GoogleGenAiSdkInstrumentor: JSON Serialization Error with `Part` Objects in system_instruction

[alexmojaki]

Originally reported in pydantic/logfire#1547

import os

from google import genai
from google.genai import types
from opentelemetry.instrumentation.google_genai import GoogleGenAiSdkInstrumentor

os.environ['OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT'] = 'true'
GoogleGenAiSdkInstrumentor().instrument()

client = genai.Client()
config = types.GenerateContentConfig(system_instruction=[types.Part.from_text(text='help')])
response = client.models.generate_content(model='gemini-2.0-flash-exp', contents=['hi'], config=config)

Traceback (most recent call last):
  File "/Users/alex/Library/Application Support/JetBrains/PyCharm2025.2/scratches/scratch_2463.py", line 12, in <module>
    response = client.models.generate_content(model='gemini-2.0-flash-exp', contents=['hi'], config=config)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/alex/work/logfire/.venv/lib/python3.12/site-packages/opentelemetry/instrumentation/google_genai/generate_content.py", line 763, in instrumented_generate_content
    helper.process_request(contents, config)
  File "/Users/alex/work/logfire/.venv/lib/python3.12/site-packages/opentelemetry/instrumentation/google_genai/generate_content.py", line 388, in process_request
    self._maybe_log_system_instruction(config=config)
  File "/Users/alex/work/logfire/.venv/lib/python3.12/site-packages/opentelemetry/instrumentation/google_genai/generate_content.py", line 567, in _maybe_log_system_instruction
    body["content"] = _to_dict(system_instruction)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/alex/work/logfire/.venv/lib/python3.12/site-packages/opentelemetry/instrumentation/google_genai/generate_content.py", line 166, in _to_dict
    return json.loads(json.dumps(value))
                      ^^^^^^^^^^^^^^^^^
  File "/Users/alex/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/json/__init__.py", line 231, in dumps
    return _default_encoder.encode(obj)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/alex/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/json/encoder.py", line 200, in encode
    chunks = self.iterencode(o, _one_shot=True)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/alex/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/json/encoder.py", line 258, in iterencode
    return _iterencode(o, 0)
           ^^^^^^^^^^^^^^^^^
  File "/Users/alex/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/json/encoder.py", line 180, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '
TypeError: Object of type Part is not JSON serializable

[alexmojaki]

This code:

def _to_dict(value: object):
    if isinstance(value, dict):
        return value
    if hasattr(value, "model_dump"):
        return value.model_dump()
    return json.loads(json.dumps(value))

handles a pydantic BaseModel but not a list of them which is the problem. Explicitly handling that case might be enough. Using a TypeAdapter would be most general.