Struct OllamaCtx 

pub struct OllamaCtx { /* private fields */ }

本地 Ollama 模型的上下文策略。

Implementations

impl OllamaCtx

pub fn new(provider: &ProviderConfig) -> Self

Trait Implementations

impl Clone for OllamaCtx

fn clone(&self) -> OllamaCtx

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl CtxBuilder for OllamaCtx

fn needs_compression(&self, conv: &Conversation, system_tokens: usize) -> bool

复用 ctx::render::needs_compression — 它的绝对 headroom 公式 ctx_window - min(13K, ctx_window/4) 在小窗口下天然偏紧 (8K Ollama → 6K threshold = 75% 触发, 比之前的 35% 晚但更接近“撑爆前一刻“的真实 headroom)。 之前的 35% hardcoded 阈值是为 4-8K Ollama 量身的早触发, 但在 16K-32K Ollama 上反而过早。新公式自适应窗口大小, 不再需要单独的 Ollama tier。

fn build_messages( &self, conv: &Conversation, system_prompt: &str, turn_reminder: &str, ) -> (Vec<Message>, ContextStats)

Build the messages array to send to the LLM for this turn.

fn compression_plan(&self, conv: &Conversation) -> Option<(String, usize)>

Produce a compression plan (content_to_summarize, messages_to_remove). Default: None (no compression).

fn truncate_tool_output(&self, result: &mut ToolResult, tool_name: &str)

Truncate a single tool output in place.

fn ctx_window(&self) -> usize

Effective token budget for this strategy.

fn name(&self) -> &'static str

Human-readable name for logging / debugging.

impl Debug for OllamaCtx

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.