee: Pinned Dependency to Stable Version
   Problem: Tests for cognee failed on Python 3.11+ due to a Pydantic v2 AttributeError inside the external cognee library.

Solution: Since we can't fix the external dependency, I pinned cognee-python to an older, stable version (0.1.26) in the package's pyproject.toml to avoid the buggy code.

[iamzimozic]

The last fix passed all the checks but when "[Merge branch 'main' into fix/postgresV2]" happened, this showed up in every unit test
1 packages had test failures:
llama-index-integrations/graph_stores/llama-index-graph-stores-kuzu
this wasn't showing up earlier. Please help.