Struct ConversationContext 

pub struct ConversationContext { /* private fields */ }

A multi-turn conversation with automatic context window management.

Each added turn is stored with its role, content, and token ids. build_tokens() concatenates all turn token ids in order, respecting the underlying ContextWindow’s token budget.

Implementations

impl ConversationContext

pub fn new(max_tokens: usize) -> Self

Create a new conversation context with the given maximum token budget.

pub fn add_turn(&mut self, role: &str, content: &str, token_ids: Vec<u32>)

Add a conversation turn.

The turn’s token ids are appended to the context window.

pub fn build_tokens(&self) -> Vec<u32>

Build a flat token sequence from all turns, respecting the window budget.

Concatenates token ids in turn order. The result is always within max_tokens after truncation.

pub fn turn_count(&self) -> usize

Number of turns added to this conversation.

pub fn total_tokens(&self) -> usize

Total token count across the current context window (after truncation).

pub fn is_full(&self) -> bool

Returns true if the context window is at its maximum capacity.

pub fn clear(&mut self)

Clear all turns and reset the context window.

pub fn last_turn(&self) -> Option<&ConversationTurn>

Reference to the most recently added turn, if any.

pub fn utilization(&self) -> f32

Utilisation of the token budget: total_tokens / max_tokens.