fix(vertex-ai-langchain): token count parsing

[hassiebp]

Important

Adds parsing for Vertex AI token details in _parse_usage_model() to extract modality-specific token counts in langchain.py.

• Behavior:
  • Adds parsing for prompt_tokens_details, candidates_tokens_details, and cache_tokens_details in _parse_usage_model() in langchain.py.
  • Extracts token_count for each modality and updates usage_model with keys input_modality_{modality}, output_modality_{modality}, and cached_modality_{modality}.

^{This description was created by for 1065560. You can customize this summary. It will automatically update as commits are pushed.}

Greptile Summary

Disclaimer: Experimental PR review

Enhanced token usage parsing for Vertex AI responses in the Langfuse Python SDK, focusing on extracting and storing token counts for different modalities.

• Added support for parsing input, output, and cached token details from Vertex AI usage models in langfuse/callback/langchain.py
• Implemented modality-specific token count storage with descriptive keys (e.g., 'input_modality_X')
• Ensured compatibility with existing OpenAI usage schema in langfuse/api/resources/ingestion/types/open_ai_usage.py
• Cleaned up usage model by removing raw token details lists after processing

[greptile-apps]

LGTM

_{1 file(s) reviewed, no comment(s)
Edit PR Review Bot Settings | Greptile}