Module utils 

Shared utilities for LLM request/response processing

This module provides common functions used across multiple providers to eliminate duplicate code and reduce allocations.

Functions

estimate_token_count

Count the number of tokens in text using tiktoken BPE.

parse_stream_event_openai_format

Parse stream event from OpenAI-compatible format

truncate_to_token_limit

Truncate text to at most max_tokens tokens.